1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.16:
3 * bug fix: two potential GC deadlocks affecting threaded builds.
4 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
5 returns T (reported by Anton Kazennikov)
6 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
7 with non-variable places
8 * improvement: the debugger will now also display local variables that
9 are only used once, for code compiled with a DEBUG optimization quality
12 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
13 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
14 * incompatible change: External-format support for FFI calls. The
15 SB-ALIEN:C-STRING no longer implies an ASCII
16 external-format. Instead the string is subject to external-format
17 conversion. Additionally return values of type SB-ALIEN:C-STRING
18 are by default of type (SIMPLE-ARRAY CHARACTER), not
19 SIMPLE-BASE-STRING. For an alien type that behaves like the old
20 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
21 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
22 * incompatible change: SB-EXT package no longer contains the
23 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
24 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
26 * incompatible change: the single-stepper is no longer available
27 on Alpha, Mips and Sparc platforms.
28 * minor incompatible change: the direct superclasses of
29 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
30 not (STANDARD-OBJECT FUNCTION). This makes the
31 class-precedence-lists of GENERIC-FUNCTION and
32 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
34 * minor incompatible change: the default stream external format on
35 non-unicode SBCL is no longer detected from the locale, but is always
37 * new feature: Add a version of evaluator that uses an interpreter instead
38 of the compiler. EVAL still uses the compiler by default, to switch it
39 to use the interpreter, set the value of the variable
40 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
41 * minor incompatible change: the single-stepper REPL has been merged
42 with the normal debugger (see the "Stepping" heading of the debugger help
43 for more details). The debugger command STEP will no longer switch
44 to the single-stepper REPL.
45 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
47 * bug fix: loading of default sysinit file works. (thanks to Leonid
49 * bug fix: better detection of circularities in the file-compiler.
50 (reported by Marco Monteiro)
51 * bug fix: the CL pathname functions now work with files that have
52 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
53 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
54 systems with Unix98 pty semantics.
55 * bug fix: ASDF-INSTALL will now work with bsd tar.
56 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
58 * bug fix: timers expiring in dead threads no longer cause a
59 type-error (reported by Paul "Nonny Mouse").
60 * bug fix: thanks to more lightweight single-stepper instrumentation,
61 code compiled with (DEBUG 3) will compile and execute significantly faster,
62 and will have more accurate type-inferencing than before
63 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
64 whose bindings are modified
65 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
66 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
67 ** replace dummy implementations of CL:MACHINE-INSTANCE and
68 CL:SOFTWARE-VERSION with proper version
70 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
71 * feature: implemented the READER-METHOD-CLASS and
72 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
74 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
76 * optimization: faster LOGCOUNT implementation on x86 and x86-64
77 (thanks to Lutz Euler)
78 * optimization: hashing of general arrays and vectors has been
79 improved. (reported by Any Fingerhut)
80 * enhancement: SB-INTROSPECT is now able to find definitions of
81 profiled functions. (thanks to Troels Henriksen)
82 * enhancement: compiler-macro expansion applies now to FUNCALL forms
83 as well, allowing compiler-macros for SETF-functions to expand.
84 * enhancement: step-instrumentation no longer wraps known
85 single-value functions with multiple-value context, allowing
86 better type inference.
87 * fixed bug #337: use of MAKE-METHOD in method combination now works
88 even in the presence of user-defined method classes. (reported by
89 Bruno Haible and Pascal Costanza)
90 * fixed bug #339(c): if there are applicable methods not part of any
91 long-form method-combination group, call INVALID-METHOD-ERROR.
92 (reported by Bruno Haible)
93 * fixed bug #361: the :FUNCTION initarg in the protocol for
94 initialization of methods can now be used to override
95 internally-produced optimized functions. (reported by Bruno
97 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
98 system-provided lambda expression no longer cause warnings about
99 unbound #:|pv-table| symbols.
100 * bug fix: improved the handling of type declarations and the
101 detection of violations for keyword arguments with non-constant
103 * bug fix: potentially erronous calls to PATHNAME and
104 MERGE-PATHNAMES were being flushed in some cases.
105 (reported by Richard Kreuter)
106 * bug fix: compiled calls to TYPEP were mishandling obsolete
107 instances. (reported by James Bielman and Attila Lendvai)
108 * bug fix: format strings with ~> without matching ~< no longer
109 trigger an AVER, but signal an understandable error instead.
110 (reported by Antonio Martinez)
111 * bug fix: specifying an output-file without a pathname-type for
112 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
113 of input-file instead of "fasl". (reported by Robert Dodier)
114 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
115 (reported by James Y Knight).
116 * bug fix: compiler-macros-function did not consider the environment
117 argument for shadowing by local functions.
118 * bug fix: compiler-macros expansion was inhibited by local INLINE
120 * bug fix: inline expansions of known functions were subject to
121 step-instrumentation in high debug policies, leading to problems
123 * bug fix: compiler failed to differentiate between different CONS
125 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
126 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
127 * bug fix: #368: incorrect use of expressed vs. upgraded array
129 * bug fix: #306a: more precise unions of array types.
130 * thread-safety improvements:
131 ** CONDITION-WAIT could return early on Linux, if the thread was
132 interrupted and subsequently continued with SIGCONT.
133 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
135 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
136 * added support for the ucs-2 external format. (contributed by Ivan
138 * minor incompatible change: pretty printing of objects of type
139 (cons symbol) is, in the default pprint-dispatch-table, now
140 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
142 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
143 called later in a class's lifetime, possibly as late as when the
144 first instance of the class is created. Previously,
145 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
146 class became finalizeable.
147 * fixed bug: FILE-POSITION sometimes returned inconsistent results
148 for multibyte external-format streams. (thanks to "vbzoli")
149 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
150 with :ALLOCATION :CLASS inherited from superclasses of the
152 * fixed bug: anonymous classes can now be created using the :NAME
153 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
154 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
155 * fixed bug: core-files saved with :EXECUTABLE T can again be
156 executed when SBCL_HOME isn't set. (reported by James Knight)
157 * fixed bug: toplevel LOCALLY forms with declarations could
158 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
159 * fixed bug: printing from several different threads using different
160 values of *print-case* could cause invalid output, due to
161 some internal special variables of the printer not being bound
162 thread-locally (reported by Max Mikhanosha)
163 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
164 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
165 (reported by Pascal Costanza)
166 * fixed bug: FINALIZE-INHERITANCE is called from
167 REINITIALIZE-INSTANCE on classes when the class has previously
168 been finalized, as required by AMOP.
169 * minor code generation optimizations:
170 ** better register allocation in CLOS dispatching functions
171 ** overflow detection when coercing signed bytes to fixnums on x86-64
172 ** is now implemented with one IMUL instruction instead of three shifts
173 ** more efficient bit-vector access on x86 and x86-64
174 ** more efficient access to raw structure slots on x86 and x86-64
175 * fixed some bugs revealed by Paul Dietz' test suite:
176 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
179 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
180 * feature: thread support on Solaris/x86, and experimental thread support
182 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
184 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
185 default method for SLOT-UNBOUND.
186 * minor incompatible change: prevent the user from specializing the
187 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
188 somewhat counter to the intent of the protocol, I (CSR) think, and
189 additionally it just doesn't work in SBCL as currently
190 implemented, thanks to optimizations (that are always valid for
191 the other three SLOT-VALUEish functions, but not for the setter).
192 * minor incompatible change: the :SB-LDB feature is now enabled by
193 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
194 the low-level debugger.
195 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
196 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
197 down to an SB-SYS:FD-STREAM, allowing direct communication in
198 more cases. Temporary files and pipes are still used as a fallback
200 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
201 on the non-reentrant C functions localtime(3) and gmtime(3).
202 * bug fix: global optimization policy was not visible in LOCALLY and
204 * bug fix: class objects can be used as specializers in methods.
205 (reported by Pascal Costanza)
206 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
207 with a type error. (reported by blitz_ on #lisp)
208 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
209 failed with a type error. (reported by Pascal Bourguignon)
210 * bug fix: merging pathnames against defaults with :DIRECTORY
211 starting with '(:RELATIVE :BACK) should preserve the :BACK.
212 (reported by James Y Knight)
213 * bug fix: saving large (>2GB) cores on x86-64 now works
214 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
215 constant index and a (simple-array (signed-byte 32)) array
216 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
217 error (patch by Robert J. Macomber)
218 * feature: TIME now displays the amount of run-time spent in GC
219 * bug fix: The debugger now does a better job of respecting
220 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
221 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
222 workaround for bug 403.)
223 * bug fix: floating point exception handling now works on FreeBSD
224 (thanks to NIIMI Satoshi)
225 * fixed some bugs revealed by Paul Dietz' test suite:
226 ** MISC.641: LET-conversion were not supposed to work in late
228 * improvements to the Win32/x86 port:
229 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
230 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
232 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
233 * new feature: source path information is generated for macro-expansion
234 errors for use in IDE's like Slime (thanks to Helmut Eller)
235 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
236 * bug fix: compiler does not loop forever on an invalid type in
238 * improvement: compilation of most CLOS applications is significantly
240 * optimization: added a limited bytecode compiler for simple toplevel
241 forms, speeding up compilation and FASL loading
242 * bug fix: the statistical profiler now properly distinguishes anonymous
245 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
246 * minor incompatible change: in sbcl-0.9.11 (but not earlier
247 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
248 into subclasses of the finalized class. Now user calls to
249 FINALIZE-INHERITANCE finalize just the one class, and calls by the
250 system return before any subclasses are finalized.
251 * minor incompatible change: The reader no longer ignores errors
252 regarding non-existent packages in #+ and #- feature tests.
253 * new feature: command line options --no-sysinit, --no-userinit to
254 inhibit loading the corresponding init files
255 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
256 for sending data through UDP sockets (thanks to François-René Rideau)
257 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
258 error is signalled instead (thanks to François-René Rideau)
259 * minor incompatible change: Error signalling behaviour of lexical
260 operations violating package locks has changed slightly. Refer to
261 documentation on package locks for details.
262 * enhancement: EVAL can process IF-expressions without resorting to the
264 * bug fix: Some operations involving SIN, COS, or TAN weren't being
265 constant-folded properly due to missing out-of-line functions.
266 (reported by Mika Pihlajamäki)
267 * bug fix: LISTEN sometimes returned T even in cases where no data was
268 immediately available from the stream
269 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
270 were reversed. (reported by Levente Mészáros)
271 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
272 when given a vector displaced to an adjustable array. (reported
274 * bug fix: derivation of float boundaries from numbers outside the
275 appropriate float range (reported by John Wiseman)
276 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
278 * bug fix: precomputing cache entries for generic functions with
279 some subclasses of specializers as yet invalid does not attempt to
280 fill a cache line with a negative offset. (reported by Levente
282 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
283 allow condition class objects as arguments to DOCUMENTATION and
284 (SETF DOCUMENTATION); only find and set documentation for
285 structure names for the STRUCTURE doc-type. (suggested by Gary
287 * improvements to the Win32/x86 port:
288 ** (user-homedir-pathname) and default initialization file
289 locations now know about the user's "Documents and Settings"
290 directory (thanks to Yaroslav Kavenchuk)
291 ** run-program is implemented (thanks to Mike Thomas)
292 ** sockets support (thanks to Timothy Ritchey)
293 ** better backtrace support (thanks to Alastair Bridgewater)
294 ** sb-grovel supported
295 ** asdf-install and sb-posix work somewhat
296 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
297 * improvements to the Solaris/x86 port:
298 ** works on Solaris 11/Solaris Express
299 ** floating-point exception handling support
300 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
301 * fixed some bugs revealed by Paul Dietz' test suite:
302 ** REMOVE-METHOD returns its generic function argument even when
303 the method is not one of the generic functions' methods.
304 ** objects declared as MEMBER types can be admissible arguments to
306 ** printing characters should simply be printed by the FORMAT ~:C
308 ** compiler failure when compiling functions with hairy constant
309 defaults for optional parameters.
310 ** compiler produces wrong code when MAYBE-INLINE-expanding a
311 function, which is already optimized.
313 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
314 * new platform: experimental support for SBCL x86/Darwin, including
315 MacOS X 10.4.5 on Intel.
316 * new feature: Unicode character names are now known to the system
317 (through CHAR-NAME and NAME-CHAR).
318 * new feature: the filesystem location of SBCL's core file is
319 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
320 * minor incompatible change: the contrib modules SB-POSIX and
321 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
322 this change is to make it easier to distribute
323 location-independent binaries.
324 * bug fix: as implied by AMOP, standardized classes no longer have
325 slots named by external symbols of public packages. (reported by
327 * optimization: calling functions via constant symbols -- as in
328 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
329 via the function object as in (FUNCALL #'FOO).
330 * optimization: on x86 and x86-64, the calling convention for the
331 case of dealing with an unknown number of values has been altered
332 to be friendlier to the prediction heuristics implemented,
333 particularly with reference to CALL and RET pairing. (thanks to
334 Alastair Bridgewater)
335 * optimization: on x86, the code for access to thread-local symbol
336 values and binding/unbinding of thread-local symbols is smaller.
337 (thanks to Alastair Bridgewater)
338 * enhancement: CONSTANTP is now able to determine constantness of
339 more complex forms, including calls to constant-foldable standardized
340 functions and some special forms beyond QUOTE.
341 * fixed bug: occasional GC crashes on Solaris/x86
342 * optimization: x86-64 supports stack allocation of results of simple
343 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
344 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
345 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
346 (thanks to James Knight)
347 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
348 by the standard (thanks to Stephen Compall)
350 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
351 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
352 be used for bundling the runtime and the core file into one
353 executable binary. This feature is not currently supported on all SBCL
354 platforms. (thanks to James Bielman and NIIMI Satoshi)
355 * new feature: a generational or ephemeral garbage collector is now
356 the default on the PowerPC platform (both Linux and Darwin). The
357 old Cheney (stop and copy) collector is a build-time option.
358 (thanks to Cyrus Harmon, after Raymond Toy)
359 * minor incompatible change: the method by which SBCL finds its
360 contributed modules has changed; it no longer relies on symbolic
361 links from an $SBCL_HOME/systems directory, but searches directly
362 in the subdirectories of $SBCL_HOME.
363 * enhancement: the dynamic heap size on the Linux/PPC platform is
364 markedly larger, even using the older Cheney garbage collector.
365 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
366 certain complicated string types. (reported by Gary King)
367 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
368 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
369 * fixed bug: DOCUMENTATION on structures no longer signals an
370 error if no structure type of the right name exists. (reported by
372 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
373 applicable non-standard primary or :AROUND method on
374 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
375 no longer cause unbound variable errors. (reported by Kevin Reid)
376 * fixed bug: metacircle resolution in cases where methods have slots
377 added before the slots from STANDARD-METHOD. (reported by Jean
379 * fixed bug: the Power PC assembly code for calling into lisp
380 sometimes computed its return address wrongly (depending on the
381 15th bit of the address, fixed at link time). This used to
382 manifest itself in a segmentation violation while building PCL.
383 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
384 many others over the years)
385 * fixed bug: ROOM no longer reports silly numbers for stack usage if
386 the stack pointer (treated as a signed integer) is negative.
387 (thanks to Peter van Eynde)
389 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
390 * new platform: experimental support for the Windows operating
391 system has been added. (thanks to Alastair Bridgewater)
392 * fixed several bugs in and robustified the PPC FFI (including
393 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
394 * bug fix: allow non-simple string symbol names (reported by Paul
396 * bug fix: interrupt handling on NetBSD (thanks to Richard M
398 * bug fix: saving a core corrupted callbacks on x86/x86-64
399 * bug fix: closed a loophole in metacircularity detection and
400 grounding in the PCL implementation of CLOS.
401 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
402 (reported by tomppa on #lisp)
403 * bug fix: clear the direction flag on Lisp -> C transitions, as
404 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
405 (reported by Andrej Grozin and Hendrik Maryns)
406 * optimization: major improvements to GC efficiency on GENCGC platforms
407 * optimization: faster implementation of EQUAL
408 * optimization: emit more efficient opcodes for some common
409 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
410 * optimization: several other minor code-generation improvements on x86-64
411 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
413 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
414 * minor incompatible change: (SETF CLASS-NAME) and (SETF
415 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
416 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
417 by ANSI as a generic function, and for consistency (SETF
418 GENERIC-FUNCTION-NAME) is treated likewise.
419 * fixed bug #233.b: make constraint propagation notice when a variable
420 value is changed after it is referenced but before it is used
421 * fixed bug #296: no more arbitrary behaviour from filenames with
422 odd characters as --load arguments. (reported by Adam Warner)
423 * fixed bug #390: :CHARACTER-SET pathname components now work as
424 expected. (reported by Tim Daly Jr)
425 * fixed bug #391: complicated :TYPE intersections in slot
426 definitions no longer cause an error in PCL internals.
427 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
428 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
429 returning the number of octets which would be written to the
430 file-stream. (thanks to Robert J. Macomber)
431 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
432 arguments to the internal stream functions. (thanks to David
434 * bug fix: stack exhaustion is now handled on the x86/SunOS
435 platform. (thanks to Daisuke Homma)
436 * bug fix: more accurate ROOM results on GENCGC platforms
437 * optimization: improved type inference for arithmetic-for
438 index variables in LOOP
439 * optimization: faster floating-point SQRT on x86-64
440 * fixed some bugs revealed by Paul Dietz' test suite:
441 ** DOCUMENTATION returns NIL instead of "" for method combinations
442 that don't have a docstring
444 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
445 * minor incompatible change: (SETF CLASS-NAME) and (SETF
446 GENERIC-FUNCTION-NAME) are no longer generic functions, and
447 therefore are not customizeable by user code (as seems to be at
448 least permitted and maybe required by AMOP). As a consolation,
449 however, the SBCL implementation of these functions now calls
450 REINITIALIZE-INSTANCE as specified by AMOP.
451 * bug fix: slot-definition documentation is propagated as per ANSI
452 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
453 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
454 argument, as do eslotd initialization methods. (from Pascal
455 Costanza's "Closer" project)
456 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
457 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
459 * bug fix: it is now possible to have more than one subclass of
460 STANDARD-GENERIC-FUNCTION without causing stack overflow.
461 (reported by Bruno Haible, Pascal Costanza and others)
462 * bug fix: the dependent update protocol now works for generic
463 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
465 * bug fix: condition-class instances corresponding to
466 DEFINE-CONDITION forms are now created eagerly. (reported by
467 Kalle Olavi Niemitalo on comp.lang.lisp)
468 * bug fix: floating point printing is more accurate in some
469 circumstances. (thanks to Simon Alexander)
470 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
471 merged with *DEFAULT-PATHNAME-DEFAULTS*.
472 * bug fix: callbacks on OS X now preserve stack-alignment.
473 (thanks to Cyrus Harmon)
474 * enhancement: the x86-64 disassembler is much better at
475 disassembling SSE instructions. (thanks to Lutz Euler)
476 * enhancement: CHECK-TYPE on a variable now allows type-inference
477 (thanks to Vincent Arkesteijn)
478 * optimization: improved performance of EUC-JP external format.
479 (thanks to NIIMI Satoshi)
480 * optimization: performance improvements to IO on file streams of
481 :ELEMENT-TYPE CHARACTER
482 * optimization: much faster memory allocation on x86-64
483 * optimization: faster garbage collections (latency and throughput)
485 * optimization: faster bignum subtraction and division on x86 and x86-64
486 (thanks to Lutz Euler)
487 * optimization: more accurate type-inference for loops that use a
488 floating point index variable or a negative step.
490 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
491 * bug fix: add a workaround to SBCL looping infinitely at startup on
492 Linux kernels with apparently buggy implementations of personality().
493 (thanks to Svein Ove Aas)
494 * bug fix: Unicode symbols are correctly printed in LDB backtraces
495 (thanks to David Lichteblau)
496 * bug fix: local bindings shadow global symbol macros for
497 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
498 * optimization: non-open coded uses of numeric comparison operators
499 (e.g. >) no longer cons when called with more than one parameter
500 on platforms supporting dynamic-extent allocation.
501 * enhancement: saving cores with foreign code loaded is now supported
502 on MIPS/Linux in addition to the previously supported platforms.
503 * bug fix: FIND-RESTART now tests for activity, not applicability when given
504 a restart object as identifier. (reported by Helmut Eller for CMUCL)
505 * bug fix: division by zero in sb-sprof when no samples were collected
506 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
507 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
508 (thanks to Alexander Kjeldaas)
509 * new feature: ignoring the return values of destructive functions
510 like NREVERSE signals a compile-time style-warning.
511 (thanks to Kevin Reid)
512 * new feature: the alignment of alien structure fields can be
513 explicitly specified. (thanks to Cyrus Harmon)
514 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
515 and dump core on SIGQUIT
517 ** incompatible change: threads do not inherit values of specials
518 from their parents (see manual)
519 ** bug fix: threads stacks belonging to dead threads are freed by the
520 next exiting thread, no need to gc to collect thread stacks anymore
521 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
522 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
523 ** bug fix: dynamic variable and thread start related gc lossage
524 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
526 * fixed some bugs revealed by Paul Dietz' test suite:
527 ** SUBTYPEP is slightly more accurate on heinously complicated
528 CONS types where some of the members have uncertain (in the
529 NIL, NIL sense) type relationships to each other.
531 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
532 no longer purified unless :PURIFY T is explicitly specified.
533 ** Non-purified cores are significantly smaller than before
535 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
536 * new feature: timers based on Zach Beane's excellent timer package
537 * added support for the following external formats: koi8-u,
538 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
539 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
540 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
541 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
542 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
543 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
544 * incompatible change: a threaded SBCL will no longer revert to
545 non-threaded mode on non-NPTL systems, but refuse to start entirely.
546 * bug fix: interrupts are disabled until startup is complete; no
547 more sigsegvs when receiving a signal to soon
548 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
550 * bug fix: add a workaround for the memory randomization features in
551 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
552 workaround will only be in effect on systems with the proc filesystem
554 * bug fix: printing objects of type HASH-TABLE signals a
555 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
557 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
558 have been running for over 50 days. (reported by Gilbert Baumann)
559 * bug fix: the logic for getting names of functions gets less
560 confused when confronded with alternate-metaclass
561 funcallable-instances. (reported by Cyrus Harmon)
562 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
563 consistent, even on internal alternate-metaclass objects.
564 * bug fix: flush closure information collected by physical
565 environment analysis prepass before the main pass. (bug reported
567 * bug fix: compiler pack phase does not modify a hashtable, which is
568 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
569 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
570 may be expanded. (reported by Alexander <malishev> on c.l.l)
571 * bug fix: no more style-warnings when compiling code using LAST
572 under high-SPEED low-SPACE settings. (reported by David Wallin
574 * fixed some incompatibilities between SBCL's MOP and the MOP
576 ** the METAOBJECT class is now implemented;
577 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
578 STANDARD-OBJECT, as required;
579 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
580 now compatible (as required by VALIDATE-SUPERCLASS); there
581 remains a consistency requirement over the presence or absence
582 of the FUNCTION class in the superclass list of finalized
583 classes; see the manual for more details;
584 ** the :AROUND method for COMPUTE-SLOTS on
585 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
586 requested slot ordering.
588 ** bug fix: parent thread now can be gc'ed even with a live
590 ** bug fix: binding a special with PROGV to no value is not
591 the same as that symbol not having been bound (thanks to
593 ** bug fix: binding specials is thread safe (thanks to
595 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
596 and SB-SPROF do not die with 'no handler for signal XX in
597 interrupt_handle_now(..)' anymore
598 ** bug fix: WITH-TIMEOUT works with multiple running threads
599 * fixed some bugs revealed by Paul Dietz' test suite:
600 ** ENSURE-GENERIC-FUNCTION should take a method class object for
601 the :method-class keyword argument.
603 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
604 * new port: the Solaris operating system on x86 processors is now
605 mostly supported, though some rough edges in the environment
606 remain. (thanks to Daisuke Homma)
607 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
608 space for its heap. (thanks to Thiemo Seufer)
609 * optimization: SBCL on MIPS platforms now supports dynamic-extent
610 closures. (thanks to Thiemo Seufer)
611 * minor incompatible change: eof selects abort in the debugger.
612 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
613 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
615 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
616 deprecated, and will go away in a future revision of SBCL.
617 * minor incompatible change: GC-ON and GC-OFF are no longer
618 implemented with a counter, it does not matter how many times gc
619 is switched on or off
620 * bug fix: discriminating functions for generic function classes
621 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
622 make invalid assumptions about method precedence order. (reported
624 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
625 platforms. (thanks to Thiemo Seufer)
626 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
627 * bug fix: correct pathname printing with printer escaping is on.
628 (thanks to Kevin Reid)
629 * bug fix: complex VOP definitions in "user-space" no longer trigger
630 package locks. (reported by Zach Beane)
631 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
632 for funcallable-instance functions, and is no different from
634 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
635 characters in the current readtable. (reported by Nicholas Neuss)
636 * bug fix: SUBTYPEP on various CONS types returns more a more
637 accurate acknowledgment of its certainty.
638 * optimizations: REMOVE-DUPLICATES now runs in linear time on
639 lists in some cases. This partially fixes bug 384.
640 * flush all standard streams before prompting in the REPL and the
642 * bug fix: signal handling and triggering gc do not conflict
643 directly anymore, in particular a high frequency sb-sprof does
644 not prevent gc from running
645 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
646 approximation for timezone and DST information between the
647 universal time epoch and the smallest negative 32-bit time_t.
648 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
649 year 1899 so long as the encoded time is non-negative.
650 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
651 the correct answer for non-constant rotate amounts when performing
652 an inline 32-bit rotation.
654 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
655 there is only one thread in the session
656 ** bug fix: memory leak for streams created in one thread and
657 written to in another
658 ** bug fix: lockup when compiled with gcc4
659 ** bug fix: race that allows the gc to be triggered when gc is
661 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
662 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
664 ** bug fix: finished threads are gc'ed properly
665 * fixed some bugs revealed by Paul Dietz' test suite:
666 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
667 has arguments with a different set of applicable methods from
668 the orignal arguments.
669 ** The type error thrown by MAP now has a correct expected-type
671 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
672 name a compiled function.
673 ** Three MISC tests where a large bignum was improperly coerced to
674 a float that couldn't represent that bignum during type
675 derivation were fixed.
676 ** SUBTYPEP can now handle types involving the negation of a
677 list-form FUNCTION type.
678 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
679 as (COMPLEX (AND RATIO (NOT FIXNUM))).
680 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
682 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
683 * New feature: Experimental support for bivalent streams: streams
684 opened with :element-type :default now allow character and binary
685 (unsigned-byte 8) I/O
686 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
687 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
689 * bug fix: no more highly sporadic "couldn't check whether ~S is
690 readable" when reading a stream and an interrupt hits in the middle
691 of a select system call
692 * compiler better recognizes complex arrays (reported by Eduardo
694 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
695 for variables on SBCL built with linkage-tables. (reported by Luis
697 * various error reporting improvements.
698 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
699 (Thanks to Thiemo Seufer)
700 * bug fix: nested reader invokations work correctly
701 * bug fix: it is possible to have simultaneous references to foreign
702 code and foreign data with the same name.
704 ** added x86-64 support
705 ** incompatible change: the threading api now works with thread
706 objects instead of thread ids
707 ** bug fix: threads are protected from signals and interruption when
708 starting up or going down
709 ** bug fix: a race where an exiting thread could lose its stack to gc
710 ** bug fix: don't halt on infinite error in threads if possible
711 ** fixed numerous gc deadlocks introduced in the pthread merge
712 ** bug fix: fixed thread safety issues in read and print
713 ** bug fix: debugger doesn't hang on session lock if interrupted at
714 an inappropriate moment
715 ** bug fix: run-program is now thread safe(r)
716 ** bug fix: inner with-recursive-lock no longer releases the mutex
717 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
718 * fixed some bugs revealed by Paul Dietz' test suite:
719 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
720 EXPECTED-TYPE slots filled.
721 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
723 ** MAKE-CONDITION accepts classes as type-designators.
724 ** COMPILE may never return NIL.
725 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
726 range before calling Unix time functions
728 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
729 * numerous signal handling fixes to increase stability
730 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
731 * minor incompatible change: we now correctly canonize default
732 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
733 as per the MOP, rather than the historical (INITARG INITFUNCTION
734 INITFORM). (reported by Bruno Haible)
735 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
736 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
737 for more information.
738 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
739 pathname is a directory pathname.
740 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
741 * dynamic space size on PPC has been increased to 768Mb. (thanks to
743 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
744 :METACLASS argument in addition to a class name. (reported by
745 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
746 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
747 (reported by Bruno Haible; patch from Gerd Moellmann)
748 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
750 * bug fix: more cleanups to the floating point exception handling on
751 x86-64 (thanks to James Knight)
752 * bug fix: the compiler does not try to constant fold calls of
753 COERCE and BIT-* functions when they return freshly constructed
754 objects. (reported by Nikodemus Siivola and Paul Dietz)
755 * optimization: improved the MIPS versions of generic (in the
756 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
757 * optimization: direct conversion of (unsigned-byte 32) to floats on
758 the PowerPC platform.
759 * optimization: structure instances with raw slots now use less
760 memory, and probably show better memory locality. (thanks to
762 * optimization: DYNAMIC-EXTENT declarations for lists and closures
763 are treated as requests for stack allocation on the x86-64,
764 Alpha-32, PPC and SPARC platforms.
765 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
766 the wrong file descriptor; implementation of SOCKET-OPEN-P.
767 (thanks to Tony Martinez)
769 ** gcing a dead thread can no longer lead to lockups
770 ** threads block signals until they are set up properly
771 ** errno is no longer shared by threads
772 ** interrupt-thread restores the eflags register on x86
773 ** fixed some lockups due to gc/thread interaction
774 * fixed some bugs revealed by Paul Dietz' test suite:
775 ** invalid dotted lists no longer raise a read error when
777 ** signal an error if a symbol that names a declaration is used
778 as the name of a type, or vice versa
779 ** allow using the (declare (typespec var*)) abbreviation for
780 (declare (type typespec var*)) with all type specifiers
781 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
782 ** always return NIL from PPRINT-POP when OBJECT is NIL
783 ** don't signal errors when pretty-printing malformed LABELS,
784 FLET or MACROLET forms
785 ** declarations in a DOLIST body are also in scope for the
787 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
789 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
792 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
793 * fixed cross-compiler leakages that prevented building a 32-bit
794 target with a 64-bit host compiler.
795 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
796 opened with :IF-EXISTS :OVERWRITE.
797 * fixed bug 281: error for an invalid qualifier in a short-form method
798 combination method is not signalled until the faulty method is called.
799 * bug fix: iteration variable type inferrer failed to deal with open
800 intervals. (reported by Alan Shields)
801 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
802 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
803 or more runtime options were provided to the sbcl binary.
804 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
805 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
806 * optimization: there's now a fast-path for fixnum arguments in the
807 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
808 * optimization: the code generated on x86-64 is more compact thanks
809 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
810 * bug fix: floating-point exception handling now partly works on
811 x86-64. (Thanks to James Knight)
812 * improvement to the MIPS backend from Thiemo Seufer: C-style
813 64-bit long long arguments and return values to alien functions
815 * fixed some bugs revealed by Paul Dietz' test suite:
816 ** the type-error signalled from WARN has a filled-in DATUM slot.
817 ** the type-error required when a stream is not associated with
818 a file has the stream as its datum.
819 ** type-errors on single-floats on x86-64 no longer have
820 :INVALID-OBJECT as the datum
821 ** the type-errors signalled for invalid function names now have
822 a correct expected type
823 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
824 for structures defined with a :TYPE. Documentation strings for
825 typed structures are no longer immediately discarded
826 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
827 broadcast streams and synonym streams. FILE-LENGTH now also works
828 on broadcast streams.
830 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
831 * incompatible change: the --noprogrammer option, deprecated since
832 version 0.7.5, has been removed. Please use the equivalent
833 --disable-debugger option instead.
834 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
836 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
837 inoperational for a while now) has been completely removed.
838 * null lexical environments are now printed as #<NULL-LEXENV>,
839 significantly reducing the amount of clutter in typical backtraces.
840 * documentation on weak pointers, finalization, and after GC hooks
841 has been added to the manual.
842 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
843 as other specialized array types, is much faster. SUBSEQ and
844 COPY-SEQ on such arrays have also been sped up.
845 * optimization: EQL is now more efficient when at least other argument
846 is known to be of type (OR FIXNUM (NOT NUMBER)).
847 * fixed bug: the runtime is now less vulnerable to changes in the
848 size of the SBCL object on OS X, and virtual memory is reserved for
849 all spaces that need to be at a fixed address.
850 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
851 * fixed bug: finalizers and after GC hooks that cause consing are now
853 * fixed bug: compiler error messages and summaries are now printed to
854 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
855 * fixed inference of the upper bound of an iteration variable.
856 (reported by Rajat Datta).
857 * fixed bug 211e: calling local functions with duplicated constant
858 keyword argument no longer causes a bogus style warning about an
860 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
861 ftype declaration to be disregarded. (reported by Dave Roberts)
862 * fixed bug 373: caused by erronous compilation of references to alien
863 variables in the runtime on ppc/darwin.
864 * fixed bug 376: CONJUGATE type deriver.
865 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
866 (reported by Baughn on #lisp)
867 * a cross-compiler bug on non-x86 platforms has been identified and
868 fixed. (thanks to Bruno Haible)
869 * improvements to the MIPS runtime code for increased stability.
870 (thanks to Thiemo Seufer)
871 * increased the maximimum compact environment size to allow
872 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
873 * partial workaround for bug 135: don't name the function we're
874 calling for hairy cases of EVAL, so as not to accumulate one environment
875 entry per call to EVAL. (reported by Kevin Reid)
876 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
877 * optimization: type testing for non-vector arrays should be faster.
878 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
880 * bug fix: setting 31st element of a bit vector to zero did not work
882 * bug fix: redefining a class definition which failed due to a
883 previous accessor / function clash now works (but see BUGS entry
884 #380 for more problems in this area). (thanks to Zach Beane)
885 * the long form of DEFINE-METHOD-COMBINATION disables method group
886 checking when given a single method group with pattern *. (thanks
888 * on x86 compiler supports stack allocation of results of simple
889 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
890 * support for building and running on Mac OS X 10.4 "Tiger" has been added
891 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
892 * fixed some bugs related to Unicode integration:
893 ** the restarts for recovering from input and output encoding
894 errors only appear when there is in fact such an error to
896 * fixed some bugs revealed by Paul Dietz' test suite:
897 ** MISC.549 and similar: late transformation of unsafe type
898 assertions into derived types caused unexpected code
900 ** SCALE-FLOAT type deriver is less wrong.
901 ** type derivers for EXP, LOG and similar functions compute result
902 types for complex arguments better.
903 ** (MISC.563) CONJUGATE type deriver works for very restricted
905 ** out-of-line type testers for character strings are available.
906 ** EQUAL compiler transform understands specialness of objects
908 ** accessing double-floats stored on the stack now works on x86-64.
909 ** debugger internals could sometimes create invalid lispobjs,
910 resulting in GC crashes.
911 ** MISC.548: type check weakening can convert required type into
913 ** initialization forms for bindings are not in scope of free special
916 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
917 * incompatible change: thread support for non-NPTL systems has
918 been removed - locking is buggy and unreliable. A threaded
919 SBCL build will now warn at startup and refuse to create
920 new threads, unless futex support is detected in the kernel
921 * incompatible change: the top level REPL now has only an ABORT
922 restart associated with it, not TOPLEVEL and ABORT as it used to.
923 TOP and TOPLEVEL are now available as debugger commands for
924 returning to the top level.
925 * incompatible change: forms evaluated in the REPL now use the
926 global optimization policy.
927 * incompatible change: user- and system-initialization files are
928 no longer processed with LOAD, but by READ and EVAL; hence the
929 global optimization policy, startup package, readtable, etc,
931 * The .fasl file format number has been incremented because of
932 various incompatible changes.
933 * internal entry point details and argument counts no longer appear
934 in backtraces unless explicitly requested by setting
935 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
936 * built-in and standard functions no longer have names like "top
937 level local call to FOO".
938 * fixed bug 32: functions defined in non-null lexical environments
939 now have more legible printed representation
940 * fixed bug 33: functions defined in non-null lexical environemnts
941 are now more amenable to inspection by INSPECT.
942 * workaround for bug 354: XEPs no longer appear in backtraces unless
943 explicitly requested.
944 * fixed bug: receiving the signal which results from attempting to
945 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
946 SIGBUS on other *BSDs) on architectures where the C stack is also
947 the Lisp stack (x86 and x86-64) from foreign code no longer leads
948 to debugger confusion or wild instability. (reported by Cheuksan
950 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
951 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
952 (reported by Lutz Euler)
953 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
954 compiled code. (reported by Paul Dietz)
955 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
956 specialized parameter to a method to something that is not TYPEP
957 the specializer is now possible.
958 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
959 face of package deletion.
960 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
961 pathname. (reported by David Tolpin on c.l.l)
962 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
963 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
964 STRING-STREAM and FILE-STREAM, respectively.
965 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
966 than previously; in addition, its test suite is now run on build.
967 (thanks to Luke Gorrie)
968 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
969 * optimization: added a immediate representation for single-floats
971 * fixed some bugs revealed by Paul Dietz' test suite:
972 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
973 ** attempting to create a package with a colliding nickname causes
974 correctable errors to be signalled.
975 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
976 ** several x86-64 backend bugs related to sign-extension of immediate
979 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
980 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
981 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
983 * fixed loading of multiply forward-referenced layouts.
984 (thanks to Cheuksan Wang)
985 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
986 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
987 (see COMPILE-FILE documentation for details of :PRINT option)
988 * fixed bugs 19 and 317: fixed-format floating point printing is
989 more accurate. This also fixes a bug reported by Adam Warner
990 related to the ~@F format directive.
991 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
993 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
994 dispatch table if the from-char is a dispatch macro character.
995 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
996 by the wordsize no longer ignore the last word. (reported by Lutz
998 * fixed bug in type checking of dynamic-extent variables. (reported
1000 * optimization: sequence traversal functions use their freedom to
1001 coerce function designators to functions.
1002 * optimization: code with many calls to CLOS methods specialized on
1003 CLOS classes has had redundant type checks removed.
1004 * optimization: added declarations to speed up operations that access
1005 the internal character database (for example STRING-UPCASE)
1006 * optimization: comparison operations between floats and sufficiently small
1007 fixnums no longer create extra rationals
1008 * fixed some bugs related to Unicode integration:
1009 ** portions of multibyte characters at the end of buffers for
1010 character-based file input are correctly transferred to the
1011 start of the buffer at the next read.
1012 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1013 passing it through to OPEN.
1014 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1015 argument given to internal calls to OPEN.
1016 ** the built-in comment readers (introduced by character sequences
1017 ";" and "#|") are more forgiving to encoding errors; they will
1018 STYLE-WARN and then attempt to resync the stream at a character
1019 boundary. (thanks to Teemu Kalvas)
1020 * fixed some bugs in the x86-64 port:
1021 ** Negative short int return values from c-calls are sign-extended
1023 ** The stack is aligned to 16-bytes for c-calls, as required by
1024 the ABI. (thanks to Cheuksan Wang)
1025 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1026 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1027 for enums. (thanks to Vincent Arkesteijn)
1028 ** Multiple small optimizations and bugfixes for floating point
1030 * fixed some bugs revealed by Paul Dietz' test suite:
1031 ** Space, Tab, Linefeed, Return and Page have the invalid
1032 secondary constituent character trait.
1033 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1035 ** WITH-INPUT-FROM-STRING should only update the index place on
1037 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1038 works more reliably.
1039 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1040 with one-dimensional arrays with fill pointers.
1041 ** TYPE-OF failed on a complex with an integer realpart and a
1043 ** compiler failure during type inference for the code of form
1044 (IF (EQL X (THE ...)) ...) (MISC.535).
1046 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1047 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1048 platform. The port supports 61-bit fixnums, large memory spaces
1049 and reloading shared object files.
1050 * enhancement: saving cores with foreign code loaded is now
1051 supported on ppc/Darwin in addition to the previously supported
1053 * enhancement: the statistical profiler now walks deeper into the
1054 call stack for more meaningful call-graphs and accrued time
1055 reports (x86/x86-64 only). It also now reports time spent in
1057 * enhancement: it is now possible to trace most individual methods
1058 of a generic function in addition to tracing the generic function
1060 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1062 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1063 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1064 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1065 produces an error. (thanks to Vincent Arkesteijn)
1066 * bug fix: NAMESTRING on pathnames with :WILD components in their
1067 directories works correctly. (thanks to Artem V. Andreev)
1068 * fixed bug 125: compiler preserves identity of closures. (reported
1070 * bug fixed: functions with &REST arguments sometimes failed with
1071 "Undefined function" when compiled with (DEBUG 3). (reported by
1073 * bug fix: overflow during compiling of setting element of a bit
1074 vector with constant index and value. (reported by Timmy Douglas)
1075 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1076 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1077 * on x86 compiler partially supports stack allocation of dynamic-extent
1079 * GO and RETURN-FROM do not check the extent of their exit points
1080 when compiled with SAFETY 0.
1081 * fixed some bugs related to Unicode integration:
1082 ** encoding and decoding errors are now much more robustly
1083 handled; it should now be possible to recover even from invalid
1084 input or output to the terminal. (thanks to Teemu Kalvas)
1085 ** provided a first cut at implementing STRING-TO-OCTETS and
1086 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1087 ** altered the SB-MD5 contributed module slightly, changing the
1088 interface just enough for it to be supportable for builds where
1089 lisp characters are not eight bits.
1090 * fixed some bugs revealed by Paul Dietz' test suite:
1091 ** the FORMATTER-generated functions for ~V[ conditionals require
1092 the correct number of arguments.
1093 ** READ-FROM-STRING returns the mandated second value when applied
1094 to displaced strings.
1095 ** the #\Rubout and #\Backspace characters are treated as invalid
1096 constituent characters by the tokenizer.
1098 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1099 * new feature: reloading changed shared object files with
1100 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1101 * new feature: references to foreign variables and functions
1102 can now be compiled and loaded before the corresponding shared
1103 object file is loaded, as long as the foreign definitions are
1104 available at runtime.
1105 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1106 supported. (thanks to Dan Debertin)
1107 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1108 just keywords. (thanks to Vincent Arkesteijn)
1109 * fixed bug #331: structure-class instances corresponding to
1110 DEFSTRUCT forms are now created eagerly.
1111 * fixed bug #345: backtraces from calls to undefined functions work
1112 on x86 as well. Related bug #61 is now also partially fixed on x86
1113 (backtraces from throws to unknown catch tags.)
1114 * bug fix: lambda-list parsing is now stricter vrt. order and number
1115 of lambda-list keywords.
1116 * bug fix: as specified by AMOP, an error is signalled if a
1117 class-option appears multiple times in a DEFCLASS form. (reported
1119 * bug fix: RANDOM can be compiled when the compiler derives the type
1120 of its numeric argument as a disjoint set of small integers.
1121 (reported by Paul Dietz)
1122 * bug fix: starting a core saved with shared objects loaded when
1123 those objects are not available no longer causes threaded SBCL to
1124 hang. (reported by Sean Ross)
1125 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1127 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1128 argument, which must always be NIL. (reported by Kalle Niemitalo)
1129 * bug fix: printing 1.0d+23 no longer results in an error.
1130 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1131 * bug fix: structure slot setters preserve evaluation order. (thanks
1133 * fixed some bugs related to Unicode integration:
1134 ** RUN-PROGRAM can allow its child to take input from a Lisp
1135 stream. (reported by Stefan Scholl)
1136 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1137 locales. (reported by Ken Causey)
1138 * fixed some bugs revealed by Paul Dietz' test suite:
1139 ** INCF, DECF and REMF evaluate their place form as specified in
1141 ** FORMATTER expands ~{ ~} iteration directives with V or #
1142 parameters correctly.
1143 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1144 consequent uses no arguments correctly.
1145 ** the system has a partial understanding of the (COMPLEX RATIO)
1147 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1150 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1151 * new feature: a build-time option (controlled by the :SB-UNICODE
1152 keyword feature, enabled by default) for building the system with
1153 support for the entire 21-bit character space defined by the
1155 * new feature: the system now has rudimentary external-format
1156 support; the primary user-visible change at this time is that
1157 characters with the high bit set (such as lower-case-e-acute) will
1158 print correctly to a terminal in a UTF-8 environment.
1159 * minor incompatible change: BASE-CHAR no longer names a class;
1160 however, CHARACTER continues to do so, as required by ANSI.
1161 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1162 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1163 has been moved to the SB-EXT package (temporarily re-exported from
1165 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1166 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1168 * fixed bug #351: better error-handlind and reporting for malformed
1170 * fixed bug #350: bignum-printing is now more memory-efficient,
1171 allowing printing of very large bignums, eg. (expt 2 10000000).
1172 (reported by Bruno Haible)
1173 * fixed bug #302: better primitive-type selection for intersection
1175 * fixed bug #308: non-graphic characters now all have names, as
1176 required. (reported by Bruno Haible)
1177 * bug fix: redefining a class with different superclasses now correctly
1178 removes it from the direct-subclasses of its previous superclasses.
1179 (reported by David Morse)
1180 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1181 new value now works. (reported by Bruno Haible)
1182 * bug fix: correct canonicalization of multiple non-standard slot
1183 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1184 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1185 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1186 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1187 now exists, an signals an error.
1188 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1189 during macroexpansion for non-iteration variables as well. (reported
1190 by Bruno Haible for CMUCL)
1191 * bug fix: Cyclic structures and unprintable objects in compiler
1192 messages no longer cause errors. (reported by Bruno Haible)
1193 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1194 and READ-FROM-STRING all now return a primary value of NIL if
1195 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1196 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1197 (reported by Bruno Haible for CMUCL)
1198 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1199 do not designate a condition. (reported by Bruno Haible for
1201 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1202 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1203 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1204 permitting resolution in favour of any of the conflicting symbols.
1205 (reported by Bruno Haible for CMUCL)
1206 * bug fix: EQUAL compiler optimizations is less aggressive on
1207 strings which can potentially compare true despite having distinct
1208 specialized array element types.
1209 * bug fix: unit enumerations can be defined without dividing by
1210 zero. (thanks to Vincent Arkesteijn)
1211 * FORMAT compile-time argument count checking has been enhanced.
1212 (report from Bruno Haible for CMUCL)
1213 * a partial workaround for the bug 262: the compiler does not try to
1214 inline-expand a local function doing RETURN-FROM from a deleted
1215 BLOCK. (thanks to Peter Denno for the bug report and to David
1216 Wragg for the simple test case)
1217 * fixed some bugs revealed by Paul Dietz' test suite:
1218 ** RENAME-PACKAGE allows all package designators as new package
1220 ** constraint propagation and lambda variable substitution are
1221 more cautious in dealing with partially deleted code.
1222 ** compiler sometimes lost reoptimization passes.
1223 ** CERROR, given a condition as condition designator, passes its
1224 remaining arguments to the continue format control without
1226 ** Case-altering FORMAT directives work correctly on non-ASCII
1228 ** The REAL type specifier handles bounds outside the floating
1229 point ranges without signalling FLOATING-POINT-OVERFLOW.
1230 ** Functions with IR1-transformations can create intercomponent
1231 references to global functions.
1232 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1234 ** FORMAT ~R treats a nil value for its first parameter correctly.
1236 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1237 * enhancement: saving cores with foreign code loaded is now
1238 supported on x86/NetBSD and sparc/Linux in addition to the previously
1239 supported platforms.
1240 * bug fix: on some platforms repeated installations caused multiple
1241 copies of HTML documentation to be installed -- should not happen
1242 any more. (reported by Stefan Scholl)
1243 * bug fix: parsing self-recursive alien record types multiple times
1244 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1245 original patch by Helmut Eller for CMUCL)
1246 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1247 (thanks to Richard Kreuter)
1248 * bug fix: defining classes whose accessors are methods on existing
1249 generic functions in other (locked) packages no longer signals
1250 bogus package lock violations. (reported by François-René Rideau)
1251 * bug fix: special variables as DEFMETHOD parameters no longer have
1252 associated bogus type declarations. (reported by David Wragg and
1254 * bug fix: read-write consistency on streams of element-type
1255 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1256 * bug fix: redefiniton of the only method of a generic function with
1257 no DEFGENERIC no longer emits a full WARNING. In addition,
1258 redefinition of generic functions with no DEFGENERIC to an
1259 incompatible lambda list now signals an error. (thanks to Zach
1261 * bug fix: DEFGENERIC now works even when there's a function of the
1262 same name in an enclosing lexical environment. (thanks to Zach
1264 * fixed compiler failure, caused by instrumenting code during
1265 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1266 * optimization: added loop analysis and improved register allocation
1267 * fixed some bugs revealed by Paul Dietz' test suite:
1268 ** POSITION on displaced vectors with non-zero displacement
1269 returns the right answer.
1270 ** (SIMPLE-STRING) is a valid type specifier for sequence
1272 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1274 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1275 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1277 ** The justification version of the FORMAT ~< directive treats
1278 non-zero minpad parameter correctly.
1280 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1281 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1282 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1283 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1284 the supported interface.
1285 * new feature: Single-stepping of code compiled with DEBUG 2 or
1286 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1287 * new feature: saving cores with foreign code loaded is now
1288 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1289 Timothy Moore's work for CMUCL)
1290 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1291 parameters to * instead of NIL if no initform is supplied.
1292 (reported by Johan Bockgaard)
1293 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1294 correctly when FUNCALL appears as the car of the form. Note:
1295 despite this FUNCALL forms are not currently subject to
1296 compiler-macro expansion. (port of Raymond Toy's fix for the
1297 same from CMUCL, reported by Johan Bockgaard)
1298 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1299 (thanks for Teemu Kalvas)
1300 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1301 case instead of terminating the iteration (thanks for Julian
1302 Squires, Sean Champ and Raymond Toy)
1303 * bug fix: incorrect expansion of defgeneric that caused a style
1304 warning. (thanks for Zach Beane)
1305 * bug fix: slot accessor effective method computation works properly
1306 for classes with multiple non-standard applicable methods on
1307 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1308 * on x86 compiler supports stack allocation of results of LIST and
1309 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1310 CMUCL implementation by Gerd Moellmann)
1311 * fixed some bugs revealed by Paul Dietz' test suite:
1312 ** FORMAT strings with both the ~<~:;~> form of the justification
1313 directive and pretty-printing directives cause an error.
1315 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1316 * incompatible change: the internal functions
1317 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1318 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1319 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1320 instead of the old functions.
1321 * new feature: on platforms where "dladdr" is available foreign
1322 function names now appear in backtraces. (based on Helmut Eller's
1324 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1325 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1327 * documentation: ASDF manual is now distributed with SBCL.
1328 * bug fix: SBCL can now load its contributed modules with REQUIRE
1329 even if the system-provided entries have been removed from
1330 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1332 * bug fix: backtraces involving undefined functions or assembly
1333 routines are more informative. (thanks to Brian Downing)
1334 * bug fix: mutually referent alien structures now work correctly.
1335 (reported by Rick Taube)
1336 * bug fix: structures defined by WITH-ALIEN can be referred to
1337 within other definitions in the same WITH-ALIEN.
1338 * bug fix: division operators (MOD, TRUNCATE and the like) with
1339 constant zero divisors and integer dividends no longer generate
1341 * bug fix: provide default methods for INPUT-STREAM-P and
1342 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1343 * bug fix: improve the detection and resolution of MOP metacycles
1344 (where effective method computation depends on the generic function
1345 having its effective method computed). (reported by Bruno Haible)
1346 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1347 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1348 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1349 * bug fix: pathnames with relative directory components can be
1350 represented relative to default pathnames.
1351 * optimization: in taking the GCD of bignums, reduce the two bignums
1352 to approximately the same size (using Euclid's algorithm) before
1353 applying the more sophisticated binary GCD. (thanks to Juho
1355 * optimization: COUNT on bitvectors now operates word-at-a-time.
1356 * optimization: ASH with a positive, but not necessarily constant,
1357 (leftwards) shift, when in a modular context, is compiled to a
1359 * fixed some bugs revealed by Paul Dietz' test suite:
1360 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1361 if the corresponding argument is NIL.
1362 ** FORMAT directives accepting a minpad parameter treat negative
1364 ** literal commas as character parameters to FORMAT directives are
1366 ** literal spaces directly after ~<Newline> directives within a
1367 format-logical-block (~:< ~@:>) do not induce :FILL-style
1368 conditional newlines.
1369 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1370 as specified: it no longer includes conditional newlines).
1371 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1373 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1374 :PER-LINE-PREFIX argument does not evaluate to a string.
1376 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1377 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1378 the manual for details; for now, package locks can be disabled by
1379 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1380 if no major problems are found then it is likely that they will be
1381 compiled in unconditionally.
1382 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1383 unsupported operators on all platforms. To load a shared library
1384 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1385 object file, link it into a shared library outside of SBCL and
1386 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1388 * bug workaround: ROOM T and the GC were not working together
1389 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1390 reporting facility aren't preserved. That reporting has been
1391 disabled, so now until and unless someone figures out how to make it
1392 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1393 * minor incompatible change: as threatened around sbcl-0.8.0, the
1394 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1395 The symbols are also exported from SB-PCL for backwards
1396 compatibility, but more so than before SB-PCL should be treated as
1397 an implementation-internal package.
1398 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1400 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1401 It is known as of this release not to work on the Alpha, however.
1402 * fixed bug #167: errors signalled due to illegal syntax in method
1403 bodies are now more legible.
1404 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1405 designators and can hence be used with TYPEP.
1406 * fixed bug #333: CHECK-TYPE now ensures that the type error
1407 signalled, if any, has the right object to be accessed by
1408 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1409 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1410 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1412 * fixed bug #334: programmatic addition of slots using specialized
1413 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1414 and :ALLOCATION :CLASS slots.
1415 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1416 float exponents. (rereported by Peter Seibel)
1417 * fixed a bug: #\Space (and other whitespace characters) are no
1418 longer considered to be macro characters in standard syntax by
1419 GET-MACRO-CHARACTER.
1420 * fixed bug: initialization of condition class metaobjects no longer
1421 causes an instance of the condition to be created. (reported by
1423 * fixed bug: it is now possible to have slots such that
1424 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1425 description is neither :INSTANCE nor :CLASS.
1426 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1427 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1428 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1429 non-standard SLOT-DEFINITION-ALLOCATION values.
1430 * fixed bugs: various race conditions handling exiting threads.
1431 CL-PPCRE's thread torture test now passes.
1432 * fixed arguably-a-bug: GC time proportional to number of threads in
1433 system even when most of them are idle
1434 * optimization: improved performance of BIT and SBIT on bit-vectors.
1435 * .fasl file incompatibility: The fasl file version number has
1436 been incremented because of changes associated with package locks.
1438 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1439 * minor incompatible change: the system no longer provides
1440 optimization hints (conditions which are TYPEP
1441 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1442 mode; these hints are still emitted when higher SPEED optimization
1444 * new contrib module: a sampling profiler (profiling by statistical
1445 sampling, rather than by instrumenting functions) is available as
1446 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1447 * the behaviour of the standard function ED is now customizeable by
1448 third parties through a hook variable: see ED's documentation
1449 string for information on the protocol.
1450 * the compiler no longer emits efficiency notes for (FUNCALL X)
1451 when the type of X is uncertain under default optimization
1453 * fixed bug 276: mutating a binding of a specialized parameter to a
1454 method to something that is not TYPEP the specializer is now
1456 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1457 exhibit the required behaviour on the x86 platform. (thanks to
1458 Peter van Eynde, Eric Marsden and Bruno Haible)
1459 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1460 even for difficult arguments. (reported by Peter Graves)
1461 * fixed bug 141a: the backquote printer now descends quoted
1463 * fixed another bug in backquote printing: no more destructive
1464 modification of the form's list structure. (reported by Brian
1466 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1467 move between its address being taken and the call to
1468 interrupt_thread, fixing a crashing race condition.
1469 * the SB-POSIX contrib implementation has been adjusted so that it
1470 no longer exhibits ridiculously poor performance when constructing
1471 instances corresponding to C structs.
1473 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1474 * minor incompatible change: the sb-grovel contrib now treats C
1475 structures as alien (in the sense of SB-ALIEN) objects rather than
1476 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1477 has implications for memory management of client code
1478 (sb-grovel-returned objects must now be manually managed) and for
1479 type safety (alien objects now have full types).
1480 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1481 used to control emission of compiler diagnostics, rather than the
1482 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1483 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1484 quality should be considered deprecated.
1485 * install.sh now installs the user manual as well
1486 * (not quite a new documentable feature, but worth considering in
1487 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1488 beginnings of a semantically meaningful condition hierarchy is
1489 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1491 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1492 designator as the defaults argument.
1493 * fixed bug: Displaced arrays whose displaced-to array has become
1494 too small now cause ARRAY-DIMENSION to signal an error, providing
1495 for safer bounds-checking. (reported by Bruno Haible)
1496 * fixed bug: DEFCLASS slot definitions with identical :READER and
1497 :WRITER names now signal a reasonable error. (reported by Thomas
1499 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1501 * fixed bug: Invalid slot specification errors now print correctly.
1502 (thanks to Zach Beane)
1503 * fixed bug 320: Shared to local slot value transfers in class
1504 redefinitions now happen corrently from superclasses as
1505 well. (reported by Bruno Haible)
1506 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1508 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1509 improper lists correctly. (reported by Bruno Haible)
1510 * fixed bug 313: source-transform for <fun-name> was erroneously
1511 applied to a call of a value of a variable with name <fun-name>.
1512 (reported by Antonio Menezes Leitao)
1513 * fixed bug 307: The obsolete instance protocol ensures that
1514 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1515 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1516 remove dead unknown-values globs from the middle of the stack.
1517 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1518 easier to resume long computations after using *BREAK-ON-SIGNALS*
1519 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1520 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1521 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1522 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1523 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1524 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1526 * fixed a (fairly theoretical) bug in string printing: if
1527 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1528 does not have array-element-type equal to the most general string
1530 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1531 function name. (thanks to Nikodemus Siivola)
1532 * fixed bug: compile-time format string checker failed on
1533 non-closed ~{. (reported by Thomas F Burdick)
1534 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1535 don't warn on legal loop code involving more than one aggregate
1536 boolean. (thanks to Nikodemus Siivola)
1537 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1538 in keyword position to MAKE-PACKAGE was not being reported as
1540 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1541 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1543 * fixed bugs 280 and 312: the checking for multiple definitions in a
1544 file is less likely to become confused by uses of inline
1546 * fixed bug: the #S reader macro performs the keyword coercion
1547 specified for slot names. (reported by Kalle Niemitalo)
1548 * fixed bug: lambda lists may contain symbols whose names start with
1549 & but are not lambda-list-keywords; their occurrence triggers a
1551 * fixed bug 321: define-method-combination argument lambda lists do
1552 not cause invalid code to be generated when &optional or &aux
1553 variables have default values. (reported by Bruno Haible)
1554 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1555 have CLOS classes; this appears also to have fixed reported
1556 instability in stack exhaustion detection.
1557 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1558 MUFFLE-WARNING when no associated restart is present is now
1560 * optimization: rearranged the expansion of various defining macros
1561 so that each expands into only one top-level form in a
1562 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1564 * optimization: used a previously little-used slot in symbols to
1565 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1567 * fixed some bugs revealed by Paul Dietz' test suite:
1568 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1570 ** Fixed an optimization bug related to inheritance of initforms
1571 from local to shared slots.
1572 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1573 ** CAST optimizer forgot to flush argument derived type.
1574 ** print/read consistency on floats is now orders of magnitude
1575 more likely. (thanks also to Bruno Haible for a similar report
1577 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1578 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1579 for objects of type REAL. Make it so.
1580 ** SXHASH is specified (infelicitously) to respect similarity,
1581 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1582 it so. (thanks to Markus Ziegler)
1583 ** on the Alpha, the compiler succeeds in compiling functions
1584 returning a known number of arguments greater than 63.
1585 ** fixed handling of invalid NIL arguments in keyword position
1587 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1589 ** provided a readably-printable representation for RANDOM-STATE
1591 ** ensured that pathnames, where they have a namestring, always
1592 print using #P"..." syntax.
1594 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1595 * Support for the forthcoming 2.0 version of the NetBSD kernel
1596 running on x86 hardware has been added. (thanks to Perry
1597 E. Metzger most immediately, and others for their past work)
1598 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1599 current release version; SBCL's OpenBSD support had been broken
1600 since about the time of OpenBSD's switch to ELF binary format.)
1601 * [placeholder for DX summary]
1602 ** user code with &REST lists declared dynamic-extent, under high
1603 speed or space and low safety and debug optimization policy.
1604 * The manual has been converted to Texinfo format and the debugger
1605 chapter from the cmucl manual has been added.
1606 * A facility has been added to extract documentation strings from
1607 sbcl and store them as Texinfo-formatted snippets for inclusion in
1608 the manual (via Texinfo's @include directive)
1609 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1610 the test case to Dave Roberts)
1611 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1612 pointers. (reported by Sean Ross)
1613 * bug fix: PROFILE output is printed nicely even for large numerical
1614 values. (thanks to Zach Beane)
1615 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1616 greater than 32 handle EOF correctly.
1617 * bug fix: on X86 an immediate argument of the IMUL instruction is
1618 correctly printed in disassembly. (thanks to Lutz Euler)
1619 * bug fix: class slots in redefined classes preserve their old
1620 values. (thanks to Bruno Haible and Nikodemus Siivola)
1621 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1622 succeeds. (reported by Marco Baringer)
1623 * fixed some bugs revealed by Paul Dietz' test suite:
1624 ** READ-SEQUENCE now works on ECHO-STREAMs.
1625 ** RATIONALIZE works more according to its specification. (thanks
1628 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1629 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1630 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1631 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1632 should matter to you only if you rebind the printer control
1633 variables and then find you want different bindings in the
1634 debugger than in the ordinary execution of your program.)
1635 * The runtime build system has been tweaked to support building
1636 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1637 assembler and linker. (thanks to Nikodemus Siivola)
1638 * Unbound, undefined, undeclared variables now trigger full
1639 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1640 more often programmer error than deliberate exploitation of undefined
1642 * optimization: the hash algorithm for strings has changed to one
1643 that is less vulnerable to spurious collisions. (thanks to Juho
1645 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1646 needless bounds checking. (thanks to Juho Snellman)
1647 * optimization: implemented multiplication as a modular
1648 (UNSIGNED-BYTE 32) operation on the PPC backend.
1649 * fixed some bugs revealed by Paul Dietz' test suite:
1650 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1652 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1653 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1654 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1656 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1658 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1660 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1661 their output stream on EOF from read.
1662 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1663 have been read to end-of-file.
1664 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1666 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1667 description of determination of which consecutive characters
1669 ** Printing the "Space" character with escaping on now yields "#\\ ",
1670 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1671 ** Reading floating-point numbers with *READ-BASE* set to a number
1672 less than 10 works correctly.
1673 ** Reading floating-point numbers with *READ-BASE* set to a number
1674 more than 10 works correctly.
1675 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1676 the readtable currently in effect.
1678 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1679 * minor incompatible change: parsing of namestrings on a physical
1680 (Unix) host has changed; numbers after the final #\. in a
1681 namestring are no longer interpreted as a version field. This is
1682 intented to be largely invisible to the user, except that the
1683 meaning of the namestring "*.*.*" has changed: it now refers to a
1684 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1685 should usually be replaced by
1686 (make-pathname :name :wild :type :wild :version :wild)
1687 with the added benefit that this is more likely to be portable.
1688 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1689 option to OPEN now signals an error if the file being opened
1690 exists; this may have an impact on existing code.
1691 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1692 signals received, on the PowerPC platforms (both Linux and
1693 Darwin). (thanks to Pierre Mai for pointing out the location of
1695 * several fixes on OS X: The system now builds and runs cleanly on
1696 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1697 (thanks to Brian Mastenbrook)
1698 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1699 second-resolution: integer multiples of 1/3600 between -24 and 24.
1700 (thanks to Vincent Arkesteijn)
1701 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1702 type when called with 1 argument; PEEK-CHAR checked type of
1703 PEEK-TYPE only after having read first character from a
1704 stream. (reported by Peter Graves)
1705 * bug fix: the garbage collector now has much better locality
1706 behaviour, and in particular no longer treats all memory as being
1707 exhausted when many small objects point to each other in a deeply
1709 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1710 recognized as being TYPEP their class.
1711 * bug fix: the PUSHNEW documentation string has been corrected.
1712 (thanks to Vincent Arkesteijn)
1713 * bug fix: defaulting of the value for the last of an atypically
1714 large number of multiple values being bound was not being
1715 performed correctly on the Alpha or PPC platforms
1716 * optimization: implemented multiplication as a modular
1717 (UNSIGNED-BYTE 32) operation on the x86 backend.
1718 * optimization: SEARCH on simple-base-strings can now be open-coded.
1719 (see also contrib/compiler-extras.lisp for inspiration for
1720 teaching the compiler about the Boyer-Moore algorithm).
1721 * value, returned by MAX (and MIN) called with several EQUALP, but
1722 not EQL, arguments now does not depend on compiler settings.
1723 * fixed some bugs revealed by Paul Dietz' test suite:
1724 ** in stack analysis liveness information is propagated from
1725 non-local entry points.
1726 ** pathwise CAST removing failed when the CAST node did not start
1728 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1729 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1731 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1732 host is already defined.
1733 ** RENAME-FILE works on streams instead of signalling an internal
1735 ** PEEK-CHAR uses the current readtable when determining whether
1736 or not a character is whitespace.
1737 ** MERGE-PATHNAMES handles the case when the pathname does not
1738 specify a name while the default-pathname specifies a version.
1739 ** Pathnames now stand a better chance of respecting print/read
1741 ** Attempting to use standardized file system operators with a
1742 pathname with invalid :DIRECTORY components signals a
1744 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1746 ** DIRECTORY on logical pathnames is more correct.
1747 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1748 signal a TYPE-ERROR if their argument is not a stream.
1749 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1750 designator argument does not designate a stream.
1751 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1752 examining the synonym.
1753 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1755 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1756 element-type larger than ([UN]SIGNED-BYTE 32).
1758 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1759 * When built with the :SB-FUTEX feature, threaded builds now take
1760 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1761 for faster/more reliable mutex and condition variable support.
1762 * Incompatible change (but one you probably shouldn't have been using
1763 anyway): the interface and code for arbitrating between multiple
1764 threads in the same user session has been redesigned.
1765 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1766 error when called without an explicit environment argument.
1767 (thanks to Nikodemus Siivola)
1768 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1769 confused about their position. (thanks to Adam Warner and Gerd
1771 * bug fix: Pretty printing backquoted forms with unquotations in the
1772 argument list position of various code constructs such as LAMBDA
1773 now works correctly. (reported by Paul Dietz)
1774 * bug fix: Pretty printing unquotations no longer loses all
1775 stream position information.
1776 * optimization: performance of string output streams is now less
1777 poor for multiple small sequence writes.
1778 * optimization: performance of CSUBTYPEP in the presence of complex
1779 expressions involving CONS and NOT many times has been improved.
1780 (reported by Paul Dietz)
1781 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1783 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1784 streams for the REPL, for the debugger, and so on. (thanks to
1786 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1787 * fixed some bugs revealed by Paul Dietz' test suite:
1788 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1789 is ignored if it is too small, rather than propagating through
1790 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1791 ** extremely complex negations of CONS types were not being
1792 sufficiently canonicalized, leading to inconsistencies in
1794 ** VALUES tranformer lost derived type.
1796 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1797 * fixed a bootstrapping bug: the build process no longer assumes
1798 that the various BOOLE-related constants have the same value in
1799 host and target lisps. (noted by Paul Dietz' test suite on an
1800 SBCL binary built from CLISP)
1801 * The system can now be dynamically linked on the MIPS platform,
1802 which enables dynamic loading of foreign code from Lisp. (thanks
1803 to Ralf Baechle for discussions on the MIPS ABI)
1804 * The system now records debugging information for its own source
1805 files in a filesystem-position-independent manner, relative to
1806 the "SYS" logical host.
1807 * fixed a compiler bug: MV-LET convertion did not check references
1808 to the "max args" entry point. (reported by Brian Downing)
1809 * tweaked disassembly notes to be less confident about proclaiming
1810 some instruction as an LRA. (thanks to Brian Downing)
1811 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1812 now each have their own history, command character, and other
1813 characteristics. (thanks to David Lichteblau)
1814 * fixed some bugs revealed by Paul Dietz' test suite:
1815 ** compiler failure in compiling LOGAND expressions including a
1817 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1818 shift greater than 32.
1819 ** FLUSH-DEST did not mark blocks for type check regeneration.
1820 ** HANDLER-CASE failed to accept declarations in handler clauses
1821 in some circumstances.
1823 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1824 * New code in contrib/sb-introspect (still probably not entirely
1825 stable yet) provides some support for smart Lisp development
1826 environments like SLIME.
1827 * The conditions signalled for errors occurring when loading .fasl
1828 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1829 in a way which should help ASDF recover gracefully.
1830 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1831 stuff has been cleaned up. If you code contrib/ stuff, this might
1832 affect you, and you can look at contrib/README, contrib/STANDARDS,
1833 and/or the 0.8.4.27 diff to check.
1834 * In full calls the compiler now does not generate checks for declared
1835 argument types for all arguments.
1836 * various threading fixes
1837 ** and some experimental patches which didn't make it into
1838 the main tree for this release, but which are shipped in
1839 contrib/experimental-thread.patch as a possible fix for some
1840 failures (deadlock, spinning...) in GC-intensive multithreaded
1842 * fixed PPC build problem (source code incompatibility of different
1843 library versions): added offsetof() hackery which attempts to divine
1844 where glibc maintainers put uc_mcontext today
1845 * fixed bug 282: compiler does not trust type assertions while passing
1846 arguments to a full call.
1847 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1848 &OPTIONAL VALUES type parameter.
1849 * fix bug 214: algorithm for noting rejected templates is now more
1850 similar to that of template seletion. (also reported by rydis on
1852 * fixed bug 141b: printing backquoted information readably and prettily
1853 inserts a space where necessary.
1854 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1855 (CHARACTER 10) are now reported as errors, rather than propagated
1856 as unknown types. (reported by piso on #lisp)
1857 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1858 with values NIL and :ERROR. (thanks to Milan Zamazal)
1859 * fixed bug 191c: CLOS now does proper keyword argument checking as
1860 described in CLHS 7.6.5 and 7.6.5.1.
1861 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1862 longer raise an error; further, using a list as a for-as-arithmetic
1863 counter now raises a meaningful error.
1864 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1865 now understood by sequence creation functions such as MAKE-SEQUENCE
1867 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1868 a STRING-INPUT-STREAM.
1869 * compiler enhancement: SIGNUM is now better able to derive the type
1871 * type declarations inside WITH-SLOTS are checked. (reported by
1873 * fixed some bugs revealed by Paul Dietz' test suite:
1874 ** incorrect optimization of TRUNCATE for a positive first
1875 argument and negative second.
1876 ** compiler failure in let-convertion during flushing dead code.
1877 ** compiler failure while deriving type of TRUNCATE on an
1878 interval, containing 0.
1879 ** ASH of a negative bignum by a negative bignum count now returns
1881 ** intersection of CONS types now canonicalizes properly, fixing
1882 inconsistencies in SUBTYPEP.
1884 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1885 * incompatible change: The --disable-debugger command line
1886 option now clobbers the debugger at a more fundamental
1887 level, by redefining #'INVOKE-DEBUGGER instead of by
1888 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1889 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1890 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1891 enter the debugger REPL and then suffer recursive errors
1892 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1893 while under the new system, BREAK in --disable-debugger mode
1894 terminates the system just as an unhandled error would.
1895 * fixed compiler performance when processing loops with a step >1;
1896 * bug fix: DOCUMENTATION now retrieves generic function
1897 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1898 support has been systematized, and now supports the methods
1899 specified by ANSI, along with a default method and a method for
1900 slot documentation. (reported by Nathan Froyd)
1901 * bug fix: effective methods associated with a generic function are
1902 no longer cached over a change of that generic function's method
1903 combination. (reported by Andreas Fuchs)
1904 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1905 names a non-existent directory. (thanks to Andreas Fuchs)
1906 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1907 the PPC platform, lead to stack corruption; this has been fixed.
1908 (reported by Rainer Joswig)
1909 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1910 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1911 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1912 arithmetic in general, is now much more reliable on the PPC
1914 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1915 platform now returns the right answer.
1916 * optimization: restored some effective method precomputation in
1917 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1918 precomputation is now tunable.
1919 * optimization: compiler-internal data structure use has been
1920 reviewed, and changes have been made that should improve the
1921 performance of the compiler by about 20%.
1922 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1923 simple-base-strings and simple-bit-vectors is improved.
1924 * optimization: the optimization of 32-bit logical and arithmetic
1925 functions introduced in version 0.8.3 on the x86 has been
1926 implemented on the mips, ppc and sparc platforms; an
1927 implementation of the same facility, but for 64-bit arithmetic,
1928 has been added for the alpha.
1929 * microoptimization: the compiler is better able to make use of the
1930 x86 LEA instruction for multiplication by constants.
1931 * bug fix: in some situations compiler did not report usage of
1932 generic arithmetic in (SPEED 3) policy.
1933 * bug 145b fix: compiler used wrong type specifier while converting
1934 MEMBER-types to numeric.
1935 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1937 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1938 index argument now works properly on the Alpha platform.
1939 * bug fix: floating point exception treatment on the Alpha platform
1941 * bug fix: FILE-POSITION works much better on string input and
1942 output streams. (thanks to Nikodemus Siivola)
1943 * bug fix: many threading/garbage collection symptoms sorted.
1944 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1945 might be pseudo-atomic.
1946 * internal change: Stopping for GC is now done with signals not
1947 ptrace. GC is now done in whichever thread wanted it, instead of
1949 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1951 * bug fix: result form in DO is not contained in the implicit
1953 * incompatible change: ICR structure is changed; the value part of
1954 CONTINUATION is now called LVAR; corresponding functions are
1955 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1956 * added type deriver for ISQRT (thanks to Robert E. Brown).
1957 * bug fix: better support for loading from the command line when an
1958 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1960 * fixed some bugs revealed by Paul Dietz' test suite:
1961 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1962 ** ROUND and FROUND now give the right answer when given very
1963 small float arguments.
1964 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1966 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1967 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1968 ** LCM should return a non-negative integer.
1969 ** PARSE-INTEGER returned the index of a terminator instead of the
1970 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1971 ** PARSE-INTEGER returned an incorrect index being applied to a
1973 ** LCM with two arguments of 0 returns 0 rather than signalling
1975 ** unsigned addition of a 32-bit constant with the high bit set no
1976 longer causes an internal compiler error on the x86.
1977 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1978 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1979 with negative last argument.
1980 ** byte specifiers with zero size and position no longer cause
1981 an error during type derivation.
1982 ** bignum multiplication on the Alpha platform now returns the
1984 * porting: The system now builds on SuSE AMD64, although it still
1985 generates a 32-bit binary.
1986 * .fasl file incompatibility: The fasl file version number has
1987 been incremented (because of the changes to internal compiler
1988 data structures referred to above).
1990 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1991 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1992 more accurately, on the Darwin kernel running on PowerPC hardware.
1993 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1994 * Compiler code deletion notes now signal a condition of type
1995 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1996 an associated MUFFLE-WARNING restart.
1997 * The compiler now performs limited argument count validation of
1998 constant format strings in FORMAT, and where appropriate in ERROR,
1999 CERROR and WARN. (thanks to Gerd Moellmann)
2000 * New ASDF-INSTALL contrib can be used for automatic download and
2001 installation of third-party Lisp code from CCLAN or other sites
2003 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2004 which forces another thread to execute a function supplied by the
2006 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2007 now accept and act upon their :ELEMENT-TYPE keyword argument.
2008 (reported by Martin Atzmueller, Edi Weitz)
2009 * bug fix: FILE-POSITION now accepts position designators up to
2010 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2011 is the greater. (thanks to Patrik Nordebo)
2012 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2013 Kalle Olavi Niemitalo)
2014 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2015 printable. (reported by Eric Marsden)
2016 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2017 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2018 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2019 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2020 (thanks to Stig Sandoe)
2021 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2022 UNREAD-CHAR now correctly includes the unread character in the
2023 target sequence. (thanks to Gerd Moellmann)
2024 * bug fix in threaded builds: the system can now be suspended and
2025 resumed by shell job control with minimal disruption.
2026 * bug fixes in times and timezones >2038AD
2027 * better handling of "where is GNU make?" problem in build scripts
2028 (thanks to Nikodemus Siivola)
2029 * new optimization: inside a named function any reference to a
2030 function with the same name is considered to be a self-reference;
2031 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2032 optimization quality.
2033 * new optimization on x86: logical functions and + now have
2034 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2035 used when the result is truncated to 32 bits.
2036 * VALUES declaration is partially enabled.
2037 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2038 * fixed some bugs revealed by Paul Dietz' test suite:
2039 ** The system now obeys the constraint imposed by
2040 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2041 types form a lattice under type intersection.
2042 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2043 ** ASSOC now ignores NIL elements in an alist.
2044 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2045 and (1+ MOST-POSITIVE-FIXNUM) answers.
2046 ** The addition of a method with invalid qualifiers to a generic
2047 function does not cause an error to be signalled immediately;
2048 a warning is signalled, and the error is generated only on
2049 calling the generic function.
2050 * changed .fasl file version number, in part to add type codes for
2051 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2052 obscure ANSI requirements
2054 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2055 * fixed bug 148: failure to inline-expand a local function left
2056 garbage, confusing the compiler.
2057 * fixed bugs 3cd: structure slot readers perform type check if the
2058 slot can have an invalid value (i.e. it is either not initialized
2059 or can be written with a less specific slot writer).
2060 * bug fix: the compiler now traps array references to elements off
2061 the end of an array; previously, the bounds checking in some
2062 circumstances could go off-by-one.
2063 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2065 * type declarations for array element types now obey the description
2066 on the CLHS page "Declaration TYPE", as per discussions on
2067 sbcl-help around 2003-05-08. This means that a declaration
2068 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2069 declaration, all references to BAR will be asserted or assumed
2070 (with THE, so dependent on compiler policy) to involve objects of
2071 type FOO. Note that no such declaration is implied in
2072 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2073 * declared types of functions from the "Conditions"
2074 chapter. (reported by Paul Dietz)
2075 * bug fix: CERROR accepts a function as its first argument.
2076 * bug fix: NTH an NTHCDR accept a bignum as index
2077 arguments. (reported by Adam Warner)
2078 * optimization: character compare routines now optimize comparing
2079 against a constant character. (reported by Gilbert Baumann)
2080 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2081 argument now works properly on the MIPS platform.
2082 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2083 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2084 declared to have a functional type, does not cause an error
2086 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2087 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2088 * fixed bug: if last continuation of a deleted block has a
2089 destination, this destination should be deleted too. (reported by
2091 * fixed a bug in the bootstrap process: the host compiler's values
2092 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2093 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2094 test case from Patrik Nordebo)
2095 * improved the ability of the disassembler on the PPC platform to
2096 provide helpful disassembly notes.
2097 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2098 the class in more cases than previously.
2099 * bug fix: FILE-POSITION now understands :START and :END for
2100 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2101 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2102 8) data. (thanks to David Lichteblau)
2103 * bug fix: it is possible to add a method to a generic function
2104 without lambda list.
2105 * bug fix: reader failed to signal END-OF-FILE inside an
2106 object representation. (reported by Nikodemus Siivola)
2107 * fixed some bugs revealed by Paul Dietz' test suite:
2108 ** LAST and [N]BUTLAST should accept a bignum.
2109 ** condition slot accessors are methods.
2110 ** (VECTOR NIL) is a subtype of STRING.
2112 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2113 * minor incompatible change: some nonsensical specialized lambda
2114 lists (used in DEFMETHOD) which were previously ignored now signal
2116 * minor incompatible change: the system is now aware of the types of
2117 variables in the COMMON-LISP package, and will signal errors for
2118 most violations of these type constraints (where previously they
2119 were silently accepted).
2120 * minor incompatible change: COMPILE-FILE now uses the freedom
2121 afforded (ANSI 3.2.2.3) to use derived function types for
2122 functions defined in the same file. This also permits the system
2123 to warn on static type mismatches and function
2124 redefinition. (Currently it does not work with high DEBUG level.)
2125 * minor incompatible change: VALUES declaration is disabled.
2126 * When issuing notes, the compiler now signals a condition of type
2127 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2128 restart for use in user handlers. It is expected that the
2129 COMPILER-NOTE condition will eventually become a condition
2130 supertype to a hierarchy of note types, which will then be
2131 handleable in a similar fashion. However, at the moment, no such
2132 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2133 but it's an implementation detail, not a classification for the
2135 * Changes in type checking closed the following bugs:
2136 ** type checking of unused values (192b, 194d, 203);
2137 ** template selection based on unsafe type assertions (192c, 236);
2138 ** type checking in branches (194bc).
2139 * A short form of VALUES type specifier has ANSI meaning (it has
2140 increased the number of situations when SBCL cannot perform type
2142 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2143 keywords or constants is permissible.
2144 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2145 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2146 classes. (thanks to Antonio Martinez)
2147 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2148 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2149 argument to a pretty-print stream. Instead, it leaves any such
2150 operations to DESCRIBE-OBJECT methods.
2151 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2153 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2154 lambda lists. (thanks to David Lichteblau)
2155 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2156 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2157 (thanks to Gerd Moellmann)
2158 * fixed bug 246: increased compilation speed of long
2159 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2161 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2163 * DEFCONSTANT now throws a condition of type
2164 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2165 constant to a non-EQL value; CONTINUE and ABORT restarts
2166 respectively change and preserve the value.
2167 * fixed bug 63: The code walker, part of the implementation of CLOS,
2168 is now better at handling symbol macros.
2169 * bug fix: There is no longer an internal implementation type named
2170 CL:LENGTH. (reported by Raymond Toy)
2171 * bug fix: In macro-like defining macros/special operators the
2172 implicit block does not enclose the lambda list.
2173 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2174 enclosed in parentheses) are not suitable as type specifiers, and
2175 their use properly signals an error now.
2176 * bug fix: An argument count mismatch for a type specifier in code
2177 being compiled no longer causes an unhandled error at compile
2178 time, but signals a compile-time warning.
2179 * fixed simple vector readable printing
2180 * bug fix: DESCRIBE takes more care over whether the class
2181 precedence list slot of a class is bound before accessing it.
2182 (reported by Markus Krummenacker)
2183 * bug fix: FORMATTER can successfully compile pretty-printer format
2184 strings which use variants of the ~* directive inside.
2185 * bug fix: SEARCH now applies its TEST predicate to the elements of
2186 the arguments in the correct order. (thanks to Wolfhard Buss)
2187 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2188 in type inference. (thanks to Robert E. Brown)
2189 * bug fix: Reading in symbols with an explicit package name of ""
2190 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2191 * fixed some bugs revealed by Paul Dietz' test suite:
2192 ** NIL is now allowed as a structure slot name.
2193 ** Arbitrary numbers, not just REALs, are allowed in certain
2194 circumstances in LOOP for-as-arithmetic clauses.
2195 ** Multiple class redefinitions before slot access no longer
2196 causes a type error.
2197 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2198 association between the name and a class.
2199 ** Generic functions with non-standard method-combination and over
2200 five methods all of which return constants no longer return NIL
2201 after the first few invocations. (thanks to Gerd Moellmann)
2202 ** CALL-NEXT-METHOD with no arguments now passes the original
2203 values of the arguments, even in the presence of assignment.
2204 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2206 ** DISASSEMBLE works with closures and funcallable instances.
2207 ** ADD-METHOD now returns the generic function, not the new method.
2208 ** FIND-METHOD signals an error if the lengths of the specializers
2209 is incompatible with the generic function, even if the ERRORP
2211 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2212 which its argument is a member.
2213 ** DEFCLASS only redefines the class named by its class-name
2214 argument if that name is the proper name of the class;
2215 otherwise, it creates a new class.
2216 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2217 of the UNBOUND-SLOT condition to the name of the slot.
2218 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2219 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2220 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2221 SLOT-MAKUNBOUND in the specified fashion.
2223 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2224 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2225 cross-compilation host. As a consequence, we can now bootstrap our
2226 way up to SBCL starting with a bare gcc toolchain and human-readable
2227 source code (first the source to CLISP, then the source to SBCL).
2228 * A contributed module containing a partial implementation of the
2229 simple-streams interface has been included. (thanks to Rudi
2231 * A contributed module implementing the RFC1321 Message Digest
2232 Algorithm, known as MD5, has been included.
2233 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2234 no longer has any effect, as the code controlled by this feature
2235 has been deleted. (As far as we know, no-one has ever built using
2236 this feature, and its semantics were confused in any case).
2237 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2238 work on conditions (as required by the ANSI specification),
2239 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2240 expected behaviour on conditions. Users should note, however,
2241 that such behaviour is not required by the ANSI specification,
2242 and so use of this behaviour may render their code unportable.
2243 * fixed some bugs revealed by Paul Dietz' test suite:
2244 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2246 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2247 ** SXHASH on condition objects no longer returns NIL.
2248 ** :ALLOCATION :CLASS slots are better treated; their values are
2249 updated on class redefinition, and initforms inherited from
2250 superclasses are applied.
2251 ** REMOVE-METHOD returns its generic function argument even when
2252 no method was removed.
2253 ** SHARED-INITIALIZE now initializes the values of the requested
2254 slots, including those with :ALLOCATION :CLASS.
2255 ** ALLOCATE-INSTANCE now works on structure classes defined via
2256 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2258 ** SLOT-EXISTS-P now works on conditions, as well as structures
2260 ** MAKE-LOAD-FORM now has the required methods on
2261 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2262 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2263 keyword, and now implements the SLOT-NAMES argument.
2264 ** methods with &OPTIONAL arguments no longer allow too many
2265 arguments to be passed in the call without error.
2266 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2267 option is consistent with the required arguments of the generic
2268 function lambda list.
2269 * bug fix: REQUIRE accepts a string designator. (Thanks to
2271 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2272 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2273 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2274 * bug fix: The compiler now checks for duplicated variables in macro
2276 * bug fix: SETQ on globals returns the correct value.
2277 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2278 not a COMPILER-ERROR (followed by some other strange error on
2279 choosing the CONTINUE restart).
2280 * bug fix: make.sh and friends are now more consistent in the way that
2281 they look for GNU "make".
2283 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2284 * experimental native threads support (on x86 Linux >=2.4 only).
2285 This is not compiled in by default: you need to add :SB-THREAD to
2286 the target features. See the "Beyond ANSI" chapter of the manual
2288 * fix for longstanding nonANSIism: The old distinction between
2289 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2290 The return value from CL:FIND-CLASS is now a CLOS class, and
2291 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2292 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2293 * An interface to the MetaObject Protocol, as described in Kiczales,
2294 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2295 MIT Press, 1991, is available from the SB-MOP package.
2296 * incompatible change: the SB-PCL package should now be considered
2297 a private implementation detail, and no longer a semi-private MOP
2299 * minor incompatible change: due to rearrangement for threads, the
2300 control stack and binding stack are now allocated at arbitrary
2301 addresses instead of being hardcoded per-port. Users affected by
2302 this probably have to be doing advanced things with shared
2303 libraries, and will know who they are.
2304 * minor incompatible change: Previously, all --eval forms used were
2305 processed with READ before any of them were processed with EVAL.
2306 Now each --eval form is processed with both READ and EVAL before
2307 the next --eval form is processed. (Thus package operations like
2308 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2309 work as the user might reasonably expect.)
2310 * minor incompatible change: *STANDARD-INPUT* is now only an
2311 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2313 * minor incompatible change: Y-OR-N-P is now character-oriented,
2314 not line oriented. Also, YES-OR-NO-P now works without errors.
2315 (thanks to Antonio Martinez)
2316 * sb-aclrepl module improvements: an integrated inspector, added
2317 repl features, and a bug fix to :trace command.
2318 * Known functions, which cannot be open coded by the backend, are
2319 considered to be able to check types of their arguments. (fixing
2320 a bug report by Nathan J. Froyd)
2321 * fixed a bug in computing method discriminating functions: It is
2322 now possible to define methods specialized on classes which have
2323 forward-referenced superclasses. (thanks to Gerd Moellmann)
2324 * fixed evaluation order in optional entries (reported by Gilbert
2326 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2327 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2328 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2329 required-by-AMOP NAME argument, as well as CLASS and
2330 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2331 * fixed bug 20: DEFMETHOD can define methods using names that are
2332 not the proper names of classes to designate class specializers.
2333 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2334 Unix file descriptors, instead of blowing up. (thanks to Antonio
2336 * Garbage collection refactoring: user-visible change is that a
2337 call to the GC function during WITHOUT-GCING will not do garbage
2338 collection until the end of the WITHOUT-GCING. If you were doing
2339 this you were probably losing anyway.
2340 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2341 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2342 TYPEP the latter but not the former.
2343 * The compiler issues a full WARNING for calls to undefined functions
2344 with names from the CL package.
2345 * MAP-INTO for a vector destination is open coded. (reported by
2346 Brian Downing on c.l.l)
2347 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2348 documentation string.
2349 * fixed some bugs revealed by Paul Dietz' test suite:
2350 ** COPY-ALIST now signals an error if its argument is a dotted
2352 ** Condition slots are now accessed more correctly in the presence
2353 of multiple initargs for a given slot.
2354 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2355 exclude restarts of the same name associated with a different
2357 ** DEFCLASS of forward-referenced classes with another
2358 forward-referenced class in the superclasses list no longer
2360 ** Condition slots are now initialized once each, not multiple
2361 times. (thanks to Gerd Moellmann)
2362 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2364 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2365 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2366 arguments contain duplicated elements.
2367 ** RESTART-CASE understands local macros.
2368 ** RESTART-CASE associates exactly its own restarts with a condition.
2369 ** ENDP in safe mode checks its argument to be of type LIST.
2370 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2371 in question is unbound.
2372 ** Optimized MAKE-INSTANCE functions no longer cause internal
2373 assertion failures in the presence of duplicate initargs.
2374 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2375 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2376 propagated even in the ctor optimized implementation of
2378 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2380 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2381 * a better implementation of SXHASH on (simple) bit vectors,
2382 measured both in execution speed and in distribution of results
2383 over the positive fixnums, has been installed. Likewise, a better
2384 implementation of EQUAL for simple bit vectors is now available.
2385 * fixed CEILING optimization for a divisor of form 2^k.
2386 * fixed bug 240 (emitting extra style warnings "using the lexical
2387 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2388 by Antonio Martinez)
2389 * fixed SXHASH, giving different results for NIL depending on type
2390 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2391 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2392 variable. (found by Rolf Wester)
2393 * SBCL does not ignore type declarations for special
2394 variables. (reported by rif on c.l.l 2003-03-05)
2395 * some bug fixes in contrib/sb-aclrepl/
2396 * fixed some bugs revealed by Paul Dietz' test suite:
2397 ** a bug in the CONS type specifier, whereby the CAR and CDR
2398 types got intertwined, has been fixed;
2399 ** the type system is now able to reason about the interaction
2400 between INTEGER and RATIO types more completely;
2401 ** APPEND, [N]REVERSE and NRECONC check that those their
2402 arguments, which must be proper lists, are really so;
2403 ** An array specialized to be unable to hold elements has been
2404 implemented, as required -- yes, really -- by ANSI;
2405 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2406 on malformed property lists;
2408 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2409 * incompatible packaging change: in line with Unix convention,
2410 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2411 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2412 * REQUIRE and PROVIDE are now optionally capable of doing something
2413 useful. See the documentation string for REQUIRE.
2414 * infrastructure for a managed SBCL contrib system: contributed
2415 modules in this release include:
2416 ** the ASDF system definition facility;
2417 ** an interface to the BSD Sockets API;
2418 ** an ACL-like convenience interface to the repl;
2419 (thanks to Kevin Rosenberg)
2420 ** an implementation of ROTATE-BYTE, with efficient implementation
2422 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2423 gives a closer approximation to the right answer than previously.
2424 (thanks to Raymond Toy)
2425 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2426 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2427 optional environment arguments, as required by ANSI.
2428 * fixed bugs in other functions taking environment objects, allowing
2429 calls with an explicit NIL environment argument to be compiled
2431 * fixed bug 228: primary return values from
2432 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2433 COMPILE or FUNCTION.
2434 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2435 :TYPE will no longer signal errors on innocuous objects.
2436 * fixed bug 231b: SETQ is better at respecting type declarations in
2437 the lexical environment.
2438 * fixed a bug in DEFCLASS: classes named by symbols with no or
2439 unprintable packages can now be defined.
2440 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2441 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2442 * fixed compiler failure related to checking types of functions.
2443 (reported by Robert E. Brown)
2444 * the compiler is now much more consistent in its error-checking
2445 treatment of bounding index arguments to sequence functions: in
2446 (SAFETY 3) code, errors will be signalled in almost all cases if
2447 invalid sequence bounding indices are passed to functions defined
2448 by ANSI to operate on sequences.
2449 * fixed a bug in the build procedure: documentation of SBCL-specific
2450 packages is now preserved and available in the final Lisp image.
2451 * lifted FDEFINITION lookup out of loops in the implementation of
2452 many list operations. (thanks to Robert E. Brown)
2453 * fixed a bug in the reader: the #n# reader macro now works for
2454 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2455 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2456 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2457 E. Brown and Tony Martinez respectively)
2458 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2459 SLOT-BOUNDP: in particular, it is now able to optimize them much
2460 better, and is now not vulnerable to having packages renamed.
2461 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2462 always called when a slot is not present in an instance. (thanks
2464 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2465 longer signal an error on unreading a character following EOF on
2466 the previous constituent stream. (thanks to Tony Martinez)
2467 * fixed some bugs revealed by Paul Dietz' test suite:
2468 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2469 not just nonnegative fixnums;
2470 ** the logical bit-array operators such as BIT-AND now accept an
2471 explicit NIL for their "opt-arg" argument (to indicate a
2472 freshly-consed result bit-array);
2473 ** ELT now signals an error on an invalid sequence index in safe
2475 ** the type system is now cleverer about negations of numeric
2476 types, and consequently understands the BIGNUM and RATIO types
2478 ** the type system is now cleverer about the interaction between
2479 INTEGER and RATIO types: while bugs still remain, many more
2480 cases are accurately computed;
2481 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2482 if it is in the last clause;
2483 ** CONSTANTLY now correctly returns a side-effect-free function in
2485 ** DECLARE is no longer treated as a special-operator; in
2486 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2487 * incremented fasl file version number due to the change in the
2488 DEFSTRUCT-SLOT-DESCRIPTION structure.
2490 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2491 * minor incompatible change: code processed by the "interpreter" or
2492 EVAL now has a compilation optimization policy of (DEBUG 2)
2493 (changed from (DEBUG 1)) to improve debuggability of interactive
2494 development, and to allow the use of the debug RETURN command in
2496 * an experimental implementation of the RETURN command for the
2497 debugger has been included. (thanks to Frederik Kuivinen)
2498 * fixed bug 62: constraints were not propagated into a loop.
2499 * fixed bug in embedded calls of SORT (reported and investigated by
2501 * fixed some bugs revealed by Paul Dietz' test suite:
2502 ** printing and reading of arrays with some dimensions having
2503 length 0 (thanks to Gerd Moellmann);
2504 ** BOA constructor with &AUX argument without a default value does
2505 not cause a type error;
2506 ** CONSTANTP now returns true for all self-evaluating objects.
2508 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2509 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2510 accessors that are related by inheritance, as specified in the
2511 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2512 Valtteri Vuorikoski)
2513 * The compiler is now able to inline functions that were defined in
2514 a complex lexical environment (e.g. inside a MACROLET).
2515 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2517 * Support for the upcoming FreeBSD-5.0 release has been included.
2518 (thanks to Dag-Erling Smorgrav)
2519 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2520 effect when it is not in a toplevel context.
2521 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2522 stand a better chance of being correct. (thanks to Gerd
2524 * fixed bug in COERCE, which now signals an error on coercing a
2525 rational to a bounded real type which excludes the expected
2527 * The compiler is now able to derive types more accurately from the
2528 COERCE and COMPILE functions.
2529 * fixed bug 223: functional binding is considered to be constant
2530 only for symbols in the CL package.
2531 * fixed bug 231: SETQ did not check the type of a variable being set
2532 (reported by Robert E. Brown)
2533 * A new optimization for MAKE-INSTANCE has been included, fixing
2534 various bugs (including relating to :ALLOCATION :CLASS slots and
2535 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2537 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2538 ** As required by ANSI, LOOP now disallows anonymous collection
2539 clauses such as COLLECT I in conjunction with aggregate boolean
2540 clauses such as THEREIS (= I 1);
2541 ** LOOP now signals an error when any variable is reused in the
2542 same loop (including the potentially useful construct analogous
2543 to WITH A = 1 WITH A = (1+ A);
2544 ** IT is only a special loop symbol within the first clause of a
2545 conditional loop clause;
2546 ** LOOP with a typed iteration variable over a hashtable now
2547 signals a type error iff it should.
2548 * fixed some other bugs revealed by Paul Dietz' test suite:
2549 ** FILE-STREAM now names the class previously known as FD-STREAM;
2550 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2551 argument) no longer signals an error;
2552 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2553 of the slot symbol, rather than using the current package
2554 ((:CONC-NAME "") continues to intern the slot's name in the
2556 * incremented fasl file version number, because of the incompatible
2557 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2558 of the new implementation of DEFINE-COMPILER-MACRO.
2560 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2561 * Support for building SBCL for MIPS platforms running in
2562 little-endian mode has now been checked in, and basic
2563 functionality on said platforms verified.
2564 * minor incompatible change: PCL now records the pathname of a file
2565 in which methods and the like are defined, rather than its
2567 * minor incompatible change: TRUENAME now considers the truename of
2568 a file naming a directory to be the pathname with :DIRECTORY
2569 component indicating that directory.
2570 * minor incompatible change: a NAMED clause in the extended form of
2571 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2572 reason for the previous behaviour is unclear.
2573 * more systematization and improvement of CLOS and MOP conformance
2574 in PCL (thanks to Gerd Moellman and Pierre Mai):
2575 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2577 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2578 primary methods with no specializers;
2579 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2581 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2582 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2583 FORWARD-REFERENCED-CLASSes; error reporting on
2584 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2586 ** SXHASH on CLOS instances now uses a slot internal to the
2587 instance to return different numbers on distinct instances,
2588 while preserving the same return value through invocations of
2590 ** DEFMETHOD signals errors when methods with longer incongruent
2591 lambda lists are added to generic functions;
2592 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2593 CLASS, as specified in AMOP;
2594 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2595 on the order returned by the primary method for classes of
2596 class STANDARD-CLASS;
2597 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2598 * fixed some bugs shown by Paul Dietz' test suite:
2599 ** DOLIST puts its body in TAGBODY;
2600 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2602 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2603 value producing form;
2604 ** if more variables are given to PROGV than values, extra
2605 variables are bound and made to have no value;
2606 ** NSUBSTITUTE on list arguments gets the right answer with
2608 ** ELT signals an error of type TYPE-ERROR when the index argument
2609 is not a valid sequence index;
2610 ** LOOP signals (at macroexpansion time) an error of type
2611 PROGRAM-ERROR when duplicate variable names are found;
2612 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2613 ** FUNCALL of special-operators now cause an error of type
2615 ** PSETQ now works as required in the presence of side-effecting
2616 symbol-macro places;
2617 ** NCONC accepts any object as its last argument;
2618 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2620 ** loop-for-as-package does not require a package to be explicitely
2622 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2624 * fixed bug 166: compiler preserves "there is a way to go"
2625 invariant when deleting code.
2626 * fixed bug 172: macro lambda lists with required arguments after
2627 &REST arguments now cause an error to be signalled. (thanks to
2629 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2630 bug. (thanks to Matthew Danish)
2631 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2633 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2634 arguments are lexically rebound. (thanks to Gerd Moellmann and
2636 * fixed bug 194: error messages are now more informative when there
2637 is no primary method applicable in a call to a generic
2638 function. (thanks to Gerd Moellmann)
2639 * fixed bug in command line argument checking (thanks to Julian
2641 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2642 (thanks to Matthew Danish)
2643 * incremented fasl file version number, because of the
2644 SXHASH-related changes in the layout of CLOS data structures
2646 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2647 * minor incompatible change: The runtime (the Unix executable named
2648 "sbcl") is now much pickier about the .core files it will load.
2649 Essentially it now requires .core files to descend from the same
2650 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2651 as the runtime does. (The intent is to prevent the crashes which
2652 can occur, and which can even be reported as mysterious failures,
2653 when people patch the sources or change the build parameters
2654 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2655 sbcl and .core files.)
2656 * fixed bug: VALUES-LIST is no longer optimized away.
2657 * fixed bug 142: The FFI conversion of C string values to Lisp
2658 string values no longer conses excessively. (thanks to Nathan
2659 Froyd porting Raymond Toy's fix to CMU CL)
2660 * began to systematize and improve MOP conformance in PCL (thanks to
2661 Nathan Froyd, Gerd Moellman and Pierre Mai):
2662 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2664 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2665 ** FINALIZE-INHERITANCE is now called on class finalization;
2666 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2667 argument precedence order.
2668 * fixed bug 202: The compiler no longer fails on functions whose
2669 derived types contradict their declared type.
2670 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2671 so it can be non-toplevel.
2672 * The fasl file version number has changed (because of the new
2673 implementation of DEFMACRO).
2674 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2675 safe code, that any length requirement by their type-specifier
2676 argument is valid. The exceptions to this are described in bug
2678 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2679 functions now check the types of their inputs as required by ANSI.
2680 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2681 introduced symbol is DECLAREd to be SPECIAL.
2682 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2683 * fixed bug: :COUNT argument to sequence functions may be negative
2684 * fixed bug: body of DO-SYMBOLS may contain declarations
2685 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2686 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2687 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2688 symbol macro only once
2689 * fixed printing of call frame when argument list is unavailable
2690 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2691 * compiler no longer signals WARNING on unknown keyword
2694 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2695 * A beta-quality port to the mips architecture running Linux,
2696 based on the old CMUCL backend, has been made. It has been tested
2697 on a big-endian kernel, and works sufficiently well to be able to
2698 rebuild itself; it has not been tested in little-endian mode.
2699 * fixed an inconsistency between gencgc.c and purify.c which made
2700 dumping/loading .core files unreliable
2701 * fixed bug 120a: The compiler now deals correctly with IFs where
2702 the consequent is the same as the alternative, instead of
2703 misderiving the return type. (thanks to Alexey Dejneka)
2704 * fixed bug 113: Logical pathnames are now dumpable (the logical
2705 host is resolved at load-time, throwing an error if it is not
2707 * fixed bug 174: FORMAT's error message is slightly clearer when a
2708 non-printing character is used in a format directive.
2709 * fixed several bugs in compiler checking of type declarations, i.e.
2710 violations of the Python "declarations are assertions" principle
2711 (thanks to Alexey Dejneka)
2712 * fixed several bugs in PCL's error checking (thanks to Gerd
2714 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2716 * fixed bug in compilation of functions as first class values
2717 (thanks to Antonio Martinez-Shotton)
2718 * The compiler's handling TYPE-ERRORs which it can prove will
2719 inevitably happen at runtime has been cleaned up and corrected
2720 in several ways. (thanks to Alexey Dejneka)
2721 * improved argument type checking for various basic arithmetic
2722 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2723 lavished on them in the past that they can be compiled in many
2724 ways in different special cases
2725 * fixed bug 181: compiler checks validity of user supplied type
2727 * cleaned up code flushing in optimization: Function calls which
2728 should signal errors for safety purposes (e.g. which ANSI says
2729 should signal errors when their arguments are of incorrect type)
2730 are no longer optimized away.
2731 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2732 * incremented fasl file version number, because changes in the
2733 implementation of sequence functions like COERCE caused
2734 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2735 in old inline expansions) to become undefined. (Actually these
2736 changes were later undone, so we might very well be binary
2737 compatible with 0.7.7 after all, but leaving the version number
2738 incremented seemed like the simplest and most conservative
2741 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2742 * An alpha-quality port to the parisc architecture running Linux,
2743 based on the old CMUCL backend, has been made. This, even more so
2744 than the other backends, should be considered still a work in
2745 progress; known problems include that the Linux kernel in 64-bit
2746 mode does not propagate the correct sigcontext structure to
2747 userspace, and consequently SBCL on a parisc64 kernel will not
2749 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2750 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2751 suppressed.) Also now that the compiler is looking at declarations
2752 in the environment, it checks optimization declarations as well,
2753 and suppresses inlining when (> DEBUG SPEED).
2754 * More fixes have been made to treatment of floating point exception
2755 treatment and other Unix signals. In particular, floating point
2756 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2757 * The detection and handling of control stack exhaustion (infinite
2758 or very deeply nested recursion) has changed. Stack exhaustion
2759 detection is now done by write-protecting pages at the OS level
2760 and applies at all optimization settings; when found, a
2761 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2762 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2763 be caught using IGNORE-ERRORS.
2764 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2765 correctly when transforming recursive LABELS functions to LETs.
2766 (thanks to Alexey Dejneka)
2767 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2768 that are names of constants or global variables.
2769 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2770 alien routines with docstrings.
2771 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2772 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2774 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2775 (thanks to Eric Marsden)
2776 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2777 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2778 * Bug fix: The SPARC backend can now compile functions involving
2779 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2780 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2781 to a functions expecting a PATHNAME-DESIGNATOR.
2782 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2783 lambda lists. (thanks to Alexey Dejneka)
2784 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2785 functions so that EXPORTing the name of the class doesn't cause
2786 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2787 to Antonio Martinez for reporting this)
2788 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2789 constant-folded without failing an assertion. (thanks to Einar
2790 Floystad Dorum for reporting this)
2791 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2792 types are now dealt with more correctly by the compiler.
2793 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2794 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2795 behaviour analogous to RENAME-FILE. This puts its behaviour more
2796 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2798 * The fasl file version number has changed again. (because of the
2799 bug fix involving the names of PCL MAKE-INSTANCE functions)
2801 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2802 * bug fix: Floating point exceptions are treated much more
2803 consistently on the x86/Linux and PPC/Linux platforms.
2804 * Array initialization with :INITIAL-ELEMENT is now much faster for
2805 cases when the compiler cannot open code the array creation, but
2806 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2807 array accesses have also seen a speed increase.
2808 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2809 specified. (This was at the root of some bad interactions between
2810 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2812 * bug fix: Internal error arguments for undefined functions are now
2813 computed correctly on the PPC/Linux platform.
2814 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2815 Raymond Toy's patch for CMU CL)
2816 * Support for the Solaris 9 operating environment has been included
2817 (thanks to Daniel Merritt)
2818 * A very ugly but hopefully complete draft of the missing FFI chapter
2819 of the manual has been created by reformatting the corresponding
2820 CMU CL manual chapter into (currently very ugly and incoherent)
2821 DocBook and bringing it up to date for SBCL behavior. Thus, the
2822 manual is now essentially complete, at least by my extreme
2823 once-and-only-once standards, whereby it's acceptable to refer to
2824 the doc strings of SB-EXT functions as the primary documentation.
2825 * The fasl file version number has changed again, due to cleanup of
2826 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2827 is no longer a static symbol.)
2829 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2830 * SBCL now builds with OpenMCL (version 0.12) as the
2831 cross-compilation host; also, more progress has been made toward
2832 bootstrapping under CLISP.
2833 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2835 * bug 158 fixed: The compiler can now deal with integer loop
2836 increments different from 1; fixing this turned out also to fix
2838 * bug 169 fixed: no more bogus warnings about using lexical bindings
2839 despite the presence of perfectly good SPECIAL declarations (thanks
2840 to David Lichteblau)
2841 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2842 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2843 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2845 * bug 180 fixed: Method combination specifications no longer ignore
2846 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2847 * bug fix: Structure type predicate functions now check their argument
2848 count as they should.
2849 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2850 correctly. (thanks to Pierre Mai)
2851 * minor incompatible change: The --noprogrammer option is deprecated
2852 in favor of the new --disable-debugger option, which is very similar.
2853 (The major difference is that it takes effect at a slightly different
2854 time at startup, causing handling of errors in --sysinit and
2855 --userinit files will be affected differently.) The
2856 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2857 been added to allow this functionality to be controlled from ordinary
2858 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2859 maintainers, who might want to run non-interactive scripts to
2860 build SBCL cores which will later be used interactively.)
2861 * minor incompatible change: The LOAD function no longer, when given
2862 a wild pathname to load, loads all files matching that pathname.
2863 Instead, an error of type FILE-ERROR is signalled.
2865 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2866 * bug 147 fixed: The compiler preserves its block link/count
2867 invariants more correctly now so that it doesn't crash. (thanks
2869 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2871 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2872 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2873 no-value-for-this-character as NIL (as specified by ANSI).
2874 * HOST-NAMESTRING on physical pathnames now returns a string that is
2875 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2876 (thanks to Christophe Rhodes)
2877 * The Alpha port handles icache flushing more correctly. (thanks to
2879 * More progress has been made toward bootstrapping under CLISP. (thanks
2880 to Christophe Rhodes)
2881 * The fasl file format has changed again, because dynamic loading
2882 on OpenBSD (which has non-ELF object files) motivated some cleanups
2883 in the way that foreign symbols are transformed and passed around.
2884 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2885 is no longer treated as whitespace by the reader, but instead as
2886 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
2887 |A
\x7fB|, instead of A as it used to.
2889 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2890 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2891 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2892 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2893 this as of 0.7.1.45, but was left out of the previous news section
2894 (thanks to Dan Barlow)
2895 * SBCL now runs on the Solaris operating system on SPARC architectures
2896 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2897 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2898 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2899 * SPARC backend cleanups, allowing builds of cores optimized for V8
2900 and V9 SPARCS, and also emission of code targeted to a particular
2901 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2903 * SBCL is closer to bootstrapping under CLISP, thanks to various
2904 fixes by Christophe Rhodes.
2905 * The fasl file format has changed again, to allow the compiler's
2906 INFO database to support symbol macros.
2907 * The user manual (in doc/) is formatted into HTML more nicely.
2908 (thanks to coreythomas)
2909 * The system is smarter about SUBTYPEP relationships, especially
2910 those involving NOT types (including types such as ATOM which are
2911 represented internally using NOT types). Thus SUBTYPEP is less
2912 likely to return (VALUES NIL NIL) in general, and in particular
2913 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2915 * The fasl file format has changed again, because the internal
2916 representation of types now includes a new slot to support the new
2917 SUBTYPEP-of-NOT-types logic.
2918 * (not a change in the main branch of SBCL, but a related prototype
2919 which can hopefully be merged into the main branch of SBCL in the
2920 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2921 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2922 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2923 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2924 dispatch macro character combinations. (thanks to Alexey Dejneka)
2925 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2926 that superficially look like logical namestrings correctly.
2927 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2929 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2930 default supplied-p) for &optional and &key arguments. (thanks to
2933 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2934 * incompatible change: The compiler is now less aggressive about
2935 tail call optimization, doing it only when (> SPACE DEBUG) or
2936 (> SPEED DEBUG). (This is an incompatible change because there are
2937 programs which relied on the old CMU-CL-style behavior to optimize
2938 away their unbounded recursion which will now die of stack overflow.)
2939 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2940 for non-GENCGC systems has been increased to 20M (since that
2941 seems much closer to the likely performance optimum for modern
2942 systems than the old 4M value was)
2943 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2944 and *DEBUG-PRINT-LEVEL*
2945 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2946 of CMU CL's support for SPARC, and various endianness and other
2947 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2948 * new syntactic sugar for the Unix command line: --load foo.bar is now
2949 an alternate notation for --eval '(load "foo.bar")'.
2951 ** The system now detects stack overflow and handles it gracefully,
2952 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2953 optimization settings. (This is a good thing in general, and
2954 its introduction in this version should be particularly timely
2955 for anyone whose code fails because of suppression of tail
2957 ** The system now hunts for the C variable "environ" in a more
2958 devious way, to avoid segfaults when the C library version differs
2959 between compile time and run time. (thanks to Christophe Rhodes)
2960 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2961 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2962 ** The compiler no longer issues bogus style warnings for undefined
2963 classes in the same source file as the DEFCLASSes which defined
2964 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2966 ** fixes in CONDITION class precedence list for undefined function
2967 errors (thanks to Alexei Dejneka)
2968 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2969 correctly. (thanks to Dan Barlow)
2970 ** portability fixes aiming at bootstrapping under CLISP (thanks
2971 to Dave McDonald and Christophe Rhodes)
2972 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2973 ** fixes in type translation and and type inference (thanks to
2975 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2976 ** various fixes in the new ports (thanks to Dan Barlow)
2977 * several changes related to debugging:
2978 ** suppression of tail recursion, as noted above
2979 ** stack overflow detection, as noted above
2980 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2981 is now the default. (For some time encapsulation has been more
2982 reliable than the breakpoint-based :ENCAPSULATE NIL
2983 implementation, at least on X86 systems; and I just noticed that
2984 encapsulation also seems closer to the spirit of the ANSI
2987 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2989 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2990 up properly again. (There was a packaging bug in 0.7.0 which
2991 left their definitions in SB-SYS::LOAD-FOREIGN and
2992 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2993 vital for most things which interface to C-level interfaces,
2994 like extensions working with sockets or databases or
2995 Perl-compatible regexes or whatever, and the need to fix
2996 this bug is the main reason that 0.7.1 was released so
2998 ** DEFGENERIC is now choosier about the methods it redefines, so that
2999 reLOADing a previously-LOADed file containing DEFGENERICs does
3000 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3001 cycle now works as it should. (thanks to Alexey Dejneka)
3002 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3003 fixed by Christophe Rhodes. (He actually submitted this patch
3004 months ago, and I delayed until after 0.7.0.)
3005 ** Bug 111 (internal compiler confusion about runtime checks on
3006 FUNCTION types) was fixed by Alexey Dejneka.
3007 * Some internal cleanups (getting rid of variables which aren't
3008 needed now that the byte interpreter is gone) caused the fasl
3009 file format number to change again.
3011 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3012 * major incompatible change: The default fasl file extension, i.e. the
3013 default extension for files produced by COMPILE-FILE, has changed
3014 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3016 ** There are many changes in the implementation of the compiler.
3017 SBCL is now essentially a compiler-only implementation of ANSI
3018 Common Lisp. EVAL still "interprets" a few special cases, but
3019 almost all the interesting cases are handled by creating
3020 a LAMBDA expression, calling COMPILE on it, then calling
3021 FUNCALL on the result.
3022 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3023 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3024 Since the code is newer, there might still be some new bugs
3025 (though not as many as before Martin Atzmueller's fixes:-). But
3026 the new code is substantially simpler and clearer, and hopefully
3027 any remaining bugs will be simpler, less fundamental, and more
3028 fixable then the bugs in the old code.
3029 ** The revised compiler is still a little unsteady on its feet.
3031 *** The debugging information it produces (particularly the names
3032 of FUNCTION objects) is sometimes much less useful than what
3033 the old compiler produced.
3034 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3035 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3036 within a MACROLET) has been temporarily weakened.
3037 ** There are new compiler optimizations for various functions:
3038 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3039 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3040 *** the math functions TRUNCATE, FLOOR, and CEILING
3041 *** the function-of-all-trades COERCE
3042 Mostly these should be transparent, but there's one
3043 potentially-annoying problem (bug 117): when the compiler
3044 inline-expands a function and does type analysis on the result,
3045 it can create control paths which have type mismatches, and
3046 when it can't prove that those control paths aren't taken,
3047 it will issue WARNINGs about the type mismatches. This is
3048 a particular problem in practice for the new sequence functions.
3049 It's not clear how this should be fixed, and for now, a
3050 workaround is given in the entry for 117 in the BUGS file.
3051 ** (Because of the interaction between the two previous items --
3052 occasional inlining problems and new inline expansions -- some
3053 of the new sequence function optimizations won't really kick in
3054 completely until debugging information, and then inlining, are
3055 straightened out in some future version.)
3056 * minor incompatible changes:
3057 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3058 DIRECTORY no longer implicitly promotes NIL slots of its
3059 pathname argument to :WILD. In particular, when you ask for the
3060 contents of a directory (which you used to be able to do without
3061 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3062 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3063 ** changes in behavior that ANSI explicitly defines to be
3064 implementation dependent:
3065 *** The new compiler-only implementation still conforms with ANSI,
3066 but acts a little different than before. Besides the obvious
3067 changes in performance tradeoffs (that the cost per form passed
3068 to EVAL has gone up, and the cost per form executed by EVAL
3069 has gone down), the behavior of the system changes a little
3070 because there are no longer any interpreted function objects.
3071 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3072 e.g. doing COMPILE on the output of interactive DEFUN is
3074 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3076 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3077 has changed from (:CL) to NIL.
3078 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3079 e.g. #\Newline and #\Tab, don't have names specified in the
3080 ANSI Common Lisp standard, is now based on their ASCII symbolic
3081 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3082 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3083 are no longer used for output.
3084 ** changes in internal implementation constants:
3085 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3086 4 million. (If your application spends a lot of time GCing and
3087 you have a lot of RAM, you might want to experiment with
3088 increasing it even more.)
3089 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3090 However, almost all old code should still continue to work without
3091 immediate update, as SB-C-CALL is now a (deprecated) nickname
3093 ** Old operator names in the style DEF-FOO are now deprecated in
3094 favor of new corresponding names DEFINE-FOO, for consistency with
3095 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3096 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3097 internal symbols, but a few supported extensions like
3098 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3099 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3100 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3101 etc. as you get deeper into recursive calls to the debugger
3102 command loop, instead of the old "5]", "5]]", "5]]]"
3103 sequence. (I was motivated to do this when squabbles between
3104 ILISP and SBCL left me very deeply nested in the debugger. In the
3105 short term, this change will probably provoke more ILISP/SBCL
3106 squabbles, but hopefully it will be an improvement in the long run.)
3107 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3108 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3109 function, since declaiming return types involving aliens is
3110 (1) annoyingly messy to do by hand and (2) vital to efficient
3111 compilation of code which calls such functions.
3112 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3113 longer reexported by the SB-EXT package. They're solely useful
3114 for alien code, so it seems more logical that you should get
3115 them from the SB-ALIEN package, not in SB-EXT.
3116 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3117 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3118 features. Instead, the code that they used to control is always
3119 built into the system.
3120 * many other bug fixes
3121 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3122 advantage of the new EVAL-WHEN stuff and to clean them up in
3123 general, and they are now more ANSI-compliant in a number of
3124 ways. Martin Atzmueller is responsible for a lot of this.
3125 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3127 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3128 *** correct ERROR type for various file operations
3129 *** some fixes for Lisp streams
3130 *** DEFMETHOD syntax checking
3131 *** changing old weird representation of debug information as
3132 strings (which, among their other deficiencies, don't transform
3133 correctly when you rename packages, and don't change their
3134 print representation when you change things like *PACKAGE*
3135 and *PRINT-LENGTH*) to symbols and lists of symbols
3136 He also made several improvements and fixed several bugs in DESCRIBE.
3137 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3139 *** misbehavior of WRITE-STRING/WRITE-LINE
3140 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3141 and several other LOOP problems as well
3142 *** DIRECTORY when similar filenames are present
3143 *** DEFGENERIC with :METHOD options
3144 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3145 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3146 *** argument ordering in FIND with :TEST option
3147 *** mishandled package designator argument in APROPOS-LIST
3148 *** various problems in the backquote readmacro
3149 *** a bug in APROPOS
3150 *** probably some others that I'm not describing very well here,
3151 since the CVS log documents them by reference to sbcl-devel
3152 messages, and the SourceForge archives aren't working well.:-(
3153 ** Dan Barlow improved the Alpha port (and is making progress on the
3154 PPC port, for those of you who think different).
3155 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3156 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3157 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3158 continues to work on the SPARC port (for those of you in a position
3159 to look down upon our little PC-compatible boxes from a great height).
3160 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3161 and uses that copy, rather than the current dynamic value, when
3162 it's trying to decide whether to truncate output. Thus e.g.
3163 (let ((*print-lines* 50))
3164 (pprint-logical-block (stream nil)
3166 (let ((*print-lines* 8))
3167 (print (aref possiblybigthings i) stream)))))
3168 should now truncate the logical block only at 50 lines, instead of
3169 often truncating it at 8 lines, as it did before.
3170 * The doc/cmucl/ directory, containing old CMU CL documentation
3171 from the time of the fork, is no longer part of the base system.
3172 SourceForge has shut down its anonymous FTP service, and with it
3173 my original plan for distributing the old CMU CL documentation
3174 there. For now, if you need these files you can download an old
3175 SBCL source release and extract them from it.
3176 * The fasl file version number changed again, for dozens of reasons,
3177 some of which are apparent above.
3179 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3180 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3181 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3182 patch, so that DISASSEMBLE output is much nicer.
3183 * The code in the SB-PROFILE package now seems reasonably stable.
3184 I still haven't decided what the final interface should look like
3185 (I'd like PROFILE to interact cleanly with TRACE, since both
3186 facilities use function encapsulation) but if you have a need
3187 for profiling now, you can probably use it successfully with
3188 the current CMU-CL-style interface.
3189 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3190 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3191 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3192 link now returns the dangling link itself, and for similar
3193 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3194 link itself. (In these cases the old code signalled an error and
3195 looped endlessly, respectively.) Thus, DIRECTORY now works even
3196 in the presence of dangling and cyclic symbolic links.
3197 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3198 is now a supported extension again, since the consensus on
3199 sbcl-devel was that it can be useful for ordinary development
3200 work, not just for debugging SBCL itself.
3201 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3202 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3203 that currently-defined functions might be redefined later with
3204 different return types.
3205 * Hash tables can be printed readably, as inspired by CMU CL code
3206 of Eric Marsden and SBCL code of Martin Atzmueller.
3207 * better error handling in CLOS method combination, thanks to
3208 Martin Atzmueller porting Pierre Mai's CMU CL patches
3209 * more overflow fixes for >16Mbyte I/O buffers
3210 * A bug in READ has been fixed, so that now a single Ctrl-D
3211 character suffices to cause end-of-file on character streams.
3212 In particular, now you only need one Ctrl-D at the command
3213 line (not two) to exit SBCL.
3214 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3216 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3217 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3218 does the right thing.
3219 * The implementation of some type tests, especially for CONDITION
3220 types, is now tidier and maybe faster, due to CMU CL code
3221 originally by Douglas Crosher, ported by Martin Atzmueller.
3222 * Some math functions have been fixed, and there are new
3223 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3224 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3225 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3226 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3227 likely to be built into sbcl-0.7.0.)
3228 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3229 CL, ported to SBCL by Martin Atzmueller)
3230 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3231 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3232 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3233 * There's a new slam.sh hack to shorten the edit/compile/debug
3234 cycle for low-level changes to SBCL itself, and a new
3235 :SB-AFTER-XC-CORE target feature to control the generation of
3236 the after-xc.core file needed by slam.sh.
3237 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3238 COMPILE-FILE is no longer supported, so that now every function
3239 gets an entry point, so that block compilation looks a little
3240 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3241 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3242 returns the number of bytes consed since the system started,
3243 rather than the number consed since the first time the function
3244 was called. (The new definition parallels ANSI functions like
3245 CL:GET-INTERNAL-RUN-TIME.)
3246 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3247 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3248 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3249 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3250 * Fasl file version numbers are now independent of the target CPU,
3251 since historically most system changes which required version
3252 number changes have affected all CPUs equally. Similarly,
3253 the byte fasl file version is now equal to the ordinary
3256 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3257 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3258 is no longer recognized. I apologize for this, because it was
3259 listed in SB-EXT as a supported extension, but I found that
3260 its existing behavior was poorly specified, as well as incorrectly
3261 specified, and it looked like too much of a mess to straighten it
3262 out. I have enough on my hands trying to get ANSI stuff to work..
3263 * many patches ported from CMU CL by Martin Atzmueller, with
3264 half a dozen bug fixes in pretty-printing and the debugger, and
3265 half a dozen others elsewhere
3266 * fixed bug 13: Floating point infinities are now supported again.
3267 They might still be a little bit flaky, but thanks to bug reports
3268 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3269 as flaky as they were.
3270 * The --noprogrammer command line option is now supported. (Its
3271 behavior is slightly different in detail from what the old man
3272 page claimed it would do, but it's still appropriate under the
3273 same circumstances that the man page talks about.)
3274 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3275 are now supported, and enabled by default. Thus, the compiler can
3276 handle many floating point and complex operations much less
3277 inefficiently. (Thus e.g. you can implement a complex FFT
3279 * The compiler now detects type mismatches between DECLAIM FTYPE
3280 and DEFUN better, and implements CHECK-TYPE more correctly, and
3281 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3282 thanks to patches from Martin Atzmueller.
3283 * various fixes to make the cross-compiler more portable to
3284 ANSI-conforming-but-different cross-compilation hosts (notably
3285 Lispworks for Windows, following bug reports from Arthur Lemmens)
3286 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3287 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3288 fixed thanks to Pierre Mai's CMU CL patches.
3289 * fixes to keep the system from overflowing internal counters when
3290 it tries to use i/o buffers larger than 16M bytes
3291 * fixed bug 45a: Various internal functions required to support
3292 complex special functions have been merged from CMU CL sources.
3293 (When I was first setting up SBCL, I misunderstood a compile-time
3294 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3295 * improved support for type intersection and union, fixing bug 12
3296 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3297 more obscure bugs as well
3298 * some steps toward byte-compiling non-performance-critical
3299 parts of the system, courtesy of patches from Martin Atzmueller
3300 * Christophe Rhodes has made some debian packages of sbcl at
3301 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3302 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3303 stable, but are nonetheless usable. When he's ready, I'd be happy
3304 to add them to the SourceForge "File Releases" section. (And if
3305 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3306 * new fasl file format version number (because of changes in
3307 internal representation of (OR ..) types to accommodate the new
3308 support for (AND ..) types, among other things)
3310 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3311 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3313 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3315 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3316 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3317 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3318 * DESCRIBE now works on structure objects again.
3319 * Most function call argument type mismatches are now handled as
3320 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3321 know whether the function will be redefined before the call is
3322 executed. (The compiler could flag local calls with full WARNINGs,
3323 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3324 it doesn't keep track of enough information to know whether calls
3325 are local in this sense.)
3326 * Compiler output is now more verbose, with messages truncated
3327 later than before. (There should be some supported way for users
3328 to override the default verbosity, but I haven't decided how to
3329 provide it yet, so this behavior is still controlled by the internal
3330 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3331 src/compiler/ir1util.lisp.)
3332 * Fasl file format version numbers have increased again, because
3333 support for the Gray streams extension changes the layout of the
3334 system's STREAM objects.
3335 * The Gray subclassable streams extension now works, thanks to a
3336 patch from Martin Atzmueller.
3337 * The full LOAD-FOREIGN extension (not just the primitive
3338 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3339 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3340 but like most other programs, it defaults to copying the Unix
3341 environment from the original process instead of starting the
3342 new process in an empty environment.
3343 * Extensions which manipulate the Unix environment now support
3344 an :ENVIRONMENT keyword option which doesn't smash case or
3345 do other bad things. The CMU-CL-style :ENV option is retained
3346 for porting convenience.
3347 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3348 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3350 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3352 * A patch from Martin Atzmueller seems to have solved the SIGINT
3353 problem, and as far as we know, signal-handling now works cleanly.
3354 (If you find any new bugs, please report them!)
3355 * The system no longer defaults Lisp source file names to types
3356 ".l", ".cl", or ".lsp", but only to ".lisp".
3357 * The compiler no longer uses special default file extensions for
3358 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3359 seems to expect a single default extension for all compiled code,
3360 and there's no compelling reason to try to stretch the standard
3361 to allow two different extensions.) Instead, byte-compiled files
3362 default to the same extension as native-compiled files.
3363 * Fasl file format version numbers have increased again, because
3364 a rearrangement of internal implementation packages made some
3365 dumped symbols in old fasl files unreadable in new cores.
3366 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3367 many fewer weird special cases.
3368 * Bug #17 (differing COMPILE-FILE behavior between logical and
3369 physical pathnames) has been fixed, and some related misbehavior too,
3370 thanks to a patch from Martin Atzmueller.
3371 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3372 by Tim Moore, ported to SBCL by Martin Atzmueller.
3373 * Martin Atzmueller fixed several filesystem-related problems,
3374 including bug #36, in part by porting CMU CL patches, which were
3375 written in part by Paul Werkowski.
3376 * More compiler warnings in src/runtime/ are gone, thanks to
3377 more patches from Martin Atzmueller.
3378 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3381 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3383 * DESCRIBE now works on CONDITION objects.
3384 * The debugger now handles errors which arise when trying to print
3385 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3387 * The build system now uses an additional file, customize-target-features.lisp,
3388 to allow local modifications to the target *FEATURES* list. (The point of
3389 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3390 debugging features enabled, without having to worry about propagating it
3391 into everyone's system when I do a "cvs update".) When no
3392 customize-target-features.lisp file exists, the target *FEATURES* list
3393 should be constructed the same way as before.
3394 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3395 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3396 warn about behavior which is undefined under the ANSI spec. Note: This
3397 is specified by ANSI, but it's not very popular with programmers.
3398 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3399 macro in the SBCL sources for an example of a workaround which you
3400 might use to make portable ANSI-standard code which does what you want.
3401 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3402 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3403 when it's used as a non-toplevel form. (This is required in order
3404 to implement the DEFCONSTANT-EQX macro.)
3405 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3406 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3407 funny things, due to bugs in the implementation of EVAL-WHEN
3408 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3409 non-toplevel DEFCONSTANTs are uncommon.)
3410 * The core file version number and fasl file version number have been
3411 incremented, because the old noncompliant DEFCONSTANT behavior involved
3412 calling functions which no longer exist, and because I also took the
3413 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3414 * fixed bug 1 (error handling before read-eval-print loop starts), and
3415 redid debugger restarts and related debugger commands somewhat while
3417 ** The QUIT debugger command is gone, since it did something
3418 rather different than the SB-EXT:QUIT command, and since it never
3419 worked properly outside the main toplevel read/eval/print loop.
3420 Invoking the new TOPLEVEL restart provides the same functionality.
3421 ** The GO debugger command is also gone, since you can just invoke
3422 the CONTINUE restart directly instead.
3423 ** The TOP debugger command is also gone, since it's redundant with the
3424 FRAME 0 command, and since it interfered with abbreviations for the
3426 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3428 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3429 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3430 some porting work by Martin Atzmueller)
3431 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3432 with a boatload of patches which he ported to SBCL. Now that those
3434 ** The system tries to make sure that its low-priority messages
3435 are prefixed by semicolons, to help people who like to use
3436 syntax highlighting in their ILISP buffer. (This patch
3437 was originally due to Raymond Toy.)
3438 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3439 patches originally written by Raymond Toy.
3440 ** The compiler understands coercion between single-value and
3441 multiple-VALUES type expressions better, getting rid of some very
3442 weird behavior, thanks to patches originally by Robert MacLachlan
3443 and Douglas Crosher.
3444 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3445 lambda lists, thanks to a patch originally by Pierre Mai.
3446 ** The system no longer bogusly warns about "abbreviated type
3448 ** The compiler gets less confused by inlining and RETURN-FROM,
3449 thanks to some patches originally by Tim Moore.
3450 ** The system no longer hangs when dumping circular lists to fasl
3451 files, thanks to a patch originally from Douglas Crosher.
3452 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3453 undefined function error.
3454 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3455 in structure slot initforms) for now, documented workaround instead:-|
3456 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3457 * fixed bug 5: added stubs for various Gray stream functions called
3458 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3459 installed, at least appropriate type errors are generated
3460 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3461 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3462 consistently in DEFMETHOD forms.
3463 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3464 it doesn't seem to affect SBCL after all
3465 * The C runtime system now builds with better optimization and many
3466 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3468 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3470 * The system is now under CVS at SourceForge (instead of the
3471 CVS repository on my home machine).
3472 * The new signal handling code has been tweaked to treat register
3473 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3474 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3475 downstream, e.g. in debug-int.lisp, has implicit dependencies
3476 on the unsignedness of integer representation of machine words,
3477 and that caused the system to bomb out with infinite regress
3478 when trying to recover from type errors involving signed values,
3479 e.g. (BUTLAST '(1 2 3) -1).)
3480 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3481 (This was one of the bugs Peter Van Eynde reported back in July.)
3482 * The system now uses code inspired by Colin Walters' O(N)
3483 implementation of MAP (from the cmucl-imp@cons.org mailing
3484 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3485 inline the MAP operation, and there is more than one
3486 sequence argument to the MAP call (so that it can't just
3487 do ETYPECASE once and for all based on the type of the
3488 single sequence argument). (The old non-inline implementation
3489 of the general M-argument sequence-of-length-N case required
3490 O(M*N*N) time when any of the sequence arguments were LISTs.)
3491 * The QUIT :UNIX-CODE keyword argument has been renamed to
3492 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3493 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3495 * Raymond Wiker's patches to port dynamic loading from Linux to
3496 FreeBSD have been added.
3497 * The BUGS file is now more nearly up to date, thanks in large part
3498 to Martin Atzmueller's review of it.
3499 * The debugger now flushes standard output streams before it begins
3500 its output ("debugger invoked" and so forth).
3501 * The core version number and fasl file version number have both
3502 been incremented, because of incompatible changes in the layout
3504 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3505 used to not be called for a saved Lisp image.)
3506 * Martin Atzmueller's version of a patch to fix a compiler crash,
3507 as posted on sbcl-devel 13 September 2000, has been installed.
3508 * Instead of installing Martin Atzmueller's patch for the
3509 compiler transform for SUBSEQ, I deleted the compiler transform,
3510 and transforms for some similar consing operations.
3511 * A bug in signal handling which kept TRACE from working on OpenBSD
3513 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3514 away by constant folding
3515 * The system now defines its address space constants in one place
3516 (in the Lisp sources), and propagates them automatically elsewhere
3517 (through GENESIS and the sbcl.h file). Therefore, patching the
3518 address map is less unnecessarily tedious and error-prone. The
3519 Lisp names of address space constants have also been systematized.
3520 * CVS tags like dollar-Header-dollar have been removed from
3521 the sources, because they have never saved me trouble and
3522 they've been source of trouble working with patches and other
3523 diff-related operations.
3524 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3525 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3527 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3529 * The system has been ported to OpenBSD.
3530 * The system now compiles with a simple "sh make.sh" on the systems
3531 that it's supported on. I.e., now you no longer need to tweak
3532 text in the target-features.lisp-expr and symlinks in src/runtime/
3533 by hand, the make.sh takes care of it for you.
3534 * The system is no longer so grossly inefficient when compiling code
3535 involving vectors implemented as general (not simple) vectors (VECTOR T),
3536 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3537 which dares to use the various sequence functions on non-simple
3538 vectors, takes less of a performance hit.
3539 * There is now a primitive type predicate VECTOR-T-P
3540 to test for the (VECTOR T) type, so that e.g.
3541 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3542 can now be compiled with some semblance of efficiency. (The old code
3543 turned the type declaration into a full call to %TYPEP at runtime!)
3544 * AREF on (VECTOR T) is still not fast, since it's still compiled
3545 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3546 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3547 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3548 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3549 without making full calls to SUBTYPEP at runtime.
3550 (Some analogous efficiency issues for non-simple vectors specialized to
3551 element types other than T, or for non-simple multidimensional arrays,
3552 have not been addressed. They could almost certainly be handled the
3553 same way if anyone is motivated to do so.)
3554 * The changes in array handling break binary compatibility, so
3555 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3556 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3557 instead of (VALUES T T).
3558 * By following the instructions that Dan Barlow posted to sbcl-devel
3559 on 2 July 2000, I was able to enable primitive dynamic object
3560 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3561 functionality is not implemented (since it calls ld to resolve
3562 library references automatically, requiring RUN-PROGRAM for its
3563 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3564 try to resolve library references) is now supported.
3565 * The system now flushes the standard output streams when it terminates,
3566 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3567 them at several other probably-convenient times, e.g. in each pass of
3568 the toplevel read-eval-print loop, and after evaluating a form given
3569 as an "--eval" command-line option. (These changes were motivated by a
3570 discussion of stream flushing issues on cmucl-imp in August 2000.)
3571 * The source transform for TYPEP of array types no longer assumes
3572 that an array whose element type is a not-yet-defined type
3573 is implemented as an array of T, but instead punts, so that the
3574 type will be interpreted at runtime.
3575 * There is now some support for cross-compiling in make.sh: each of
3576 the phases of make.sh has its own script. (This should be transparent
3577 to people doing ordinary, non-cross-compile builds.)
3578 * Since my laptop doesn't have hundreds of megabytes of memory like
3579 my desktop machine, I became more motivated to do some items on
3580 my to-do list in order to reduce the size of the system a little:
3581 ** Arrange for various needed-only-at-cold-init things to be
3582 uninterned after cold init. To support this, those things have
3583 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3584 symbols with such names are now uninterned after cold init).
3585 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3586 fasl files for cold load.
3587 ** Remove the old compiler structure pooling code (which used to
3588 be conditional on the target feature :SB-ALLOC) completely.
3589 ** Redo the representation of some data in cold init to be more compact.
3590 (I also looked into supporting byte compiled code at bootstrap time,
3591 which would probably reduce the size of the system a lot, but that
3592 looked too complicated, so I punted for now.)
3593 * The maximum signal nesting depth in the src/runtime/ support code has
3594 been reduced from 4096 to 256. (I don't know any reason for the very
3595 large old value. If the new smaller value turns out to break something,
3596 I'll probably just bump it back up.)
3597 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3599 * Many, many bugs reported by Peter Van Eynde have been added to
3600 the BUGS list; some have even been fixed.
3601 * While enabling dynamic object file loading, I tried to make the
3602 code easier to understand, renaming various functions and variables
3603 with less ambiguous names, and changing some function calling
3604 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3605 * While trying to figure out how to do the OpenBSD port, I tried to
3606 clean up some of the code in src/runtime/. In particular, I dropped
3607 support for non-POSIX signal handling, added various comments,
3608 tweaked the code to reduce the number of compilation warnings, and
3609 renamed some files to increase consistency.
3610 * To support the new automatic configuration functionality in make.sh,
3611 the source file target-features.lisp-expr has been replaced with the
3612 source file base-target-features.lisp-expr and the machine-generated
3613 file local-target-features.lisp-expr.
3614 * fixed a stupid quoting error in make.sh so that using CMU CL
3615 "lisp -batch" as cross-compilation host works again
3617 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3619 * DESCRIBE no longer tries to call itself recursively to describe
3620 bound/fbound values, so that it no longer fails on symbols which are
3621 bound to themselves (like keywords, T, and NIL).
3622 * DESCRIBE now works on generic functions.
3623 * The printer now prints less-screwed-up representations of closures
3624 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3625 * A private symbol is used instead of the :EMPTY keyword previously
3626 used to mark empty slots in hash tables. Thus
3627 (DEFVAR *HT* (MAKE-HASH-TABLE))
3628 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3629 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3630 now does what ANSI says that it should. (You can still get
3631 similar noncompliant behavior if bang on the hash table
3632 implementation with all the symbols you get back from
3633 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3634 This breaks binary compatibility, since tests for equality to
3635 :EMPTY are wired into things like the macroexpansion of
3636 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3638 * There's now a minimal placeholder implementation for CL:STEP,
3639 as required by ANSI.
3640 * An obscure bug in the interaction of the normal compiler, the byte
3641 compiler, inlining, and structure predicates has been patched
3642 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3643 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3645 * Missing ordinary arguments in a macro call are now detected even
3646 when the macro lambda list contains &KEY or &REST.
3647 * The debugger no longer complains about encountering the top of the
3648 stack when you type "FRAME 0" to explicitly instruct it to go to
3649 the top of the stack. And it now prints the frame you request even
3650 if it's the current frame (instead of saying "You are here.").
3651 * As specified by ANSI, the system now always prints keywords
3652 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3654 * The default initial SIZE of HASH-TABLEs is now smaller.
3655 * Type information from CLOS class dispatch is now propagated
3656 into DEFMETHOD bodies, so that e.g.
3657 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3659 is now basically equivalent to
3660 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3661 (DECLARE (TYPE SINGLE-FLOAT X))
3663 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3664 operation, without having to do run-time type dispatch.
3665 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3666 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3667 * Since I don't seem to be making big file reorganizations very often
3668 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3669 is overflowing my ability to conveniently back them up), I've finally
3670 checked the system into CVS. (The CVS repository is on my home system,
3671 not at SourceForge -- putting it on SourceForge might come later.)
3672 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3673 high-level GC-NOTIFY-FOO functions send their output. (There's
3674 still very little control of where low-level verbose GC functions
3675 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3676 less than it used to -- the GC-NOTIFY-FOO functions are now under
3677 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3678 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3679 in only one place in the source code, and propagates it automatically
3680 everywhere that it's needed. Thus e.g. when I bump the version from
3681 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3682 * The C source files now include boilerplate legalese and documentation
3683 at the head of each file (just as the Lisp source files already did).
3684 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3685 to his page will be replaced with a link to his new CLiki service.
3687 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3689 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3690 * The build process now looks for GNU make under the default name "gmake",
3691 instead of "make" as it used to. If GNU make is not available as "gmake"
3692 on your system, you can change this default behavior by setting the
3693 GNUMAKE environment variable.
3694 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3695 can build without error under CMU CL.
3697 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3699 * There is now a partial SBCL user manual (with some new text and some
3700 text cribbed from the CMU CL manual).
3701 * The beginnings of a profiler have been added (starting with the
3702 CMU CL profiler and simplifying and cleaning up). Eventually the
3703 main interface should be through the TRACE macro, but for now,
3704 it's still accessed through vaguely CMU-CL-style functions and macros
3705 exported from the package SB-PROFILE.
3706 * Some problems left over from porting CMU CL to the new
3707 cross-compilation bootstrap process have been cleaned up:
3708 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3709 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3710 part when getting rid of the compiler *BACKEND* stuff.)
3711 ** The value of *NULL-TYPE* was screwed up, because it was
3712 being initialized before the type system knew the final
3713 definition of the 'NULL type. This screwed up several key
3714 optimizations in the compiler, causing inefficiency in all sorts
3715 of places. (I found it because I wanted to understand why
3716 GET-INTERNAL-RUN-TIME was consing.)
3717 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3718 PROCLAIM FTYPE information. Unfortunately this broke binary
3719 compatibility again, since now the forms output by DEFGENERIC
3720 to refer to functions which didn't exist in 0.6.3.
3721 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3722 can use the new (as of 0.6.3) transform for SOME into MAP into
3724 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3725 values to (INTEGER 0 1000000), so that the time code will no longer
3726 occasionally get blown up by Linux returning 1000000 microseconds
3727 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3728 its output conform to the ANSI spec. (Alas, this makes its output
3729 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3730 * A full call to MAP NIL with a single sequence argument no longer conses.
3731 * fixes to problems pointed out by Martin Atzmueller:
3732 * The manual page no longer talks about multiprocessing as though
3733 it were currently supported.
3734 * The ILISP support patches have been removed from the distribution,
3735 because as of version 5.10.1, ILISP now supports SBCL without us
3736 having to maintain patches.
3737 * added a modified version of Raymond Toy's recent CMU CL patch for
3738 EQUALP comparison of HASH-TABLE
3740 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3742 * The system still can't cross-compile itself with
3743 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3744 entails), but at least it can compile more of itself that way
3745 than it used to be able to, and various buglets which were uncovered
3746 by trying to cross-compile itself that way have now been fixed.
3747 * This release breaks binary compatibility again. This time
3748 at least I've incremented the FASL file format version to 2, so that the
3749 problem can be detected reliably instead of just causing weird errors.
3750 * various new style warnings:
3751 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3752 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3753 ** using the lexical binding of a variable named in the *FOO* style
3754 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3755 as specified by ANSI.
3756 * *RANDOM-STATE* is no longer automatically initialized from
3757 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3758 default behavior of the system is to repeat its behavior every time
3759 it's run. If you'd like to change this behavior, you can always
3760 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3761 old convention there was no comparably easy way to get the system to
3762 repeat its behavior every time it was run.
3763 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3764 FTYPE declarations has been removed, in favor of their ANSI
3765 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3766 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3767 the types of their sequence arguments can be determined at compile time.
3768 This is done through a new open code expansion for MAP which eliminates
3769 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3770 argument types can be determined at compile time.
3771 * The optimizer now transforms COERCE into an identity operation when it
3772 can prove that the coerced object is already of the correct type. (This
3773 can be a win for machine generated code, including the output of other
3774 optimization transforms, such as the MAP transform above.)
3775 * Credit information has been moved from source file headers into CREDITS.
3776 * Source file headers have been made more standard.
3777 * The CASE macro now compiles without complaining even when it has
3780 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3782 * (Note that the way that the PCL macroexpansions were rewritten
3783 to accommodate the change in DEFGENERIC below breaks binary
3784 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3785 not run under sbcl-0.6.2. Once we get out of alpha releases,
3786 i.e. hit release 1.0.0, we'll probably try to maintain binary
3787 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3788 and sbcl-1.4.4. Until then, however, it might be fairly common
3789 for maintenance releases to break binary compatibility.)
3790 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3792 * The reporting of unhandled signals has been changed to print some
3793 explanatory text as well as the report form. (Previously only
3794 the report form was printed.)
3795 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3796 it defines, so that the compiler no longer issues undefined function
3797 warnings for compiled-but-not-yet-loaded generic functions.
3798 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3799 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3800 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3801 section of the ANSI spec are supported.
3802 * The "" nickname for the "KEYWORD" package has been removed.
3803 The reader still handles symbol tokens which begin with a package marker
3804 as keywords, but it doesn't expose its mechanism for doing so in the
3805 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3806 * The system now issues STYLE-WARNINGs for contradictory TYPE
3807 proclamations. (Warnings for contradictory FTYPE proclamations would
3808 be nice too, but those can't be done usefully unless the type system
3809 is made smarter about FUNCTION types.)
3810 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3811 systematized, so that "*target-*.lisp is supposed to exist only on the
3812 target and imply that there's a related file which exists on the
3813 host, and *host-*.lisp is supposed to exist only on the host and imply
3814 that there's a related file which exists on the target. This involves a
3815 lot of renaming. Hopefully the acute confusion caused by the renaming
3816 will be justified by the reduction in chronic confusion..
3817 ** runtime-type.lisp -> early-target-type.lisp
3818 ** target-type.lisp -> late-target-type.lisp
3819 ** early-host-format.lisp -> early-format.lisp
3820 ** late-host-format.lisp -> late-format.lisp
3821 ** host-error.lisp -> misc-error.lisp
3822 ** early-error.lisp -> early-target-error.lisp
3823 ** late-error.lisp -> late-target-error.lisp
3824 ** host-defboot.lisp -> early-defboot.lisp
3825 ** code/misc.lisp -> code/target-misc.lisp
3826 ** code/host-misc.lisp -> code/misc.lisp
3827 ** code/numbers.lisp -> code/target-numbers.lisp
3828 ** code/early-numbers.lisp -> numbers.lisp
3829 ** early-host-type.lisp -> early-type.lisp
3830 ** late-host-type.lisp -> late-type.lisp
3831 ** host-typep.lisp -> typep.lisp
3832 ** load.lisp -> target-load.lisp
3833 ** host-load.lisp -> load.lisp
3834 ** host-disassem.lisp -> disassem.lisp
3835 ** host-insts.lisp -> insts.lisp
3836 ** byte-comp.lisp -> target-byte-comp.lisp
3837 ** host-byte-comp.lisp -> byte-comp.lisp
3838 ** host-signal.lisp -> signal.lisp
3839 ** host-defstruct.lisp -> defstruct.lisp
3840 ** late-target-type.lisp -> deftypes-for-target.lisp
3841 Furthermore, several other previously target-only files foo.lisp (e.g.
3842 hash-table.lisp and random.lisp) have been split into a target-and-host
3843 foo.lisp file and a target-only target-foo.lisp file, with their key type
3844 definitions in the target-and-host part, so that the cross-compiler will
3845 know more about target types.
3846 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3847 gone away. In their place are various *BACKEND-FOO* variables
3848 corresponding to the slots of the old structure.
3849 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3850 into the target SBCL has been fixed.
3851 * The chill.lisp system for loading cold code into a running SBCL
3853 * Support for the CMU CL "scavenger hook" extension has been removed.
3854 (It was undocumented and unused in the CMU CL sources that SBCL was
3855 derived from, and stale in sbcl-0.6.1.)
3856 * Various errors in the cross-compiler type system were detected
3857 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3858 (enabling various consistency checks). Many of them were fixed,
3859 but some hard problems remain, so the compiler is back to
3860 running without *TYPE-SYSTEM-INITIALIZED* for now.
3861 * As part of the cross-compiler type system cleanup, I implemented
3862 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3863 * I have started adding UNCROSS calls throughout the type system
3864 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3865 on cross-compiler input files will be able to go away, and various
3867 * CONSTANTP now returns true for quoted forms (as explicitly required
3870 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3872 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3873 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3874 (MAKE-LIST -1). In the longer term, it ought to become true
3875 that declarations are assertions even at SAFETY 1. For now, it's not
3876 quite true even at SAFETY 3, but it's at least more nearly true..
3877 (Note that this change seems to increases the size of the system by
3878 O(5%) and to decrease the speed of the compiler by 20% or more.)
3879 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3880 for the problem of printing dozens of lines of distracting information
3881 about low-level system machinery as part of the top stack frame
3882 on entry to the debugger when an undefined function was called.
3883 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3884 is not reset to COMMON-LISP-USER.
3885 * Compilation of stuff related to dyncount.lisp has been made conditional
3886 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3887 smaller. The various dyncount-related symbols have been moved into
3888 a new "SB-DYNCOUNT" package.
3889 * tty-inspect.lisp has been renamed to inspect.lisp.
3890 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3891 feature has gone away. (When we eventually port to other flavors of
3892 libc and/or Unix, we'll try to make the differences between flavors
3893 invisible at the user level.)
3894 * Various other *FEATURES* tags, and/or their associated conditionals,
3895 have been removed if obsolescent, or given better documentation, or
3896 sometimes given more-mnemonic names.
3898 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3900 * tidied up "make.sh" script
3901 * tidied up system directory structure
3902 * better "clean.sh" behavior
3903 * added doc/FOR-CMUCL-DEVELOPERS
3904 * many many small tweaks to output format, e.g. removing possibly-confusing
3905 trailing #\. character in DESCRIBE-INSTANCE
3906 * (EQUALP #\A 'A) no longer signals an error.
3907 * new hashing code, including EQUALP hashing
3908 * tidied up Lisp initialization and toplevel
3909 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3910 * command line argument processing
3911 * added POSIX-GETENV function to deal with Unix-ish environment variables
3912 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3913 terminating SBCL on EOF
3914 * non-verbose GC by default
3915 * There is no more "sbcl" shell script; the sbcl file is now the C
3916 runtime executable (just like CMU CL).
3917 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3919 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3920 symbol and package information as Lisp native symbol and package objects
3921 instead of strings naming symbols and strings naming packages. This way,
3922 whenever packages are renamed (as in warm init), debug information is
3923 transformed along with everything else.
3924 * tweaked the optimization policy declarations which control the building
3925 of SBCL itself. Now, among other things, the system no longer saves
3926 source location debugging information. (This helps two problems at once
3927 by reducing SBCL size and by keeping SBCL from trying to look for its
3928 sources -- which may not exist -- when reporting errors.)
3929 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3930 debugging and testing purposes
3931 * cleaned up printing, making the printer call PRINT-OBJECT for
3932 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3933 methods, giving nearly-ANSI behavior
3934 * converted almost all special variables to use *FOO* naming convention
3935 * deleted PARSE-TIME functionality, since it can be done portably
3936 * moved some files out of cold init into warm init
3937 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3939 * regularized formatting of source files
3940 * added an install.sh script
3941 * fixed ridiculous memory usage of cross-compiler by making
3942 compiler/alloc.lisp not try to do pooling unless it can hook
3943 itself into the GC of the cross-compilation host. Now the system
3944 builds nicely on my old laptop.
3945 * added :SB-ALLOC in target-features.lisp-expr
3946 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3947 was not implemented)
3948 * re-did condition handling and note reporting in the compiler. Notes
3949 are no longer handled by signalling conditions. Style warnings
3950 and warnings are handled more correctly and reported in such a way
3951 that it's easy to find one or the other in your output (so that you
3952 can e.g. figure out which of many problems caused COMPILE-FILE to
3954 * changed the severity of several compiler warnings from full WARNING
3955 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3956 compiler note reporting so that it doesn't use the condition system
3957 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3958 COMPILE-FILE command)
3959 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3960 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3961 no longer supported, since their implementation was deeply intertwingled
3962 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3963 * removed bogus "support" for compiler macros named (SETF FOO), and
3964 removed the compiler macro for SETF INFO (but only after making a fool
3965 of myself on the cmucl-imp mailing list by posting a bogus patch for
3966 DEFINE-COMPILER-MACRO..)
3967 * Compiled files containing forms which have side effects on the Lisp
3968 reader (such as DEFPACKAGE forms) are now handled more correctly.
3969 (Compiler queuing of top level lambdas has been suppressed by setting
3970 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3971 * deleted various currently-unused source files, e.g. gengc.lisp. They
3972 may be added back at some point e.g. when porting to other architectures,
3973 but until they are it's distracting to distribute them and to try to
3975 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3976 there were so many of them they're just distractions, and UNCROSS is
3977 known to be able to handle the current sources
3978 * moved PROFILE functionality into TRACE, so that it will be clear
3979 how the wrapping and unwrapping of functions when you profile them
3980 interacts with the wrapping and unwrapping of functions when you
3981 trace them. (Actually, the functionality isn't there yet, but at least
3982 the interface specification is there. Hopefully, the functionality will
3983 arrive with some future maintenance release.)
3984 * removed host-oops.lisp
3985 * changed signature of QUIT function to allow UNIX-CODE argument
3986 * fixed READ-SEQUENCE bug
3987 * tweaked verbose GC output so that it looks more like the progress
3988 output that ANSI specifies for functions like LOAD
3989 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3990 * added <http://sbcl.sourceforge.com> to the banner information printed by