1.0.9.50: O(1) weak pointer scavenging

[sbcl/simd.git] / NEWS

;;;; -*- coding: utf-8; -*-
changes in sbcl-1.0.10 relative to sbcl-1.0.9:
  * minor incompatible change: the MSI installer on Windows no longer
    associates .lisp and .fasl files with the installed SBCL.
  * minor incompatible change: :UNIX is no longer present in *FEATURES*
    on Windows. (thanks to Luis Oliviera)
  * optimization: scavenging weak pointers is now more efficient,
    requiring O(1) instead of O(N) per weak pointer to identify
    scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
  * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
    names, when the class of the instance is a direct instance
    STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
    as slow as the constant slot-name case.)
  * optimization: member type construction is now O(N) instead
    of O(N^2).
  * optimization: UNION and NUNION are now O(N+M) for large
    inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
    EQUALP.
  * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
    now more readable in environments like Slime which display it.
    (thanks to Tobias C. Rittweiler)
  * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
    was able to infer the correct argument type for the object on which
    the CAS operation was being performed.

changes in sbcl-1.0.9 relative to sbcl-1.0.8:
  * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
  * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
    indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
    is called.
  * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
    and generic function now print the object identity for anonymous
    instances. (thanks to Attila Lendvai)
  * enhancement: as an extension to MOP, SBCL now supports SETF of
    STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
    (thanks to Attila Lendvai)
  * enhancement: sb-cover has an alternate annotation mode which puts
    the form annotations at the car of the form, instead of annotating
    it completely. The benefit of this mode is that it shows better
    which forms were instrumented by the compiler.
  * bug fix: new compiler transforms for MEMBER and ASSOC were affected
    by printer control variables. (reported by Dan Corkill)
  * bug fix: system leaked memory when delayed output was performed by
    the OS in smaller chunks then expected. (thanks to David Smith)
  * bug fix: system leaked memory when file streams were not closed 
    properly.
  * bug fix: large objects written to slow streams that were modified
    after the write could end up with the modified state written to
    the underlying file descriptor.
  * bug fix: multiple threads operating in parallel on the same stream
    could cause buffer-overflows.
  * bug fix: source location information is stored correctly 
    (broken since 1.0.6). This bug would generally show up as the
    Slime debugger higlighting the wrong form when the sldb-show-source
    command was used.
  * bug fix: Branch forms are again annotated as branches in the sb-cover
    annotations.
  * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
    on x86-64. (reported by Gregory Vanuxem)
  * bug fix: EQUALP could return wrong results for structures with raw
    slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
    word). (reported by Vjacheslav Fyodorov)
  * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
    by Pierre Mai)

changes in sbcl-1.0.8 relative to sbcl-1.0.7:
  * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
    atomic compare-and-swap operations on threaded platforms.
  * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
    allows assining a global minimum value to optimization qualities
    (overriding proclamations and declarations).
  * enhancement: closed over variables can be stack-allocated on x86
    and x86-64.
  * performance bug fix: GETHASH and (SETF GETHASH) are once again
    non-consing.
  * optimization: slot definition lookup is now O(1). This speeds up
    eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
  * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
  * optimization: ASSOC and MEMBER can now be open-coded for all
    combinations of keyword arguments when second argument is constant
    and SPEED >= SPACE. In other cases a specialized version is
    selected.
  * bug fix: using obsoleted structure instances with TYPEP and
    generic functions now signals a sensible error.
  * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
    (reported by Kristoffer Kvello)
  * bug fix: backtrace construction is now more careful when making
    lisp-objects from pointers on the stack, to avoid creating bogus
    objects that can be seen by the GC.
  * bug fix: defaulting of values in contexts expecting more than 7
    variables now works on x86-64.  (reported by Christopher Laux)
  * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
    thread safe.
  * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
    as the property-list of a symbol.
  * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
    in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL 
    situations requested, are once again file-compileable.  (reported
    by Sascha Wilde)

changes in sbcl-1.0.7 relative to sbcl-1.0.6:
  * MOP improvement: support for user-defined subclasses of
    SB-MOP:SPECIALIZER has been enhanced.  The experimental interface
    function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
    the expansion of DEFMETHOD, is responsible for generating a form
    which creates a list of specializers when evaluated.  Additional
    functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
    debugging and introspective support.
  * minor incompatible change: the (unsupported) spinlock interface
    has changed: free spinlock now has the value NIL, and a held spinlock
    has the owning thread as its value.
  * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
    WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
    for details.
  * enhancement: name of a socket-stream is now "a socket" instead of
    "a constant string".
  * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)  
  * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
  * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
    for generic functions. (Thanks to Tobias C. Rittweiler)
  * optimization: bignum printing speed has been improved by 20-40%
    (depending on the bignum size.)
  * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
    on Linux.
  * bug fix: the cache used by the CLOS to store precomputed effective
    methods, slot offsets, and constant return values is now thread and
    interrupt safe.
  * bug fix: generic function dispatch function updating is now thread
    and interrupt safe (in the sense that the known issues have been
    fixed.)
  * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
  * bug fix: interning EQL-specializers is now thread and interrupt safe.
  * bug fix: asdf systems with dependencies to the SB-POSIX or 
    SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
  * bug fix: interrupt safety of applicable method computation has been
    improved.

changes in sbcl-1.0.6 relative to sbcl-1.0.5:
  * new contrib: sb-cover, an experimental code coverage tool, is included
    as a contrib module.
  * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
    significantly faster.
  * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
    produce more efficient code on x86-64 if the compiler can determine
    that (+ INDEX <constant>) does not require a bounds check and FOO
    has an element type at least 8 bits wide.
  * enhancement: a new, experimental synchronous timeout facility is
    provided. Refer to SB-SYS:WITH-DEADLINE for details.
  * enhancement: when a symbol name conflict error arises, the
    conflicting symbols are always printed with a package prefix.
    (thanks to Kevin Reid)
  * enhancement: stepping is now once again supported on the SPARC and
    MIPS platforms.  (It is also now more likely to work on CheneyGC
    builds on the PPC.)
  * enhancement: sb-sprof can now also track and report accurate call
    counts.
  * bug fixes: the treatment of non-standard subclasses of
    SB-MOP:SPECIALIZER is more correct.
  * incompatible change: PURIFY no longer copies the data from the 
    dynamic space into the static and read-only spaces on platforms
    that use the generational garbage collector
  * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
    interrupt safe.
  * bug fix: GC race condition occasionally resulting in crashes with
    the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
    fixed.
  * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
    declared ignored.
  * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
    a STYLE-WARNING to be signalled (regression from 1.0.4.)
  * bug fix: an asynchronous interrupt could previously leave the
    system running with GC inhibited.
  * bug fix: a DECLARE form evaluated at top-level now causes an error
    rather than silently (or verbosely) returning NIL.
  * bug fix: trying to dissassemble functions compiled at high DEBUG could
    result in a error being signalled due to source form lookup errors.
    (reported by Peter Graves)

changes in sbcl-1.0.5 relative to sbcl-1.0.4:
  * incompatible change: removed writer methods for host-ent-name,
    host-ent-addresses -- changing the values did not update the DNS
    information anyway.
  * minor incompatible change: changed experimental JOIN-THREAD interface
  * documentation: the manual now lists reader and writer methods
    in class slot documentation sections. (thanks to Richard M Kreuter)
  * documentation: unwinding from asyncronous events has been
    documented as unsafe.
  * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
    in multithreaded application code.
  * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
    platforms. (thanks to James Anderson for the optimization hint)
  * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
    more cases.
  * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
    variants no longer cons.
  * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
    their NOT- variants no longer cons.
  * optimization: EQUAL hash tables no longer use SXHASH for objects
    of all data types, but instead use an EQL hash for types for which
    EQUAL is the same as EQL.
  * optimization: the non-inlined generic versions of AREF and (SETF AREF)
    are significantly faster.
  * optimization: new STRING-OUTPUT-STREAM implementation is much
    faster for large outputs and conses 30% less on average.
  * enhancement: XREF information is now collected for references made
    to global variables using SYMBOL-VALUE and a constant argument.
  * enhancement: SIGINT now causes a specific condition
    SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
  * bug fix: FILE-POSITION used to signal an error in some cases where
    ANSI requires it to return NIL.
  * bug fix: ADJUST-ARRAY is now interrupt-safe.
  * bug fix: adding and removing fd-handlers is now interrupt-safe.
  * bug fix: inlined calls to C now ensure 16byte stack alignment on
    x86/Darwin.
  * bug fix: bad type declaration in the CLOS implementation has
    been fixed. (reported by James Anderson)
  * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
    (thanks to Lutz Euler)
  * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
    required by the CLHS. (reported by Samium Gromoff)
  * bug fix: dead unbound variable references now signal an error.
    (reported by Marco Monteiro)
  * bug fix: / with an unused value was being deleted in safe code.
    (thanks to Marco Monteiro and Kevin Reid)
  * bug fix: number of characters that can be written onto a single
    line in a file is unlimited.
  * bug fix: some GC deadlocks caused by asynchronous interrupts have
    been fixed by inhibiting interrupts for when GC is disbled.
  * bug fix: some interrupt safety issues with GETHASH, (SETF
    GETHASH), CLRHASH and REMHASH have been fixed.
  * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
    specifier no longer causes infinite recursion.
  * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a 
    bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
  * bug fix: an &environment argument with the correct variable information
    is passed to macros that are expanded during byte compilation.
    (reported by Samium Gromoff)
  * bug fix: a logic error was causing the hash function for bignums to
    have excessive amounts of collisions. (reported by Faré Rideau)
  * bug fix: modifying the contents of an array could change the return
    value of SXHASH on that array, which is only allowed for strings 
    and bit vectors. (bug introduced in 0.9.16)
  * improvement: the x86-64/darwin port now passes all tests (except
    for the debugger tests) but should still be considered
    experimental until this is fixed.
  * improvement: a style-warning is signaled for CASE (etc) clauses with
    duplicate keys. (thanks to Kevin Reid)
  * improvement: macos/x86 and macos/x86-64 now use mach exception
    handlers for dealing with illegal instructions (for trapping and
    error handling) and memory protection violations (for GC).

changes in sbcl-1.0.4 relative to sbcl-1.0.3:
  * new platform: experimental support for x86-64/darwin (MacOS).
  * incompatible change: the thread-safe (on most platforms) getaddrinfo
    and getnameinfo sockets functions are used instead of gethostbyaddr
    and gethostbyname, on platforms where the newer functions are available.
    As a result, the ALIASES field of HOST-ENT will always be NIL on these
    platforms.
  * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
    don't print the startup banner, but behave as if --noinform was passed
    as a command line argument. (thanks to Kevin Reid)
  * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
  * optimization: code using alien values with undeclared types is much faster.
  * optimization: the compiler is now able to open code SEARCH in more cases.
  * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
  * bug fix: using standardized COMMON-LISP special variables as loop
    variables no longer signals bogus package lock violations. (reported
    by Eric Marsden)
  * bug fix: declaring local loop variables to be of a range-limited type
    such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
    (reported by Andras Simon)
  * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
    bugs remain on x86-64.)
  * bug fix: the #= and ## reader macros now interact reasonably with
    funcallable instances.
  * bug fix: type-checks for function arguments were compiled using the
    compiler policy settings of the wrong lexical environment (in 1.0.2
    and 1.0.3).
  * bug fix: SHADOW accepts characters as string designators, as required
    by the spec (thanks to Eric Marsden)
  * bug fix: fixed GC safety issues when foreign functions are called with
    non-base strings as arguments
  * bug fix: more consistent error output for fatal-compiler-warnings, like
    reader errors
  * bug fix: functions with proper names were showing up as NIL in
    backtraces (reported by Edi Weitz, regression in 1.0.2)

changes in sbcl-1.0.3 relative to sbcl-1.0.2:
  * new platform: NetBSD/PPC.  (thanks to Aymeric Vincent)
  * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
    produce more efficient code on the x86 if the compiler can determine
    that (+ INDEX <constant>) does not require a bounds check and FOO
    has an element type at least 8 bits wide.
  * bug fix: references to undefined variables are handled the same way
    in toplevel forms as in the normal compiler
  * bug fix: the build scripts again work with non-bash /bin/sh (thanks
    to Magnus Henoch)
  * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
    (thanks to Jon Buller)
  * improvement: faster compilation times for complex functions
  * improvement: added readlink support to SB-POSIX (thanks to Richard
    M Kreuter)

changes in sbcl-1.0.2 relative to sbcl-1.0.1:
  * improvement: experimental support for mach exception handling on
    x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
    to use.
  * improvement: support for GBK external format.
    (thanks to Chun Tian (binghe))
  * improvement: the debugger now displays variables that have been closed
    over, in code compiled with (DEBUG 2) or higher
  * improvement: support for executable cores on NetBSD (thanks to 
    Richard Kreuter)
  * new feature: added a RESTART-FRAME debugger command
  * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can 
    be used to provide an implementation for FILE-POSITION on Gray streams
    (thanks to Eric Marsden)
  * new feature: add syslog, openlog and closelog support to SB-POSIX 
    (thanks to Richard Kreuter)
  * optimization: the function call overhead in code compiled with 
    a high DEBUG optimization setting is significantly 
  * bug fix: an error is signaled for attempts to use READ-SEQUENCE
    for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
    (thanks to Tony Martinez)
  * bug fix: the initforms for DEFMETHOD &AUX parameters are only
    evaluated once (reported by Kevin Reid)
  * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
    works again (thanks to Kilian Sprotte)
  * bug fix: an error is signaled for tagbodies with duplicate tags 
    (thanks to Stephen Wilson)
  * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
  * bug fix: Win32 port can now handle foreign code unwinding Lisp
    stack frames from alien callbacks.
  * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)  
  * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
    65k pages (thanks to David Woodhouse)
  * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)

changes in sbcl-1.0.1 relative to sbcl-1.0:
  * new platform: FreeBSD/x86-64, including support for threading.
  * new feature: the compiler stores cross-referencing information
    abount function calls (who-calls), macroexpansion (who-macroexpands)
    and special variables (who-binds, who-sets, who-references) for code
    compiled with (< SPACE 3). This information is available through the
    sb-introspect contrib.
  * new feature: users may subclass SEQUENCE, and have instances of
    these classes interoperate with standard Common Lisp functions if
    a number of methods are defined.  (This feature is experimental
    and the interface subject to change based on feedback from SBCL
    users and the general community)
  * improvement: sb-sprof traces call stacks to an arbitrary depth on
    x86/x86-64, rather than the previous fixed depth of 8
  * improvement: another pthread back-end of mutex "pthread-futex".
  * bug fix: non-ascii command-line arguments are processed correctly 
    (thanks to Yaroslav Kavenchuk)
  * bug fix: non-required arguments were not passed correctly when a method
    defined using DEFMETHOD was called from a mop-generated method using
    CALL-NEXT-METHOD (reported by Pascal Costanza)
  * bug fix: recursion is now permitted in accessors through
    SLOT-UNBOUND.  (reported by Pascal Costanza)
  * bug fix: an error was signaled at startup if the HOME environment 
    variable was defined, but had an empty value (reported by Peter Van Eynde)
  * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
    signal a PROGRAM-ERROR, not a BUG.
  * bug fix: SB-POSIX:READDIR works when built with large file support.
  * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
    to Marco Monteiro)
  * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
    for better automatic indentation support. (thanks to Matt Pillsbury)
  * optimization: loading generic functions no longer takes O(n^2) time,
    proportional to the amount of methods in the generic function
    (reported by Todd Sabin and Jeremy Brown)        
  * optimization: the FIND and POSITION family of sequence functions
    are significantly faster on arrays whose element types have been
    declared.
  * improvements to the Windows port:
    ** Intermittent heap corruption problems have been fixed. (thanks
       to Alastair Bridgewater)
    ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
       work on Windows.
    ** Lisp is able to unwind foreign exception frames from alien
       callbacks. (thanks to Alastair Bridgewater)

changes in sbcl-1.0 relative to sbcl-0.9.18:
  * improvement: experimental support for threading on FreeBSD/x86.
    (thanks to NIIMI Satoshi)
  * improvement: runtime option --dynamic-space-size can be used
    to set the size of the dynamic space reserved on startup.
  * improvement: floating point modes in effect are now saved in
    core, and restored on startup.
  * improvement: GET-INTERNAL-REAL-TIME now reports the time since
    startup, not time since first call to GET-INTERNAL-REAL-TIME.
  * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
    threads are not running after *SAVE-HOOKS* have run.
  * improvement: writes to CLOS instance slots are type-checked in code
    compiled with (SAFETY 3)
  * improvement: floating-point exception handling on FreeBSD (thanks to
    NIIMI Satoshi)
  * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
    (thanks to Zach Beane)
  * improvement: support for files larger than 2GB for CL streams and SB-POSIX
    on Linux/x86
  * improvement: added support for the Shift-JIS external format.
    (contributed by NIIMI Satoshi)
  * improvement: callbacks are supported on Linux/PPC.  (thanks to
    Joshua Ross)
  * bug fix: compiler bug triggered by a (non-standard) VALUES
    declaration in a LET* was fixed. (reported by Kaersten Poeck)
  * bug fix: file compiler no longer confuses validated and already
    dumped structurres. (reported by Kaersten Poeck)
  * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
    fill-pointer signals a type-error as required. (thanks to
    Lars Brinkhoff)
  * bug fix: disassemly of funcallable instances works.
  * bug fix: single stepping on PPC.
  * bug fix: fix thread-safety problems in the type system (generally 
    manifesting as nonsensical errors like "STRING is a bad type specifier 
    for sequences" or "The value 1 is not of type FIXNUM")
  * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
    (reported by Josip Gracin)
  * bug fix: an error is signaled for attempts to displace arrays with
    incompatible element types (thanks to Mario Mommer)
  * bug fix: more correct handling of wide characters in debug info
    (bug reported by Attila Lendvai and fixed by Juho Snellman)
  * optimization: method calls with &OPTIONAL or &KEY arguments are faster
    and don't cause extra consing
  * optimization: MAP and MAP-INTO are significantly faster on vectors
    whose elements types have been declared.
  * Improvements to SB-SPROF:
    ** Support for allocation profiling
    ** Reduced profiling overhead, especially for long profiling runs
  * Improvements to the Windows port:
    ** floating point exceptions are now reported correctly.
    ** stack exhaustion detection works partially.
    ** more accurate GET-INTERNAL-REAL-TIME.
    ** better RUN-PROGRAM behaviour in case of errors in the child
       process.
    ** PROBE-FILE now simplifies pathnames correctly.
    ** DIRECTORY now works correctly with :WILD-INFERIORS.
    ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
       1st 1970 (although time timezone data will be unreliable).
    ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.

changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
  * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
    cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
    Max-Gerd Retzlaff)
  * bug fix: two potential GC deadlocks affecting threaded builds.
  * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
    returns T (reported by Anton Kazennikov)
  * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
    with non-variable places
  * bug fix: remove a race condition in the setting of
    funcallable-instance functions, this should make threaded CLOS
    code more stable against memory faults.
  * bug fix: corruption of specials when unbinding is interrupted by an
    asynchronous unwind (reported by Hannu Koivisto)
  * improvement: the debugger will now also display local variables that
    are only used once, for code compiled with a DEBUG optimization quality
    of 2 or higher.

changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
  * feature: weak hash tables, see MAKE-HASH-TABLE documentation
  * incompatible change: External-format support for FFI calls.  The
    SB-ALIEN:C-STRING no longer implies an ASCII
    external-format. Instead the string is subject to external-format
    conversion. Additionally return values of type SB-ALIEN:C-STRING
    are by default of type (SIMPLE-ARRAY CHARACTER), not
    SIMPLE-BASE-STRING. For an alien type that behaves like the old
    SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
    :ASCII :ELEMENT-TYPE BASE-CHAR).  (thanks to Yaroslav Kavenchuk)
  * incompatible change: SB-EXT package no longer contains the
    following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
    *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
    *ERROR-PRINT-LINES*
  * incompatible change: the single-stepper is no longer available 
    on Alpha, Mips and Sparc platforms.
  * minor incompatible change: the direct superclasses of
    SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
    not (STANDARD-OBJECT FUNCTION).  This makes the
    class-precedence-lists of GENERIC-FUNCTION and
    STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
    1.4.4.5.
  * minor incompatible change: the default stream external format on
    non-unicode SBCL is no longer detected from the locale, but is always
    ISO-8859-1
  * new feature: Add a version of evaluator that uses an interpreter instead
    of the compiler. EVAL still uses the compiler by default, to switch it
    to use the interpreter, set the value of the variable 
    SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
  * minor incompatible change: the single-stepper REPL has been merged
    with the normal debugger (see the "Stepping" heading of the debugger help
    for more details). The debugger command STEP will no longer switch
    to the single-stepper REPL. 
  * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
    for a type now works.
  * bug fix: loading of default sysinit file works. (thanks to Leonid
    Slobodov)
  * bug fix: better detection of circularities in the file-compiler.
    (reported by Marco Monteiro)
  * bug fix: the CL pathname functions now work with files that have
    non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
  * bug fix: The :PTY argument for RUN-PROGRAM will now work on 
    systems with Unix98 pty semantics.
  * bug fix: ASDF-INSTALL will now work with bsd tar.
  * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
    Gracin).
  * bug fix: timers expiring in dead threads no longer cause a
    type-error (reported by Paul "Nonny Mouse"). 
  * bug fix: thanks to more lightweight single-stepper instrumentation, 
    code compiled with (DEBUG 3) will compile and execute significantly faster,
    and will have more accurate type-inferencing than before
  * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
    whose bindings are modified
  * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
    ** bug fix: arguments to RUN-PROGRAM are escaped correctly
    ** replace dummy implementations of CL:MACHINE-INSTANCE and 
       CL:SOFTWARE-VERSION with proper version

changes in sbcl-0.9.16 relative to sbcl-0.9.15:
  * feature: implemented the READER-METHOD-CLASS and
    WRITER-METHOD-CLASS portion of the Class Initialization Protocol
    as specified by AMOP.
  * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
    no longer exists.
  * optimization: faster LOGCOUNT implementation on x86 and x86-64
    (thanks to Lutz Euler)
  * optimization: hashing of general arrays and vectors has been
    improved. (reported by Any Fingerhut)
  * enhancement: SB-INTROSPECT is now able to find definitions of
    profiled functions. (thanks to Troels Henriksen)
  * enhancement: compiler-macro expansion applies now to FUNCALL forms
    as well, allowing compiler-macros for SETF-functions to expand.
  * enhancement: step-instrumentation no longer wraps known
    single-value functions with multiple-value context, allowing
    better type inference.
  * fixed bug #337: use of MAKE-METHOD in method combination now works
    even in the presence of user-defined method classes.  (reported by
    Bruno Haible and Pascal Costanza)
  * fixed bug #339(c): if there are applicable methods not part of any
    long-form method-combination group, call INVALID-METHOD-ERROR.
    (reported by Bruno Haible)
  * fixed bug #361: the :FUNCTION initarg in the protocol for
    initialization of methods can now be used to override
    internally-produced optimized functions.  (reported by Bruno
    Haible)
  * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
    system-provided lambda expression no longer cause warnings about
    unbound #:|pv-table| symbols.
  * bug fix: improved the handling of type declarations and the
    detection of violations for keyword arguments with non-constant
    defaults.
  * bug fix: potentially erronous calls to PATHNAME and
    MERGE-PATHNAMES were being flushed in some cases.
    (reported by Richard Kreuter)
  * bug fix: compiled calls to TYPEP were mishandling obsolete
    instances.  (reported by James Bielman and Attila Lendvai)
  * bug fix: format strings with ~> without matching ~< no longer
    trigger an AVER, but signal an understandable error instead.
    (reported by Antonio Martinez)
  * bug fix: specifying an output-file without a pathname-type for
    COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
    of input-file instead of "fasl". (reported by Robert Dodier)
  * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
    (reported by James Y Knight).
  * bug fix: compiler-macros-function did not consider the environment
    argument for shadowing by local functions.
  * bug fix: compiler-macros expansion was inhibited by local INLINE 
    declarations.
  * bug fix: inline expansions of known functions were subject to
    step-instrumentation in high debug policies, leading to problems
    with type-inference.
  * bug fix: compiler failed to differentiate between different CONS
    types in some cases.
  * bug fix: fixed input, output and error redirection in RUN-PROGRAM
    for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
  * bug fix: #368: incorrect use of expressed vs. upgraded array
    element type.
  * bug fix: #306a: more precise unions of array types.
  * thread-safety improvements:
    ** CONDITION-WAIT could return early on Linux, if the thread was
       interrupted and subsequently continued with SIGCONT.
    ** STABLE-SORT and ADJUST-ARRAY were not reentrant.

changes in sbcl-0.9.15 relative to sbcl-0.9.14:
  * added support for the ucs-2 external format.  (contributed by Ivan
    Boldyrev)
  * minor incompatible change: pretty printing of objects of type
    (cons symbol) is, in the default pprint-dispatch-table, now
    sensitive to whether the symbol satisfies FBOUNDP.  (thanks to
    Marcus Pearce)
  * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
    called later in a class's lifetime, possibly as late as when the
    first instance of the class is created.  Previously,
    SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
    class became finalizeable.
  * fixed bug: FILE-POSITION sometimes returned inconsistent results
    for multibyte external-format streams.  (thanks to "vbzoli")
  * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
    with :ALLOCATION :CLASS inherited from superclasses of the
    original class.
  * fixed bug: anonymous classes can now be created using the :NAME
    initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
    AMOP.  (reported by Leonid Slobodov on comp.lang.lisp)
  * fixed bug: core-files saved with :EXECUTABLE T can again be 
    executed when SBCL_HOME isn't set. (reported by James Knight)
  * fixed bug: toplevel LOCALLY forms with declarations could 
    occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
  * fixed bug: printing from several different threads using different
    values of *print-case* could cause invalid output, due to 
    some internal special variables of the printer not being bound
    thread-locally (reported by Max Mikhanosha)
  * fixed bug: SPECIALIZER metaobjects (including anonymous classes
    and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
    (reported by Pascal Costanza)
  * fixed bug: FINALIZE-INHERITANCE is called from
    REINITIALIZE-INSTANCE on classes when the class has previously
    been finalized, as required by AMOP.
  * minor code generation optimizations:
    ** better register allocation in CLOS dispatching functions
    ** overflow detection when coercing signed bytes to fixnums on x86-64 
    ** is now implemented with one IMUL instruction instead of three shifts
    ** more efficient bit-vector access on x86 and x86-64
    ** more efficient access to raw structure slots on x86 and x86-64
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
       return its argument.

changes in sbcl-0.9.14 relative to sbcl-0.9.13:
  * feature: thread support on Solaris/x86, and experimental thread support
    on OS X/x86.
  * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
    of heap.
  * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
    default method for SLOT-UNBOUND.
  * minor incompatible change: prevent the user from specializing the
    new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS.  It's
    somewhat counter to the intent of the protocol, I (CSR) think, and
    additionally it just doesn't work in SBCL as currently
    implemented, thanks to optimizations (that are always valid for
    the other three SLOT-VALUEish functions, but not for the setter).
  * minor incompatible change: the :SB-LDB feature is now enabled by
    default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
    the low-level debugger.
  * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
    from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
    down to an SB-SYS:FD-STREAM, allowing direct communication in
    more cases. Temporary files and pipes are still used as a fallback
    strategy.
  * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
    on the non-reentrant C functions localtime(3) and gmtime(3).
  * bug fix: global optimization policy was not visible in LOCALLY and 
    MACROLET forms.
  * bug fix: class objects can be used as specializers in methods.
    (reported by Pascal Costanza)
  * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
    with a type error. (reported by blitz_ on #lisp)
  * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
    failed with a type error.  (reported by Pascal Bourguignon)
  * bug fix: merging pathnames against defaults with :DIRECTORY
    starting with '(:RELATIVE :BACK) should preserve the :BACK.
    (reported by James Y Knight)
  * bug fix: saving large (>2GB) cores on x86-64 now works
  * bug fix: a x86-64 backend bug when compiling (setf aref) with a 
    constant index and a (simple-array (signed-byte 32)) array
  * bug fix: NAME-CHAR on an invalid symbol no longer signals an
    error (patch by Robert J. Macomber)
  * feature: TIME now displays the amount of run-time spent in GC
  * bug fix: The debugger now does a better job of respecting 
    (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
    when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
    workaround for bug 403.) 
  * bug fix: floating point exception handling now works on FreeBSD
    (thanks to NIIMI Satoshi)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** MISC.641: LET-conversion were not supposed to work in late
       compilation stages.
  * improvements to the Win32/x86 port:
    ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
    ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.

changes in sbcl-0.9.13 relative to sbcl-0.9.12:
  * new feature: source path information is generated for macro-expansion
    errors for use in IDE's like Slime (thanks to Helmut Eller)
  * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
  * bug fix: compiler does not loop forever on an invalid type in
    TYPEP.
  * improvement: compilation of most CLOS applications is significantly
    faster
  * optimization: added a limited bytecode compiler for simple toplevel
    forms, speeding up compilation and FASL loading
  * bug fix: the statistical profiler now properly distinguishes anonymous
    functions

changes in sbcl-0.9.12 relative to sbcl-0.9.11:
  * minor incompatible change: in sbcl-0.9.11 (but not earlier
    versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
    into subclasses of the finalized class.  Now user calls to
    FINALIZE-INHERITANCE finalize just the one class, and calls by the
    system return before any subclasses are finalized.
  * minor incompatible change: The reader no longer ignores errors
    regarding non-existent packages in #+ and #- feature tests.
  * new feature: command line options --no-sysinit, --no-userinit to
    inhibit loading the corresponding init files
  * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS, 
    for sending data through UDP sockets (thanks to François-René Rideau)
  * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
    error is signalled instead (thanks to François-René Rideau)
  * minor incompatible change: Error signalling behaviour of lexical
    operations violating package locks has changed slightly. Refer to
    documentation on package locks for details.
  * enhancement: EVAL can process IF-expressions without resorting to the
    compiler.
  * bug fix: Some operations involving SIN, COS, or TAN weren't being
    constant-folded properly due to missing out-of-line functions.
    (reported by Mika Pihlajamäki)
  * bug fix: LISTEN sometimes returned T even in cases where no data was
    immediately available from the stream
  * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
    were reversed. (reported by Levente Mészáros)
  * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
    when given a vector displaced to an adjustable array.  (reported
    by Utz-Uwe Haus)
  * bug fix: derivation of float boundaries from numbers outside the
    appropriate float range (reported by John Wiseman)
  * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
    list.
  * bug fix: precomputing cache entries for generic functions with
    some subclasses of specializers as yet invalid does not attempt to
    fill a cache line with a negative offset.  (reported by Levente
    Mészároz)
  * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
    allow condition class objects as arguments to DOCUMENTATION and
    (SETF DOCUMENTATION); only find and set documentation for
    structure names for the STRUCTURE doc-type.  (suggested by Gary
    King)
  * improvements to the Win32/x86 port:
    ** (user-homedir-pathname) and default initialization file
       locations now know about the user's "Documents and Settings"
       directory (thanks to Yaroslav Kavenchuk)
    ** run-program is implemented (thanks to Mike Thomas)
    ** sockets support (thanks to Timothy Ritchey)
    ** better backtrace support (thanks to Alastair Bridgewater)
    ** sb-grovel supported
    ** asdf-install and sb-posix work somewhat
    ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
  * improvements to the Solaris/x86 port:
    ** works on Solaris 11/Solaris Express
    ** floating-point exception handling support
    ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** REMOVE-METHOD returns its generic function argument even when
       the method is not one of the generic functions' methods.
    ** objects declared as MEMBER types can be admissible arguments to 
       structure accessors.
    ** printing characters should simply be printed by the FORMAT ~:C
       directive.
    ** compiler failure when compiling functions with hairy constant
       defaults for optional parameters.
    ** compiler produces wrong code when MAYBE-INLINE-expanding a
       function, which is already optimized.

changes in sbcl-0.9.11 relative to sbcl-0.9.10:
  * new platform: experimental support for SBCL x86/Darwin, including
    MacOS X 10.4.5 on Intel.
  * new feature: Unicode character names are now known to the system
    (through CHAR-NAME and NAME-CHAR).
  * new feature: the filesystem location of SBCL's core file is
    exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
  * minor incompatible change: the contrib modules SB-POSIX and
    SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
    this change is to make it easier to distribute
    location-independent binaries.
  * bug fix: as implied by AMOP, standardized classes no longer have
    slots named by external symbols of public packages.  (reported by
    Pascal Costanza)
  * optimization: calling functions via constant symbols -- as in 
    (FUNCALL 'FOO) -- is now roughly as efficient as calling them
    via the function object as in (FUNCALL #'FOO).
  * optimization: on x86 and x86-64, the calling convention for the
    case of dealing with an unknown number of values has been altered
    to be friendlier to the prediction heuristics implemented,
    particularly with reference to CALL and RET pairing.  (thanks to
    Alastair Bridgewater)
  * optimization: on x86, the code for access to thread-local symbol
    values and binding/unbinding of thread-local symbols is smaller.
    (thanks to Alastair Bridgewater)
  * enhancement: CONSTANTP is now able to determine constantness of
    more complex forms, including calls to constant-foldable standardized
    functions and some special forms beyond QUOTE.
  * fixed bug: occasional GC crashes on Solaris/x86
  * optimization: x86-64 supports stack allocation of results of simple
    calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
  * enchancement: the PROCESS-INPUT and -OUTPUT streams created by 
    SB-EXT:RUN-PROGRAM can be used for both character and byte IO
    (thanks to James Knight)
  * fixed bug: CL:LISTEN always returns NIL at end of file, as required
    by the standard (thanks to Stephen Compall)

changes in sbcl-0.9.10 relative to sbcl-0.9.9:
  * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
    be used for bundling the runtime and the core file into one 
    executable binary. This feature is not currently supported on all SBCL
    platforms.  (thanks to James Bielman and NIIMI Satoshi)
  * new feature: a generational or ephemeral garbage collector is now
    the default on the PowerPC platform (both Linux and Darwin).  The
    old Cheney (stop and copy) collector is a build-time option.
    (thanks to Cyrus Harmon, after Raymond Toy)
  * minor incompatible change: the method by which SBCL finds its
    contributed modules has changed; it no longer relies on symbolic
    links from an $SBCL_HOME/systems directory, but searches directly
    in the subdirectories of $SBCL_HOME.
  * enhancement: the dynamic heap size on the Linux/PPC platform is
    markedly larger, even using the older Cheney garbage collector.
  * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
    certain complicated string types.  (reported by Gary King)
  * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
    character codes.  (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
  * fixed bug: DOCUMENTATION on structures no longer signals an 
    error if no structure type of the right name exists.  (reported by
    Glenn Ehrlich)
  * fixed bug: compiled calls to MAKE-INSTANCE where there is an
    applicable non-standard primary or :AROUND method on
    INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
    no longer cause unbound variable errors.  (reported by Kevin Reid)
  * fixed bug: metacircle resolution in cases where methods have slots
    added before the slots from STANDARD-METHOD.  (reported by Jean
    Bresson)
  * fixed bug: the Power PC assembly code for calling into lisp
    sometimes computed its return address wrongly (depending on the
    15th bit of the address, fixed at link time).  This used to
    manifest itself in a segmentation violation while building PCL.
    (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
    many others over the years)
  * fixed bug: ROOM no longer reports silly numbers for stack usage if
    the stack pointer (treated as a signed integer) is negative.
    (thanks to Peter van Eynde)

changes in sbcl-0.9.9 relative to sbcl-0.9.8:
  * new platform: experimental support for the Windows operating
    system has been added.  (thanks to Alastair Bridgewater)
  * fixed several bugs in and robustified the PPC FFI (including
    callbacks).  (thanks to Cyrus Harmon and Heiner Schwarte)
  * bug fix: allow non-simple string symbol names (reported by Paul
    Dietz)
  * bug fix: interrupt handling on NetBSD (thanks to Richard M
    Kreuter)
  * bug fix: saving a core corrupted callbacks on x86/x86-64
  * bug fix: closed a loophole in metacircularity detection and
    grounding in the PCL implementation of CLOS.
  * bug fix: TRUENAME on "/" no longer returns a relative pathname.
    (reported by tomppa on #lisp)
  * bug fix: clear the direction flag on Lisp -> C transitions, as 
    required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
    (reported by Andrej Grozin and Hendrik Maryns)
  * optimization: major improvements to GC efficiency on GENCGC platforms
  * optimization: faster implementation of EQUAL
  * optimization: emit more efficient opcodes for some common 
    immediate->register MOV instructions on x86-64.  (thanks to Lutz Euler)
  * optimization: several other minor code-generation improvements on x86-64
  * fixed segfaults on x86 FreeBSD 7-current.  (thanks to NIIMI Satoshi)

changes in sbcl-0.9.8 relative to sbcl-0.9.7:
  * minor incompatible change: (SETF CLASS-NAME) and (SETF
    GENERIC-FUNCTION-NAME) are generic functions once more (reverting
    the change in the 0.9.7 release).  (SETF CLASS-NAME) is specified
    by ANSI as a generic function, and for consistency (SETF
    GENERIC-FUNCTION-NAME) is treated likewise.
  * fixed bug #233.b: make constraint propagation notice when a variable
    value is changed after it is referenced but before it is used
  * fixed bug #296: no more arbitrary behaviour from filenames with
    odd characters as --load arguments.  (reported by Adam Warner)
  * fixed bug #390: :CHARACTER-SET pathname components now work as
    expected.  (reported by Tim Daly Jr)
  * fixed bug #391: complicated :TYPE intersections in slot
    definitions no longer cause an error in PCL internals.
  * fixed bug #392: plugged a hole in the obsolete-instance protocol.
  * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
    returning the number of octets which would be written to the
    file-stream.  (thanks to Robert J. Macomber)
  * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
    arguments to the internal stream functions.  (thanks to David
    Lichteblau)
  * bug fix: stack exhaustion is now handled on the x86/SunOS
    platform.  (thanks to Daisuke Homma)
  * bug fix: more accurate ROOM results on GENCGC platforms
  * optimization: improved type inference for arithmetic-for
    index variables in LOOP
  * optimization: faster floating-point SQRT on x86-64
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** DOCUMENTATION returns NIL instead of "" for method combinations
       that don't have a docstring
\a
changes in sbcl-0.9.7 relative to sbcl-0.9.6:
  * minor incompatible change: (SETF CLASS-NAME) and (SETF
    GENERIC-FUNCTION-NAME) are no longer generic functions, and
    therefore are not customizeable by user code (as seems to be at
    least permitted and maybe required by AMOP).  As a consolation,
    however, the SBCL implementation of these functions now calls
    REINITIALIZE-INSTANCE as specified by AMOP.
  * bug fix: slot-definition documentation is propagated as per ANSI
    7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
    COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
    argument, as do eslotd initialization methods.  (from Pascal
    Costanza's "Closer" project)
  * bug fix: REINITIALIZE-INSTANCE on generic functions calls
    COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
    specified by AMOP.
  * bug fix: it is now possible to have more than one subclass of
    STANDARD-GENERIC-FUNCTION without causing stack overflow.
    (reported by Bruno Haible, Pascal Costanza and others)
  * bug fix: the dependent update protocol now works for generic
    functions.  (thanks to Gerd Moellmann; reported by Bruno Haible
    and Pascal Costanza)
  * bug fix: condition-class instances corresponding to
    DEFINE-CONDITION forms are now created eagerly.  (reported by
    Kalle Olavi Niemitalo on comp.lang.lisp)
  * bug fix: floating point printing is more accurate in some
    circumstances.  (thanks to Simon Alexander)
  * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
    merged with *DEFAULT-PATHNAME-DEFAULTS*.
  * bug fix: callbacks on OS X now preserve stack-alignment.
    (thanks to Cyrus Harmon)
  * enhancement: the x86-64 disassembler is much better at
    disassembling SSE instructions.  (thanks to Lutz Euler)
  * enhancement: CHECK-TYPE on a variable now allows type-inference
    (thanks to Vincent Arkesteijn)
  * optimization: improved performance of EUC-JP external format.
    (thanks to NIIMI Satoshi)
  * optimization: performance improvements to IO on file streams of
    :ELEMENT-TYPE CHARACTER
  * optimization: much faster memory allocation on x86-64
  * optimization: faster garbage collections (latency and throughput)
    on gencgc
  * optimization: faster bignum subtraction and division on x86 and x86-64 
    (thanks to Lutz Euler)
  * optimization: more accurate type-inference for loops that use a
    floating point index variable or a negative step.

changes in sbcl-0.9.6 relative to sbcl-0.9.5:
  * bug fix: add a workaround to SBCL looping infinitely at startup on
    Linux kernels with apparently buggy implementations of personality().
    (thanks to Svein Ove Aas)
  * bug fix: Unicode symbols are correctly printed in LDB backtraces
    (thanks to David Lichteblau)
  * bug fix: local bindings shadow global symbol macros for
    macroexpansion.  (reported by Matthew D Swank on comp.lang.lisp)
  * optimization: non-open coded uses of numeric comparison operators
    (e.g. >) no longer cons when called with more than one parameter
    on platforms supporting dynamic-extent allocation.
  * enhancement: saving cores with foreign code loaded is now supported
    on MIPS/Linux in addition to the previously supported platforms.
  * bug fix: FIND-RESTART now tests for activity, not applicability when given
    a restart object as identifier. (reported by Helmut Eller for CMUCL)
  * bug fix: division by zero in sb-sprof when no samples were collected
  * bug fix: a race when a slow to arrive sigprof signal killed sbcl
  * bug fix: asdf-install uses CRLF as required by the HTTP spec.
    (thanks to Alexander Kjeldaas)
  * new feature: ignoring the return values of destructive functions
    like NREVERSE signals a compile-time style-warning.
    (thanks to Kevin Reid)
  * new feature: the alignment of alien structure fields can be
    explicitly specified. (thanks to Cyrus Harmon)
  * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
    and dump core on SIGQUIT
  * threads
    ** incompatible change: threads do not inherit values of specials
       from their parents (see manual)
    ** bug fix: threads stacks belonging to dead threads are freed by the
       next exiting thread, no need to gc to collect thread stacks anymore
    ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
    ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
    ** bug fix: dynamic variable and thread start related gc lossage
    ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
       suspended for gc
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** SUBTYPEP is slightly more accurate on heinously complicated
       CONS types where some of the members have uncertain (in the
       NIL, NIL sense) type relationships to each other.
  * GENCGC
    ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
       no longer purified unless :PURIFY T is explicitly specified.
    ** Non-purified cores are significantly smaller than before

changes in sbcl-0.9.5 relative to sbcl-0.9.4:
  * new feature: timers based on Zach Beane's excellent timer package
  * added support for the following external formats: koi8-u,
    x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
    cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
    iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
    iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
    iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
    cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
  * incompatible change: a threaded SBCL will no longer revert to
    non-threaded mode on non-NPTL systems, but refuse to start entirely.
  * bug fix: interrupts are disabled until startup is complete; no
    more sigsegvs when receiving a signal to soon
  * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
    platforms
  * bug fix: add a workaround for the memory randomization features in
    Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
    workaround will only be in effect on systems with the proc filesystem
    mounted.
  * bug fix: printing objects of type HASH-TABLE signals a
    PRINT-NOT-READABLE error when *READ-EVAL* is NIL.  (reported by
    Faré Rideau)
  * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
    have been running for over 50 days. (reported by Gilbert Baumann)
  * bug fix: the logic for getting names of functions gets less
    confused when confronded with alternate-metaclass
    funcallable-instances.  (reported by Cyrus Harmon)
  * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
    consistent, even on internal alternate-metaclass objects.
  * bug fix: flush closure information collected by physical
    environment analysis prepass before the main pass. (bug reported
    by Vasile Rotaru)
  * bug fix: compiler pack phase does not modify a hashtable, which is
    iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
  * bug fix: backquote does not optimize (LIST* x y) to CONS when x
    may be expanded. (reported by Alexander <malishev> on c.l.l)
  * bug fix: no more style-warnings when compiling code using LAST
    under high-SPEED low-SPACE settings.  (reported by David Wallin
    aka froog on #lisp)
  * fixed some incompatibilities between SBCL's MOP and the MOP
    specified by AMOP:
    ** the METAOBJECT class is now implemented;
    ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
       STANDARD-OBJECT, as required;
    ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
       now compatible (as required by VALIDATE-SUPERCLASS); there
       remains a consistency requirement over the presence or absence
       of the FUNCTION class in the superclass list of finalized
       classes; see the manual for more details;
    ** the :AROUND method for COMPUTE-SLOTS on
       FUNCALLABLE-STANDARD-CLASS now honours the primary method's
       requested slot ordering.
  * threads
    ** bug fix: parent thread now can be gc'ed even with a live
       child thread
    ** bug fix: binding a special with PROGV to no value is not
       the same as that symbol not having been bound (thanks to
       Hannu Koivisto)
    ** bug fix: binding specials is thread safe (thanks to
       Hannu Koivisto)
    ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
       and SB-SPROF do not die with 'no handler for signal XX in
       interrupt_handle_now(..)' anymore
    ** bug fix: WITH-TIMEOUT works with multiple running threads
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** ENSURE-GENERIC-FUNCTION should take a method class object for
       the :method-class keyword argument.

changes in sbcl-0.9.4 relative to sbcl-0.9.3:
  * new port: the Solaris operating system on x86 processors is now
    mostly supported, though some rough edges in the environment
    remain.  (thanks to Daisuke Homma)
  * enhancement: SBCL on MIPS platforms now has a much larger dynamic
    space for its heap.  (thanks to Thiemo Seufer)
  * optimization: SBCL on MIPS platforms now supports dynamic-extent
    closures.  (thanks to Thiemo Seufer)
  * minor incompatible change: eof selects abort in the debugger.
  * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
    *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
    is disabled.
  * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
    deprecated, and will go away in a future revision of SBCL.
  * minor incompatible change: GC-ON and GC-OFF are no longer
    implemented with a counter, it does not matter how many times gc
    is switched on or off
  * bug fix: discriminating functions for generic function classes
    with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
    make invalid assumptions about method precedence order.  (reported
    by Bruno Haible)
  * bug fix: TRUNCATE now correctly signals division by zero on MIPS
    platforms.  (thanks to Thiemo Seufer)
  * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
  * bug fix: correct pathname printing with printer escaping is on.
    (thanks to Kevin Reid)
  * bug fix: complex VOP definitions in "user-space" no longer trigger
    package locks.  (reported by Zach Beane)
  * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
    for funcallable-instance functions, and is no different from
    regular LAMBDA.
  * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
    characters in the current readtable.  (reported by Nicholas Neuss)
  * bug fix: SUBTYPEP on various CONS types returns more a more
    accurate acknowledgment of its certainty.
  * optimizations: REMOVE-DUPLICATES now runs in linear time on
    lists in some cases.  This partially fixes bug 384.
  * flush all standard streams before prompting in the REPL and the
    debugger.
  * bug fix: signal handling and triggering gc do not conflict
    directly anymore, in particular a high frequency sb-sprof does
    not prevent gc from running
  * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
    approximation for timezone and DST information between the
    universal time epoch and the smallest negative 32-bit time_t.
  * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
    year 1899 so long as the encoded time is non-negative.
  * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
    the correct answer for non-constant rotate amounts when performing
    an inline 32-bit rotation.
  * threads
    ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
       there is only one thread in the session
    ** bug fix: memory leak for streams created in one thread and
       written to in another
    ** bug fix: lockup when compiled with gcc4
    ** bug fix: race that allows the gc to be triggered when gc is
       inhibited
    ** bug fix: one less memory fault in INTERRUPT-THREAD, again
    ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
       queue is full
    ** bug fix: finished threads are gc'ed properly
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** CALL-NEXT-METHOD signals an error (in safe code) when the call
       has arguments with a different set of applicable methods from
       the orignal arguments.
    ** The type error thrown by MAP now has a correct expected-type
       cell.
    ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
       name a compiled function.
    ** Three MISC tests where a large bignum was improperly coerced to
       a float that couldn't represent that bignum during type
       derivation were fixed.
    ** SUBTYPEP can now handle types involving the negation of a
       list-form FUNCTION type.
    ** SUBTYPEP also now handles certain COMPLEX type specifiers such
       as (COMPLEX (AND RATIO (NOT FIXNUM))).
    ** READ-BYTE and WRITE-BYTE no longer take stream designators.

changes in sbcl-0.9.3 relative to sbcl-0.9.2:
  * New feature: Experimental support for bivalent streams: streams
    opened with :element-type :default now allow character and binary
    (unsigned-byte 8) I/O
  * Support for the koi8-r external format.  (thanks to Ivan Boldyrev)
  * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
    to Zach Beane)
  * bug fix: no more highly sporadic "couldn't check whether ~S is
    readable" when reading a stream and an interrupt hits in the middle
    of a select system call
  * compiler better recognizes complex arrays (reported by Eduardo
    Muñoz)
  * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
    for variables on SBCL built with linkage-tables. (reported by Luis
    Oliveira)
  * various error reporting improvements.
  * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
    (Thanks to Thiemo Seufer)
  * bug fix: nested reader invokations work correctly
  * bug fix: it is possible to have simultaneous references to foreign
    code and foreign data with the same name.
  * threads
    ** added x86-64 support
    ** incompatible change: the threading api now works with thread
       objects instead of thread ids
    ** bug fix: threads are protected from signals and interruption when
       starting up or going down
    ** bug fix: a race where an exiting thread could lose its stack to gc
    ** bug fix: don't halt on infinite error in threads if possible
    ** fixed numerous gc deadlocks introduced in the pthread merge
    ** bug fix: fixed thread safety issues in read and print
    ** bug fix: debugger doesn't hang on session lock if interrupted at
       an inappropriate moment
    ** bug fix: run-program is now thread safe(r)
    ** bug fix: inner with-recursive-lock no longer releases the mutex
  * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** TYPE-ERRORs from signalled by COERCE now have DATUM and
       EXPECTED-TYPE slots filled.
    ** MULTIPLE-VALUE-SETQ always returns the primary value of the
       values form.
    ** MAKE-CONDITION accepts classes as type-designators.
    ** COMPILE may never return NIL.
    ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
       range before calling Unix time functions

changes in sbcl-0.9.2 relative to sbcl-0.9.1:
  * numerous signal handling fixes to increase stability
  * Support for EUC-JP external format.  (thanks to NIIMI Satoshi)
  * minor incompatible change: we now correctly canonize default
    initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
    as per the MOP, rather than the historical (INITARG INITFUNCTION
    INITFORM).  (reported by Bruno Haible)
  * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
    :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
    for more information.
  * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
    pathname is a directory pathname.
  * SB-SPROF now works (more) reliably on non-GENCGC platforms.
  * dynamic space size on PPC has been increased to 768Mb. (thanks to
    Cyrus Harmon)
  * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
    :METACLASS argument in addition to a class name.  (reported by
    Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
  * RESTART-CASE can now be wrapped around CALL-METHOD forms.
    (reported by Bruno Haible; patch from Gerd Moellmann)
  * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to 
    Sascha Wilde)
  * bug fix: more cleanups to the floating point exception handling on
    x86-64 (thanks to James Knight)
  * bug fix: the compiler does not try to constant fold calls of
    COERCE and BIT-* functions when they return freshly constructed
    objects. (reported by Nikodemus Siivola and Paul Dietz)
  * optimization: improved the MIPS versions of generic (in the
    generic sense) arithmetic routines.  (thanks to Thiemo Seufer)
  * optimization: direct conversion of (unsigned-byte 32) to floats on
    the PowerPC platform.
  * optimization: structure instances with raw slots now use less
    memory, and probably show better memory locality.  (thanks to
    David Lichteblau)
  * optimization: DYNAMIC-EXTENT declarations for lists and closures
    are treated as requests for stack allocation on the x86-64,
    Alpha-32, PPC and SPARC platforms.
  * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
    the wrong file descriptor; implementation of SOCKET-OPEN-P.
    (thanks to Tony Martinez)
  * threads
    ** gcing a dead thread can no longer lead to lockups
    ** threads block signals until they are set up properly
    ** errno is no longer shared by threads
    ** interrupt-thread restores the eflags register on x86
    ** fixed some lockups due to gc/thread interaction
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** invalid dotted lists no longer raise a read error when 
       *READ-SUPPRESS* is T
    ** signal an error if a symbol that names a declaration is used
       as the name of a type, or vice versa
    ** allow using the (declare (typespec var*)) abbreviation for 
       (declare (type typespec var*)) with all type specifiers
    ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
    ** always return NIL from PPRINT-POP when OBJECT is NIL
    ** don't signal errors when pretty-printing malformed LABELS, 
       FLET or MACROLET forms
    ** declarations in a DOLIST body are also in scope for the 
       DOLIST return-form
    ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
       and OUTPUT-FILE
    ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is 
       always evaluated

changes in sbcl-0.9.1 relative to sbcl-0.9.0:
  * fixed cross-compiler leakages that prevented building a 32-bit
    target with a 64-bit host compiler.
  * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
    opened with :IF-EXISTS :OVERWRITE.
  * fixed bug 281: error for an invalid qualifier in a short-form method
    combination method is not signalled until the faulty method is called.
  * bug fix: iteration variable type inferrer failed to deal with open
    intervals. (reported by Alan Shields)
  * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
    "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
    or more runtime options were provided to the sbcl binary.
  * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
  * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
  * optimization: there's now a fast-path for fixnum arguments in the
    generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
  * optimization: the code generated on x86-64 is more compact thanks
    to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
  * bug fix: floating-point exception handling now partly works on
    x86-64.  (Thanks to James Knight)
  * improvement to the MIPS backend from Thiemo Seufer: C-style
    64-bit long long arguments and return values to alien functions
    are now supported.
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** the type-error signalled from WARN has a filled-in DATUM slot.
    ** the type-error required when a stream is not associated with
       a file has the stream as its datum.
    ** type-errors on single-floats on x86-64 no longer have 
       :INVALID-OBJECT as the datum
    ** the type-errors signalled for invalid function names now have
       a correct expected type
    ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
       for structures defined with a :TYPE. Documentation strings for
       typed structures are no longer immediately discarded
    ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on 
       broadcast streams and synonym streams. FILE-LENGTH now also works
       on broadcast streams.

changes in sbcl-0.9.0 relative to sbcl-0.8.21:
  * incompatible change: the --noprogrammer option, deprecated since
    version 0.7.5, has been removed.  Please use the equivalent
    --disable-debugger option instead.
  * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
    interrupts enabled.
  * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
    inoperational for a while now) has been completely removed.
  * null lexical environments are now printed as #<NULL-LEXENV>, 
    significantly reducing the amount of clutter in typical backtraces.
  * documentation on weak pointers, finalization, and after GC hooks
    has been added to the manual.
  * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
    as other specialized array types, is much faster.  SUBSEQ and
    COPY-SEQ on such arrays have also been sped up.
  * optimization: EQL is now more efficient when at least other argument 
    is known to be of type (OR FIXNUM (NOT NUMBER)).
  * fixed bug: the runtime is now less vulnerable to changes in the
    size of the SBCL object on OS X, and virtual memory is reserved for
    all spaces that need to be at a fixed address.
  * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
  * fixed bug: finalizers and after GC hooks that cause consing are now
    safe.
  * fixed bug: compiler error messages and summaries are now printed to 
    *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
  * fixed inference of the upper bound of an iteration variable.
    (reported by Rajat Datta).
  * fixed bug 211e: calling local functions with duplicated constant
    keyword argument no longer causes a bogus style warning about an
    unused variable.
  * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
    ftype declaration to be disregarded. (reported by Dave Roberts)
  * fixed bug 373: caused by erronous compilation of references to alien
    variables in the runtime on ppc/darwin.
  * fixed bug 376: CONJUGATE type deriver.
  * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
    (reported by Baughn on #lisp)
  * a cross-compiler bug on non-x86 platforms has been identified and
    fixed.  (thanks to Bruno Haible)
  * improvements to the MIPS runtime code for increased stability.
    (thanks to Thiemo Seufer)
  * increased the maximimum compact environment size to allow
    purify on images with large amounts of functions. (thanks to Cyrus Harmon)
  * partial workaround for bug 135: don't name the function we're
    calling for hairy cases of EVAL, so as not to accumulate one environment
    entry per call to EVAL.  (reported by Kevin Reid)
  * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
  * optimization: type testing for non-vector arrays should be faster.
  * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
    on x86-64
  * bug fix: setting 31st element of a bit vector to zero did not work
    on Alpha-32.
  * bug fix: redefining a class definition which failed due to a
    previous accessor / function clash now works (but see BUGS entry
    #380 for more problems in this area).  (thanks to Zach Beane)
  * the long form of DEFINE-METHOD-COMBINATION disables method group
    checking when given a single method group with pattern *.  (thanks
    to Wendall Marvel)
  * on x86 compiler supports stack allocation of results of simple
    calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
  * support for building and running on Mac OS X 10.4 "Tiger" has been added
    ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
  * fixed some bugs related to Unicode integration:
    ** the restarts for recovering from input and output encoding
       errors only appear when there is in fact such an error to
       handle.
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** MISC.549 and similar: late transformation of unsafe type
       assertions into derived types caused unexpected code
       transformations.
    ** SCALE-FLOAT type deriver is less wrong.
    ** type derivers for EXP, LOG and similar functions compute result
       types for complex arguments better.
    ** (MISC.563) CONJUGATE type deriver works for very restricted
       complex types.
    ** out-of-line type testers for character strings are available.
    ** EQUAL compiler transform understands specialness of objects
       of type BIT-VECTOR.
    ** accessing double-floats stored on the stack now works on x86-64.
    ** debugger internals could sometimes create invalid lispobjs, 
       resulting in GC crashes.
    ** MISC.548: type check weakening can convert required type into
       optional.
    ** initialization forms for bindings are not in scope of free special
       declarations.

changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
  * incompatible change: thread support for non-NPTL systems has
    been removed - locking is buggy and unreliable.  A threaded 
    SBCL build will now warn at startup and refuse to create
    new threads, unless futex support is detected in the kernel
  * incompatible change: the top level REPL now has only an ABORT
    restart associated with it, not TOPLEVEL and ABORT as it used to.
    TOP and TOPLEVEL are now available as debugger commands for 
    returning to the top level.
  * incompatible change: forms evaluated in the REPL now use the
    global optimization policy.
  * incompatible change: user- and system-initialization files are
    no longer processed with LOAD, but by READ and EVAL; hence the
    global optimization policy, startup package, readtable, etc, 
    can be set by them.
  * The .fasl file format number has been incremented because of 
    various incompatible changes.
  * internal entry point details and argument counts no longer appear
    in backtraces unless explicitly requested by setting
    SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
  * built-in and standard functions no longer have names like "top
    level local call to FOO".
  * fixed bug 32: functions defined in non-null lexical environments
    now have more legible printed representation
  * fixed bug 33: functions defined in non-null lexical environemnts
    are now more amenable to inspection by INSPECT.
  * workaround for bug 354: XEPs no longer appear in backtraces unless
    explicitly requested.
  * fixed bug: receiving the signal which results from attempting to
    write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
    SIGBUS on other *BSDs) on architectures where the C stack is also
    the Lisp stack (x86 and x86-64) from foreign code no longer leads
    to debugger confusion or wild instability.  (reported by Cheuksan
    Wang)
  * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
    notes when operating on objects known to be SIMPLE-BIT-VECTORs.
    (reported by Lutz Euler)
  * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
    compiled code. (reported by Paul Dietz)
  * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
    specialized parameter to a method to something that is not TYPEP
    the specializer is now possible.
  * fixed bug: the MAKE-INSTANCE optimization is now correct in the
    face of package deletion.
  * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
    pathname. (reported by David Tolpin on c.l.l)
  * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
    STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
    STRING-STREAM and FILE-STREAM, respectively.
  * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
    than previously; in addition, its test suite is now run on build.
    (thanks to Luke Gorrie)
  * a more robust x86-64 disassembler. (thanks to Lutz Euler)
  * optimization: added a immediate representation for single-floats 
    on x86-64
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** MISC.564: defined out-of-line version of %ATAN2 on x86.
    ** attempting to create a package with a colliding nickname causes
       correctable errors to be signalled.
    ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
    ** several x86-64 backend bugs related to sign-extension of immediate
       operands.

changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
  * fixed inspection of specialized arrays. (thanks to Simon Alexander)
  * fixed disassembly of SHLD and SHRD on x86.  (thanks to David
    Lichteblau)
  * fixed loading of multiply forward-referenced layouts. 
    (thanks to Cheuksan Wang)
  * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
    output is now directed to *STANDARD-OUTPUT* as specified by ANSI. 
    (see COMPILE-FILE documentation for details of :PRINT option)
  * fixed bugs 19 and 317: fixed-format floating point printing is
    more accurate.  This also fixes a bug reported by Adam Warner
    related to the ~@F format directive.
  * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
    Hanche-Olsen)
  * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
    dispatch table if the from-char is a dispatch macro character.
  * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
    by the wordsize no longer ignore the last word.  (reported by Lutz
    Euler)
  * fixed bug in type checking of dynamic-extent variables. (reported
    by Svein Ove Aas)
  * optimization: sequence traversal functions use their freedom to
    coerce function designators to functions.
  * optimization: code with many calls to CLOS methods specialized on
    CLOS classes has had redundant type checks removed.
  * optimization: added declarations to speed up operations that access
    the internal character database (for example STRING-UPCASE)
  * optimization: comparison operations between floats and sufficiently small
    fixnums no longer create extra rationals
  * fixed some bugs related to Unicode integration:
    ** portions of multibyte characters at the end of buffers for
       character-based file input are correctly transferred to the
       start of the buffer at the next read.
    ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
       passing it through to OPEN.
    ** LOAD on source files likewise passes any EXTERNAL-FORMAT
       argument given to internal calls to OPEN.
    ** the built-in comment readers (introduced by character sequences
       ";" and "#|") are more forgiving to encoding errors; they will
       STYLE-WARN and then attempt to resync the stream at a character
       boundary.  (thanks to Teemu Kalvas)
  * fixed some bugs in the x86-64 port:
    ** Negative short int return values from c-calls are sign-extended
       correctly.
    ** The stack is aligned to 16-bytes for c-calls, as required by
       the ABI. (thanks to Cheuksan Wang)
    ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
    ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
       for enums. (thanks to Vincent Arkesteijn)
    ** Multiple small optimizations and bugfixes for floating point
       operations.
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** Space, Tab, Linefeed, Return and Page have the invalid
       secondary constituent character trait.
    ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
       syntax.
    ** WITH-INPUT-FROM-STRING should only update the index place on
       normal termination.
    ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
       works more reliably.
    ** Bit-array operations (BIT-AND and similar) worked incorrectly
       with one-dimensional arrays with fill pointers.
    ** TYPE-OF failed on a complex with an integer realpart and a
       RATIO imagpart.
    ** compiler failure during type inference for the code of form
       (IF (EQL X (THE ...)) ...) (MISC.535).

changes in sbcl-0.8.19 relative to sbcl-0.8.18:
  * new port: SBCL now works in native 64-bit mode on x86-64/Linux
    platform. The port supports 61-bit fixnums, large memory spaces
    and reloading shared object files.
  * enhancement: saving cores with foreign code loaded is now
    supported on ppc/Darwin in addition to the previously supported
    platforms.
  * enhancement: the statistical profiler now walks deeper into the 
    call stack for more meaningful call-graphs and accrued time
    reports (x86/x86-64 only). It also now reports time spent in
    foreign functions.
  * enhancement: it is now possible to trace most individual methods
    of a generic function in addition to tracing the generic function
    itself.
  * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
    to MAKE-INSTANCE.
  * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
    SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
  * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
    produces an error.  (thanks to Vincent Arkesteijn)
  * bug fix: NAMESTRING on pathnames with :WILD components in their
    directories works correctly.  (thanks to Artem V. Andreev)
  * fixed bug 125: compiler preserves identity of closures. (reported
    by Gabe Garza)
  * bug fixed: functions with &REST arguments sometimes failed with
    "Undefined function" when compiled with (DEBUG 3). (reported by
    Robert J. Macomber)
  * bug fix: overflow during compiling of setting element of a bit
    vector with constant index and value. (reported by Timmy Douglas)
  * build fix: fixed the dependence on *LOAD-PATHNAME* and
    *COMPILE-FILE-PATHNAME* being absolute pathnames.
  * on x86 compiler partially supports stack allocation of dynamic-extent
    closures.
  * GO and RETURN-FROM do not check the extent of their exit points
    when compiled with SAFETY 0.
  * fixed some bugs related to Unicode integration:
    ** encoding and decoding errors are now much more robustly
       handled; it should now be possible to recover even from invalid
       input or output to the terminal.  (thanks to Teemu Kalvas)
    ** provided a first cut at implementing STRING-TO-OCTETS and
       OCTETS-TO-STRING.  (thanks to Robert J. Macomber)
    ** altered the SB-MD5 contributed module slightly, changing the
       interface just enough for it to be supportable for builds where
       lisp characters are not eight bits.
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** the FORMATTER-generated functions for ~V[ conditionals require
       the correct number of arguments.
    ** READ-FROM-STRING returns the mandated second value when applied
       to displaced strings.
    ** the #\Rubout and #\Backspace characters are treated as invalid
       constituent characters by the tokenizer.

changes in sbcl-0.8.18 relative to sbcl-0.8.17:
  * new feature: reloading changed shared object files with
    LOAD-SHARED-OBJECT now causes the new definitions to take effect.
  * new feature: references to foreign variables and functions
    can now be compiled and loaded before the corresponding shared 
    object file is loaded, as long as the foreign definitions are
    available at runtime.
  * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
    supported.  (thanks to Dan Debertin)
  * SB-ALIEN enums can now be represented in Lisp by any symbols, not
    just keywords.  (thanks to Vincent Arkesteijn)
  * fixed bug #331: structure-class instances corresponding to
    DEFSTRUCT forms are now created eagerly.
  * fixed bug #345: backtraces from calls to undefined functions work
    on x86 as well. Related bug #61 is now also partially fixed on x86
    (backtraces from throws to unknown catch tags.)
  * bug fix: lambda-list parsing is now stricter vrt. order and number
    of lambda-list keywords.
  * bug fix: as specified by AMOP, an error is signalled if a
    class-option appears multiple times in a DEFCLASS form. (reported
    by Bruno Haible)
  * bug fix: RANDOM can be compiled when the compiler derives the type
    of its numeric argument as a disjoint set of small integers.
    (reported by Paul Dietz)
  * bug fix: starting a core saved with shared objects loaded when
    those objects are not available no longer causes threaded SBCL to 
    hang. (reported by Sean Ross)
  * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
    TRACE).
  * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
    argument, which must always be NIL. (reported by Kalle Niemitalo)
  * bug fix: printing 1.0d+23 no longer results in an error.
    (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
  * bug fix: structure slot setters preserve evaluation order. (thanks
    to Gabor Melis)
  * fixed some bugs related to Unicode integration:
    ** RUN-PROGRAM can allow its child to take input from a Lisp
       stream.  (reported by Stefan Scholl)
    ** ASDF-INSTALL successfully downloads .tar.gz files in all
       locales.  (reported by Ken Causey)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** INCF, DECF and REMF evaluate their place form as specified in
       CLtS 5.1.3.
    ** FORMATTER expands ~{ ~} iteration directives with V or #
       parameters correctly.
    ** FORMATTER deals with the ~@[ ~] conditional directive where the
       consequent uses no arguments correctly.
    ** the system has a partial understanding of the (COMPLEX RATIO)
       type specifier.
    ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
       required.

changes in sbcl-0.8.17 relative to sbcl-0.8.16:
  * new feature: a build-time option (controlled by the :SB-UNICODE
    keyword feature, enabled by default) for building the system with
    support for the entire 21-bit character space defined by the
    Unicode consortium.
  * new feature: the system now has rudimentary external-format
    support; the primary user-visible change at this time is that
    characters with the high bit set (such as lower-case-e-acute) will
    print correctly to a terminal in a UTF-8 environment.
  * minor incompatible change: BASE-CHAR no longer names a class;
    however, CHARACTER continues to do so, as required by ANSI.
  * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
    are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
    has been moved to the SB-EXT package (temporarily re-exported from
    SB-DEBUG).
  * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
    are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
    instead. 
  * fixed bug #351: better error-handlind and reporting for malformed
    LET and LET* forms.
  * fixed bug #350: bignum-printing is now more memory-efficient,
    allowing printing of very large bignums, eg. (expt 2 10000000).
    (reported by Bruno Haible)
  * fixed bug #302: better primitive-type selection for intersection
    types.
  * fixed bug #308: non-graphic characters now all have names, as
    required.  (reported by Bruno Haible)
  * bug fix: redefining a class with different superclasses now correctly
    removes it from the direct-subclasses of its previous superclasses.
    (reported by David Morse)
  * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
    new value now works. (reported by Bruno Haible)
  * bug fix: correct canonicalization of multiple non-standard slot
    options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
  * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
    is not yet finalized, as required by AMOP. (reported by Bruno Haible)
  * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
    now exists, an signals an error.
  * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
    during macroexpansion for non-iteration variables as well. (reported 
    by Bruno Haible for CMUCL)
  * bug fix: Cyclic structures and unprintable objects in compiler
    messages no longer cause errors. (reported by Bruno Haible)
  * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
    and READ-FROM-STRING all now return a primary value of NIL if
    *READ-SUPPRESS* is true.  (reported by Bruno Haible for CMUCL)
  * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
    (reported by Bruno Haible for CMUCL)
  * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
    do not designate a condition.  (reported by Bruno Haible for
    CMUCL)
  * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
    SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
    name conflict situations in CLHS 11.1.1.2.5, and provide a restart
    permitting resolution in favour of any of the conflicting symbols.
    (reported by Bruno Haible for CMUCL)
  * bug fix: EQUAL compiler optimizations is less aggressive on
    strings which can potentially compare true despite having distinct
    specialized array element types.
  * bug fix: unit enumerations can be defined without dividing by
    zero.  (thanks to Vincent Arkesteijn)
  * FORMAT compile-time argument count checking has been enhanced.
    (report from Bruno Haible for CMUCL)
  * a partial workaround for the bug 262: the compiler does not try to
    inline-expand a local function doing RETURN-FROM from a deleted
    BLOCK. (thanks to Peter Denno for the bug report and to David
    Wragg for the simple test case)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** RENAME-PACKAGE allows all package designators as new package
       names.
    ** constraint propagation and lambda variable substitution are
       more cautious in dealing with partially deleted code.
    ** compiler sometimes lost reoptimization passes.
    ** CERROR, given a condition as condition designator, passes its
       remaining arguments to the continue format control without
       complaint.
    ** Case-altering FORMAT directives work correctly on non-ASCII
       characters.
    ** The REAL type specifier handles bounds outside the floating
       point ranges without signalling FLOATING-POINT-OVERFLOW.
    ** Functions with IR1-transformations can create intercomponent
       references to global functions.
    ** NIL parameter to the FORMAT directive ~^ means `unsupplied
       parameter'.
    ** FORMAT ~R treats a nil value for its first parameter correctly.

changes in sbcl-0.8.16 relative to sbcl-0.8.15:
  * enhancement: saving cores with foreign code loaded is now
    supported on x86/NetBSD and sparc/Linux in addition to the previously
    supported platforms.
  * bug fix: on some platforms repeated installations caused multiple
    copies of HTML documentation to be installed -- should not happen
    any more. (reported by Stefan Scholl)
  * bug fix: parsing self-recursive alien record types multiple times
    no longer causes infinite recursion. (reported by Thomas F. Burdick,
    original patch by Helmut Eller for CMUCL)
  * bug fix: stack-exhaustion detection works now on NetBSD as well.
    (thanks to Richard Kreuter)
  * bug fix: defining classes whose accessors are methods on existing
    generic functions in other (locked) packages no longer signals
    bogus package lock violations. (reported by François-René Rideau)
  * bug fix: special variables as DEFMETHOD parameters no longer have
    associated bogus type declarations. (reported by David Wragg and
    Bruno Haible)
  * bug fix: read-write consistency on streams of element-type
    (SIGNED-BYTE N) for N > 32.  (reported by Bruno Haible for CMUCL)
  * bug fix: redefiniton of the only method of a generic function with
    no DEFGENERIC no longer emits a full WARNING.  In addition,
    redefinition of generic functions with no DEFGENERIC to an
    incompatible lambda list now signals an error.  (thanks to Zach
    Beane)
  * bug fix: DEFGENERIC now works even when there's a function of the
    same name in an enclosing lexical environment.  (thanks to Zach
    Beane)
  * fixed compiler failure, caused by instrumenting code during
    IR1-optimization.  (Debian bug report #273606 by Gabor Melis)
  * optimization: added loop analysis and improved register allocation
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** POSITION on displaced vectors with non-zero displacement
       returns the right answer.
    ** (SIMPLE-STRING) is a valid type specifier for sequence
       creators.
    ** *PRINT-LEVEL* handling for slotless structures is pedantically
       correct.
    ** PPRINT-INDENT accepts a request for an indentation of any REAL.
    ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
       correct amounts.
    ** The justification version of the FORMAT ~< directive treats
       non-zero minpad parameter correctly.

changes in sbcl-0.8.15 relative to sbcl-0.8.14:
  * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
    SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
    SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
    the supported interface.
  * new feature: Single-stepping of code compiled with DEBUG 2 or
    higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
  * new feature: saving cores with foreign code loaded is now
    supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
    Timothy Moore's work for CMUCL)
  * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
    parameters to * instead of NIL if no initform is supplied.
    (reported by Johan Bockgaard)
  * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
    correctly when FUNCALL appears as the car of the form. Note:
    despite this FUNCALL forms are not currently subject to
    compiler-macro expansion. (port of Raymond Toy's fix for the
    same from CMUCL, reported by Johan Bockgaard)
  * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
    (thanks for Teemu Kalvas)
  * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
    case instead of terminating the iteration (thanks for Julian
    Squires, Sean Champ and Raymond Toy)
  * bug fix: incorrect expansion of defgeneric that caused a style
    warning. (thanks for Zach Beane)
  * bug fix: slot accessor effective method computation works properly
    for classes with multiple non-standard applicable methods on
    SB-MOP:SLOT-VALUE-USING-CLASS.  (reported by Ralf Mattes)
  * on x86 compiler supports stack allocation of results of LIST and
    LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
    CMUCL implementation by Gerd Moellmann)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** FORMAT strings with both the ~<~:;~> form of the justification
       directive and pretty-printing directives cause an error.

changes in sbcl-0.8.14 relative to sbcl-0.8.13:
  * incompatible change: the internal functions
    SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
    logical operations on (UNSIGNED-BYTE 32) values, have been renamed
    to SB-KERNEL:WORD-LOGICAL-FOO.  Modular arithmetic should be used
    instead of the old functions.
  * new feature: on platforms where "dladdr" is available foreign
    function names now appear in backtraces.  (based on Helmut Eller's
    work for CMUCL)
  * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY, 
    SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user 
    manual. 
  * documentation: ASDF manual is now distributed with SBCL.
  * bug fix: SBCL can now load its contributed modules with REQUIRE
    even if the system-provided entries have been removed from
    ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
    detected.
  * bug fix: backtraces involving undefined functions or assembly
    routines are more informative.  (thanks to Brian Downing)
  * bug fix: mutually referent alien structures now work correctly.
    (reported by Rick Taube)
  * bug fix: structures defined by WITH-ALIEN can be referred to
    within other definitions in the same WITH-ALIEN.
  * bug fix: division operators (MOD, TRUNCATE and the like) with
    constant zero divisors and integer dividends no longer generate
    left shifts.
  * bug fix: provide default methods for INPUT-STREAM-P and
    OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
  * bug fix: improve the detection and resolution of MOP metacycles
    (where effective method computation depends on the generic function
    having its effective method computed).  (reported by Bruno Haible)
  * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
    SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
    MAKE-INSTANCE constructors.  (reported by Axel Schairer for cmucl)
  * bug fix: pathnames with relative directory components can be
    represented relative to default pathnames.
  * optimization: in taking the GCD of bignums, reduce the two bignums
    to approximately the same size (using Euclid's algorithm) before
    applying the more sophisticated binary GCD.  (thanks to Juho
    Snellman)
  * optimization: COUNT on bitvectors now operates word-at-a-time.
  * optimization: ASH with a positive, but not necessarily constant,
    (leftwards) shift, when in a modular context, is compiled to a
    hardware shift.
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** FORMAT variable parameters ("~V<char>") are defaulted properly
       if the corresponding argument is NIL.
    ** FORMAT directives accepting a minpad parameter treat negative
       values as 0.
    ** literal commas as character parameters to FORMAT directives are
       parsed correctly.
    ** literal spaces directly after ~<Newline> directives within a
       format-logical-block (~:< ~@:>) do not induce :FILL-style
       conditional newlines.
    ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
       as specified: it no longer includes conditional newlines).
    ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
       *PRINT-ESCAPE*).
    ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
       :PER-LINE-PREFIX argument does not evaluate to a string.

changes in sbcl-0.8.13 relative to sbcl-0.8.12:
  * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
    the manual for details; for now, package locks can be disabled by
    removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
    if no major problems are found then it is likely that they will be
    compiled in unconditionally.
  * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
    unsupported operators on all platforms.  To load a shared library
    into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT.  To load a non-shared
    object file, link it into a shared library outside of SBCL and
    load it using SB-ALIEN:LOAD-SHARED-OBJECT.  (incidentally fixes
    bug #79)
  * bug workaround: ROOM T and the GC were not working together
    reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
    reporting facility aren't preserved. That reporting has been
    disabled, so now until and unless someone figures out how to make it
    work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
  * minor incompatible change: as threatened around sbcl-0.8.0, the
    home package of MOP-related symbols is now SB-MOP, not SB-PCL.
    The symbols are also exported from SB-PCL for backwards
    compatibility, but more so than before SB-PCL should be treated as
    an implementation-internal package.
  * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
    of earlier versions
  * the SB-SPROF contrib now works on (most) non-x86 architectures.
    It is known as of this release not to work on the Alpha, however.
  * fixed bug #167: errors signalled due to illegal syntax in method 
    bodies are now more legible.
  * fixed bug #338: instances of EQL-SPECIFIER are now valid type
    designators and can hence be used with TYPEP.
  * fixed bug #333: CHECK-TYPE now ensures that the type error
    signalled, if any, has the right object to be accessed by
    TYPE-ERROR-DATUM.  (reported by Tony Martinez)
  * fixed bug #340: SETF of VALUES obeys the specification in ANSI
    5.1.2.3 for multiple-value place subforms.  (reported by Kalle
    Olavi Niemetalo)
  * fixed bug #334: programmatic addition of slots using specialized
    methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
    and :ALLOCATION :CLASS slots.
  * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
    float exponents.  (rereported by Peter Seibel)
  * fixed a bug: #\Space (and other whitespace characters) are no
    longer considered to be macro characters in standard syntax by
    GET-MACRO-CHARACTER.
  * fixed bug: initialization of condition class metaobjects no longer
    causes an instance of the condition to be created.  (reported by
    Marco Baringer)
  * fixed bug: it is now possible to have slots such that
    SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
    description is neither :INSTANCE nor :CLASS.
  * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
    user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
    SLOT-VALUE-USING-CLASS), and no longer causes errors with
    non-standard SLOT-DEFINITION-ALLOCATION values.
  * fixed bugs: various race conditions handling exiting threads.  
    CL-PPCRE's thread torture test now passes.
  * fixed arguably-a-bug: GC time proportional to number of threads in 
    system even when most of them are idle
  * optimization: improved performance of BIT and SBIT on bit-vectors.
  * .fasl file incompatibility: The fasl file version number has
    been incremented because of changes associated with package locks.

changes in sbcl-0.8.12 relative to sbcl-0.8.11:
  * minor incompatible change: the system no longer provides
    optimization hints (conditions which are TYPEP
    SB-EXT:COMPILER-NOTE) for conforming code in default compilation
    mode; these hints are still emitted when higher SPEED optimization
    is requested.
  * new contrib module: a sampling profiler (profiling by statistical
    sampling, rather than by instrumenting functions) is available as
    the SB-SPROF contrib.  (thanks to Gerd Moellmann and Juho Snellman)
  * the behaviour of the standard function ED is now customizeable by
    third parties through a hook variable: see ED's documentation
    string for information on the protocol.
  * the compiler no longer emits efficiency notes for (FUNCALL X)
    when the type of X is uncertain under default optimization
    settings.
  * fixed bug 276: mutating a binding of a specialized parameter to a
    method to something that is not TYPEP the specializer is now
    possible.
  * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
    exhibit the required behaviour on the x86 platform.  (thanks to
    Peter van Eynde, Eric Marsden and Bruno Haible)
  * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
    even for difficult arguments.  (reported by Peter Graves)
  * fixed bug 141a: the backquote printer now descends quoted
    structure.
  * fixed another bug in backquote printing: no more destructive
    modification of the form's list structure.  (reported by Brian
    Downing)
  * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
    move between its address being taken and the call to
    interrupt_thread, fixing a crashing race condition.
  * the SB-POSIX contrib implementation has been adjusted so that it
    no longer exhibits ridiculously poor performance when constructing
    instances corresponding to C structs.

changes in sbcl-0.8.11 relative to sbcl-0.8.10:
  * minor incompatible change: the sb-grovel contrib now treats C
    structures as alien (in the sense of SB-ALIEN) objects rather than
    as undistinguished (simple-array (unsigned-byte 8) (*))s.  This
    has implications for memory management of client code
    (sb-grovel-returned objects must now be manually managed) and for
    type safety (alien objects now have full types).
  * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
    used to control emission of compiler diagnostics, rather than the
    SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality.  See the manual for
    documentation on this feature.  The SB-EXT:INHIBIT-WARNINGS
    quality should be considered deprecated.
  * install.sh now installs the user manual as well
  * (not quite a new documentable feature, but worth considering in
    the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
    beginnings of a semantically meaningful condition hierarchy is
    under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
    IDEs.
  * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
    designator as the defaults argument.
  * fixed bug: Displaced arrays whose displaced-to array has become
    too small now cause ARRAY-DIMENSION to signal an error, providing
    for safer bounds-checking. (reported by Bruno Haible)
  * fixed bug: DEFCLASS slot definitions with identical :READER and
    :WRITER names now signal a reasonable error. (reported by Thomas
    Burdick)
  * fixed bug: CLOSE :ABORT T on appending stream no longer causes
    file deletion.
  * fixed bug: Invalid slot specification errors now print correctly. 
    (thanks to Zach Beane)
  * fixed bug 320: Shared to local slot value transfers in class 
    redefinitions now happen corrently from superclasses as
    well. (reported by Bruno Haible)
  * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
    Haible)
  * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
    improper lists correctly. (reported by Bruno Haible)
  * fixed bug 313: source-transform for <fun-name> was erroneously
    applied to a call of a value of a variable with name <fun-name>.
    (reported by Antonio Menezes Leitao)
  * fixed bug 307: The obsolete instance protocol ensures that
    subclasses are properly obsoleted.  (thanks to Nikodemus Siivola)
  * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
    remove dead unknown-values globs from the middle of the stack.
  * added a new restart to *BREAK-ON-SIGNALS* handling to make it
    easier to resume long computations after using *BREAK-ON-SIGNALS*
    to diagnose and fix failures.  (thanks to Nikodemus Siivola)
  * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
    CLISP test suite): WRITE-TO-STRING is not constant-foldable.
  * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
    is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
    MEMBER and UNION types have likewise been fixed.  (thanks to Bruno
    Haible)
  * fixed a (fairly theoretical) bug in string printing: if
    *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
    does not have array-element-type equal to the most general string
    type.
  * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
    function name. (thanks to Nikodemus Siivola)
  * fixed bug: compile-time format string checker failed on
    non-closed ~{.  (reported by Thomas F Burdick)
  * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
    don't warn on legal loop code involving more than one aggregate
    boolean.  (thanks to Nikodemus Siivola)
  * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
    in keyword position to MAKE-PACKAGE was not being reported as
    invalid.
  * fixed bug: as reported by Juan Ripoll on cmucl-imp,
    MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
    keywords.
  * fixed bugs 280 and 312: the checking for multiple definitions in a
    file is less likely to become confused by uses of inline
    functions.
  * fixed bug: the #S reader macro performs the keyword coercion
    specified for slot names.  (reported by Kalle Niemitalo)
  * fixed bug: lambda lists may contain symbols whose names start with
    & but are not lambda-list-keywords; their occurrence triggers a
    STYLE-WARNING.
  * fixed bug 321: define-method-combination argument lambda lists do
    not cause invalid code to be generated when &optional or &aux
    variables have default values.  (reported by Bruno Haible)
  * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
    have CLOS classes; this appears also to have fixed reported
    instability in stack exhaustion detection.
  * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
    MUFFLE-WARNING when no associated restart is present is now
    printable.
  * optimization: rearranged the expansion of various defining macros
    so that each expands into only one top-level form in a
    :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
    approximately 10%.
  * optimization: used a previously little-used slot in symbols to
    cache SXHASH values, yielding a 5-10% compiler speedup.  (thanks
    to Juho Snellman)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** MAKE-INSTANCES-OBSOLETE returns the class name when called with 
       a symbol.
    ** Fixed an optimization bug related to inheritance of initforms
       from local to shared slots.
    ** FILE-POSITION works as specified on BROADCAST-STREAMs.
    ** CAST optimizer forgot to flush argument derived type.
    ** print/read consistency on floats is now orders of magnitude
       more likely.  (thanks also to Bruno Haible for a similar report
       and discussions)
    ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
    ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
       for objects of type REAL.  Make it so.
    ** SXHASH is specified (infelicitously) to respect similarity,
       which means that (SXHASH 0.0) must equal (SXHASH -0.0).  Make
       it so.  (thanks to Markus Ziegler)
    ** on the Alpha, the compiler succeeds in compiling functions
       returning a known number of arguments greater than 63.
    ** fixed handling of invalid NIL arguments in keyword position
       in local calls.
    ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
       true.
    ** provided a readably-printable representation for RANDOM-STATE
       objects.
    ** ensured that pathnames, where they have a namestring, always
       print using #P"..." syntax.

changes in sbcl-0.8.10 relative to sbcl-0.8.9:
  * Support for the forthcoming 2.0 version of the NetBSD kernel
    running on x86 hardware has been added.  (thanks to Perry
    E. Metzger most immediately, and others for their past work)
  * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
    current release version; SBCL's OpenBSD support had been broken 
    since about the time of OpenBSD's switch to ELF binary format.)
  * [placeholder for DX summary]
    ** user code with &REST lists declared dynamic-extent, under high
       speed or space and low safety and debug optimization policy.
  * The manual has been converted to Texinfo format and the debugger
    chapter from the cmucl manual has been added.
  * A facility has been added to extract documentation strings from
    sbcl and store them as Texinfo-formatted snippets for inclusion in
    the manual (via Texinfo's @include directive)
  * bug fix: compiler emitted division in optimized DEREF.  (thanks for
    the test case to Dave Roberts)
  * bug fix: multidimensional simple arrays loaded from FASLs had fill
    pointers.  (reported by Sean Ross)
  * bug fix: PROFILE output is printed nicely even for large numerical
    values.  (thanks to Zach Beane)
  * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
    greater than 32 handle EOF correctly.
  * bug fix: on X86 an immediate argument of the IMUL instruction is
    correctly printed in disassembly. (thanks to Lutz Euler)
  * bug fix: class slots in redefined classes preserve their old
    values.  (thanks to Bruno Haible and Nikodemus Siivola)
  * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
    succeeds.  (reported by Marco Baringer)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** READ-SEQUENCE now works on ECHO-STREAMs.
    ** RATIONALIZE works more according to its specification.  (thanks
       to Bruno Haible)

changes in sbcl-0.8.9 relative to sbcl-0.8.8:
  * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
    *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
    general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
    should matter to you only if you rebind the printer control
    variables and then find you want different bindings in the 
    debugger than in the ordinary execution of your program.)
  * The runtime build system has been tweaked to support building
    (on SPARC/SunOS) using a C compiler which invokes Sun's own
    assembler and linker.  (thanks to Nikodemus Siivola)
  * Unbound, undefined, undeclared variables now trigger full
    WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
    more often programmer error than deliberate exploitation of undefined
    behaviour.
  * optimization: the hash algorithm for strings has changed to one
    that is less vulnerable to spurious collisions.  (thanks to Juho
    Snellman)
  * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
    needless bounds checking.  (thanks to Juho Snellman)
  * optimization: implemented multiplication as a modular
    (UNSIGNED-BYTE 32) operation on the PPC backend.
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** ADJUST-ARRAY now copies the datum in a zero rank array if
       required.
    ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
    ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
    ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
       element-type NIL.
    ** CLEAR-INPUT accepts NIL and T for its (optional) stream
       argument.
    ** Ratios can now be printed correctly with *PRINT-BASE* bound to
       2, 8 or 16.
    ** ECHO-STREAMs no longer attempt to echo the end of file value to
       their output stream on EOF from read.
    ** CONCATENATED-STREAM-STREAMS discards constituent streams which
       have been read to end-of-file.
    ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
       STRING-STREAMS.
    ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
       description of determination of which consecutive characters
       constitute a word.
    ** Printing the "Space" character with escaping on now yields "#\\ ", 
       rather than "#\\Space", as mandated by ANSI 22.1.3.2.
    ** Reading floating-point numbers with *READ-BASE* set to a number
       less than 10 works correctly.
    ** Reading floating-point numbers with *READ-BASE* set to a number
       more than 10 works correctly.
    ** Printing with *PRINT-READABLY* targets the standard readtable, not
       the readtable currently in effect.

changes in sbcl-0.8.8 relative to sbcl-0.8.7:
  * minor incompatible change: parsing of namestrings on a physical
    (Unix) host has changed; numbers after the final #\. in a
    namestring are no longer interpreted as a version field.  This is
    intented to be largely invisible to the user, except that the
    meaning of the namestring "*.*.*" has changed: it now refers to a
    pathname with :TYPE :WILD :NAME #<pattern "*.*">.  This namestring
    should usually be replaced by 
      (make-pathname :name :wild :type :wild :version :wild)
    with the added benefit that this is more likely to be portable.
    As a consequence of this change, the :IF-EXISTS :NEW-VERSION
    option to OPEN now signals an error if the file being opened
    exists; this may have an impact on existing code.
  * fixed bug 190: RUN-PROGRAM should now work properly, respecting
    signals received, on the PowerPC platforms (both Linux and
    Darwin).  (thanks to Pierre Mai for pointing out the location of
    the error)
  * several fixes on OS X: The system now builds and runs cleanly on
    Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
    (thanks to Brian Mastenbrook)
  * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
    second-resolution: integer multiples of 1/3600 between -24 and 24.
    (thanks to Vincent Arkesteijn)
  * bug fix: functions =, /=, <, <=, >, >= did not check the argument
    type when called with 1 argument; PEEK-CHAR checked type of
    PEEK-TYPE only after having read first character from a
    stream.  (reported by Peter Graves)
  * bug fix: the garbage collector now has much better locality
    behaviour, and in particular no longer treats all memory as being
    exhausted when many small objects point to each other in a deeply
    recursive manner.
  * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
    recognized as being TYPEP their class.
  * bug fix: the PUSHNEW documentation string has been corrected.
    (thanks to Vincent Arkesteijn)
  * bug fix: defaulting of the value for the last of an atypically
    large number of multiple values being bound was not being
    performed correctly on the Alpha or PPC platforms
  * optimization: implemented multiplication as a modular
    (UNSIGNED-BYTE 32) operation on the x86 backend.
  * optimization: SEARCH on simple-base-strings can now be open-coded.
    (see also contrib/compiler-extras.lisp for inspiration for
    teaching the compiler about the Boyer-Moore algorithm).
  * value, returned by MAX (and MIN) called with several EQUALP, but
    not EQL, arguments now does not depend on compiler settings.
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** in stack analysis liveness information is propagated from
       non-local entry points.
    ** pathwise CAST removing failed when the CAST node did not start
       a block.
    ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
       OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
       stream.
    ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
       host is already defined.
    ** RENAME-FILE works on streams instead of signalling an internal
       type error.
    ** PEEK-CHAR uses the current readtable when determining whether
       or not a character is whitespace.
    ** MERGE-PATHNAMES handles the case when the pathname does not
       specify a name while the default-pathname specifies a version.
    ** Pathnames now stand a better chance of respecting print/read
       consistency.
    ** Attempting to use standardized file system operators with a
       pathname with invalid :DIRECTORY components signals a
       FILE-ERROR.
    ** OPEN :DIRECTION :IO no longer fails to work on non-existent
       files.
    ** DIRECTORY on logical pathnames is more correct.
    ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
       signal a TYPE-ERROR if their argument is not a stream.
    ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
       designator argument does not designate a stream.
    ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
       examining the synonym.
    ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
       work as specified.
    ** OPEN and WITH-OPEN-STREAM allow opening streams with
       element-type larger than ([UN]SIGNED-BYTE 32).

changes in sbcl-0.8.7 relative to sbcl-0.8.6:
  * When built with the :SB-FUTEX feature, threaded builds now take 
    advantage of the "fast userspace mutex" facility in Linux kernel 2.6
    for faster/more reliable mutex and condition variable support.
  * Incompatible change (but one you probably shouldn't have been using
    anyway): the interface and code for arbitrating between multiple
    threads in the same user session has been redesigned.
  * bug fix: GET-SETF-EXPANSION no longer throws an internal type
    error when called without an explicit environment argument.
    (thanks to Nikodemus Siivola)
  * bug fix: buffered :DIRECTION :IO streams are less likely to become
    confused about their position.  (thanks to Adam Warner and Gerd 
    Moellmann)
  * bug fix: Pretty printing backquoted forms with unquotations in the
    argument list position of various code constructs such as LAMBDA
    now works correctly.  (reported by Paul Dietz)
  * bug fix: Pretty printing unquotations no longer loses all
    stream position information.
  * optimization: performance of string output streams is now less
    poor for multiple small sequence writes.
  * optimization: performance of CSUBTYPEP in the presence of complex
    expressions involving CONS and NOT many times has been improved.
    (reported by Paul Dietz)
  * ASDF-INSTALL bug fix: now parses *PROXY* properly.  (thanks to
    Sean Ross)
  * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
    streams for the REPL, for the debugger, and so on.  (thanks to
    David Licteblau)
  * DEFINE-CODITION is more efficient.  (thanks to Brian Mastenbrook)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
       is ignored if it is too small, rather than propagating through
       to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
    ** extremely complex negations of CONS types were not being
       sufficiently canonicalized, leading to inconsistencies in
       SUBTYPEP.
    ** VALUES tranformer lost derived type.

changes in sbcl-0.8.6 relative to sbcl-0.8.5:
  * fixed a bootstrapping bug: the build process no longer assumes
    that the various BOOLE-related constants have the same value in
    host and target lisps.  (noted by Paul Dietz' test suite on an
    SBCL binary built from CLISP)
  * The system can now be dynamically linked on the MIPS platform,
    which enables dynamic loading of foreign code from Lisp.  (thanks
    to Ralf Baechle for discussions on the MIPS ABI)
  * The system now records debugging information for its own source
    files in a filesystem-position-independent manner, relative to 
    the "SYS" logical host.
  * fixed a compiler bug: MV-LET convertion did not check references
    to the "max args" entry point.  (reported by Brian Downing)
  * tweaked disassembly notes to be less confident about proclaiming
    some instruction as an LRA.  (thanks to Brian Downing)
  * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
    now each have their own history, command character, and other
    characteristics.  (thanks to David Lichteblau)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** compiler failure in compiling LOGAND expressions including a
       constant 0
    ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
       shift greater than 32.
    ** FLUSH-DEST did not mark blocks for type check regeneration.
    ** HANDLER-CASE failed to accept declarations in handler clauses
       in some circumstances.

changes in sbcl-0.8.5 relative to sbcl-0.8.4:
  * New code in contrib/sb-introspect (still probably not entirely
    stable yet) provides some support for smart Lisp development 
    environments like SLIME.
  * The conditions signalled for errors occurring when loading .fasl
    files have been systematized (inheriting from SB-EXT:INVALID-FASL)
    in a way which should help ASDF recover gracefully.
  * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS* 
    stuff has been cleaned up. If you code contrib/ stuff, this might 
    affect you, and you can look at contrib/README, contrib/STANDARDS, 
    and/or the 0.8.4.27 diff to check.
  * In full calls the compiler now does not generate checks for declared
    argument types for all arguments.
  * various threading fixes
    ** and some experimental patches which didn't make it into 
       the main tree for this release, but which are shipped in 
       contrib/experimental-thread.patch as a possible fix for some
       failures (deadlock, spinning...) in GC-intensive multithreaded
       applications.
  * fixed PPC build problem (source code incompatibility of different 
    library versions): added offsetof() hackery which attempts to divine 
    where glibc maintainers put uc_mcontext today 
  * fixed bug 282: compiler does not trust type assertions while passing
    arguments to a full call.
  * fixed bug 261: compiler allows NIL or "no value" to be accepted for
    &OPTIONAL VALUES type parameter.
  * fix bug 214: algorithm for noting rejected templates is now more
    similar to that of template seletion. (also reported by rydis on
    #lisp)
  * fixed bug 141b: printing backquoted information readably and prettily
    inserts a space where necessary.
  * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
    (CHARACTER 10) are now reported as errors, rather than propagated
    as unknown types.  (reported by piso on #lisp)
  * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
    with values NIL and :ERROR.  (thanks to Milan Zamazal)
  * fixed bug 191c: CLOS now does proper keyword argument checking as
    described in CLHS 7.6.5 and 7.6.5.1.
  * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
    longer raise an error; further, using a list as a for-as-arithmetic
    counter now raises a meaningful error.
  * fixed bug 213a: even fairly unreasonable CONS type specifiers are
    now understood by sequence creation functions such as MAKE-SEQUENCE
    and COERCE.
  * fixed bug 46k: READ-BYTE now signals an error when asked to read from
    a STRING-INPUT-STREAM.
  * compiler enhancement: SIGNUM is now better able to derive the type
    of its result.
  * type declarations inside WITH-SLOTS are checked.  (reported by
    salex on #lisp)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** incorrect optimization of TRUNCATE for a positive first
       argument and negative second.
    ** compiler failure in let-convertion during flushing dead code.
    ** compiler failure while deriving type of TRUNCATE on an
       interval, containing 0.
    ** ASH of a negative bignum by a negative bignum count now returns
       -1, not 0.
    ** intersection of CONS types now canonicalizes properly, fixing
       inconsistencies in SUBTYPEP.

changes in sbcl-0.8.4 relative to sbcl-0.8.3:
  * incompatible change: The --disable-debugger command line
    option now clobbers the debugger at a more fundamental
    level, by redefining #'INVOKE-DEBUGGER instead of by 
    rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
    is specified by ANSI to ignore *DEBUGGER-HOOK* and
    INVOKE-DEBUGGER regardless. Under the old system, BREAK would
    enter the debugger REPL and then suffer recursive errors
    because *DEBUG-IO* is also messed up in --disable-debugger mode;
    while under the new system, BREAK in --disable-debugger mode
    terminates the system just as an unhandled error would.
  * fixed compiler performance when processing loops with a step >1;
  * bug fix: DOCUMENTATION now retrieves generic function
    documentation.  Also, DOCUMENTATION and (SETF DOCUMENTATION)
    support has been systematized, and now supports the methods
    specified by ANSI, along with a default method and a method for
    slot documentation.  (reported by Nathan Froyd)
  * bug fix: effective methods associated with a generic function are
    no longer cached over a change of that generic function's method
    combination.  (reported by Andreas Fuchs)
  * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
    names a non-existent directory.  (thanks to Andreas Fuchs)
  * bug fix: ROUND and TRUNCATE could, under certain circumstances on
    the PPC platform, lead to stack corruption; this has been fixed.
    (reported by Rainer Joswig)
  * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
    no longer ever returns 1 instead of 0.  (thanks to Lars Brinkhoff)
  * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
    arithmetic in general, is now much more reliable on the PPC
    platform.
  * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha 
    platform now returns the right answer.
  * optimization: restored some effective method precomputation in
    CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
    precomputation is now tunable.
  * optimization: compiler-internal data structure use has been
    reviewed, and changes have been made that should improve the
    performance of the compiler by about 20%.
  * optimization: performance of FILL (and :INITIAL-ELEMENT) on
    simple-base-strings and simple-bit-vectors is improved.
  * optimization: the optimization of 32-bit logical and arithmetic
    functions introduced in version 0.8.3 on the x86 has been
    implemented on the mips, ppc and sparc platforms; an
    implementation of the same facility, but for 64-bit arithmetic,
    has been added for the alpha.
  * microoptimization: the compiler is better able to make use of the
    x86 LEA instruction for multiplication by constants.
  * bug fix: in some situations compiler did not report usage of
    generic arithmetic in (SPEED 3) policy.
  * bug 145b fix: compiler used wrong type specifier while converting
    MEMBER-types to numeric.
  * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
    McNaught)
  * bug fix: (SETF AREF) on byte-sized-element arrays with constant 
    index argument now works properly on the Alpha platform.
  * bug fix: floating point exception treatment on the Alpha platform
    is improved.
  * bug fix: FILE-POSITION works much better on string input and 
    output streams.  (thanks to Nikodemus Siivola)
  * bug fix: many threading/garbage collection symptoms sorted.  
    SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that 
    might be pseudo-atomic.
  * internal change: Stopping for GC is now done with signals not 
    ptrace. GC is now done in whichever thread wanted it, instead of
    in the parent. 
  * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers 
    work again.
  * bug fix: result form in DO is not contained in the implicit
    TAGBODY.
  * incompatible change: ICR structure is changed; the value part of
    CONTINUATION is now called LVAR; corresponding functions are
    renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
  * added type deriver for ISQRT (thanks to Robert E. Brown).
  * bug fix: better support for loading from the command line when an
    initialization file sets (READTABLE-CASE *READTABLE*).  (thanks 
    to Adam Warner)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
    ** ROUND and FROUND now give the right answer when given very
       small float arguments.
    ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
       circumstances.
    ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
    ** (GCD 0 <negative-integer>) returned <negative-integer>.
    ** LCM should return a non-negative integer.
    ** PARSE-INTEGER returned the index of a terminator instead of the
       upper bounding index of a substring in case :JUNK-ALLOWED NIL.
    ** PARSE-INTEGER returned an incorrect index being applied to a
       displaced string.
    ** LCM with two arguments of 0 returns 0 rather than signalling
       DIVISION-BY-ZERO.
    ** unsigned addition of a 32-bit constant with the high bit set no
       longer causes an internal compiler error on the x86.
    ** LOGBITP accepts a non-negative bignum as its INDEX argument.
    ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
       with negative last argument.
    ** byte specifiers with zero size and position no longer cause
       an error during type derivation.
    ** bignum multiplication on the Alpha platform now returns the
       right answer.
  * porting: The system now builds on SuSE AMD64, although it still
    generates a 32-bit binary.
  * .fasl file incompatibility: The fasl file version number has
    been incremented (because of the changes to internal compiler
    data structures referred to above).

changes in sbcl-0.8.3 relative to sbcl-0.8.2:
  * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
    more accurately, on the Darwin kernel running on PowerPC hardware.
    (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
  * Compiler code deletion notes now signal a condition of type
    SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
    an associated MUFFLE-WARNING restart.
  * The compiler now performs limited argument count validation of
    constant format strings in FORMAT, and where appropriate in ERROR, 
    CERROR and WARN.  (thanks to Gerd Moellmann)
  * New ASDF-INSTALL contrib can be used for automatic download and 
    installation of third-party Lisp code from CCLAN or other sites
    via CLiki.
  * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
    which forces another thread to execute a function supplied by the
    caller.
  * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
    now accept and act upon their :ELEMENT-TYPE keyword argument.
    (reported by Martin Atzmueller, Edi Weitz)
  * bug fix: FILE-POSITION now accepts position designators up to
    ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
    is the greater.  (thanks to Patrik Nordebo)
  * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
    Kalle Olavi Niemitalo)
  * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
    printable.  (reported by Eric Marsden)
  * bug fix in sb-posix: mmap() now works on systems with a 64-bit
    off_t, including Darwin and FreeBSD.  (thanks to Andreas Fuchs)
  * x86 bug fix in control stack exhaustion checking: now shows backtrace
  * bug fix in WITH-TIMEOUT: now the body can have more than one form.
    (thanks to Stig Sandoe)
  * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
    UNREAD-CHAR now correctly includes the unread character in the
    target sequence.  (thanks to Gerd Moellmann)
  * bug fix in threaded builds: the system can now be suspended and
    resumed by shell job control with minimal disruption.
  * bug fixes in times and timezones >2038AD
  * better handling of "where is GNU make?" problem in build scripts
    (thanks to Nikodemus Siivola)
  * new optimization: inside a named function any reference to a
    function with the same name is considered to be a self-reference;
    this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
    optimization quality.
  * new optimization on x86: logical functions and + now have
    optimized (UNSIGNED-BYTE 32) versions, which are automatically
    used when the result is truncated to 32 bits.
  * VALUES declaration is partially enabled.
  * fixes in SB-GROVEL (thanks to Andreas Fuchs)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** The system now obeys the constraint imposed by
       UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
       types form a lattice under type intersection.
    ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
    ** ASSOC now ignores NIL elements in an alist.
    ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
       and (1+ MOST-POSITIVE-FIXNUM) answers.
    ** The addition of a method with invalid qualifiers to a generic
       function does not cause an error to be signalled immediately;
       a warning is signalled, and the error is generated only on
       calling the generic function.
  * changed .fasl file version number, in part to add type codes for
    new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by 
    obscure ANSI requirements

changes in sbcl-0.8.2 relative to sbcl-0.8.1:
  * fixed bug 148: failure to inline-expand a local function left
    garbage, confusing the compiler.
  * fixed bugs 3cd: structure slot readers perform type check if the
    slot can have an invalid value (i.e. it is either not initialized
    or can be written with a less specific slot writer).
  * bug fix: the compiler now traps array references to elements off
    the end of an array; previously, the bounds checking in some
    circumstances could go off-by-one.
  * improved MACHINE-VERSION, especially on Linux (thanks to Lars
    Brinkhoff)
  * type declarations for array element types now obey the description
    on the CLHS page "Declaration TYPE", as per discussions on
    sbcl-help around 2003-05-08.  This means that a declaration 
    (TYPE (ARRAY FOO) BAR) means that, within the scope of the
    declaration, all references to BAR will be asserted or assumed
    (with THE, so dependent on compiler policy) to involve objects of
    type FOO.  Note that no such declaration is implied in 
    (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
  * declared types of functions from the "Conditions"
    chapter. (reported by Paul Dietz)
  * bug fix: CERROR accepts a function as its first argument.
  * bug fix: NTH an NTHCDR accept a bignum as index
    arguments. (reported by Adam Warner)
  * optimization: character compare routines now optimize comparing
    against a constant character. (reported by Gilbert Baumann)
  * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
    argument now works properly on the MIPS platform.
  * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
  * repeated evaluation of the same DEFSTRUCT, a slot of which is
    declared to have a functional type, does not cause an error
    anymore.
  * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
    many :DEFAULT-INITARGS. (reported by Istvan Marko)
  * fixed bug: if last continuation of a deleted block has a
    destination, this destination should be deleted too. (reported by
    ohler on #lisp)
  * fixed a bug in the bootstrap process: the host compiler's values
    of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
    into the newly-built SBCL. (reported by Eric Marsden on #lisp,
    test case from Patrik Nordebo)
  * improved the ability of the disassembler on the PPC platform to
    provide helpful disassembly notes.
  * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
    the class in more cases than previously.
  * bug fix: FILE-POSITION now understands :START and :END for
    STRING-INPUT-STREAMs.  (thanks to Nikodemus Siivola)
  * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
    8) data.  (thanks to David Lichteblau)
  * bug fix: it is possible to add a method to a generic function
    without lambda list.
  * bug fix: reader failed to signal END-OF-FILE inside an
    object representation. (reported by Nikodemus Siivola)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** LAST and [N]BUTLAST should accept a bignum.
    ** condition slot accessors are methods.
    ** (VECTOR NIL) is a subtype of STRING.

changes in sbcl-0.8.1 relative to sbcl-0.8.0:
  * minor incompatible change: some nonsensical specialized lambda
    lists (used in DEFMETHOD) which were previously ignored now signal
    errors.
  * minor incompatible change: the system is now aware of the types of
    variables in the COMMON-LISP package, and will signal errors for
    most violations of these type constraints (where previously they
    were silently accepted).
  * minor incompatible change: COMPILE-FILE now uses the freedom
    afforded (ANSI 3.2.2.3) to use derived function types for
    functions defined in the same file. This also permits the system
    to warn on static type mismatches and function
    redefinition.  (Currently it does not work with high DEBUG level.)
  * minor incompatible change: VALUES declaration is disabled.
  * When issuing notes, the compiler now signals a condition of type
    SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
    restart for use in user handlers.  It is expected that the
    COMPILER-NOTE condition will eventually become a condition
    supertype to a hierarchy of note types, which will then be
    handleable in a similar fashion. However, at the moment, no such
    note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
    but it's an implementation detail, not a classification for the
    purpose above.)
  * Changes in type checking closed the following bugs:
    ** type checking of unused values (192b, 194d, 203);
    ** template selection based on unsafe type assertions (192c, 236);
    ** type checking in branches (194bc).
  * A short form of VALUES type specifier has ANSI meaning (it has
    increased the number of situations when SBCL cannot perform type
    checking).
  * fixed bug in DEFSTRUCT: once again, naming structure slots with
    keywords or constants is permissible.
  * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
    defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
    classes.  (thanks to Antonio Martinez)
  * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
    outputs FRESH-LINE or TERPRI, and no longer converts its stream
    argument to a pretty-print stream. Instead, it leaves any such 
    operations to DESCRIBE-OBJECT methods.
  * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
    by Teemu Kalvas)
  * bug fix: NIL is now a valid destructuring argument in DEFMACRO
    lambda lists. (thanks to David Lichteblau)
  * bug fix: Defining a generic function with a :METHOD-CLASS being a
    subclass of STANDARD-METHOD no longer causes stack exhaustion.
    (thanks to Gerd Moellmann)
  * fixed bug 246: increased compilation speed of long
    MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
    integer argument)
  * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
    has been included.
  * DEFCONSTANT now throws a condition of type
    SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
    constant to a non-EQL value; CONTINUE and ABORT restarts
    respectively change and preserve the value. 
  * fixed bug 63: The code walker, part of the implementation of CLOS,
    is now better at handling symbol macros.
  * bug fix: There is no longer an internal implementation type named
    CL:LENGTH. (reported by Raymond Toy)
  * bug fix: In macro-like defining macros/special operators the
    implicit block does not enclose the lambda list.
  * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
    enclosed in parentheses) are not suitable as type specifiers, and
    their use properly signals an error now.
  * bug fix: An argument count mismatch for a type specifier in code
    being compiled no longer causes an unhandled error at compile
    time, but signals a compile-time warning.
  * fixed simple vector readable printing
  * bug fix: DESCRIBE takes more care over whether the class
    precedence list slot of a class is bound before accessing it.
    (reported by Markus Krummenacker)
  * bug fix: FORMATTER can successfully compile pretty-printer format
    strings which use variants of the ~* directive inside.
  * bug fix: SEARCH now applies its TEST predicate to the elements of
    the arguments in the correct order. (thanks to Wolfhard Buss)
  * fixed bug 235b: The compiler uses return types of MAPCAR and friends
    in type inference. (thanks to Robert E. Brown)
  * bug fix: Reading in symbols with an explicit package name of ""
    (e.g. '||::FOO) now works correctly.  (reported by Henrik Motakef)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** NIL is now allowed as a structure slot name.
    ** Arbitrary numbers, not just REALs, are allowed in certain
       circumstances in LOOP for-as-arithmetic clauses.
    ** Multiple class redefinitions before slot access no longer
       causes a type error.
    ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
       association between the name and a class.
    ** Generic functions with non-standard method-combination and over
       five methods all of which return constants no longer return NIL
       after the first few invocations. (thanks to Gerd Moellmann)
    ** CALL-NEXT-METHOD with no arguments now passes the original
       values of the arguments, even in the presence of assignment.
    ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
       object.
    ** DISASSEMBLE works with closures and funcallable instances.
    ** ADD-METHOD now returns the generic function, not the new method.
    ** FIND-METHOD signals an error if the lengths of the specializers
       is incompatible with the generic function, even if the ERRORP
       argument is true.
    ** TYPE-OF returns recognizeable subtypes of all built-in-types of
       which its argument is a member.
    ** DEFCLASS only redefines the class named by its class-name
       argument if that name is the proper name of the class;
       otherwise, it creates a new class.
    ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
       of the UNBOUND-SLOT condition to the name of the slot.
    ** (SETF (AREF bv 0) ...) did not work for bit vectors.
    ** SLOT-UNBOUND and SLOT-MISSING now have their return values
       treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
       SLOT-MAKUNBOUND in the specified fashion.

changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
  * SBCL now builds using CLISP (version of late April 2003 from CVS) as
    cross-compilation host. As a consequence, we can now bootstrap our
    way up to SBCL starting with a bare gcc toolchain and human-readable
    source code (first the source to CLISP, then the source to SBCL).
  * A contributed module containing a partial implementation of the
    simple-streams interface has been included.  (thanks to Rudi
    Schlatte)
  * A contributed module implementing the RFC1321 Message Digest
    Algorithm, known as MD5, has been included.
  * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
    no longer has any effect, as the code controlled by this feature
    has been deleted.  (As far as we know, no-one has ever built using
    this feature, and its semantics were confused in any case).
  * minor incompatible change: As a consequence of making SLOT-EXISTS-P
    work on conditions (as required by the ANSI specification),
    SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
    expected behaviour on conditions.  Users should note, however,
    that such behaviour is not required by the ANSI specification,
    and so use of this behaviour may render their code unportable.
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
       types.
    ** &ENVIRONMENT parameter in macro lambda list is bound first.
    ** SXHASH on condition objects no longer returns NIL.
    ** :ALLOCATION :CLASS slots are better treated; their values are
       updated on class redefinition, and initforms inherited from
       superclasses are applied.
    ** REMOVE-METHOD returns its generic function argument even when
       no method was removed.
    ** SHARED-INITIALIZE now initializes the values of the requested
       slots, including those with :ALLOCATION :CLASS.
    ** ALLOCATE-INSTANCE now works on structure classes defined via
       DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
       STRUCTURE-CLASS).
    ** SLOT-EXISTS-P now works on conditions, as well as structures
       and CLOS instances.
    ** MAKE-LOAD-FORM now has the required methods on
       STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
    ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
       keyword, and now implements the SLOT-NAMES argument.
    ** methods with &OPTIONAL arguments no longer allow too many
       arguments to be passed in the call without error.
    ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
       option is consistent with the required arguments of the generic
       function lambda list.
  * bug fix: REQUIRE accepts a string designator. (Thanks to 
    Antonio Martinez.)
  * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
    SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
    specified-by-AMOP lambda list of (CLASS &REST INITARGS).
  * bug fix: The compiler now checks for duplicated variables in macro
    lambda lists.
  * bug fix: SETQ on globals returns the correct value.
  * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
    not a COMPILER-ERROR (followed by some other strange error on
    choosing the CONTINUE restart).
  * bug fix: make.sh and friends are now more consistent in the way that
    they look for GNU "make".

changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
  * experimental native threads support (on x86 Linux >=2.4 only).
    This is not compiled in by default: you need to add :SB-THREAD to
    the target features.  See the "Beyond ANSI" chapter of the manual
    for details.
  * fix for longstanding nonANSIism: The old distinction between
    CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
    The return value from CL:FIND-CLASS is now a CLOS class, and
    likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
    CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
  * An interface to the MetaObject Protocol, as described in Kiczales,
    des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
    MIT Press, 1991, is available from the SB-MOP package.
  * incompatible change: the SB-PCL package should now be considered
    a private implementation detail, and no longer a semi-private MOP
    interface.
  * minor incompatible change: due to rearrangement for threads, the
    control stack and binding stack are now allocated at arbitrary
    addresses instead of being hardcoded per-port.  Users affected by
    this probably have to be doing advanced things with shared
    libraries, and will know who they are.
  * minor incompatible change: Previously, all --eval forms used were
    processed with READ before any of them were processed with EVAL.
    Now each --eval form is processed with both READ and EVAL before
    the next --eval form is processed. (Thus package operations like
    sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
    work as the user might reasonably expect.)
  * minor incompatible change: *STANDARD-INPUT* is now only an
    INPUT-STREAM, not a BIDIRECTIONAL-STREAM.  (thanks to Antonio
    Martinez)
  * minor incompatible change: Y-OR-N-P is now character-oriented,
    not line oriented.  Also, YES-OR-NO-P now works without errors.
    (thanks to Antonio Martinez)
  * sb-aclrepl module improvements: an integrated inspector, added
    repl features, and a bug fix to :trace command.
  * Known functions, which cannot be open coded by the backend, are
    considered to be able to check types of their arguments. (fixing
    a bug report by Nathan J. Froyd)
  * fixed a bug in computing method discriminating functions: It is
    now possible to define methods specialized on classes which have
    forward-referenced superclasses. (thanks to Gerd Moellmann)
  * fixed evaluation order in optional entries (reported by Gilbert
    Baumann)
  * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
    specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
  * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
    required-by-AMOP NAME argument, as well as CLASS and
    DIRECT-SLOT-DEFINITIONS.  (thanks to Kevin Rosenberg)
  * fixed bug 20: DEFMETHOD can define methods using names that are
    not the proper names of classes to designate class specializers.
  * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
    Unix file descriptors, instead of blowing up. (thanks to Antonio
    Martinez)
  * Garbage collection refactoring: user-visible change is that a
    call to the GC function during WITHOUT-GCING will not do garbage
    collection until the end of the WITHOUT-GCING.  If you were doing
    this you were probably losing anyway.
  * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
    (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
    TYPEP the latter but not the former.
  * The compiler issues a full WARNING for calls to undefined functions
    with names from the CL package.
  * MAP-INTO for a vector destination is open coded.  (reported by
    Brian Downing on c.l.l)
  * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
    documentation string.
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** COPY-ALIST now signals an error if its argument is a dotted
       list.
    ** Condition slots are now accessed more correctly in the presence
       of multiple initargs for a given slot.
    ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
       exclude restarts of the same name associated with a different
       condition.
    ** DEFCLASS of forward-referenced classes with another
       forward-referenced class in the superclasses list no longer
       causes an error.
    ** Condition slots are now initialized once each, not multiple
       times. (thanks to Gerd Moellmann)
    ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
       Gerd Moellmann)
    ** &WHOLE and &REST arguments in macro lambda lists are patterns.
    ** NSET-EXCLUSIVE-OR does not return extra elements when its
       arguments contain duplicated elements.
    ** RESTART-CASE understands local macros.
    ** RESTART-CASE associates exactly its own restarts with a condition.
    ** ENDP in safe mode checks its argument to be of type LIST.
    ** COPY-SYMBOL in a threaded build no longer fails when the symbol
       in question is unbound.
    ** Optimized MAKE-INSTANCE functions no longer cause internal
       assertion failures in the presence of duplicate initargs.
    ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
    ** Side-effectful :DEFAULT-INITARGS have their side-effects
       propagated even in the ctor optimized implementation of
       MAKE-INSTANCE.
    ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.

changes in sbcl-0.7.14 relative to sbcl-0.7.13:
  * a better implementation of SXHASH on (simple) bit vectors,
    measured both in execution speed and in distribution of results
    over the positive fixnums, has been installed.  Likewise, a better
    implementation of EQUAL for simple bit vectors is now available.
  * fixed CEILING optimization for a divisor of form 2^k.
  * fixed bug 240 (emitting extra style warnings "using the lexical
    binding of the symbol *XXX*" for &OPTIONAL arguments).  (reported
    by Antonio Martinez)
  * fixed SXHASH, giving different results for NIL depending on type
    declarations (SYMBOL or LIST).  (thanks to Gerd Moellmann)
  * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
    variable.  (found by Rolf Wester)
  * SBCL does not ignore type declarations for special
    variables.  (reported by rif on c.l.l 2003-03-05)
  * some bug fixes in contrib/sb-aclrepl/
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** a bug in the CONS type specifier, whereby the CAR and CDR
       types got intertwined, has been fixed;
    ** the type system is now able to reason about the interaction
       between INTEGER and RATIO types more completely;
    ** APPEND, [N]REVERSE and NRECONC check that those their
       arguments, which must be proper lists, are really so;
    ** An array specialized to be unable to hold elements has been
       implemented, as required -- yes, really -- by ANSI;
    ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
       on malformed property lists;

changes in sbcl-0.7.13 relative to sbcl-0.7.12:
  * incompatible packaging change: in line with Unix convention, 
    SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core 
    if it's not in $SBCL_HOME.  It also sets SBCL_HOME to match.
  * REQUIRE and PROVIDE are now optionally capable of doing something
    useful. See the documentation string for REQUIRE.
  * infrastructure for a managed SBCL contrib system: contributed 
    modules in this release include:
    ** the ASDF system definition facility;
    ** an interface to the BSD Sockets API;
    ** an ACL-like convenience interface to the repl; 
       (thanks to Kevin Rosenberg)
    ** an implementation of ROTATE-BYTE, with efficient implementation
       on x86 hardware;
  * fixed a bug in LOG, so that LOG of a rational argument near 1 now
    gives a closer approximation to the right answer than previously.
    (thanks to Raymond Toy)
  * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
    UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
    optional environment arguments, as required by ANSI.
  * fixed bugs in other functions taking environment objects, allowing
    calls with an explicit NIL environment argument to be compiled
    without error.
  * fixed bug 228: primary return values from
    FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
    COMPILE or FUNCTION.
  * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
    :TYPE will no longer signal errors on innocuous objects.
  * fixed bug 231b: SETQ is better at respecting type declarations in
    the lexical environment.
  * fixed a bug in DEFCLASS: classes named by symbols with no or
    unprintable packages can now be defined.
  * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
    carelessly renamed to :TEST-FUN.  (thanks to Robert E. Brown)
  * fixed compiler failure related to checking types of functions.
    (reported by Robert E. Brown)
  * the compiler is now much more consistent in its error-checking
    treatment of bounding index arguments to sequence functions: in
    (SAFETY 3) code, errors will be signalled in almost all cases if
    invalid sequence bounding indices are passed to functions defined
    by ANSI to operate on sequences.
  * fixed a bug in the build procedure: documentation of SBCL-specific
    packages is now preserved and available in the final Lisp image.
  * lifted FDEFINITION lookup out of loops in the implementation of
    many list operations.  (thanks to Robert E. Brown)
  * fixed a bug in the reader: the #n# reader macro now works for
    objects of type STANDARD-OBJECT.  (reported by Tony Martinez)
  * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
    and that READ-DELIMITED-LIST returns a LIST.  (thanks to Robert
    E. Brown and Tony Martinez respectively)
  * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
    SLOT-BOUNDP: in particular, it is now able to optimize them much
    better, and is now not vulnerable to having packages renamed.
    Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
    always called when a slot is not present in an instance.  (thanks
    to Gerd Moellmann)
  * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
    longer signal an error on unreading a character following EOF on
    the previous constituent stream.  (thanks to Tony Martinez)
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
       not just nonnegative fixnums;
    ** the logical bit-array operators such as BIT-AND now accept an
       explicit NIL for their "opt-arg" argument (to indicate a
       freshly-consed result bit-array);
    ** ELT now signals an error on an invalid sequence index in safe
       code;
    ** the type system is now cleverer about negations of numeric
       types, and consequently understands the BIGNUM and RATIO types
       better;
    ** the type system is now cleverer about the interaction between
       INTEGER and RATIO types: while bugs still remain, many more
       cases are accurately computed;
    ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
       if it is in the last clause;
    ** CONSTANTLY now correctly returns a side-effect-free function in
       all cases;
    ** DECLARE is no longer treated as a special-operator; in
       particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
  * incremented fasl file version number due to the change in the
    DEFSTRUCT-SLOT-DESCRIPTION structure.

changes in sbcl-0.7.12 relative to sbcl-0.7.11:
  * minor incompatible change: code processed by the "interpreter" or
    EVAL now has a compilation optimization policy of (DEBUG 2)
    (changed from (DEBUG 1)) to improve debuggability of interactive
    development, and to allow the use of the debug RETURN command in
    such code.
  * an experimental implementation of the RETURN command for the
    debugger has been included.  (thanks to Frederik Kuivinen)
  * fixed bug 62: constraints were not propagated into a loop.
  * fixed bug in embedded calls of SORT (reported and investigated by
    Wolfgang Jenkner).
  * fixed some bugs revealed by Paul Dietz' test suite:
    ** printing and reading of arrays with some dimensions having
       length 0 (thanks to Gerd Moellmann);
    ** BOA constructor with &AUX argument without a default value does
       not cause a type error;
    ** CONSTANTP now returns true for all self-evaluating objects.

changes in sbcl-0.7.11 relative to sbcl-0.7.10:
  * fixed bug 127: DEFSTRUCT now does not clobber old structure
    accessors that are related by inheritance, as specified in the
    :CONC-NAME section of the specification of DEFSTRUCT.  (thanks to
    Valtteri Vuorikoski)
  * The compiler is now able to inline functions that were defined in
    a complex lexical environment (e.g. inside a MACROLET).
  * fixed bug in DESCRIBE, which now works on rank-0 arrays.  (thanks
    to Lutz Euler)
  * Support for the upcoming FreeBSD-5.0 release has been included.
    (thanks to Dag-Erling Smorgrav)
  * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
    effect when it is not in a toplevel context.
  * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
    stand a better chance of being correct.  (thanks to Gerd
    Moellmann)
  * fixed bug in COERCE, which now signals an error on coercing a
    rational to a bounded real type which excludes the expected
    answer.
  * The compiler is now able to derive types more accurately from the
    COERCE and COMPILE functions.
  * fixed bug 223: functional binding is considered to be constant
    only for symbols in the CL package.
  * fixed bug 231: SETQ did not check the type of a variable being set
    (reported by Robert E. Brown)
  * A new optimization for MAKE-INSTANCE has been included, fixing
    various bugs (including relating to :ALLOCATION :CLASS slots and
    :DEFAULT-INITARGS over-eager evalueation).  (thanks to Gerd
    Moellmann)
  * fixed some LOOP bugs revealed by Paul Dietz' test suite:
    ** As required by ANSI, LOOP now disallows anonymous collection
       clauses such as COLLECT I in conjunction with aggregate boolean
       clauses such as THEREIS (= I 1);
    ** LOOP now signals an error when any variable is reused in the
       same loop (including the potentially useful construct analogous
       to WITH A = 1 WITH A = (1+ A);
    ** IT is only a special loop symbol within the first clause of a
       conditional loop clause;
    ** LOOP with a typed iteration variable over a hashtable now
       signals a type error iff it should.
  * fixed some other bugs revealed by Paul Dietz' test suite:
    ** FILE-STREAM now names the class previously known as FD-STREAM;
    ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
       argument) no longer signals an error;
    ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
       of the slot symbol, rather than using the current package
       ((:CONC-NAME "") continues to intern the slot's name in the
       current package);
  * incremented fasl file version number, because of the incompatible
    change to the DEFSTRUCT-DESCRIPTION structure, and again because
    of the new implementation of DEFINE-COMPILER-MACRO.

changes in sbcl-0.7.10 relative to sbcl-0.7.9:
  * Support for building SBCL for MIPS platforms running in
    little-endian mode has now been checked in, and basic
    functionality on said platforms verified.
  * minor incompatible change: PCL now records the pathname of a file
    in which methods and the like are defined, rather than its
    truename.
  * minor incompatible change: TRUENAME now considers the truename of
    a file naming a directory to be the pathname with :DIRECTORY
    component indicating that directory.
  * minor incompatible change: a NAMED clause in the extended form of
    LOOP no longer causes a BLOCK named NIL to surround the LOOP.  The
    reason for the previous behaviour is unclear.
  * more systematization and improvement of CLOS and MOP conformance
    in PCL (thanks to Gerd Moellman and Pierre Mai):
    ** the standard ANSI CL generic function NO-NEXT-METHOD is now
       implemented;
    ** DEFINE-METHOD-COMBINATION no longer signals an error for
       primary methods with no specializers;
    ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
       implemented;
    ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
       and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
       FORWARD-REFERENCED-CLASSes; error reporting on
       CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
       has been improved;
    ** SXHASH on CLOS instances now uses a slot internal to the
       instance to return different numbers on distinct instances,
       while preserving the same return value through invocations of
       CHANGE-CLASS;
    ** DEFMETHOD signals errors when methods with longer incongruent
       lambda lists are added to generic functions;
    ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
       CLASS, as specified in AMOP;
    ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
       on the order returned by the primary method for classes of
       class STANDARD-CLASS;
    ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
  * fixed some bugs shown by Paul Dietz' test suite:
    ** DOLIST puts its body in TAGBODY;
    ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
       correct order;
    ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
       value producing form;
    ** if more variables are given to PROGV than values, extra
       variables are bound and made to have no value;
    ** NSUBSTITUTE on list arguments gets the right answer with
       :FROM-END;
    ** ELT signals an error of type TYPE-ERROR when the index argument
       is not a valid sequence index;
    ** LOOP signals (at macroexpansion time) an error of type
       PROGRAM-ERROR when duplicate variable names are found;
    ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
    ** FUNCALL of special-operators now cause an error of type
       UNDEFINED-FUNCTION;
    ** PSETQ now works as required in the presence of side-effecting
       symbol-macro places;
    ** NCONC accepts any object as its last argument;
    ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
       Gerd Moellman)
    ** loop-for-as-package does not require a package to be explicitely
       specified;
    ** LOOP WITH now treats NIL in the d-var-spec correctly as an
       ignored binding.
  * fixed bug 166: compiler preserves "there is a way to go"
    invariant when deleting code.
  * fixed bug 172: macro lambda lists with required arguments after
    &REST arguments now cause an error to be signalled.  (thanks to
    Matthew Danish)
  * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
    bug. (thanks to Matthew Danish)
  * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
    Baumann)
  * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
    arguments are lexically rebound. (thanks to Gerd Moellmann and
    Pierre Mai)
  * fixed bug 194: error messages are now more informative when there
    is no primary method applicable in a call to a generic
    function. (thanks to Gerd Moellmann)
  * fixed bug in command line argument checking (thanks to Julian
    Fondren)
  * fixed bug in COUNT-IF, making it handle :FROM-END correctly
    (thanks to Matthew Danish)
  * incremented fasl file version number, because of the
    SXHASH-related changes in the layout of CLOS data structures

changes in sbcl-0.7.9 relative to sbcl-0.7.8:
  * minor incompatible change: The runtime (the Unix executable named
    "sbcl") is now much pickier about the .core files it will load.
    Essentially it now requires .core files to descend from the same
    build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
    as the runtime does. (The intent is to prevent the crashes which
    can occur, and which can even be reported as mysterious failures,
    when people patch the sources or change the build parameters
    without changing LISP-IMPLEMENTATION-VERSION, then mix and match
    sbcl and .core files.)
  * fixed bug: VALUES-LIST is no longer optimized away.
  * fixed bug 142: The FFI conversion of C string values to Lisp
    string values no longer conses excessively. (thanks to Nathan
    Froyd porting Raymond Toy's fix to CMU CL)
  * began to systematize and improve MOP conformance in PCL (thanks to
    Nathan Froyd, Gerd Moellman and Pierre Mai):
    ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
       itself;
    ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
    ** FINALIZE-INHERITANCE is now called on class finalization;
    ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
       argument precedence order.
  * fixed bug 202: The compiler no longer fails on functions whose
    derived types contradict their declared type.
  * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
    so it can be non-toplevel.
  * The fasl file version number has changed (because of the new
    implementation of DEFMACRO).
  * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
    safe code, that any length requirement by their type-specifier
    argument is valid.  The exceptions to this are described in bug
    213.
  * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
    functions now check the types of their inputs as required by ANSI.
  * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
    introduced symbol is DECLAREd to be SPECIAL.
  * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
  * fixed bug: :COUNT argument to sequence functions may be negative
  * fixed bug: body of DO-SYMBOLS may contain declarations
  * fixed bug: PUSHNEW now evaluates its arguments from left to right
    (reported by Paul F. Dietz, fixed by Gerd Moellman)
  * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
    symbol macro only once
  * fixed printing of call frame when argument list is unavailable
  * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
  * compiler no longer signals WARNING on unknown keyword
    :ALLOW-OTHER-KEYS

changes in sbcl-0.7.8 relative to sbcl-0.7.7:
  * A beta-quality port to the mips architecture running Linux,
    based on the old CMUCL backend, has been made.  It has been tested
    on a big-endian kernel, and works sufficiently well to be able to
    rebuild itself; it has not been tested in little-endian mode.
  * fixed an inconsistency between gencgc.c and purify.c which made
    dumping/loading .core files unreliable
  * fixed bug 120a: The compiler now deals correctly with IFs where
    the consequent is the same as the alternative, instead of
    misderiving the return type. (thanks to Alexey Dejneka)
  * fixed bug 113: Logical pathnames are now dumpable (the logical
    host is resolved at load-time, throwing an error if it is not
    found).
  * fixed bug 174: FORMAT's error message is slightly clearer when a
    non-printing character is used in a format directive.
  * fixed several bugs in compiler checking of type declarations, i.e.
    violations of the Python "declarations are assertions" principle
    (thanks to Alexey Dejneka)
  * fixed several bugs in PCL's error checking (thanks to Gerd
    Moellmann)
  * fixed bug: printing of FILE-ERROR (thanks to Antonio
    Martinez-Shotton)
  * fixed bug in compilation of functions as first class values
    (thanks to Antonio Martinez-Shotton)
  * The compiler's handling TYPE-ERRORs which it can prove will
    inevitably happen at runtime has been cleaned up and corrected
    in several ways. (thanks to Alexey Dejneka)
  * improved argument type checking for various basic arithmetic
    operations (MAX, +, LOGXOR, etc.) which have had so much TLC
    lavished on them in the past that they can be compiled in many
    ways in different special cases
  * fixed bug 181: compiler checks validity of user supplied type
    specifiers
  * cleaned up code flushing in optimization: Function calls which
    should signal errors for safety purposes (e.g. which ANSI says
    should signal errors when their arguments are of incorrect type)
    are no longer optimized away.
  * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
  * incremented fasl file version number, because changes in the
    implementation of sequence functions like COERCE caused 
    internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
    in old inline expansions) to become undefined. (Actually these
    changes were later undone, so we might very well be binary
    compatible with 0.7.7 after all, but leaving the version number
    incremented seemed like the simplest and most conservative
    thing to do.)

changes in sbcl-0.7.7 relative to sbcl-0.7.6:
  * An alpha-quality port to the parisc architecture running Linux,
    based on the old CMUCL backend, has been made.  This, even more so
    than the other backends, should be considered still a work in
    progress; known problems include that the Linux kernel in 64-bit
    mode does not propagate the correct sigcontext structure to
    userspace, and consequently SBCL on a parisc64 kernel will not
    work yet.
  * fixed bug 189: The compiler now respects NOTINLINE declarations for
    functions declared in FLET and LABELS. (I.e. "LET conversion" is
    suppressed.) Also now that the compiler is looking at declarations
    in the environment, it checks optimization declarations as well,
    and suppresses inlining when (> DEBUG SPEED).
  * More fixes have been made to treatment of floating point exception
    treatment and other Unix signals.  In particular, floating point
    exceptions no longer cause Bus errors on the SPARC/Linux platform.
  * The detection and handling of control stack exhaustion (infinite
    or very deeply nested recursion) has changed.  Stack exhaustion
    detection is now done by write-protecting pages at the OS level
    and applies at all optimization settings; when found, a
    SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
    STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
    be caught using IGNORE-ERRORS.
  * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
    correctly when transforming recursive LABELS functions to LETs.
    (thanks to Alexey Dejneka)
  * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
    that are names of constants or global variables.
  * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
    alien routines with docstrings.
  * Bug 184 fixed: Division of ratios by the integer 0 now signals an
    error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
    Raymond Toy)
  * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
    (thanks to Eric Marsden)
  * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
    object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
  * Bug fix: The SPARC backend can now compile functions involving
    LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
  * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
    to a functions expecting a PATHNAME-DESIGNATOR.
  * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
    lambda lists. (thanks to Alexey Dejneka)
  * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
    functions so that EXPORTing the name of the class doesn't cause
    MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
    to Antonio Martinez for reporting this)
  * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
    constant-folded without failing an assertion. (thanks to Einar
    Floystad Dorum for reporting this)
  * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
    types are now dealt with more correctly by the compiler.
  * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
    OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
    behaviour analogous to RENAME-FILE.  This puts its behaviour more
    in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
    Marco Antinotti)
  * The fasl file version number has changed again. (because of the
    bug fix involving the names of PCL MAKE-INSTANCE functions)

changes in sbcl-0.7.6 relative to sbcl-0.7.5:
  * bug fix: Floating point exceptions are treated much more
    consistently on the x86/Linux and PPC/Linux platforms.
  * Array initialization with :INITIAL-ELEMENT is now much faster for
    cases when the compiler cannot open code the array creation, but
    does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
    array accesses have also seen a speed increase.
  * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
    specified. (This was at the root of some bad interactions between
    SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
    reporting the bug.)
  * bug fix: Internal error arguments for undefined functions are now
    computed correctly on the PPC/Linux platform.
  * bug fix: Bad &REST syntax is now checked correctly. (thanks to 
    Raymond Toy's patch for CMU CL)
  * Support for the Solaris 9 operating environment has been included
    (thanks to Daniel Merritt)
  * A very ugly but hopefully complete draft of the missing FFI chapter
    of the manual has been created by reformatting the corresponding
    CMU CL manual chapter into (currently very ugly and incoherent)
    DocBook and bringing it up to date for SBCL behavior. Thus, the
    manual is now essentially complete, at least by my extreme
    once-and-only-once standards, whereby it's acceptable to refer to
    the doc strings of SB-EXT functions as the primary documentation.
  * The fasl file version number has changed again, due to cleanup of
    (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
    is no longer a static symbol.)

changes in sbcl-0.7.5 relative to sbcl-0.7.4:
  * SBCL now builds with OpenMCL (version 0.12) as the
    cross-compilation host; also, more progress has been made toward
    bootstrapping under CLISP.
  * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
    Alpha architecture.
  * bug 158 fixed: The compiler can now deal with integer loop
    increments different from 1; fixing this turned out also to fix
    bug 164.
  * bug 169 fixed: no more bogus warnings about using lexical bindings
    despite the presence of perfectly good SPECIAL declarations (thanks
    to David Lichteblau)
  * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
    accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
  * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
    characters in them.
  * bug 180 fixed: Method combination specifications no longer ignore
    the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
  * bug fix: Structure type predicate functions now check their argument
    count as they should.
  * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
    correctly. (thanks to Pierre Mai)
  * minor incompatible change: The --noprogrammer option is deprecated
    in favor of the new --disable-debugger option, which is very similar.
    (The major difference is that it takes effect at a slightly different
    time at startup, causing handling of errors in --sysinit and
    --userinit files will be affected differently.) The
    SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
    been added to allow this functionality to be controlled from ordinary
    Lisp code. (ENABLE-DEBUGGER should help people like the Debian
    maintainers, who might want to run non-interactive scripts to
    build SBCL cores which will later be used interactively.)
  * minor incompatible change: The LOAD function no longer, when given
    a wild pathname to load, loads all files matching that pathname.
    Instead, an error of type FILE-ERROR is signalled.

changes in sbcl-0.7.4 relative to sbcl-0.7.3:
  * bug 147 fixed: The compiler preserves its block link/count
    invariants more correctly now so that it doesn't crash. (thanks
    to Alexey Dejneka)
  * Dynamic loading of object files in OpenBSD is now supported. (thanks
    to Pierre Mai)
  * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
  * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
    no-value-for-this-character as NIL (as specified by ANSI).
  * HOST-NAMESTRING on physical pathnames now returns a string that is
    valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
    (thanks to Christophe Rhodes)
  * The Alpha port handles icache flushing more correctly. (thanks to
    Dan Barlow)
  * More progress has been made toward bootstrapping under CLISP. (thanks
    to Christophe Rhodes)
  * The fasl file format has changed again, because dynamic loading
    on OpenBSD (which has non-ELF object files) motivated some cleanups
    in the way that foreign symbols are transformed and passed around.
  * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
    is no longer treated as whitespace by the reader, but instead as
    an ordinary character. Thus e.g. (READ-FROM-STRING "A\x7fB") returns
    |A\x7fB|, instead of A as it used to.

changes in sbcl-0.7.3 relative to sbcl-0.7.2:
  * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
    Froyd porting CMU CL code originally by Douglas Thomas Crosher)
  * SBCL now runs on the PPC archtiecture under Linux. It actually did
    this as of 0.7.1.45, but was left out of the previous news section
    (thanks to Dan Barlow)
  * SBCL now runs on the Solaris operating system on SPARC architectures
    (thanks to Christophe Rhodes's port of the CMUCL runtime)
  * cleanups to the runtime on SPARC, both Linux and Solaris, and for
    gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
  * SPARC backend cleanups, allowing builds of cores optimized for V8
    and V9 SPARCS, and also emission of code targeted to a particular
    backend chosen at runtime (thanks to Christophe Rhodes and Raymond
    Toy)
  * SBCL is closer to bootstrapping under CLISP, thanks to various
    fixes by Christophe Rhodes.
  * The fasl file format has changed again, to allow the compiler's
    INFO database to support symbol macros.
  * The user manual (in doc/) is formatted into HTML more nicely.
    (thanks to coreythomas)
  * The system is smarter about SUBTYPEP relationships, especially
    those involving NOT types (including types such as ATOM which are
    represented internally using NOT types). Thus SUBTYPEP is less
    likely to return (VALUES NIL NIL) in general, and in particular
    bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
    Christophe Rhodes)
  * The fasl file format has changed again, because the internal
    representation of types now includes a new slot to support the new
    SUBTYPEP-of-NOT-types logic.
  * (not a change in the main branch of SBCL, but a related prototype
    which can hopefully be merged into the main branch of SBCL in the
    future:) Brian Spilsbury has produced a Unicode-enabled variant of
    sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
      <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
  * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
    dispatch macro character combinations. (thanks to Alexey Dejneka)
  * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings 
    that superficially look like logical namestrings correctly.
  * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
    can deal with.
  * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
    default supplied-p) for &optional and &key arguments. (thanks to
    Martin Atzmueller)

changes in sbcl-0.7.2 relative to sbcl-0.7.1:
  * incompatible change: The compiler is now less aggressive about
    tail call optimization, doing it only when (> SPACE DEBUG) or
    (> SPEED DEBUG). (This is an incompatible change because there are
    programs which relied on the old CMU-CL-style behavior to optimize
    away their unbounded recursion which will now die of stack overflow.)
  * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
    for non-GENCGC systems has been increased to 20M (since that
    seems much closer to the likely performance optimum for modern
    systems than the old 4M value was)
  * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
    and *DEBUG-PRINT-LEVEL*
  * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
    of CMU CL's support for SPARC, and various endianness and other 
    SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
  * new syntactic sugar for the Unix command line: --load foo.bar is now
    an alternate notation for --eval '(load "foo.bar")'.
  * bug fixes:
    ** The system now detects stack overflow and handles it gracefully,
       at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
       optimization settings. (This is a good thing in general, and
       its introduction in this version should be particularly timely
       for anyone whose code fails because of suppression of tail
       recursion!)
    ** The system now hunts for the C variable "environ" in a more
       devious way, to avoid segfaults when the C library version differs
       between compile time and run time. (thanks to Christophe Rhodes)
    ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
       and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
    ** The compiler no longer issues bogus style warnings for undefined
       classes in the same source file as the DEFCLASSes which defined
       them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
       for fixing this)
    ** fixes in CONDITION class precedence list for undefined function
       errors (thanks to Alexei Dejneka)
    ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and 
       correctly. (thanks to Dan Barlow)
    ** portability fixes aiming at bootstrapping under CLISP (thanks
       to Dave McDonald and Christophe Rhodes)
    ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
    ** fixes in type translation and and type inference (thanks to
       Christophe Rhodes)
    ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
    ** various fixes in the new ports (thanks to Dan Barlow)
  * several changes related to debugging:
    ** suppression of tail recursion, as noted above
    ** stack overflow detection, as noted above
    ** The default implementation of TRACE has changed. :ENCAPSULATE T
       is now the default. (For some time encapsulation has been more
       reliable than the breakpoint-based :ENCAPSULATE NIL
       implementation, at least on X86 systems; and I just noticed that
       encapsulation also seems closer to the spirit of the ANSI
       specification.)

changes in sbcl-0.7.1 relative to sbcl-0.7.0:
* mostly bug fixes:
  ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
     up properly again. (There was a packaging bug in 0.7.0 which
     left their definitions in SB-SYS::LOAD-FOREIGN and 
     SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
     vital for most things which interface to C-level interfaces,
     like extensions working with sockets or databases or
     Perl-compatible regexes or whatever, and the need to fix
     this bug is the main reason that 0.7.1 was released so
     soon after 0.7.0.)
  ** DEFGENERIC is now choosier about the methods it redefines, so that
     reLOADing a previously-LOADed file containing DEFGENERICs does
     the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
     cycle now works as it should. (thanks to Alexey Dejneka)
  ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
     fixed by Christophe Rhodes. (He actually submitted this patch
     months ago, and I delayed until after 0.7.0.)
  ** Bug 111 (internal compiler confusion about runtime checks on
     FUNCTION types) was fixed by Alexey Dejneka.
* Some internal cleanups (getting rid of variables which aren't
  needed now that the byte interpreter is gone) caused the fasl
  file format number to change again.

changes in sbcl-0.7.0 relative to sbcl-0.6.13:
* major incompatible change: The default fasl file extension, i.e. the
  default extension for files produced by COMPILE-FILE, has changed
  to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".) 
* compiler changes:
  ** There are many changes in the implementation of the compiler.
     SBCL is now essentially a compiler-only implementation of ANSI
     Common Lisp. EVAL still "interprets" a few special cases, but 
     almost all the interesting cases are handled by creating
     a LAMBDA expression, calling COMPILE on it, then calling 
     FUNCALL on the result.
  ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
     various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
     Since the code is newer, there might still be some new bugs
     (though not as many as before Martin Atzmueller's fixes:-). But 
     the new code is substantially simpler and clearer, and hopefully
     any remaining bugs will be simpler, less fundamental, and more
     fixable then the bugs in the old code.
  ** The revised compiler is still a little unsteady on its feet.
     In particular,
     *** The debugging information it produces (particularly the names
         of FUNCTION objects) is sometimes much less useful than what
         the old compiler produced.
     *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
         then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
         within a MACROLET) has been temporarily weakened.
  ** There are new compiler optimizations for various functions:
     *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
         FIND-IF-NOT, POSITION-IF-NOT, and FILL
     *** the math functions TRUNCATE, FLOOR, and CEILING
     *** the function-of-all-trades COERCE
     Mostly these should be transparent, but there's one
     potentially-annoying problem (bug 117): when the compiler
     inline-expands a function and does type analysis on the result,
     it can create control paths which have type mismatches, and
     when it can't prove that those control paths aren't taken,
     it will issue WARNINGs about the type mismatches. This is 
     a particular problem in practice for the new sequence functions.
     It's not clear how this should be fixed, and for now, a
     workaround is given in the entry for 117 in the BUGS file.
  ** (Because of the interaction between the two previous items --
     occasional inlining problems and new inline expansions -- some
     of the new sequence function optimizations won't really kick in
     completely until debugging information, and then inlining, are
     straightened out in some future version.)
* minor incompatible changes:
  ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
     DIRECTORY no longer implicitly promotes NIL slots of its
     pathname argument to :WILD. In particular, when you ask for the
     contents of a directory (which you used to be able to do without
     explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
     explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
  ** changes in behavior that ANSI explicitly defines to be
     implementation dependent:
     *** The new compiler-only implementation still conforms with ANSI,
         but acts a little different than before. Besides the obvious
         changes in performance tradeoffs (that the cost per form passed
         to EVAL has gone up, and the cost per form executed by EVAL
         has gone down), the behavior of the system changes a little
         because there are no longer any interpreted function objects.
         COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
         e.g. doing COMPILE on the output of interactive DEFUN is
         now a no-op.
     *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
         from 100 to 1000.
     *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
         has changed from (:CL) to NIL.
     *** The CHAR-NAME of unprintable ASCII characters which, unlike
         e.g. #\Newline and #\Tab, don't have names specified in the
         ANSI Common Lisp standard, is now based on their ASCII symbolic
         names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
         (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
         are no longer used for output.
  ** changes in internal implementation constants:
     *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
         4 million. (If your application spends a lot of time GCing and
         you have a lot of RAM, you might want to experiment with
         increasing it even more.)
  ** The SB-C-CALL package has been merged into the SB-ALIEN package.
     However, almost all old code should still continue to work without
     immediate update, as SB-C-CALL is now a (deprecated) nickname
     for SB-ALIEN.
  ** Old operator names in the style DEF-FOO are now deprecated in
     favor of new corresponding names DEFINE-FOO, for consistency with
     the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
     DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
     internal symbols, but a few supported extensions like
     SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
     DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
  ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
     etc. as you get deeper into recursive calls to the debugger
     command loop, instead of the old "5]", "5]]", "5]]]"
     sequence. (I was motivated to do this when squabbles between
     ILISP and SBCL left me very deeply nested in the debugger. In the
     short term, this change will probably provoke more ILISP/SBCL 
     squabbles, but hopefully it will be an improvement in the long run.)
  ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
     name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
     function, since declaiming return types involving aliens is
     (1) annoyingly messy to do by hand and (2) vital to efficient
     compilation of code which calls such functions.
  ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
     longer reexported by the SB-EXT package. They're solely useful
     for alien code, so it seems more logical that you should get
     them from the SB-ALIEN package, not in SB-EXT.
  ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
     :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
     features. Instead, the code that they used to control is always
     built into the system.
* many other bug fixes
  ** DEFSTRUCT and DEFCLASS have been substantially updated to take
     advantage of the new EVAL-WHEN stuff and to clean them up in 
     general, and they are now more ANSI-compliant in a number of
     ways. Martin Atzmueller is responsible for a lot of this.
  ** Besides the cleanups discussed above, Martin Atzmueller fixed
     several other bugs:
     *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
     *** correct ERROR type for various file operations
     *** some fixes for Lisp streams
     *** DEFMETHOD syntax checking
     *** changing old weird representation of debug information as
         strings (which, among their other deficiencies, don't transform
         correctly when you rename packages, and don't change their
         print representation when you change things like *PACKAGE*
         and *PRINT-LENGTH*) to symbols and lists of symbols
     He also made several improvements and fixed several bugs in DESCRIBE.
  ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
     discovered himself:
     *** misbehavior of WRITE-STRING/WRITE-LINE
     *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
         and several other LOOP problems as well
     *** DIRECTORY when similar filenames are present
     *** DEFGENERIC with :METHOD options
     *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
     *** bug in the optimization of ARRAY-ELEMENT-TYPE
     *** argument ordering in FIND with :TEST option
     *** mishandled package designator argument in APROPOS-LIST
     *** various problems in the backquote readmacro
     *** a bug in APROPOS
     *** probably some others that I'm not describing very well here,
         since the CVS log documents them by reference to sbcl-devel
         messages, and the SourceForge archives aren't working well.:-(
  ** Dan Barlow improved the Alpha port (and is making progress on the
     PPC port, for those of you who think different).
  ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
     Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
     contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
     continues to work on the SPARC port (for those of you in a position
     to look down upon our little PC-compatible boxes from a great height). 
  ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
     and uses that copy, rather than the current dynamic value, when
     it's trying to decide whether to truncate output. Thus e.g.
       (let ((*print-lines* 50))
         (pprint-logical-block (stream nil)
           (dotimes (i 10)
             (let ((*print-lines* 8))
               (print (aref possiblybigthings i) stream)))))
     should now truncate the logical block only at 50 lines, instead of
     often truncating it at 8 lines, as it did before.
* The doc/cmucl/ directory, containing old CMU CL documentation
  from the time of the fork, is no longer part of the base system.
  SourceForge has shut down its anonymous FTP service, and with it
  my original plan for distributing the old CMU CL documentation
  there. For now, if you need these files you can download an old
  SBCL source release and extract them from it.
* The fasl file version number changed again, for dozens of reasons,
  some of which are apparent above.

changes in sbcl-0.6.13 relative to sbcl-0.6.12:
* a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
* Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
  patch, so that DISASSEMBLE output is much nicer.
* The code in the SB-PROFILE package now seems reasonably stable.
  I still haven't decided what the final interface should look like
  (I'd like PROFILE to interact cleanly with TRACE, since both
  facilities use function encapsulation) but if you have a need
  for profiling now, you can probably use it successfully with
  the current CMU-CL-style interface.
* Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
  ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
  Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
  link now returns the dangling link itself, and for similar
  reasons, TRUENAME on a cyclic symbolic link returns the cyclic
  link itself. (In these cases the old code signalled an error and
  looped endlessly, respectively.) Thus, DIRECTORY now works even
  in the presence of dangling and cyclic symbolic links.
* Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
  is now a supported extension again, since the consensus on
  sbcl-devel was that it can be useful for ordinary development
  work, not just for debugging SBCL itself.
* The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
  NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
  that currently-defined functions might be redefined later with 
  different return types. 
* Hash tables can be printed readably, as inspired by CMU CL code
  of Eric Marsden and SBCL code of Martin Atzmueller.
* better error handling in CLOS method combination, thanks to 
  Martin Atzmueller porting Pierre Mai's CMU CL patches
* more overflow fixes for >16Mbyte I/O buffers
* A bug in READ has been fixed, so that now a single Ctrl-D
  character suffices to cause end-of-file on character streams.
  In particular, now you only need one Ctrl-D at the command
  line (not two) to exit SBCL.
* fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for 
  undisplaced arrays.
* fixed bug 107 (reported as a CMU CL bug by Erik Naggum on 
  comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
  does the right thing.
* The implementation of some type tests, especially for CONDITION
  types, is now tidier and maybe faster, due to CMU CL code
  originally by Douglas Crosher, ported by Martin Atzmueller.
* Some math functions have been fixed, and there are new
  optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
  thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
* (There are also some new optimizers in contrib/*-extras.lisp. Those
  aren't built into sbcl-0.6.13, but are a sneak preview of what's
  likely to be built into sbcl-0.7.0.)
* A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
  CL, ported to SBCL by Martin Atzmueller)
* DESCRIBE now gives more information in some cases. (Pierre Mai's
  patch to CMU CL, ported to SBCL by Martin Atzmueller)
* Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
* There's a new slam.sh hack to shorten the edit/compile/debug
  cycle for low-level changes to SBCL itself, and a new
  :SB-AFTER-XC-CORE target feature to control the generation of
  the after-xc.core file needed by slam.sh.
* minor incompatible change: The ENTRY-POINTS &KEY argument to 
  COMPILE-FILE is no longer supported, so that now every function
  gets an entry point, so that block compilation looks a little
  more like the plain vanilla ANSI section 3.2.2.3 scheme.
* minor incompatible change: SB-EXT:GET-BYTES-CONSED now
  returns the number of bytes consed since the system started,
  rather than the number consed since the first time the function
  was called. (The new definition parallels ANSI functions like
  CL:GET-INTERNAL-RUN-TIME.)
* minor incompatible change: The old CMU-CL-style DIRECTORY options,
  i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
  supported. Now DIRECTORY always does the abstract Common-Lisp-y
  thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
* Fasl file version numbers are now independent of the target CPU,
  since historically most system changes which required version
  number changes have affected all CPUs equally. Similarly, 
  the byte fasl file version is now equal to the ordinary
  fasl file version.

changes in sbcl-0.6.12 relative to sbcl-0.6.11:
* incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
  is no longer recognized. I apologize for this, because it was
  listed in SB-EXT as a supported extension, but I found that
  its existing behavior was poorly specified, as well as incorrectly
  specified, and it looked like too much of a mess to straighten it
  out. I have enough on my hands trying to get ANSI stuff to work..
* many patches ported from CMU CL by Martin Atzmueller, with 
  half a dozen bug fixes in pretty-printing and the debugger, and
  half a dozen others elsewhere
* fixed bug 13: Floating point infinities are now supported again.
  They might still be a little bit flaky, but thanks to bug reports
  from Nathan Froyd and CMU CL patches from Raymond Toy they're not
  as flaky as they were.
* The --noprogrammer command line option is now supported. (Its
  behavior is slightly different in detail from what the old man
  page claimed it would do, but it's still appropriate under the
  same circumstances that the man page talks about.)
* The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
  are now supported, and enabled by default. Thus, the compiler can
  handle many floating point and complex operations much less
  inefficiently. (Thus e.g. you can implement a complex FFT
  without consing!)
* The compiler now detects type mismatches between DECLAIM FTYPE 
  and DEFUN better, and implements CHECK-TYPE more correctly, and
  SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
  thanks to patches from Martin Atzmueller.
* various fixes to make the cross-compiler more portable to
  ANSI-conforming-but-different cross-compilation hosts (notably
  Lispworks for Windows, following bug reports from Arthur Lemmens)
* A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
  ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
  fixed thanks to Pierre Mai's CMU CL patches.
* fixes to keep the system from overflowing internal counters when
  it tries to use i/o buffers larger than 16M bytes
* fixed bug 45a: Various internal functions required to support
  complex special functions have been merged from CMU CL sources.
  (When I was first setting up SBCL, I misunderstood a compile-time
  conditional #-OLD-SPECFUN, and so accidentally deleted them.)
* improved support for type intersection and union, fixing bug 12
  (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
  more obscure bugs as well
* some steps toward byte-compiling non-performance-critical
  parts of the system, courtesy of patches from Martin Atzmueller
* Christophe Rhodes has made some debian packages of sbcl at
  <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
  From his sbcl-devel e-mail of 2001-04-08 they're not completely
  stable, but are nonetheless usable. When he's ready, I'd be happy
  to add them to the SourceForge "File Releases" section. (And if
  anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
* new fasl file format version number (because of changes in 
  internal representation of (OR ..) types to accommodate the new
  support for (AND ..) types, among other things)

changes in sbcl-0.6.11 relative to sbcl-0.6.10:
* Martin Atzmueller pointed out that bugs #9 and #25 are gone in
  current SBCL.
* bug 34 fixed by Martin Atzmueller: dumping/loading instances works
  better
* fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE, 
  and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
  types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
* DESCRIBE now works on structure objects again.
* Most function call argument type mismatches are now handled as
  STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
  know whether the function will be redefined before the call is
  executed. (The compiler could flag local calls with full WARNINGs,
  as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
  it doesn't keep track of enough information to know whether calls
  are local in this sense.)
* Compiler output is now more verbose, with messages truncated
  later than before. (There should be some supported way for users
  to override the default verbosity, but I haven't decided how to 
  provide it yet, so this behavior is still controlled by the internal
  SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
  src/compiler/ir1util.lisp.)
* Fasl file format version numbers have increased again, because
  support for the Gray streams extension changes the layout of the
  system's STREAM objects.
* The Gray subclassable streams extension now works, thanks to a 
  patch from Martin Atzmueller.
* The full LOAD-FOREIGN extension (not just the primitive
  LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
* The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
  but like most other programs, it defaults to copying the Unix
  environment from the original process instead of starting the
  new process in an empty environment.
* Extensions which manipulate the Unix environment now support
  an :ENVIRONMENT keyword option which doesn't smash case or 
  do other bad things. The CMU-CL-style :ENV option is retained
  for porting convenience.
* LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
  as per Daniel Barlow's suggestion and Martin Atzmueller's patch

changes in sbcl-0.6.10 relative to sbcl-0.6.9:

* A patch from Martin Atzmueller seems to have solved the SIGINT
  problem, and as far as we know, signal-handling now works cleanly.
  (If you find any new bugs, please report them!)
* The system no longer defaults Lisp source file names to types
  ".l", ".cl", or ".lsp", but only to ".lisp". 
* The compiler no longer uses special default file extensions for
  byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
  seems to expect a single default extension for all compiled code,
  and there's no compelling reason to try to stretch the standard
  to allow two different extensions.) Instead, byte-compiled files
  default to the same extension as native-compiled files.
* Fasl file format version numbers have increased again, because
  a rearrangement of internal implementation packages made some 
  dumped symbols in old fasl files unreadable in new cores.
* DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
  many fewer weird special cases.
* Bug #17 (differing COMPILE-FILE behavior between logical and 
  physical pathnames) has been fixed, and some related misbehavior too,
  thanks to a patch from Martin Atzmueller.
* Bug #30 (reader problems) is gone, thanks to a CMU CL patch
  by Tim Moore, ported to SBCL by Martin Atzmueller.
* Martin Atzmueller fixed several filesystem-related problems,
  including bug #36, in part by porting CMU CL patches, which were
  written in part by Paul Werkowski.
* More compiler warnings in src/runtime/ are gone, thanks to 
  more patches from Martin Atzmueller.
* Martin Atzmueller pointed out that bug 37 was fixed by his patches
  some time ago.

changes in sbcl-0.6.9 relative to sbcl-0.6.8:

* DESCRIBE now works on CONDITION objects.
* The debugger now handles errors which arise when trying to print
  *DEBUG-CONDITION*, so that it's less likely to fall into infinite
  regress.
* The build system now uses an additional file, customize-target-features.lisp,
  to allow local modifications to the target *FEATURES* list. (The point of
  this is that now I can set up a custom configuration, e.g. with :SB-SHOW
  debugging features enabled, without having to worry about propagating it
  into everyone's system when I do a "cvs update".) When no
  customize-target-features.lisp file exists, the target *FEATURES* list
  should be constructed the same way as before.
* fixed bugs in DEFCONSTANT ANSI-compatibility:
  ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to 
     warn about behavior which is undefined under the ANSI spec. Note: This
     is specified by ANSI, but it's not very popular with programmers.
     If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
     macro in the SBCL sources for an example of a workaround which you
     might use to make portable ANSI-standard code which does what you want.
  ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
     pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
     when it's used as a non-toplevel form. (This is required in order
     to implement the DEFCONSTANT-EQX macro.)
  ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
  ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
     funny things, due to bugs in the implementation of EVAL-WHEN
     (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
     non-toplevel DEFCONSTANTs are uncommon.)
* The core file version number and fasl file version number have been 
  incremented, because the old noncompliant DEFCONSTANT behavior involved
  calling functions which no longer exist, and because I also took the
  opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
* fixed bug 1 (error handling before read-eval-print loop starts), and
  redid debugger restarts and related debugger commands somewhat while
  doing so:
  ** The QUIT debugger command is gone, since it did something
     rather different than the SB-EXT:QUIT command, and since it never
     worked properly outside the main toplevel read/eval/print loop.
     Invoking the new TOPLEVEL restart provides the same functionality.
  ** The GO debugger command is also gone, since you can just invoke
     the CONTINUE restart directly instead.
  ** The TOP debugger command is also gone, since it's redundant with the
     FRAME 0 command, and since it interfered with abbreviations for the
     TOPLEVEL restart.
* The system now recovers better from non-PACKAGE values of the *PACKAGE*
  variable.
* The system now understands compound CONS types (e.g. (CONS FIXNUM T))
  as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
  some porting work by Martin Atzmueller)
* Martin Atzmueller reviewed the CMU CL mailing lists and came back
  with a boatload of patches which he ported to SBCL. Now that those
  have been applied,
  ** The system tries to make sure that its low-priority messages
     are prefixed by semicolons, to help people who like to use
     syntax highlighting in their ILISP buffer. (This patch
     was originally due to Raymond Toy.)
  ** The system now optimizes INTEGER-LENGTH better, thanks to more
     patches originally written by Raymond Toy.
  ** The compiler understands coercion between single-value and 
     multiple-VALUES type expressions better, getting rid of some very
     weird behavior, thanks to patches originally by Robert MacLachlan
     and Douglas Crosher.
  ** The system understands ANSI-style non-KEYWORD &KEY arguments in 
     lambda lists, thanks to a patch originally by Pierre Mai.
  ** The system no longer bogusly warns about "abbreviated type
     declarations".
  ** The compiler gets less confused by inlining and RETURN-FROM,
     thanks to some patches originally by Tim Moore.
  ** The system no longer hangs when dumping circular lists to fasl
     files, thanks to a patch originally from Douglas Crosher.
* Martin Atzmueller also fixed ROOM, so that it no longer fails with an
  undefined function error.
* gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
  in structure slot initforms) for now, documented workaround instead:-|
* fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
* fixed bug 5: added stubs for various Gray stream functions called
  in the not-a-CL:STREAM case, so that even when Gray streams aren't
  installed, at least appropriate type errors are generated
* fixed bug 8: better reporting of various PROGRAM-ERRORs
* fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
  consistently in DEFMETHOD forms. 
* removed bug 21 from BUGS, since Martin Atzmueller points out that 
  it doesn't seem to affect SBCL after all
* The C runtime system now builds with better optimization and many
  fewer warnings, thanks to lots of cleanups by Martin Atzmueller.

changes in sbcl-0.6.8 relative to sbcl-0.6.7:

* The system is now under CVS at SourceForge (instead of the
  CVS repository on my home machine).
* The new signal handling code has been tweaked to treat register
  contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
  instead of (SIGNED-BYTE 32), as the C header files have it. (Code 
  downstream, e.g. in debug-int.lisp, has implicit dependencies
  on the unsignedness of integer representation of machine words, 
  and that caused the system to bomb out with infinite regress
  when trying to recover from type errors involving signed values,
  e.g. (BUTLAST '(1 2 3) -1).)
* (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
  (This was one of the bugs Peter Van Eynde reported back in July.)
* The system now uses code inspired by Colin Walters' O(N)
  implementation of MAP (from the cmucl-imp@cons.org mailing
  list, 2 September 2000) when it can't use a DEFTRANSFORM to
  inline the MAP operation, and there is more than one
  sequence argument to the MAP call (so that it can't just
  do ETYPECASE once and for all based on the type of the
  single sequence argument). (The old non-inline implementation
  of the general M-argument sequence-of-length-N case required
  O(M*N*N) time when any of the sequence arguments were LISTs.)
* The QUIT :UNIX-CODE keyword argument has been renamed to 
  QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
* Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
  have been added.
* Raymond Wiker's patches to port dynamic loading from Linux to
  FreeBSD have been added.
* The BUGS file is now more nearly up to date, thanks in large part
  to Martin Atzmueller's review of it.
* The debugger now flushes standard output streams before it begins
  its output ("debugger invoked" and so forth).
* The core version number and fasl file version number have both
  been incremented, because of incompatible changes in the layout
  of static symbols.
* FINISH-OUTPUT is now called more consistently on QUIT. (It
  used to not be called for a saved Lisp image.)
* Martin Atzmueller's version of a patch to fix a compiler crash,
  as posted on sbcl-devel 13 September 2000, has been installed.
* Instead of installing Martin Atzmueller's patch for the
  compiler transform for SUBSEQ, I deleted the compiler transform,
  and transforms for some similar consing operations.
* A bug in signal handling which kept TRACE from working on OpenBSD
  has been fixed.
* added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
  away by constant folding
* The system now defines its address space constants in one place
  (in the Lisp sources), and propagates them automatically elsewhere
  (through GENESIS and the sbcl.h file). Therefore, patching the
  address map is less unnecessarily tedious and error-prone. The
  Lisp names of address space constants have also been systematized.
* CVS tags like dollar-Header-dollar have been removed from
  the sources, because they have never saved me trouble and
  they've been source of trouble working with patches and other
  diff-related operations.
* fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by 
  ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)

changes in sbcl-0.6.7 relative to sbcl-0.6.6:

* The system has been ported to OpenBSD.
* The system now compiles with a simple "sh make.sh" on the systems
  that it's supported on. I.e., now you no longer need to tweak 
  text in the target-features.lisp-expr and symlinks in src/runtime/
  by hand, the make.sh takes care of it for you.
* The system is no longer so grossly inefficient when compiling code
  involving vectors implemented as general (not simple) vectors (VECTOR T),
  so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
  which dares to use the various sequence functions on non-simple 
  vectors, takes less of a performance hit.
  * There is now a primitive type predicate VECTOR-T-P
    to test for the (VECTOR T) type, so that e.g.
        (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
    can now be compiled with some semblance of efficiency. (The old code
    turned the type declaration into a full call to %TYPEP at runtime!)
  * AREF on (VECTOR T) is still not fast, since it's still compiled
    as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
    ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
    efficiently. (The old version made full calls to SUBTYPEP at runtime!)
  * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
    without making full calls to SUBTYPEP at runtime.
  (Some analogous efficiency issues for non-simple vectors specialized to
  element types other than T, or for non-simple multidimensional arrays,
  have not been addressed. They could almost certainly be handled the
  same way if anyone is motivated to do so.)
* The changes in array handling break binary compatibility, so
  *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
* (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
  instead of (VALUES T T).
* By following the instructions that Dan Barlow posted to sbcl-devel
  on 2 July 2000, I was able to enable primitive dynamic object
  file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
  functionality is not implemented (since it calls ld to resolve
  library references automatically, requiring RUN-PROGRAM for its
  implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
  try to resolve library references) is now supported.
* The system now flushes the standard output streams when it terminates,
  unless QUIT is used with the RECKLESSLY-P option set. It also flushes
  them at several other probably-convenient times, e.g. in each pass of
  the toplevel read-eval-print loop, and after evaluating a form given
  as an "--eval" command-line option. (These changes were motivated by a
  discussion of stream flushing issues on cmucl-imp in August 2000.) 
* The source transform for TYPEP of array types no longer assumes
  that an array whose element type is a not-yet-defined type 
  is implemented as an array of T, but instead punts, so that the 
  type will be interpreted at runtime.
* There is now some support for cross-compiling in make.sh: each of
  the phases of make.sh has its own script. (This should be transparent
  to people doing ordinary, non-cross-compile builds.)
* Since my laptop doesn't have hundreds of megabytes of memory like
  my desktop machine, I became more motivated to do some items on
  my to-do list in order to reduce the size of the system a little:
  ** Arrange for various needed-only-at-cold-init things to be
     uninterned after cold init. To support this, those things have
     been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
     symbols with such names are now uninterned after cold init).
  ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
     fasl files for cold load.
  ** Remove the old compiler structure pooling code (which used to 
     be conditional on the target feature :SB-ALLOC) completely.
  ** Redo the representation of some data in cold init to be more compact.
  (I also looked into supporting byte compiled code at bootstrap time,
  which would probably reduce the size of the system a lot, but that
  looked too complicated, so I punted for now.)
* The maximum signal nesting depth in the src/runtime/ support code has
  been reduced from 4096 to 256. (I don't know any reason for the very
  large old value. If the new smaller value turns out to break something,
  I'll probably just bump it back up.)
* PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
  as per ANSI.
* Many, many bugs reported by Peter Van Eynde have been added to
  the BUGS list; some have even been fixed.
* While enabling dynamic object file loading, I tried to make the 
  code easier to understand, renaming various functions and variables
  with less ambiguous names, and changing some function calling
  conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
* While trying to figure out how to do the OpenBSD port, I tried to 
  clean up some of the code in src/runtime/. In particular, I dropped
  support for non-POSIX signal handling, added various comments, 
  tweaked the code to reduce the number of compilation warnings, and
  renamed some files to increase consistency.
* To support the new automatic configuration functionality in make.sh,
  the source file target-features.lisp-expr has been replaced with the
  source file base-target-features.lisp-expr and the machine-generated
  file local-target-features.lisp-expr.
* fixed a stupid quoting error in make.sh so that using CMU CL
  "lisp -batch" as cross-compilation host works again

changes in sbcl-0.6.6 relative to sbcl-0.6.5:

* DESCRIBE no longer tries to call itself recursively to describe
  bound/fbound values, so that it no longer fails on symbols which are
  bound to themselves (like keywords, T, and NIL).
* DESCRIBE now works on generic functions.
* The printer now prints less-screwed-up representations of closures
  (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
* A private symbol is used instead of the :EMPTY keyword previously 
  used to mark empty slots in hash tables. Thus 
        (DEFVAR *HT* (MAKE-HASH-TABLE))
        (SETF (GETHASH :EMPTY *HT*) :EMPTY)
        (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
  now does what ANSI says that it should. (You can still get 
  similar noncompliant behavior if bang on the hash table
  implementation with all the symbols you get back from 
  DO-ALL-SYMBOLS, but at least that's a little harder to do.)
  This breaks binary compatibility, since tests for equality to 
  :EMPTY are wired into things like the macroexpansion of 
  WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
  implementations.
* There's now a minimal placeholder implementation for CL:STEP,
  as required by ANSI.
* An obscure bug in the interaction of the normal compiler, the byte
  compiler, inlining, and structure predicates has been patched
  by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
  :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
  mailing list).
* Missing ordinary arguments in a macro call are now detected even
  when the macro lambda list contains &KEY or &REST.
* The debugger no longer complains about encountering the top of the
  stack when you type "FRAME 0" to explicitly instruct it to go to
  the top of the stack. And it now prints the frame you request even
  if it's the current frame (instead of saying "You are here.").
* As specified by ANSI, the system now always prints keywords
  as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
  KEYWORD package.
* The default initial SIZE of HASH-TABLEs is now smaller.
* Type information from CLOS class dispatch is now propagated
  into DEFMETHOD bodies, so that e.g.
        (DEFMETHOD FOO ((X SINGLE-FLOAT))
          (+ X 123.0))
  is now basically equivalent to 
        (DEFMETHOD FOO ((X SINGLE-FLOAT))
          (DECLARE (TYPE SINGLE-FLOAT X))
          (+ X 123.0))
  and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
  operation, without having to do run-time type dispatch.
* The macroexpansion of DEFMETHOD has been tweaked so that it has
  reasonable behavior when arguments are declared IGNORE or IGNORABLE.
* Since I don't seem to be making big file reorganizations very often
  any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
  is overflowing my ability to conveniently back them up), I've finally
  checked the system into CVS. (The CVS repository is on my home system,
  not at SourceForge -- putting it on SourceForge might come later.)
* SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the 
  high-level GC-NOTIFY-FOO functions send their output. (There's
  still very little control of where low-level verbose GC functions
  send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
  less than it used to -- the GC-NOTIFY-FOO functions are now under
  the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
* The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
  in only one place in the source code, and propagates it automatically
  everywhere that it's needed. Thus e.g. when I bump the version from
  0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
* The C source files now include boilerplate legalese and documentation
  at the head of each file (just as the Lisp source files already did).
* At Dan Barlow's suggestion, the hyperlink from the SBCL website
  to his page will be replaced with a link to his new CLiki service.

changes in sbcl-0.6.5 relative to sbcl-0.6.4:

* Raymond Wiker's patches to port the system to FreeBSD have been merged.
* The build process now looks for GNU make under the default name "gmake",
  instead of "make" as it used to. If GNU make is not available as "gmake"
  on your system, you can change this default behavior by setting the 
  GNUMAKE environment variable.
* Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system 
  can build without error under CMU CL.

changes in sbcl-0.6.4 relative to sbcl-0.6.3:

* There is now a partial SBCL user manual (with some new text and some
  text cribbed from the CMU CL manual). 
* The beginnings of a profiler have been added (starting with the
  CMU CL profiler and simplifying and cleaning up). Eventually the
  main interface should be through the TRACE macro, but for now, 
  it's still accessed through vaguely CMU-CL-style functions and macros
  exported from the package SB-PROFILE.
* Some problems left over from porting CMU CL to the new
  cross-compilation bootstrap process have been cleaned up:
  ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
     instead of SB!XC:DEFMACRO, compounded by an oversight on my 
     part when getting rid of the compiler *BACKEND* stuff.)
  ** The value of *NULL-TYPE* was screwed up, because it was
     being initialized before the type system knew the final
     definition of the 'NULL type. This screwed up several key
     optimizations in the compiler, causing inefficiency in all sorts
     of places. (I found it because I wanted to understand why
     GET-INTERNAL-RUN-TIME was consing.) 
* fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
  PROCLAIM FTYPE information. Unfortunately this broke binary 
  compatibility again, since now the forms output by DEFGENERIC
  to refer to functions which didn't exist in 0.6.3.
* added declarations so that SB-PCL::USE-CACHING-DFUN-P 
  can use the new (as of 0.6.3) transform for SOME into MAP into 
  inline code
* changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
  values to (INTEGER 0 1000000), so that the time code will no longer
  occasionally get blown up by Linux returning 1000000 microseconds
* PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
  its output conform to the ANSI spec. (Alas, this makes its output
  uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
* A full call to MAP NIL with a single sequence argument no longer conses.
* fixes to problems pointed out by Martin Atzmueller:
  * The manual page no longer talks about multiprocessing as though
    it were currently supported.
  * The ILISP support patches have been removed from the distribution,
    because as of version 5.10.1, ILISP now supports SBCL without us
    having to maintain patches.
* added a modified version of Raymond Toy's recent CMU CL patch for
  EQUALP comparison of HASH-TABLE

changes in sbcl-0.6.3 relative to sbcl-0.6.2:

* The system still can't cross-compile itself with
  *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
  entails), but at least it can compile more of itself that way
  than it used to be able to, and various buglets which were uncovered
  by trying to cross-compile itself that way have now been fixed.
* This release breaks binary compatibility again. This time 
  at least I've incremented the FASL file format version to 2, so that the
  problem can be detected reliably instead of just causing weird errors.
* various new style warnings:
  ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
  ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
  ** using the lexical binding of a variable named in the *FOO* style
* DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
  as specified by ANSI.
* *RANDOM-STATE* is no longer automatically initialized from 
  (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
  default behavior of the system is to repeat its behavior every time
  it's run. If you'd like to change this behavior, you can always
  explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the 
  old convention there was no comparably easy way to get the system to 
  repeat its behavior every time it was run.
* Support for the pre-CLTL2 interpretation of FUNCTION declarations as
  FTYPE declarations has been removed, in favor of their ANSI
  interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
* The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
  the types of their sequence arguments can be determined at compile time.
  This is done through a new open code expansion for MAP which eliminates
  consing for (MAP NIL ..), and reduces consing otherwise, when sequence
  argument types can be determined at compile time.
* The optimizer now transforms COERCE into an identity operation when it 
  can prove that the coerced object is already of the correct type. (This 
  can be a win for machine generated code, including the output of other
  optimization transforms, such as the MAP transform above.)
* Credit information has been moved from source file headers into CREDITS.
* Source file headers have been made more standard.
* The CASE macro now compiles without complaining even when it has
  no clauses.

changes in sbcl-0.6.2 relative to sbcl-0.6.1:

* (Note that the way that the PCL macroexpansions were rewritten
  to accommodate the change in DEFGENERIC below breaks binary
  compatibility. That is, fasl files compiled under sbcl-0.6.1 may
  not run under sbcl-0.6.2. Once we get out of alpha releases,
  i.e. hit release 1.0.0, we'll probably try to maintain binary
  compatibility between maintenance releases, e.g. between sbcl-1.4.3
  and sbcl-1.4.4. Until then, however, it might be fairly common
  for maintenance releases to break binary compatibility.)
* A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
  has been fixed.
* The reporting of unhandled signals has been changed to print some
  explanatory text as well as the report form. (Previously only
  the report form was printed.)
* The macroexpansion for DEFGENERIC now DECLAIMs the function that
  it defines, so that the compiler no longer issues undefined function
  warnings for compiled-but-not-yet-loaded generic functions. 
* The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
  and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
  and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
  section of the ANSI spec are supported.
* The "" nickname for the "KEYWORD" package has been removed. 
  The reader still handles symbol tokens which begin with a package marker
  as keywords, but it doesn't expose its mechanism for doing so in the
  (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
* The system now issues STYLE-WARNINGs for contradictory TYPE 
  proclamations. (Warnings for contradictory FTYPE proclamations would
  be nice too, but those can't be done usefully unless the type system
  is made smarter about FUNCTION types.)
* The names of source files "*host-*.lisp" and "*target-*.lisp" have been 
  systematized, so that "*target-*.lisp is supposed to exist only on the
  target and imply that there's a related file which exists on the
  host, and *host-*.lisp is supposed to exist only on the host and imply
  that there's a related file which exists on the target. This involves a
  lot of renaming. Hopefully the acute confusion caused by the renaming
  will be justified by the reduction in chronic confusion..
  ** runtime-type.lisp    -> early-target-type.lisp
  ** target-type.lisp     -> late-target-type.lisp
  ** early-host-format.lisp -> early-format.lisp
  ** late-host-format.lisp -> late-format.lisp
  ** host-error.lisp      -> misc-error.lisp
  ** early-error.lisp     -> early-target-error.lisp
  ** late-error.lisp      -> late-target-error.lisp
  ** host-defboot.lisp    -> early-defboot.lisp
  ** code/misc.lisp       -> code/target-misc.lisp
  ** code/host-misc.lisp  -> code/misc.lisp
  ** code/numbers.lisp    -> code/target-numbers.lisp
  ** code/early-numbers.lisp -> numbers.lisp
  ** early-host-type.lisp -> early-type.lisp
  ** late-host-type.lisp  -> late-type.lisp
  ** host-typep.lisp      -> typep.lisp
  ** load.lisp            -> target-load.lisp
  ** host-load.lisp       -> load.lisp
  ** host-disassem.lisp   -> disassem.lisp
  ** host-insts.lisp      -> insts.lisp
  ** byte-comp.lisp       -> target-byte-comp.lisp
  ** host-byte-comp.lisp  -> byte-comp.lisp
  ** host-signal.lisp     -> signal.lisp
  ** host-defstruct.lisp  -> defstruct.lisp
  ** late-target-type.lisp -> deftypes-for-target.lisp
  Furthermore, several other previously target-only files foo.lisp (e.g.
  hash-table.lisp and random.lisp) have been split into a target-and-host
  foo.lisp file and a target-only target-foo.lisp file, with their key type
  definitions in the target-and-host part, so that the cross-compiler will
  know more about target types.
* DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have 
  gone away. In their place are various *BACKEND-FOO* variables
  corresponding to the slots of the old structure.
* A bug which caused the SB-COLD bootstrap-time package to be propagated
  into the target SBCL has been fixed.
* The chill.lisp system for loading cold code into a running SBCL
  now works better.
* Support for the CMU CL "scavenger hook" extension has been removed.
  (It was undocumented and unused in the CMU CL sources that SBCL was
  derived from, and stale in sbcl-0.6.1.)
* Various errors in the cross-compiler type system were detected
  by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
  (enabling various consistency checks). Many of them were fixed,
  but some hard problems remain, so the compiler is back to 
  running without *TYPE-SYSTEM-INITIALIZED* for now.
* As part of the cross-compiler type system cleanup, I implemented
  DEF!TYPE and got rid of early-ugly-duplicates.lisp.
* I have started adding UNCROSS calls throughout the type system
  and the INFO database. (Thus perhaps eventually the blanket UNCROSS
  on cross-compiler input files will be able to go away, and various
  kludges with it).
* CONSTANTP now returns true for quoted forms (as explicitly required
  by the ANSI spec).

changes in sbcl-0.6.1 relative to sbcl-0.6.0:

* changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
  fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
  (MAKE-LIST -1). In the longer term, it ought to become true
  that declarations are assertions even at SAFETY 1. For now, it's not
  quite true even at SAFETY 3, but it's at least more nearly true..
  (Note that this change seems to increases the size of the system by
  O(5%) and to decrease the speed of the compiler by 20% or more.)
* changed ALIEN printing to be much more abbreviated, as a short-term fix
  for the problem of printing dozens of lines of distracting information
  about low-level system machinery as part of the top stack frame
  on entry to the debugger when an undefined function was called.
* tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
  is not reset to COMMON-LISP-USER. 
* Compilation of stuff related to dyncount.lisp has been made conditional
  on the :SB-DYNCOUNT target feature, so that the ordinary core system is
  smaller. The various dyncount-related symbols have been moved into
  a new "SB-DYNCOUNT" package.
* tty-inspect.lisp has been renamed to inspect.lisp.
* unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
  feature has gone away. (When we eventually port to other flavors of 
  libc and/or Unix, we'll try to make the differences between flavors
  invisible at the user level.)
* Various other *FEATURES* tags, and/or their associated conditionals,
  have been removed if obsolescent, or given better documentation, or
  sometimes given more-mnemonic names.

changes in sbcl-0.6.0 relative to sbcl-0.5.0:

* tidied up "make.sh" script
* tidied up system directory structure
* better "clean.sh" behavior
* added doc/FOR-CMUCL-DEVELOPERS
* many many small tweaks to output format, e.g. removing possibly-confusing
  trailing #\. character in DESCRIBE-INSTANCE
* (EQUALP #\A 'A) no longer signals an error.
* new hashing code, including EQUALP hashing
* tidied up Lisp initialization and toplevel
* initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
* command line argument processing
* added POSIX-GETENV function to deal with Unix-ish environment variables
* more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
  terminating SBCL on EOF
* non-verbose GC by default
* There is no more "sbcl" shell script; the sbcl file is now the C
  runtime executable (just like CMU CL).
* removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and 
  FOP-POP-FOR-EFFECT
* tweaked debug-info.lisp and debug-int.lisp to make the debugger store
  symbol and package information as Lisp native symbol and package objects
  instead of strings naming symbols and strings naming packages. This way,
  whenever packages are renamed (as in warm init), debug information is
  transformed along with everything else.
* tweaked the optimization policy declarations which control the building
  of SBCL itself. Now, among other things, the system no longer saves
  source location debugging information. (This helps two problems at once
  by reducing SBCL size and by keeping SBCL from trying to look for its
  sources -- which may not exist -- when reporting errors.)
* added src/cold/chill.lisp, to let SBCL read its own cold sources for
  debugging and testing purposes
* cleaned up printing, making the printer call PRINT-OBJECT for
  instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
  methods, giving nearly-ANSI behavior
* converted almost all special variables to use *FOO* naming convention
* deleted PARSE-TIME functionality, since it can be done portably
* moved some files out of cold init into warm init
* deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
  with (VALUES) forms
* regularized formatting of source files
* added an install.sh script
* fixed ridiculous memory usage of cross-compiler by making
  compiler/alloc.lisp not try to do pooling unless it can hook
  itself into the GC of the cross-compilation host. Now the system
  builds nicely on my old laptop.
* added :SB-ALLOC in target-features.lisp-expr
* deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
  was not implemented)
* re-did condition handling and note reporting in the compiler. Notes
  are no longer handled by signalling conditions. Style warnings
  and warnings are handled more correctly and reported in such a way
  that it's easy to find one or the other in your output (so that you
  can e.g. figure out which of many problems caused COMPILE-FILE to 
  return FAILURE-P).
* changed the severity of several compiler warnings from full WARNING
  to STYLE-WARNING in order to conform with the ANSI spec; also changed
  compiler note reporting so that it doesn't use the condition system
  at all (and hence affects neither FAILURE-P nor WARNINGS-P in the 
  COMPILE-FILE command)
* made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
  function. As a consequence, START-BLOCK and END-BLOCK declarations are
  no longer supported, since their implementation was deeply intertwingled
  with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
* removed bogus "support" for compiler macros named (SETF FOO), and
  removed the compiler macro for SETF INFO (but only after making a fool
  of myself on the cmucl-imp mailing list by posting a bogus patch for
  DEFINE-COMPILER-MACRO..)
* Compiled files containing forms which have side effects on the Lisp
  reader (such as DEFPACKAGE forms) are now handled more correctly.
  (Compiler queuing of top level lambdas has been suppressed by setting
  *TOP-LEVEL-LAMBDA-MAX* to 0. )
* deleted various currently-unused source files, e.g. gengc.lisp. They
  may be added back at some point e.g. when porting to other architectures,
  but until they are it's distracting to distribute them and to try to
  maintain them.
* deleted "UNCROSS couldn't recurse through.." style warnings, since 
  there were so many of them they're just distractions, and UNCROSS is
  known to be able to handle the current sources
* moved PROFILE functionality into TRACE, so that it will be clear
  how the wrapping and unwrapping of functions when you profile them
  interacts with the wrapping and unwrapping of functions when you
  trace them. (Actually, the functionality isn't there yet, but at least
  the interface specification is there. Hopefully, the functionality will
  arrive with some future maintenance release.)
* removed host-oops.lisp
* changed signature of QUIT function to allow UNIX-CODE argument
* fixed READ-SEQUENCE bug
* tweaked verbose GC output so that it looks more like the progress
  output that ANSI specifies for functions like LOAD
* set up the system on sourceforge.com, with home pages, mailing lists, etc.
* added <http://sbcl.sourceforge.com> to the banner information printed by
  the sbcl executable