1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
3 * new feature: source path information is generated for macro-expansion
4 errors for use in IDE's like Slime (thanks to Helmut Eller)
5 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
6 * bug fix: compiler does not loop forever on an invalid type in
8 * improvement: compilation of most CLOS applications is significantly
10 * optimization: added a limited bytecode compiler for simple toplevel
11 forms, speeding up compilation and FASL loading
13 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
14 * minor incompatible change: in sbcl-0.9.11 (but not earlier
15 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
16 into subclasses of the finalized class. Now user calls to
17 FINALIZE-INHERITANCE finalize just the one class, and calls by the
18 system return before any subclasses are finalized.
19 * minor incompatible change: The reader no longer ignores errors
20 regarding non-existent packages in #+ and #- feature tests.
21 * new feature: command line options --no-sysinit, --no-userinit to
22 inhibit loading the corresponding init files
23 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
24 for sending data through UDP sockets (thanks to François-René Rideau)
25 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
26 error is signalled instead (thanks to François-René Rideau)
27 * minor incompatible change: Error signalling behaviour of lexical
28 operations violating package locks has changed slightly. Refer to
29 documentation on package locks for details.
30 * enhancement: EVAL can process IF-expressions without resorting to the
32 * bug fix: Some operations involving SIN, COS, or TAN weren't being
33 constant-folded properly due to missing out-of-line functions.
34 (reported by Mika Pihlajamäki)
35 * bug fix: LISTEN sometimes returned T even in cases where no data was
36 immediately available from the stream
37 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
38 were reversed. (reported by Levente Mészáros)
39 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
40 when given a vector displaced to an adjustable array. (reported
42 * bug fix: derivation of float boundaries from numbers outside the
43 appropriate float range (reported by John Wiseman)
44 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
46 * bug fix: precomputing cache entries for generic functions with
47 some subclasses of specializers as yet invalid does not attempt to
48 fill a cache line with a negative offset. (reported by Levente
50 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
51 allow condition class objects as arguments to DOCUMENTATION and
52 (SETF DOCUMENTATION); only find and set documentation for
53 structure names for the STRUCTURE doc-type. (suggested by Gary
55 * improvements to the Win32/x86 port:
56 ** (user-homedir-pathname) and default initialization file
57 locations now know about the user's "Documents and Settings"
58 directory (thanks to Yaroslav Kavenchuk)
59 ** run-program is implemented (thanks to Mike Thomas)
60 ** sockets support (thanks to Timothy Ritchey)
61 ** better backtrace support (thanks to Alastair Bridgewater)
62 ** sb-grovel supported
63 ** asdf-install and sb-posix work somewhat
64 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
65 * improvements to the Solaris/x86 port:
66 ** works on Solaris 11/Solaris Express
67 ** floating-point exception handling support
68 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
69 * fixed some bugs revealed by Paul Dietz' test suite:
70 ** REMOVE-METHOD returns its generic function argument even when
71 the method is not one of the generic functions' methods.
72 ** objects declared as MEMBER types can be admissible arguments to
74 ** printing characters should simply be printed by the FORMAT ~:C
76 ** compiler failure when compiling functions with hairy constant
77 defaults for optional parameters.
78 ** compiler produces wrong code when MAYBE-INLINE-expanding a
79 function, which is already optimized.
81 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
82 * new platform: experimental support for SBCL x86/Darwin, including
83 MacOS X 10.4.5 on Intel.
84 * new feature: Unicode character names are now known to the system
85 (through CHAR-NAME and NAME-CHAR).
86 * new feature: the filesystem location of SBCL's core file is
87 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
88 * minor incompatible change: the contrib modules SB-POSIX and
89 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
90 this change is to make it easier to distribute
91 location-independent binaries.
92 * bug fix: as implied by AMOP, standardized classes no longer have
93 slots named by external symbols of public packages. (reported by
95 * optimization: calling functions via constant symbols -- as in
96 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
97 via the function object as in (FUNCALL #'FOO).
98 * optimization: on x86 and x86-64, the calling convention for the
99 case of dealing with an unknown number of values has been altered
100 to be friendlier to the prediction heuristics implemented,
101 particularly with reference to CALL and RET pairing. (thanks to
102 Alastair Bridgewater)
103 * optimization: on x86, the code for access to thread-local symbol
104 values and binding/unbinding of thread-local symbols is smaller.
105 (thanks to Alastair Bridgewater)
106 * enhancement: CONSTANTP is now able to determine constantness of
107 more complex forms, including calls to constant-foldable standardized
108 functions and some special forms beyond QUOTE.
109 * fixed bug: occasional GC crashes on Solaris/x86
110 * optimization: x86-64 supports stack allocation of results of simple
111 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
112 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
113 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
114 (thanks to James Knight)
115 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
116 by the standard (thanks to Stephen Compall)
118 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
119 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
120 be used for bundling the runtime and the core file into one
121 executable binary. This feature is not currently supported on all SBCL
122 platforms. (thanks to James Bielman and NIIMI Satoshi)
123 * new feature: a generational or ephemeral garbage collector is now
124 the default on the PowerPC platform (both Linux and Darwin). The
125 old Cheney (stop and copy) collector is a build-time option.
126 (thanks to Cyrus Harmon, after Raymond Toy)
127 * minor incompatible change: the method by which SBCL finds its
128 contributed modules has changed; it no longer relies on symbolic
129 links from an $SBCL_HOME/systems directory, but searches directly
130 in the subdirectories of $SBCL_HOME.
131 * enhancement: the dynamic heap size on the Linux/PPC platform is
132 markedly larger, even using the older Cheney garbage collector.
133 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
134 certain complicated string types. (reported by Gary King)
135 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
136 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
137 * fixed bug: DOCUMENTATION on structures no longer signals an
138 error if no structure type of the right name exists. (reported by
140 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
141 applicable non-standard primary or :AROUND method on
142 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
143 no longer cause unbound variable errors. (reported by Kevin Reid)
144 * fixed bug: metacircle resolution in cases where methods have slots
145 added before the slots from STANDARD-METHOD. (reported by Jean
147 * fixed bug: the Power PC assembly code for calling into lisp
148 sometimes computed its return address wrongly (depending on the
149 15th bit of the address, fixed at link time). This used to
150 manifest itself in a segmentation violation while building PCL.
151 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
152 many others over the years)
153 * fixed bug: ROOM no longer reports silly numbers for stack usage if
154 the stack pointer (treated as a signed integer) is negative.
155 (thanks to Peter van Eynde)
157 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
158 * new platform: experimental support for the Windows operating
159 system has been added. (thanks to Alastair Bridgewater)
160 * fixed several bugs in and robustified the PPC FFI (including
161 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
162 * bug fix: allow non-simple string symbol names (reported by Paul
164 * bug fix: interrupt handling on NetBSD (thanks to Richard M
166 * bug fix: saving a core corrupted callbacks on x86/x86-64
167 * bug fix: closed a loophole in metacircularity detection and
168 grounding in the PCL implementation of CLOS.
169 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
170 (reported by tomppa on #lisp)
171 * bug fix: clear the direction flag on Lisp -> C transitions, as
172 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
173 (reported by Andrej Grozin and Hendrik Maryns)
174 * optimization: major improvements to GC efficiency on GENCGC platforms
175 * optimization: faster implementation of EQUAL
176 * optimization: emit more efficient opcodes for some common
177 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
178 * optimization: several other minor code-generation improvements on x86-64
179 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
181 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
182 * minor incompatible change: (SETF CLASS-NAME) and (SETF
183 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
184 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
185 by ANSI as a generic function, and for consistency (SETF
186 GENERIC-FUNCTION-NAME) is treated likewise.
187 * fixed bug #233.b: make constraint propagation notice when a variable
188 value is changed after it is referenced but before it is used
189 * fixed bug #296: no more arbitrary behaviour from filenames with
190 odd characters as --load arguments. (reported by Adam Warner)
191 * fixed bug #390: :CHARACTER-SET pathname components now work as
192 expected. (reported by Tim Daly Jr)
193 * fixed bug #391: complicated :TYPE intersections in slot
194 definitions no longer cause an error in PCL internals.
195 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
196 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
197 returning the number of octets which would be written to the
198 file-stream. (thanks to Robert J. Macomber)
199 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
200 arguments to the internal stream functions. (thanks to David
202 * bug fix: stack exhaustion is now handled on the x86/SunOS
203 platform. (thanks to Daisuke Homma)
204 * bug fix: more accurate ROOM results on GENCGC platforms
205 * optimization: improved type inference for arithmetic-for
206 index variables in LOOP
207 * optimization: faster floating-point SQRT on x86-64
208 * fixed some bugs revealed by Paul Dietz' test suite:
209 ** DOCUMENTATION returns NIL instead of "" for method combinations
210 that don't have a docstring
212 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
213 * minor incompatible change: (SETF CLASS-NAME) and (SETF
214 GENERIC-FUNCTION-NAME) are no longer generic functions, and
215 therefore are not customizeable by user code (as seems to be at
216 least permitted and maybe required by AMOP). As a consolation,
217 however, the SBCL implementation of these functions now calls
218 REINITIALIZE-INSTANCE as specified by AMOP.
219 * bug fix: slot-definition documentation is propagated as per ANSI
220 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
221 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
222 argument, as do eslotd initialization methods. (from Pascal
223 Costanza's "Closer" project)
224 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
225 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
227 * bug fix: it is now possible to have more than one subclass of
228 STANDARD-GENERIC-FUNCTION without causing stack overflow.
229 (reported by Bruno Haible, Pascal Costanza and others)
230 * bug fix: the dependent update protocol now works for generic
231 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
233 * bug fix: condition-class instances corresponding to
234 DEFINE-CONDITION forms are now created eagerly. (reported by
235 Kalle Olavi Niemitalo on comp.lang.lisp)
236 * bug fix: floating point printing is more accurate in some
237 circumstances. (thanks to Simon Alexander)
238 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
239 merged with *DEFAULT-PATHNAME-DEFAULTS*.
240 * bug fix: callbacks on OS X now preserve stack-alignment.
241 (thanks to Cyrus Harmon)
242 * enhancement: the x86-64 disassembler is much better at
243 disassembling SSE instructions. (thanks to Lutz Euler)
244 * enhancement: CHECK-TYPE on a variable now allows type-inference
245 (thanks to Vincent Arkesteijn)
246 * optimization: improved performance of EUC-JP external format.
247 (thanks to NIIMI Satoshi)
248 * optimization: performance improvements to IO on file streams of
249 :ELEMENT-TYPE CHARACTER
250 * optimization: much faster memory allocation on x86-64
251 * optimization: faster garbage collections (latency and throughput)
253 * optimization: faster bignum subtraction and division on x86 and x86-64
254 (thanks to Lutz Euler)
255 * optimization: more accurate type-inference for loops that use a
256 floating point index variable or a negative step.
258 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
259 * bug fix: add a workaround to SBCL looping infinitely at startup on
260 Linux kernels with apparently buggy implementations of personality().
261 (thanks to Svein Ove Aas)
262 * bug fix: Unicode symbols are correctly printed in LDB backtraces
263 (thanks to David Lichteblau)
264 * bug fix: local bindings shadow global symbol macros for
265 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
266 * optimization: non-open coded uses of numeric comparison operators
267 (e.g. >) no longer cons when called with more than one parameter
268 on platforms supporting dynamic-extent allocation.
269 * enhancement: saving cores with foreign code loaded is now supported
270 on MIPS/Linux in addition to the previously supported platforms.
271 * bug fix: FIND-RESTART now tests for activity, not applicability when given
272 a restart object as identifier. (reported by Helmut Eller for CMUCL)
273 * bug fix: division by zero in sb-sprof when no samples were collected
274 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
275 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
276 (thanks to Alexander Kjeldaas)
277 * new feature: ignoring the return values of destructive functions
278 like NREVERSE signals a compile-time style-warning.
279 (thanks to Kevin Reid)
280 * new feature: the alignment of alien structure fields can be
281 explicitly specified. (thanks to Cyrus Harmon)
282 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
283 and dump core on SIGQUIT
285 ** incompatible change: threads do not inherit values of specials
286 from their parents (see manual)
287 ** bug fix: threads stacks belonging to dead threads are freed by the
288 next exiting thread, no need to gc to collect thread stacks anymore
289 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
290 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
291 ** bug fix: dynamic variable and thread start related gc lossage
292 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
294 * fixed some bugs revealed by Paul Dietz' test suite:
295 ** SUBTYPEP is slightly more accurate on heinously complicated
296 CONS types where some of the members have uncertain (in the
297 NIL, NIL sense) type relationships to each other.
299 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
300 no longer purified unless :PURIFY T is explicitly specified.
301 ** Non-purified cores are significantly smaller than before
303 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
304 * new feature: timers based on Zach Beane's excellent timer package
305 * added support for the following external formats: koi8-u,
306 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
307 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
308 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
309 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
310 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
311 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
312 * incompatible change: a threaded SBCL will no longer revert to
313 non-threaded mode on non-NPTL systems, but refuse to start entirely.
314 * bug fix: interrupts are disabled until startup is complete; no
315 more sigsegvs when receiving a signal to soon
316 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
318 * bug fix: add a workaround for the memory randomization features in
319 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
320 workaround will only be in effect on systems with the proc filesystem
322 * bug fix: printing objects of type HASH-TABLE signals a
323 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
325 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
326 have been running for over 50 days. (reported by Gilbert Baumann)
327 * bug fix: the logic for getting names of functions gets less
328 confused when confronded with alternate-metaclass
329 funcallable-instances. (reported by Cyrus Harmon)
330 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
331 consistent, even on internal alternate-metaclass objects.
332 * bug fix: flush closure information collected by physical
333 environment analysis prepass before the main pass. (bug reported
335 * bug fix: compiler pack phase does not modify a hashtable, which is
336 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
337 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
338 may be expanded. (reported by Alexander <malishev> on c.l.l)
339 * bug fix: no more style-warnings when compiling code using LAST
340 under high-SPEED low-SPACE settings. (reported by David Wallin
342 * fixed some incompatibilities between SBCL's MOP and the MOP
344 ** the METAOBJECT class is now implemented;
345 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
346 STANDARD-OBJECT, as required;
347 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
348 now compatible (as required by VALIDATE-SUPERCLASS); there
349 remains a consistency requirement over the presence or absence
350 of the FUNCTION class in the superclass list of finalized
351 classes; see the manual for more details;
352 ** the :AROUND method for COMPUTE-SLOTS on
353 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
354 requested slot ordering.
356 ** bug fix: parent thread now can be gc'ed even with a live
358 ** bug fix: binding a special with PROGV to no value is not
359 the same as that symbol not having been bound (thanks to
361 ** bug fix: binding specials is thread safe (thanks to
363 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
364 and SB-SPROF do not die with 'no handler for signal XX in
365 interrupt_handle_now(..)' anymore
366 ** bug fix: WITH-TIMEOUT works with multiple running threads
367 * fixed some bugs revealed by Paul Dietz' test suite:
368 ** ENSURE-GENERIC-FUNCTION should take a method class object for
369 the :method-class keyword argument.
371 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
372 * new port: the Solaris operating system on x86 processors is now
373 mostly supported, though some rough edges in the environment
374 remain. (thanks to Daisuke Homma)
375 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
376 space for its heap. (thanks to Thiemo Seufer)
377 * optimization: SBCL on MIPS platforms now supports dynamic-extent
378 closures. (thanks to Thiemo Seufer)
379 * minor incompatible change: eof selects abort in the debugger.
380 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
381 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
383 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
384 deprecated, and will go away in a future revision of SBCL.
385 * minor incompatible change: GC-ON and GC-OFF are no longer
386 implemented with a counter, it does not matter how many times gc
387 is switched on or off
388 * bug fix: discriminating functions for generic function classes
389 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
390 make invalid assumptions about method precedence order. (reported
392 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
393 platforms. (thanks to Thiemo Seufer)
394 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
395 * bug fix: correct pathname printing with printer escaping is on.
396 (thanks to Kevin Reid)
397 * bug fix: complex VOP definitions in "user-space" no longer trigger
398 package locks. (reported by Zach Beane)
399 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
400 for funcallable-instance functions, and is no different from
402 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
403 characters in the current readtable. (reported by Nicholas Neuss)
404 * bug fix: SUBTYPEP on various CONS types returns more a more
405 accurate acknowledgment of its certainty.
406 * optimizations: REMOVE-DUPLICATES now runs in linear time on
407 lists in some cases. This partially fixes bug 384.
408 * flush all standard streams before prompting in the REPL and the
410 * bug fix: signal handling and triggering gc do not conflict
411 directly anymore, in particular a high frequency sb-sprof does
412 not prevent gc from running
413 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
414 approximation for timezone and DST information between the
415 universal time epoch and the smallest negative 32-bit time_t.
416 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
417 year 1899 so long as the encoded time is non-negative.
418 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
419 the correct answer for non-constant rotate amounts when performing
420 an inline 32-bit rotation.
422 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
423 there is only one thread in the session
424 ** bug fix: memory leak for streams created in one thread and
425 written to in another
426 ** bug fix: lockup when compiled with gcc4
427 ** bug fix: race that allows the gc to be triggered when gc is
429 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
430 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
432 ** bug fix: finished threads are gc'ed properly
433 * fixed some bugs revealed by Paul Dietz' test suite:
434 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
435 has arguments with a different set of applicable methods from
436 the orignal arguments.
437 ** The type error thrown by MAP now has a correct expected-type
439 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
440 name a compiled function.
441 ** Three MISC tests where a large bignum was improperly coerced to
442 a float that couldn't represent that bignum during type
443 derivation were fixed.
444 ** SUBTYPEP can now handle types involving the negation of a
445 list-form FUNCTION type.
446 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
447 as (COMPLEX (AND RATIO (NOT FIXNUM))).
448 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
450 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
451 * New feature: Experimental support for bivalent streams: streams
452 opened with :element-type :default now allow character and binary
453 (unsigned-byte 8) I/O
454 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
455 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
457 * bug fix: no more highly sporadic "couldn't check whether ~S is
458 readable" when reading a stream and an interrupt hits in the middle
459 of a select system call
460 * compiler better recognizes complex arrays (reported by Eduardo
462 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
463 for variables on SBCL built with linkage-tables. (reported by Luis
465 * various error reporting improvements.
466 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
467 (Thanks to Thiemo Seufer)
468 * bug fix: nested reader invokations work correctly
469 * bug fix: it is possible to have simultaneous references to foreign
470 code and foreign data with the same name.
472 ** added x86-64 support
473 ** incompatible change: the threading api now works with thread
474 objects instead of thread ids
475 ** bug fix: threads are protected from signals and interruption when
476 starting up or going down
477 ** bug fix: a race where an exiting thread could lose its stack to gc
478 ** bug fix: don't halt on infinite error in threads if possible
479 ** fixed numerous gc deadlocks introduced in the pthread merge
480 ** bug fix: fixed thread safety issues in read and print
481 ** bug fix: debugger doesn't hang on session lock if interrupted at
482 an inappropriate moment
483 ** bug fix: run-program is now thread safe(r)
484 ** bug fix: inner with-recursive-lock no longer releases the mutex
485 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
486 * fixed some bugs revealed by Paul Dietz' test suite:
487 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
488 EXPECTED-TYPE slots filled.
489 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
491 ** MAKE-CONDITION accepts classes as type-designators.
492 ** COMPILE may never return NIL.
493 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
494 range before calling Unix time functions
496 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
497 * numerous signal handling fixes to increase stability
498 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
499 * minor incompatible change: we now correctly canonize default
500 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
501 as per the MOP, rather than the historical (INITARG INITFUNCTION
502 INITFORM). (reported by Bruno Haible)
503 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
504 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
505 for more information.
506 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
507 pathname is a directory pathname.
508 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
509 * dynamic space size on PPC has been increased to 768Mb. (thanks to
511 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
512 :METACLASS argument in addition to a class name. (reported by
513 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
514 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
515 (reported by Bruno Haible; patch from Gerd Moellmann)
516 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
518 * bug fix: more cleanups to the floating point exception handling on
519 x86-64 (thanks to James Knight)
520 * bug fix: the compiler does not try to constant fold calls of
521 COERCE and BIT-* functions when they return freshly constructed
522 objects. (reported by Nikodemus Siivola and Paul Dietz)
523 * optimization: improved the MIPS versions of generic (in the
524 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
525 * optimization: direct conversion of (unsigned-byte 32) to floats on
526 the PowerPC platform.
527 * optimization: structure instances with raw slots now use less
528 memory, and probably show better memory locality. (thanks to
530 * optimization: DYNAMIC-EXTENT declarations for lists and closures
531 are treated as requests for stack allocation on the x86-64,
532 Alpha-32, PPC and SPARC platforms.
533 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
534 the wrong file descriptor; implementation of SOCKET-OPEN-P.
535 (thanks to Tony Martinez)
537 ** gcing a dead thread can no longer lead to lockups
538 ** threads block signals until they are set up properly
539 ** errno is no longer shared by threads
540 ** interrupt-thread restores the eflags register on x86
541 ** fixed some lockups due to gc/thread interaction
542 * fixed some bugs revealed by Paul Dietz' test suite:
543 ** invalid dotted lists no longer raise a read error when
545 ** signal an error if a symbol that names a declaration is used
546 as the name of a type, or vice versa
547 ** allow using the (declare (typespec var*)) abbreviation for
548 (declare (type typespec var*)) with all type specifiers
549 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
550 ** always return NIL from PPRINT-POP when OBJECT is NIL
551 ** don't signal errors when pretty-printing malformed LABELS,
552 FLET or MACROLET forms
553 ** declarations in a DOLIST body are also in scope for the
555 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
557 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
560 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
561 * fixed cross-compiler leakages that prevented building a 32-bit
562 target with a 64-bit host compiler.
563 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
564 opened with :IF-EXISTS :OVERWRITE.
565 * fixed bug 281: error for an invalid qualifier in a short-form method
566 combination method is not signalled until the faulty method is called.
567 * bug fix: iteration variable type inferrer failed to deal with open
568 intervals. (reported by Alan Shields)
569 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
570 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
571 or more runtime options were provided to the sbcl binary.
572 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
573 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
574 * optimization: there's now a fast-path for fixnum arguments in the
575 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
576 * optimization: the code generated on x86-64 is more compact thanks
577 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
578 * bug fix: floating-point exception handling now partly works on
579 x86-64. (Thanks to James Knight)
580 * improvement to the MIPS backend from Thiemo Seufer: C-style
581 64-bit long long arguments and return values to alien functions
583 * fixed some bugs revealed by Paul Dietz' test suite:
584 ** the type-error signalled from WARN has a filled-in DATUM slot.
585 ** the type-error required when a stream is not associated with
586 a file has the stream as its datum.
587 ** type-errors on single-floats on x86-64 no longer have
588 :INVALID-OBJECT as the datum
589 ** the type-errors signalled for invalid function names now have
590 a correct expected type
591 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
592 for structures defined with a :TYPE. Documentation strings for
593 typed structures are no longer immediately discarded
594 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
595 broadcast streams and synonym streams. FILE-LENGTH now also works
596 on broadcast streams.
598 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
599 * incompatible change: the --noprogrammer option, deprecated since
600 version 0.7.5, has been removed. Please use the equivalent
601 --disable-debugger option instead.
602 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
604 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
605 inoperational for a while now) has been completely removed.
606 * null lexical environments are now printed as #<NULL-LEXENV>,
607 significantly reducing the amount of clutter in typical backtraces.
608 * documentation on weak pointers, finalization, and after GC hooks
609 has been added to the manual.
610 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
611 as other specialized array types, is much faster. SUBSEQ and
612 COPY-SEQ on such arrays have also been sped up.
613 * optimization: EQL is now more efficient when at least other argument
614 is known to be of type (OR FIXNUM (NOT NUMBER)).
615 * fixed bug: the runtime is now less vulnerable to changes in the
616 size of the SBCL object on OS X, and virtual memory is reserved for
617 all spaces that need to be at a fixed address.
618 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
619 * fixed bug: finalizers and after GC hooks that cause consing are now
621 * fixed bug: compiler error messages and summaries are now printed to
622 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
623 * fixed inference of the upper bound of an iteration variable.
624 (reported by Rajat Datta).
625 * fixed bug 211e: calling local functions with duplicated constant
626 keyword argument no longer causes a bogus style warning about an
628 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
629 ftype declaration to be disregarded. (reported by Dave Roberts)
630 * fixed bug 373: caused by erronous compilation of references to alien
631 variables in the runtime on ppc/darwin.
632 * fixed bug 376: CONJUGATE type deriver.
633 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
634 (reported by Baughn on #lisp)
635 * a cross-compiler bug on non-x86 platforms has been identified and
636 fixed. (thanks to Bruno Haible)
637 * improvements to the MIPS runtime code for increased stability.
638 (thanks to Thiemo Seufer)
639 * increased the maximimum compact environment size to allow
640 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
641 * partial workaround for bug 135: don't name the function we're
642 calling for hairy cases of EVAL, so as not to accumulate one environment
643 entry per call to EVAL. (reported by Kevin Reid)
644 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
645 * optimization: type testing for non-vector arrays should be faster.
646 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
648 * bug fix: setting 31st element of a bit vector to zero did not work
650 * bug fix: redefining a class definition which failed due to a
651 previous accessor / function clash now works (but see BUGS entry
652 #380 for more problems in this area). (thanks to Zach Beane)
653 * the long form of DEFINE-METHOD-COMBINATION disables method group
654 checking when given a single method group with pattern *. (thanks
656 * on x86 compiler supports stack allocation of results of simple
657 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
658 * support for building and running on Mac OS X 10.4 "Tiger" has been added
659 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
660 * fixed some bugs related to Unicode integration:
661 ** the restarts for recovering from input and output encoding
662 errors only appear when there is in fact such an error to
664 * fixed some bugs revealed by Paul Dietz' test suite:
665 ** MISC.549 and similar: late transformation of unsafe type
666 assertions into derived types caused unexpected code
668 ** SCALE-FLOAT type deriver is less wrong.
669 ** type derivers for EXP, LOG and similar functions compute result
670 types for complex arguments better.
671 ** (MISC.563) CONJUGATE type deriver works for very restricted
673 ** out-of-line type testers for character strings are available.
674 ** EQUAL compiler transform understands specialness of objects
676 ** accessing double-floats stored on the stack now works on x86-64.
677 ** debugger internals could sometimes create invalid lispobjs,
678 resulting in GC crashes.
679 ** MISC.548: type check weakening can convert required type into
681 ** initialization forms for bindings are not in scope of free special
684 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
685 * incompatible change: thread support for non-NPTL systems has
686 been removed - locking is buggy and unreliable. A threaded
687 SBCL build will now warn at startup and refuse to create
688 new threads, unless futex support is detected in the kernel
689 * incompatible change: the top level REPL now has only an ABORT
690 restart associated with it, not TOPLEVEL and ABORT as it used to.
691 TOP and TOPLEVEL are now available as debugger commands for
692 returning to the top level.
693 * incompatible change: forms evaluated in the REPL now use the
694 global optimization policy.
695 * incompatible change: user- and system-initialization files are
696 no longer processed with LOAD, but by READ and EVAL; hence the
697 global optimization policy, startup package, readtable, etc,
699 * The .fasl file format number has been incremented because of
700 various incompatible changes.
701 * internal entry point details and argument counts no longer appear
702 in backtraces unless explicitly requested by setting
703 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
704 * built-in and standard functions no longer have names like "top
705 level local call to FOO".
706 * fixed bug 32: functions defined in non-null lexical environments
707 now have more legible printed representation
708 * fixed bug 33: functions defined in non-null lexical environemnts
709 are now more amenable to inspection by INSPECT.
710 * workaround for bug 354: XEPs no longer appear in backtraces unless
711 explicitly requested.
712 * fixed bug: receiving the signal which results from attempting to
713 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
714 SIGBUS on other *BSDs) on architectures where the C stack is also
715 the Lisp stack (x86 and x86-64) from foreign code no longer leads
716 to debugger confusion or wild instability. (reported by Cheuksan
718 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
719 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
720 (reported by Lutz Euler)
721 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
722 compiled code. (reported by Paul Dietz)
723 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
724 specialized parameter to a method to something that is not TYPEP
725 the specializer is now possible.
726 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
727 face of package deletion.
728 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
729 pathname. (reported by David Tolpin on c.l.l)
730 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
731 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
732 STRING-STREAM and FILE-STREAM, respectively.
733 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
734 than previously; in addition, its test suite is now run on build.
735 (thanks to Luke Gorrie)
736 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
737 * optimization: added a immediate representation for single-floats
739 * fixed some bugs revealed by Paul Dietz' test suite:
740 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
741 ** attempting to create a package with a colliding nickname causes
742 correctable errors to be signalled.
743 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
744 ** several x86-64 backend bugs related to sign-extension of immediate
747 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
748 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
749 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
751 * fixed loading of multiply forward-referenced layouts.
752 (thanks to Cheuksan Wang)
753 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
754 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
755 (see COMPILE-FILE documentation for details of :PRINT option)
756 * fixed bugs 19 and 317: fixed-format floating point printing is
757 more accurate. This also fixes a bug reported by Adam Warner
758 related to the ~@F format directive.
759 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
761 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
762 dispatch table if the from-char is a dispatch macro character.
763 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
764 by the wordsize no longer ignore the last word. (reported by Lutz
766 * fixed bug in type checking of dynamic-extent variables. (reported
768 * optimization: sequence traversal functions use their freedom to
769 coerce function designators to functions.
770 * optimization: code with many calls to CLOS methods specialized on
771 CLOS classes has had redundant type checks removed.
772 * optimization: added declarations to speed up operations that access
773 the internal character database (for example STRING-UPCASE)
774 * optimization: comparison operations between floats and sufficiently small
775 fixnums no longer create extra rationals
776 * fixed some bugs related to Unicode integration:
777 ** portions of multibyte characters at the end of buffers for
778 character-based file input are correctly transferred to the
779 start of the buffer at the next read.
780 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
781 passing it through to OPEN.
782 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
783 argument given to internal calls to OPEN.
784 ** the built-in comment readers (introduced by character sequences
785 ";" and "#|") are more forgiving to encoding errors; they will
786 STYLE-WARN and then attempt to resync the stream at a character
787 boundary. (thanks to Teemu Kalvas)
788 * fixed some bugs in the x86-64 port:
789 ** Negative short int return values from c-calls are sign-extended
791 ** The stack is aligned to 16-bytes for c-calls, as required by
792 the ABI. (thanks to Cheuksan Wang)
793 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
794 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
795 for enums. (thanks to Vincent Arkesteijn)
796 ** Multiple small optimizations and bugfixes for floating point
798 * fixed some bugs revealed by Paul Dietz' test suite:
799 ** Space, Tab, Linefeed, Return and Page have the invalid
800 secondary constituent character trait.
801 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
803 ** WITH-INPUT-FROM-STRING should only update the index place on
805 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
807 ** Bit-array operations (BIT-AND and similar) worked incorrectly
808 with one-dimensional arrays with fill pointers.
809 ** TYPE-OF failed on a complex with an integer realpart and a
811 ** compiler failure during type inference for the code of form
812 (IF (EQL X (THE ...)) ...) (MISC.535).
814 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
815 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
816 platform. The port supports 61-bit fixnums, large memory spaces
817 and reloading shared object files.
818 * enhancement: saving cores with foreign code loaded is now
819 supported on ppc/Darwin in addition to the previously supported
821 * enhancement: the statistical profiler now walks deeper into the
822 call stack for more meaningful call-graphs and accrued time
823 reports (x86/x86-64 only). It also now reports time spent in
825 * enhancement: it is now possible to trace most individual methods
826 of a generic function in addition to tracing the generic function
828 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
830 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
831 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
832 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
833 produces an error. (thanks to Vincent Arkesteijn)
834 * bug fix: NAMESTRING on pathnames with :WILD components in their
835 directories works correctly. (thanks to Artem V. Andreev)
836 * fixed bug 125: compiler preserves identity of closures. (reported
838 * bug fixed: functions with &REST arguments sometimes failed with
839 "Undefined function" when compiled with (DEBUG 3). (reported by
841 * bug fix: overflow during compiling of setting element of a bit
842 vector with constant index and value. (reported by Timmy Douglas)
843 * build fix: fixed the dependence on *LOAD-PATHNAME* and
844 *COMPILE-FILE-PATHNAME* being absolute pathnames.
845 * on x86 compiler partially supports stack allocation of dynamic-extent
847 * GO and RETURN-FROM do not check the extent of their exit points
848 when compiled with SAFETY 0.
849 * fixed some bugs related to Unicode integration:
850 ** encoding and decoding errors are now much more robustly
851 handled; it should now be possible to recover even from invalid
852 input or output to the terminal. (thanks to Teemu Kalvas)
853 ** provided a first cut at implementing STRING-TO-OCTETS and
854 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
855 ** altered the SB-MD5 contributed module slightly, changing the
856 interface just enough for it to be supportable for builds where
857 lisp characters are not eight bits.
858 * fixed some bugs revealed by Paul Dietz' test suite:
859 ** the FORMATTER-generated functions for ~V[ conditionals require
860 the correct number of arguments.
861 ** READ-FROM-STRING returns the mandated second value when applied
862 to displaced strings.
863 ** the #\Rubout and #\Backspace characters are treated as invalid
864 constituent characters by the tokenizer.
866 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
867 * new feature: reloading changed shared object files with
868 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
869 * new feature: references to foreign variables and functions
870 can now be compiled and loaded before the corresponding shared
871 object file is loaded, as long as the foreign definitions are
872 available at runtime.
873 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
874 supported. (thanks to Dan Debertin)
875 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
876 just keywords. (thanks to Vincent Arkesteijn)
877 * fixed bug #331: structure-class instances corresponding to
878 DEFSTRUCT forms are now created eagerly.
879 * fixed bug #345: backtraces from calls to undefined functions work
880 on x86 as well. Related bug #61 is now also partially fixed on x86
881 (backtraces from throws to unknown catch tags.)
882 * bug fix: lambda-list parsing is now stricter vrt. order and number
883 of lambda-list keywords.
884 * bug fix: as specified by AMOP, an error is signalled if a
885 class-option appears multiple times in a DEFCLASS form. (reported
887 * bug fix: RANDOM can be compiled when the compiler derives the type
888 of its numeric argument as a disjoint set of small integers.
889 (reported by Paul Dietz)
890 * bug fix: starting a core saved with shared objects loaded when
891 those objects are not available no longer causes threaded SBCL to
892 hang. (reported by Sean Ross)
893 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
895 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
896 argument, which must always be NIL. (reported by Kalle Niemitalo)
897 * bug fix: printing 1.0d+23 no longer results in an error.
898 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
899 * bug fix: structure slot setters preserve evaluation order. (thanks
901 * fixed some bugs related to Unicode integration:
902 ** RUN-PROGRAM can allow its child to take input from a Lisp
903 stream. (reported by Stefan Scholl)
904 ** ASDF-INSTALL successfully downloads .tar.gz files in all
905 locales. (reported by Ken Causey)
906 * fixed some bugs revealed by Paul Dietz' test suite:
907 ** INCF, DECF and REMF evaluate their place form as specified in
909 ** FORMATTER expands ~{ ~} iteration directives with V or #
910 parameters correctly.
911 ** FORMATTER deals with the ~@[ ~] conditional directive where the
912 consequent uses no arguments correctly.
913 ** the system has a partial understanding of the (COMPLEX RATIO)
915 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
918 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
919 * new feature: a build-time option (controlled by the :SB-UNICODE
920 keyword feature, enabled by default) for building the system with
921 support for the entire 21-bit character space defined by the
923 * new feature: the system now has rudimentary external-format
924 support; the primary user-visible change at this time is that
925 characters with the high bit set (such as lower-case-e-acute) will
926 print correctly to a terminal in a UTF-8 environment.
927 * minor incompatible change: BASE-CHAR no longer names a class;
928 however, CHARACTER continues to do so, as required by ANSI.
929 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
930 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
931 has been moved to the SB-EXT package (temporarily re-exported from
933 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
934 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
936 * fixed bug #351: better error-handlind and reporting for malformed
938 * fixed bug #350: bignum-printing is now more memory-efficient,
939 allowing printing of very large bignums, eg. (expt 2 10000000).
940 (reported by Bruno Haible)
941 * fixed bug #302: better primitive-type selection for intersection
943 * fixed bug #308: non-graphic characters now all have names, as
944 required. (reported by Bruno Haible)
945 * bug fix: redefining a class with different superclasses now correctly
946 removes it from the direct-subclasses of its previous superclasses.
947 (reported by David Morse)
948 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
949 new value now works. (reported by Bruno Haible)
950 * bug fix: correct canonicalization of multiple non-standard slot
951 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
952 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
953 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
954 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
955 now exists, an signals an error.
956 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
957 during macroexpansion for non-iteration variables as well. (reported
958 by Bruno Haible for CMUCL)
959 * bug fix: Cyclic structures and unprintable objects in compiler
960 messages no longer cause errors. (reported by Bruno Haible)
961 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
962 and READ-FROM-STRING all now return a primary value of NIL if
963 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
964 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
965 (reported by Bruno Haible for CMUCL)
966 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
967 do not designate a condition. (reported by Bruno Haible for
969 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
970 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
971 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
972 permitting resolution in favour of any of the conflicting symbols.
973 (reported by Bruno Haible for CMUCL)
974 * bug fix: EQUAL compiler optimizations is less aggressive on
975 strings which can potentially compare true despite having distinct
976 specialized array element types.
977 * bug fix: unit enumerations can be defined without dividing by
978 zero. (thanks to Vincent Arkesteijn)
979 * FORMAT compile-time argument count checking has been enhanced.
980 (report from Bruno Haible for CMUCL)
981 * a partial workaround for the bug 262: the compiler does not try to
982 inline-expand a local function doing RETURN-FROM from a deleted
983 BLOCK. (thanks to Peter Denno for the bug report and to David
984 Wragg for the simple test case)
985 * fixed some bugs revealed by Paul Dietz' test suite:
986 ** RENAME-PACKAGE allows all package designators as new package
988 ** constraint propagation and lambda variable substitution are
989 more cautious in dealing with partially deleted code.
990 ** compiler sometimes lost reoptimization passes.
991 ** CERROR, given a condition as condition designator, passes its
992 remaining arguments to the continue format control without
994 ** Case-altering FORMAT directives work correctly on non-ASCII
996 ** The REAL type specifier handles bounds outside the floating
997 point ranges without signalling FLOATING-POINT-OVERFLOW.
998 ** Functions with IR1-transformations can create intercomponent
999 references to global functions.
1000 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1002 ** FORMAT ~R treats a nil value for its first parameter correctly.
1004 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1005 * enhancement: saving cores with foreign code loaded is now
1006 supported on x86/NetBSD and sparc/Linux in addition to the previously
1007 supported platforms.
1008 * bug fix: on some platforms repeated installations caused multiple
1009 copies of HTML documentation to be installed -- should not happen
1010 any more. (reported by Stefan Scholl)
1011 * bug fix: parsing self-recursive alien record types multiple times
1012 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1013 original patch by Helmut Eller for CMUCL)
1014 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1015 (thanks to Richard Kreuter)
1016 * bug fix: defining classes whose accessors are methods on existing
1017 generic functions in other (locked) packages no longer signals
1018 bogus package lock violations. (reported by François-René Rideau)
1019 * bug fix: special variables as DEFMETHOD parameters no longer have
1020 associated bogus type declarations. (reported by David Wragg and
1022 * bug fix: read-write consistency on streams of element-type
1023 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1024 * bug fix: redefiniton of the only method of a generic function with
1025 no DEFGENERIC no longer emits a full WARNING. In addition,
1026 redefinition of generic functions with no DEFGENERIC to an
1027 incompatible lambda list now signals an error. (thanks to Zach
1029 * bug fix: DEFGENERIC now works even when there's a function of the
1030 same name in an enclosing lexical environment. (thanks to Zach
1032 * fixed compiler failure, caused by instrumenting code during
1033 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1034 * optimization: added loop analysis and improved register allocation
1035 * fixed some bugs revealed by Paul Dietz' test suite:
1036 ** POSITION on displaced vectors with non-zero displacement
1037 returns the right answer.
1038 ** (SIMPLE-STRING) is a valid type specifier for sequence
1040 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1042 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1043 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1045 ** The justification version of the FORMAT ~< directive treats
1046 non-zero minpad parameter correctly.
1048 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1049 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1050 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1051 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1052 the supported interface.
1053 * new feature: Single-stepping of code compiled with DEBUG 2 or
1054 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1055 * new feature: saving cores with foreign code loaded is now
1056 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1057 Timothy Moore's work for CMUCL)
1058 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1059 parameters to * instead of NIL if no initform is supplied.
1060 (reported by Johan Bockgaard)
1061 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1062 correctly when FUNCALL appears as the car of the form. Note:
1063 despite this FUNCALL forms are not currently subject to
1064 compiler-macro expansion. (port of Raymond Toy's fix for the
1065 same from CMUCL, reported by Johan Bockgaard)
1066 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1067 (thanks for Teemu Kalvas)
1068 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1069 case instead of terminating the iteration (thanks for Julian
1070 Squires, Sean Champ and Raymond Toy)
1071 * bug fix: incorrect expansion of defgeneric that caused a style
1072 warning. (thanks for Zach Beane)
1073 * bug fix: slot accessor effective method computation works properly
1074 for classes with multiple non-standard applicable methods on
1075 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1076 * on x86 compiler supports stack allocation of results of LIST and
1077 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1078 CMUCL implementation by Gerd Moellmann)
1079 * fixed some bugs revealed by Paul Dietz' test suite:
1080 ** FORMAT strings with both the ~<~:;~> form of the justification
1081 directive and pretty-printing directives cause an error.
1083 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1084 * incompatible change: the internal functions
1085 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1086 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1087 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1088 instead of the old functions.
1089 * new feature: on platforms where "dladdr" is available foreign
1090 function names now appear in backtraces. (based on Helmut Eller's
1092 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1093 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1095 * documentation: ASDF manual is now distributed with SBCL.
1096 * bug fix: SBCL can now load its contributed modules with REQUIRE
1097 even if the system-provided entries have been removed from
1098 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1100 * bug fix: backtraces involving undefined functions or assembly
1101 routines are more informative. (thanks to Brian Downing)
1102 * bug fix: mutually referent alien structures now work correctly.
1103 (reported by Rick Taube)
1104 * bug fix: structures defined by WITH-ALIEN can be referred to
1105 within other definitions in the same WITH-ALIEN.
1106 * bug fix: division operators (MOD, TRUNCATE and the like) with
1107 constant zero divisors and integer dividends no longer generate
1109 * bug fix: provide default methods for INPUT-STREAM-P and
1110 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1111 * bug fix: improve the detection and resolution of MOP metacycles
1112 (where effective method computation depends on the generic function
1113 having its effective method computed). (reported by Bruno Haible)
1114 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1115 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1116 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1117 * bug fix: pathnames with relative directory components can be
1118 represented relative to default pathnames.
1119 * optimization: in taking the GCD of bignums, reduce the two bignums
1120 to approximately the same size (using Euclid's algorithm) before
1121 applying the more sophisticated binary GCD. (thanks to Juho
1123 * optimization: COUNT on bitvectors now operates word-at-a-time.
1124 * optimization: ASH with a positive, but not necessarily constant,
1125 (leftwards) shift, when in a modular context, is compiled to a
1127 * fixed some bugs revealed by Paul Dietz' test suite:
1128 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1129 if the corresponding argument is NIL.
1130 ** FORMAT directives accepting a minpad parameter treat negative
1132 ** literal commas as character parameters to FORMAT directives are
1134 ** literal spaces directly after ~<Newline> directives within a
1135 format-logical-block (~:< ~@:>) do not induce :FILL-style
1136 conditional newlines.
1137 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1138 as specified: it no longer includes conditional newlines).
1139 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1141 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1142 :PER-LINE-PREFIX argument does not evaluate to a string.
1144 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1145 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1146 the manual for details; for now, package locks can be disabled by
1147 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1148 if no major problems are found then it is likely that they will be
1149 compiled in unconditionally.
1150 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1151 unsupported operators on all platforms. To load a shared library
1152 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1153 object file, link it into a shared library outside of SBCL and
1154 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1156 * bug workaround: ROOM T and the GC were not working together
1157 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1158 reporting facility aren't preserved. That reporting has been
1159 disabled, so now until and unless someone figures out how to make it
1160 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1161 * minor incompatible change: as threatened around sbcl-0.8.0, the
1162 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1163 The symbols are also exported from SB-PCL for backwards
1164 compatibility, but more so than before SB-PCL should be treated as
1165 an implementation-internal package.
1166 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1168 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1169 It is known as of this release not to work on the Alpha, however.
1170 * fixed bug #167: errors signalled due to illegal syntax in method
1171 bodies are now more legible.
1172 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1173 designators and can hence be used with TYPEP.
1174 * fixed bug #333: CHECK-TYPE now ensures that the type error
1175 signalled, if any, has the right object to be accessed by
1176 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1177 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1178 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1180 * fixed bug #334: programmatic addition of slots using specialized
1181 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1182 and :ALLOCATION :CLASS slots.
1183 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1184 float exponents. (rereported by Peter Seibel)
1185 * fixed a bug: #\Space (and other whitespace characters) are no
1186 longer considered to be macro characters in standard syntax by
1187 GET-MACRO-CHARACTER.
1188 * fixed bug: initialization of condition class metaobjects no longer
1189 causes an instance of the condition to be created. (reported by
1191 * fixed bug: it is now possible to have slots such that
1192 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1193 description is neither :INSTANCE nor :CLASS.
1194 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1195 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1196 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1197 non-standard SLOT-DEFINITION-ALLOCATION values.
1198 * fixed bugs: various race conditions handling exiting threads.
1199 CL-PPCRE's thread torture test now passes.
1200 * fixed arguably-a-bug: GC time proportional to number of threads in
1201 system even when most of them are idle
1202 * optimization: improved performance of BIT and SBIT on bit-vectors.
1203 * .fasl file incompatibility: The fasl file version number has
1204 been incremented because of changes associated with package locks.
1206 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1207 * minor incompatible change: the system no longer provides
1208 optimization hints (conditions which are TYPEP
1209 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1210 mode; these hints are still emitted when higher SPEED optimization
1212 * new contrib module: a sampling profiler (profiling by statistical
1213 sampling, rather than by instrumenting functions) is available as
1214 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1215 * the behaviour of the standard function ED is now customizeable by
1216 third parties through a hook variable: see ED's documentation
1217 string for information on the protocol.
1218 * the compiler no longer emits efficiency notes for (FUNCALL X)
1219 when the type of X is uncertain under default optimization
1221 * fixed bug 276: mutating a binding of a specialized parameter to a
1222 method to something that is not TYPEP the specializer is now
1224 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1225 exhibit the required behaviour on the x86 platform. (thanks to
1226 Peter van Eynde, Eric Marsden and Bruno Haible)
1227 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1228 even for difficult arguments. (reported by Peter Graves)
1229 * fixed bug 141a: the backquote printer now descends quoted
1231 * fixed another bug in backquote printing: no more destructive
1232 modification of the form's list structure. (reported by Brian
1234 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1235 move between its address being taken and the call to
1236 interrupt_thread, fixing a crashing race condition.
1237 * the SB-POSIX contrib implementation has been adjusted so that it
1238 no longer exhibits ridiculously poor performance when constructing
1239 instances corresponding to C structs.
1241 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1242 * minor incompatible change: the sb-grovel contrib now treats C
1243 structures as alien (in the sense of SB-ALIEN) objects rather than
1244 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1245 has implications for memory management of client code
1246 (sb-grovel-returned objects must now be manually managed) and for
1247 type safety (alien objects now have full types).
1248 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1249 used to control emission of compiler diagnostics, rather than the
1250 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1251 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1252 quality should be considered deprecated.
1253 * install.sh now installs the user manual as well
1254 * (not quite a new documentable feature, but worth considering in
1255 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1256 beginnings of a semantically meaningful condition hierarchy is
1257 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1259 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1260 designator as the defaults argument.
1261 * fixed bug: Displaced arrays whose displaced-to array has become
1262 too small now cause ARRAY-DIMENSION to signal an error, providing
1263 for safer bounds-checking. (reported by Bruno Haible)
1264 * fixed bug: DEFCLASS slot definitions with identical :READER and
1265 :WRITER names now signal a reasonable error. (reported by Thomas
1267 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1269 * fixed bug: Invalid slot specification errors now print correctly.
1270 (thanks to Zach Beane)
1271 * fixed bug 320: Shared to local slot value transfers in class
1272 redefinitions now happen corrently from superclasses as
1273 well. (reported by Bruno Haible)
1274 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1276 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1277 improper lists correctly. (reported by Bruno Haible)
1278 * fixed bug 313: source-transform for <fun-name> was erroneously
1279 applied to a call of a value of a variable with name <fun-name>.
1280 (reported by Antonio Menezes Leitao)
1281 * fixed bug 307: The obsolete instance protocol ensures that
1282 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1283 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1284 remove dead unknown-values globs from the middle of the stack.
1285 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1286 easier to resume long computations after using *BREAK-ON-SIGNALS*
1287 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1288 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1289 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1290 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1291 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1292 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1294 * fixed a (fairly theoretical) bug in string printing: if
1295 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1296 does not have array-element-type equal to the most general string
1298 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1299 function name. (thanks to Nikodemus Siivola)
1300 * fixed bug: compile-time format string checker failed on
1301 non-closed ~{. (reported by Thomas F Burdick)
1302 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1303 don't warn on legal loop code involving more than one aggregate
1304 boolean. (thanks to Nikodemus Siivola)
1305 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1306 in keyword position to MAKE-PACKAGE was not being reported as
1308 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1309 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1311 * fixed bugs 280 and 312: the checking for multiple definitions in a
1312 file is less likely to become confused by uses of inline
1314 * fixed bug: the #S reader macro performs the keyword coercion
1315 specified for slot names. (reported by Kalle Niemitalo)
1316 * fixed bug: lambda lists may contain symbols whose names start with
1317 & but are not lambda-list-keywords; their occurrence triggers a
1319 * fixed bug 321: define-method-combination argument lambda lists do
1320 not cause invalid code to be generated when &optional or &aux
1321 variables have default values. (reported by Bruno Haible)
1322 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1323 have CLOS classes; this appears also to have fixed reported
1324 instability in stack exhaustion detection.
1325 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1326 MUFFLE-WARNING when no associated restart is present is now
1328 * optimization: rearranged the expansion of various defining macros
1329 so that each expands into only one top-level form in a
1330 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1332 * optimization: used a previously little-used slot in symbols to
1333 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1335 * fixed some bugs revealed by Paul Dietz' test suite:
1336 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1338 ** Fixed an optimization bug related to inheritance of initforms
1339 from local to shared slots.
1340 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1341 ** CAST optimizer forgot to flush argument derived type.
1342 ** print/read consistency on floats is now orders of magnitude
1343 more likely. (thanks also to Bruno Haible for a similar report
1345 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1346 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1347 for objects of type REAL. Make it so.
1348 ** SXHASH is specified (infelicitously) to respect similarity,
1349 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1350 it so. (thanks to Markus Ziegler)
1351 ** on the Alpha, the compiler succeeds in compiling functions
1352 returning a known number of arguments greater than 63.
1353 ** fixed handling of invalid NIL arguments in keyword position
1355 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1357 ** provided a readably-printable representation for RANDOM-STATE
1359 ** ensured that pathnames, where they have a namestring, always
1360 print using #P"..." syntax.
1362 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1363 * Support for the forthcoming 2.0 version of the NetBSD kernel
1364 running on x86 hardware has been added. (thanks to Perry
1365 E. Metzger most immediately, and others for their past work)
1366 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1367 current release version; SBCL's OpenBSD support had been broken
1368 since about the time of OpenBSD's switch to ELF binary format.)
1369 * [placeholder for DX summary]
1370 ** user code with &REST lists declared dynamic-extent, under high
1371 speed or space and low safety and debug optimization policy.
1372 * The manual has been converted to Texinfo format and the debugger
1373 chapter from the cmucl manual has been added.
1374 * A facility has been added to extract documentation strings from
1375 sbcl and store them as Texinfo-formatted snippets for inclusion in
1376 the manual (via Texinfo's @include directive)
1377 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1378 the test case to Dave Roberts)
1379 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1380 pointers. (reported by Sean Ross)
1381 * bug fix: PROFILE output is printed nicely even for large numerical
1382 values. (thanks to Zach Beane)
1383 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1384 greater than 32 handle EOF correctly.
1385 * bug fix: on X86 an immediate argument of the IMUL instruction is
1386 correctly printed in disassembly. (thanks to Lutz Euler)
1387 * bug fix: class slots in redefined classes preserve their old
1388 values. (thanks to Bruno Haible and Nikodemus Siivola)
1389 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1390 succeeds. (reported by Marco Baringer)
1391 * fixed some bugs revealed by Paul Dietz' test suite:
1392 ** READ-SEQUENCE now works on ECHO-STREAMs.
1393 ** RATIONALIZE works more according to its specification. (thanks
1396 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1397 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1398 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1399 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1400 should matter to you only if you rebind the printer control
1401 variables and then find you want different bindings in the
1402 debugger than in the ordinary execution of your program.)
1403 * The runtime build system has been tweaked to support building
1404 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1405 assembler and linker. (thanks to Nikodemus Siivola)
1406 * Unbound, undefined, undeclared variables now trigger full
1407 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1408 more often programmer error than deliberate exploitation of undefined
1410 * optimization: the hash algorithm for strings has changed to one
1411 that is less vulnerable to spurious collisions. (thanks to Juho
1413 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1414 needless bounds checking. (thanks to Juho Snellman)
1415 * optimization: implemented multiplication as a modular
1416 (UNSIGNED-BYTE 32) operation on the PPC backend.
1417 * fixed some bugs revealed by Paul Dietz' test suite:
1418 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1420 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1421 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1422 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1424 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1426 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1428 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1429 their output stream on EOF from read.
1430 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1431 have been read to end-of-file.
1432 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1434 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1435 description of determination of which consecutive characters
1437 ** Printing the "Space" character with escaping on now yields "#\\ ",
1438 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1439 ** Reading floating-point numbers with *READ-BASE* set to a number
1440 less than 10 works correctly.
1441 ** Reading floating-point numbers with *READ-BASE* set to a number
1442 more than 10 works correctly.
1443 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1444 the readtable currently in effect.
1446 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1447 * minor incompatible change: parsing of namestrings on a physical
1448 (Unix) host has changed; numbers after the final #\. in a
1449 namestring are no longer interpreted as a version field. This is
1450 intented to be largely invisible to the user, except that the
1451 meaning of the namestring "*.*.*" has changed: it now refers to a
1452 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1453 should usually be replaced by
1454 (make-pathname :name :wild :type :wild :version :wild)
1455 with the added benefit that this is more likely to be portable.
1456 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1457 option to OPEN now signals an error if the file being opened
1458 exists; this may have an impact on existing code.
1459 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1460 signals received, on the PowerPC platforms (both Linux and
1461 Darwin). (thanks to Pierre Mai for pointing out the location of
1463 * several fixes on OS X: The system now builds and runs cleanly on
1464 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1465 (thanks to Brian Mastenbrook)
1466 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1467 second-resolution: integer multiples of 1/3600 between -24 and 24.
1468 (thanks to Vincent Arkesteijn)
1469 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1470 type when called with 1 argument; PEEK-CHAR checked type of
1471 PEEK-TYPE only after having read first character from a
1472 stream. (reported by Peter Graves)
1473 * bug fix: the garbage collector now has much better locality
1474 behaviour, and in particular no longer treats all memory as being
1475 exhausted when many small objects point to each other in a deeply
1477 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1478 recognized as being TYPEP their class.
1479 * bug fix: the PUSHNEW documentation string has been corrected.
1480 (thanks to Vincent Arkesteijn)
1481 * bug fix: defaulting of the value for the last of an atypically
1482 large number of multiple values being bound was not being
1483 performed correctly on the Alpha or PPC platforms
1484 * optimization: implemented multiplication as a modular
1485 (UNSIGNED-BYTE 32) operation on the x86 backend.
1486 * optimization: SEARCH on simple-base-strings can now be open-coded.
1487 (see also contrib/compiler-extras.lisp for inspiration for
1488 teaching the compiler about the Boyer-Moore algorithm).
1489 * value, returned by MAX (and MIN) called with several EQUALP, but
1490 not EQL, arguments now does not depend on compiler settings.
1491 * fixed some bugs revealed by Paul Dietz' test suite:
1492 ** in stack analysis liveness information is propagated from
1493 non-local entry points.
1494 ** pathwise CAST removing failed when the CAST node did not start
1496 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1497 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1499 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1500 host is already defined.
1501 ** RENAME-FILE works on streams instead of signalling an internal
1503 ** PEEK-CHAR uses the current readtable when determining whether
1504 or not a character is whitespace.
1505 ** MERGE-PATHNAMES handles the case when the pathname does not
1506 specify a name while the default-pathname specifies a version.
1507 ** Pathnames now stand a better chance of respecting print/read
1509 ** Attempting to use standardized file system operators with a
1510 pathname with invalid :DIRECTORY components signals a
1512 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1514 ** DIRECTORY on logical pathnames is more correct.
1515 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1516 signal a TYPE-ERROR if their argument is not a stream.
1517 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1518 designator argument does not designate a stream.
1519 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1520 examining the synonym.
1521 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1523 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1524 element-type larger than ([UN]SIGNED-BYTE 32).
1526 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1527 * When built with the :SB-FUTEX feature, threaded builds now take
1528 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1529 for faster/more reliable mutex and condition variable support.
1530 * Incompatible change (but one you probably shouldn't have been using
1531 anyway): the interface and code for arbitrating between multiple
1532 threads in the same user session has been redesigned.
1533 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1534 error when called without an explicit environment argument.
1535 (thanks to Nikodemus Siivola)
1536 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1537 confused about their position. (thanks to Adam Warner and Gerd
1539 * bug fix: Pretty printing backquoted forms with unquotations in the
1540 argument list position of various code constructs such as LAMBDA
1541 now works correctly. (reported by Paul Dietz)
1542 * bug fix: Pretty printing unquotations no longer loses all
1543 stream position information.
1544 * optimization: performance of string output streams is now less
1545 poor for multiple small sequence writes.
1546 * optimization: performance of CSUBTYPEP in the presence of complex
1547 expressions involving CONS and NOT many times has been improved.
1548 (reported by Paul Dietz)
1549 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1551 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1552 streams for the REPL, for the debugger, and so on. (thanks to
1554 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1555 * fixed some bugs revealed by Paul Dietz' test suite:
1556 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1557 is ignored if it is too small, rather than propagating through
1558 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1559 ** extremely complex negations of CONS types were not being
1560 sufficiently canonicalized, leading to inconsistencies in
1562 ** VALUES tranformer lost derived type.
1564 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1565 * fixed a bootstrapping bug: the build process no longer assumes
1566 that the various BOOLE-related constants have the same value in
1567 host and target lisps. (noted by Paul Dietz' test suite on an
1568 SBCL binary built from CLISP)
1569 * The system can now be dynamically linked on the MIPS platform,
1570 which enables dynamic loading of foreign code from Lisp. (thanks
1571 to Ralf Baechle for discussions on the MIPS ABI)
1572 * The system now records debugging information for its own source
1573 files in a filesystem-position-independent manner, relative to
1574 the "SYS" logical host.
1575 * fixed a compiler bug: MV-LET convertion did not check references
1576 to the "max args" entry point. (reported by Brian Downing)
1577 * tweaked disassembly notes to be less confident about proclaiming
1578 some instruction as an LRA. (thanks to Brian Downing)
1579 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1580 now each have their own history, command character, and other
1581 characteristics. (thanks to David Lichteblau)
1582 * fixed some bugs revealed by Paul Dietz' test suite:
1583 ** compiler failure in compiling LOGAND expressions including a
1585 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1586 shift greater than 32.
1587 ** FLUSH-DEST did not mark blocks for type check regeneration.
1588 ** HANDLER-CASE failed to accept declarations in handler clauses
1589 in some circumstances.
1591 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1592 * New code in contrib/sb-introspect (still probably not entirely
1593 stable yet) provides some support for smart Lisp development
1594 environments like SLIME.
1595 * The conditions signalled for errors occurring when loading .fasl
1596 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1597 in a way which should help ASDF recover gracefully.
1598 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1599 stuff has been cleaned up. If you code contrib/ stuff, this might
1600 affect you, and you can look at contrib/README, contrib/STANDARDS,
1601 and/or the 0.8.4.27 diff to check.
1602 * In full calls the compiler now does not generate checks for declared
1603 argument types for all arguments.
1604 * various threading fixes
1605 ** and some experimental patches which didn't make it into
1606 the main tree for this release, but which are shipped in
1607 contrib/experimental-thread.patch as a possible fix for some
1608 failures (deadlock, spinning...) in GC-intensive multithreaded
1610 * fixed PPC build problem (source code incompatibility of different
1611 library versions): added offsetof() hackery which attempts to divine
1612 where glibc maintainers put uc_mcontext today
1613 * fixed bug 282: compiler does not trust type assertions while passing
1614 arguments to a full call.
1615 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1616 &OPTIONAL VALUES type parameter.
1617 * fix bug 214: algorithm for noting rejected templates is now more
1618 similar to that of template seletion. (also reported by rydis on
1620 * fixed bug 141b: printing backquoted information readably and prettily
1621 inserts a space where necessary.
1622 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1623 (CHARACTER 10) are now reported as errors, rather than propagated
1624 as unknown types. (reported by piso on #lisp)
1625 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1626 with values NIL and :ERROR. (thanks to Milan Zamazal)
1627 * fixed bug 191c: CLOS now does proper keyword argument checking as
1628 described in CLHS 7.6.5 and 7.6.5.1.
1629 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1630 longer raise an error; further, using a list as a for-as-arithmetic
1631 counter now raises a meaningful error.
1632 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1633 now understood by sequence creation functions such as MAKE-SEQUENCE
1635 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1636 a STRING-INPUT-STREAM.
1637 * compiler enhancement: SIGNUM is now better able to derive the type
1639 * type declarations inside WITH-SLOTS are checked. (reported by
1641 * fixed some bugs revealed by Paul Dietz' test suite:
1642 ** incorrect optimization of TRUNCATE for a positive first
1643 argument and negative second.
1644 ** compiler failure in let-convertion during flushing dead code.
1645 ** compiler failure while deriving type of TRUNCATE on an
1646 interval, containing 0.
1647 ** ASH of a negative bignum by a negative bignum count now returns
1649 ** intersection of CONS types now canonicalizes properly, fixing
1650 inconsistencies in SUBTYPEP.
1652 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1653 * incompatible change: The --disable-debugger command line
1654 option now clobbers the debugger at a more fundamental
1655 level, by redefining #'INVOKE-DEBUGGER instead of by
1656 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1657 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1658 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1659 enter the debugger REPL and then suffer recursive errors
1660 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1661 while under the new system, BREAK in --disable-debugger mode
1662 terminates the system just as an unhandled error would.
1663 * fixed compiler performance when processing loops with a step >1;
1664 * bug fix: DOCUMENTATION now retrieves generic function
1665 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1666 support has been systematized, and now supports the methods
1667 specified by ANSI, along with a default method and a method for
1668 slot documentation. (reported by Nathan Froyd)
1669 * bug fix: effective methods associated with a generic function are
1670 no longer cached over a change of that generic function's method
1671 combination. (reported by Andreas Fuchs)
1672 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1673 names a non-existent directory. (thanks to Andreas Fuchs)
1674 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1675 the PPC platform, lead to stack corruption; this has been fixed.
1676 (reported by Rainer Joswig)
1677 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1678 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1679 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1680 arithmetic in general, is now much more reliable on the PPC
1682 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1683 platform now returns the right answer.
1684 * optimization: restored some effective method precomputation in
1685 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1686 precomputation is now tunable.
1687 * optimization: compiler-internal data structure use has been
1688 reviewed, and changes have been made that should improve the
1689 performance of the compiler by about 20%.
1690 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1691 simple-base-strings and simple-bit-vectors is improved.
1692 * optimization: the optimization of 32-bit logical and arithmetic
1693 functions introduced in version 0.8.3 on the x86 has been
1694 implemented on the mips, ppc and sparc platforms; an
1695 implementation of the same facility, but for 64-bit arithmetic,
1696 has been added for the alpha.
1697 * microoptimization: the compiler is better able to make use of the
1698 x86 LEA instruction for multiplication by constants.
1699 * bug fix: in some situations compiler did not report usage of
1700 generic arithmetic in (SPEED 3) policy.
1701 * bug 145b fix: compiler used wrong type specifier while converting
1702 MEMBER-types to numeric.
1703 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1705 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1706 index argument now works properly on the Alpha platform.
1707 * bug fix: floating point exception treatment on the Alpha platform
1709 * bug fix: FILE-POSITION works much better on string input and
1710 output streams. (thanks to Nikodemus Siivola)
1711 * bug fix: many threading/garbage collection symptoms sorted.
1712 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1713 might be pseudo-atomic.
1714 * internal change: Stopping for GC is now done with signals not
1715 ptrace. GC is now done in whichever thread wanted it, instead of
1717 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1719 * bug fix: result form in DO is not contained in the implicit
1721 * incompatible change: ICR structure is changed; the value part of
1722 CONTINUATION is now called LVAR; corresponding functions are
1723 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1724 * added type deriver for ISQRT (thanks to Robert E. Brown).
1725 * bug fix: better support for loading from the command line when an
1726 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1728 * fixed some bugs revealed by Paul Dietz' test suite:
1729 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1730 ** ROUND and FROUND now give the right answer when given very
1731 small float arguments.
1732 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1734 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1735 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1736 ** LCM should return a non-negative integer.
1737 ** PARSE-INTEGER returned the index of a terminator instead of the
1738 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1739 ** PARSE-INTEGER returned an incorrect index being applied to a
1741 ** LCM with two arguments of 0 returns 0 rather than signalling
1743 ** unsigned addition of a 32-bit constant with the high bit set no
1744 longer causes an internal compiler error on the x86.
1745 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1746 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1747 with negative last argument.
1748 ** byte specifiers with zero size and position no longer cause
1749 an error during type derivation.
1750 ** bignum multiplication on the Alpha platform now returns the
1752 * porting: The system now builds on SuSE AMD64, although it still
1753 generates a 32-bit binary.
1754 * .fasl file incompatibility: The fasl file version number has
1755 been incremented (because of the changes to internal compiler
1756 data structures referred to above).
1758 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1759 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1760 more accurately, on the Darwin kernel running on PowerPC hardware.
1761 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1762 * Compiler code deletion notes now signal a condition of type
1763 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1764 an associated MUFFLE-WARNING restart.
1765 * The compiler now performs limited argument count validation of
1766 constant format strings in FORMAT, and where appropriate in ERROR,
1767 CERROR and WARN. (thanks to Gerd Moellmann)
1768 * New ASDF-INSTALL contrib can be used for automatic download and
1769 installation of third-party Lisp code from CCLAN or other sites
1771 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1772 which forces another thread to execute a function supplied by the
1774 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1775 now accept and act upon their :ELEMENT-TYPE keyword argument.
1776 (reported by Martin Atzmueller, Edi Weitz)
1777 * bug fix: FILE-POSITION now accepts position designators up to
1778 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1779 is the greater. (thanks to Patrik Nordebo)
1780 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1781 Kalle Olavi Niemitalo)
1782 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1783 printable. (reported by Eric Marsden)
1784 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1785 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1786 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1787 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1788 (thanks to Stig Sandoe)
1789 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1790 UNREAD-CHAR now correctly includes the unread character in the
1791 target sequence. (thanks to Gerd Moellmann)
1792 * bug fix in threaded builds: the system can now be suspended and
1793 resumed by shell job control with minimal disruption.
1794 * bug fixes in times and timezones >2038AD
1795 * better handling of "where is GNU make?" problem in build scripts
1796 (thanks to Nikodemus Siivola)
1797 * new optimization: inside a named function any reference to a
1798 function with the same name is considered to be a self-reference;
1799 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1800 optimization quality.
1801 * new optimization on x86: logical functions and + now have
1802 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1803 used when the result is truncated to 32 bits.
1804 * VALUES declaration is partially enabled.
1805 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1806 * fixed some bugs revealed by Paul Dietz' test suite:
1807 ** The system now obeys the constraint imposed by
1808 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1809 types form a lattice under type intersection.
1810 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1811 ** ASSOC now ignores NIL elements in an alist.
1812 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1813 and (1+ MOST-POSITIVE-FIXNUM) answers.
1814 ** The addition of a method with invalid qualifiers to a generic
1815 function does not cause an error to be signalled immediately;
1816 a warning is signalled, and the error is generated only on
1817 calling the generic function.
1818 * changed .fasl file version number, in part to add type codes for
1819 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1820 obscure ANSI requirements
1822 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1823 * fixed bug 148: failure to inline-expand a local function left
1824 garbage, confusing the compiler.
1825 * fixed bugs 3cd: structure slot readers perform type check if the
1826 slot can have an invalid value (i.e. it is either not initialized
1827 or can be written with a less specific slot writer).
1828 * bug fix: the compiler now traps array references to elements off
1829 the end of an array; previously, the bounds checking in some
1830 circumstances could go off-by-one.
1831 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1833 * type declarations for array element types now obey the description
1834 on the CLHS page "Declaration TYPE", as per discussions on
1835 sbcl-help around 2003-05-08. This means that a declaration
1836 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1837 declaration, all references to BAR will be asserted or assumed
1838 (with THE, so dependent on compiler policy) to involve objects of
1839 type FOO. Note that no such declaration is implied in
1840 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1841 * declared types of functions from the "Conditions"
1842 chapter. (reported by Paul Dietz)
1843 * bug fix: CERROR accepts a function as its first argument.
1844 * bug fix: NTH an NTHCDR accept a bignum as index
1845 arguments. (reported by Adam Warner)
1846 * optimization: character compare routines now optimize comparing
1847 against a constant character. (reported by Gilbert Baumann)
1848 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1849 argument now works properly on the MIPS platform.
1850 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1851 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1852 declared to have a functional type, does not cause an error
1854 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1855 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1856 * fixed bug: if last continuation of a deleted block has a
1857 destination, this destination should be deleted too. (reported by
1859 * fixed a bug in the bootstrap process: the host compiler's values
1860 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1861 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1862 test case from Patrik Nordebo)
1863 * improved the ability of the disassembler on the PPC platform to
1864 provide helpful disassembly notes.
1865 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1866 the class in more cases than previously.
1867 * bug fix: FILE-POSITION now understands :START and :END for
1868 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1869 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1870 8) data. (thanks to David Lichteblau)
1871 * bug fix: it is possible to add a method to a generic function
1872 without lambda list.
1873 * bug fix: reader failed to signal END-OF-FILE inside an
1874 object representation. (reported by Nikodemus Siivola)
1875 * fixed some bugs revealed by Paul Dietz' test suite:
1876 ** LAST and [N]BUTLAST should accept a bignum.
1877 ** condition slot accessors are methods.
1878 ** (VECTOR NIL) is a subtype of STRING.
1880 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1881 * minor incompatible change: some nonsensical specialized lambda
1882 lists (used in DEFMETHOD) which were previously ignored now signal
1884 * minor incompatible change: the system is now aware of the types of
1885 variables in the COMMON-LISP package, and will signal errors for
1886 most violations of these type constraints (where previously they
1887 were silently accepted).
1888 * minor incompatible change: COMPILE-FILE now uses the freedom
1889 afforded (ANSI 3.2.2.3) to use derived function types for
1890 functions defined in the same file. This also permits the system
1891 to warn on static type mismatches and function
1892 redefinition. (Currently it does not work with high DEBUG level.)
1893 * minor incompatible change: VALUES declaration is disabled.
1894 * When issuing notes, the compiler now signals a condition of type
1895 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1896 restart for use in user handlers. It is expected that the
1897 COMPILER-NOTE condition will eventually become a condition
1898 supertype to a hierarchy of note types, which will then be
1899 handleable in a similar fashion. However, at the moment, no such
1900 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1901 but it's an implementation detail, not a classification for the
1903 * Changes in type checking closed the following bugs:
1904 ** type checking of unused values (192b, 194d, 203);
1905 ** template selection based on unsafe type assertions (192c, 236);
1906 ** type checking in branches (194bc).
1907 * A short form of VALUES type specifier has ANSI meaning (it has
1908 increased the number of situations when SBCL cannot perform type
1910 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1911 keywords or constants is permissible.
1912 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1913 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1914 classes. (thanks to Antonio Martinez)
1915 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1916 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1917 argument to a pretty-print stream. Instead, it leaves any such
1918 operations to DESCRIBE-OBJECT methods.
1919 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1921 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1922 lambda lists. (thanks to David Lichteblau)
1923 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1924 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1925 (thanks to Gerd Moellmann)
1926 * fixed bug 246: increased compilation speed of long
1927 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1929 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1931 * DEFCONSTANT now throws a condition of type
1932 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1933 constant to a non-EQL value; CONTINUE and ABORT restarts
1934 respectively change and preserve the value.
1935 * fixed bug 63: The code walker, part of the implementation of CLOS,
1936 is now better at handling symbol macros.
1937 * bug fix: There is no longer an internal implementation type named
1938 CL:LENGTH. (reported by Raymond Toy)
1939 * bug fix: In macro-like defining macros/special operators the
1940 implicit block does not enclose the lambda list.
1941 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1942 enclosed in parentheses) are not suitable as type specifiers, and
1943 their use properly signals an error now.
1944 * bug fix: An argument count mismatch for a type specifier in code
1945 being compiled no longer causes an unhandled error at compile
1946 time, but signals a compile-time warning.
1947 * fixed simple vector readable printing
1948 * bug fix: DESCRIBE takes more care over whether the class
1949 precedence list slot of a class is bound before accessing it.
1950 (reported by Markus Krummenacker)
1951 * bug fix: FORMATTER can successfully compile pretty-printer format
1952 strings which use variants of the ~* directive inside.
1953 * bug fix: SEARCH now applies its TEST predicate to the elements of
1954 the arguments in the correct order. (thanks to Wolfhard Buss)
1955 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1956 in type inference. (thanks to Robert E. Brown)
1957 * bug fix: Reading in symbols with an explicit package name of ""
1958 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1959 * fixed some bugs revealed by Paul Dietz' test suite:
1960 ** NIL is now allowed as a structure slot name.
1961 ** Arbitrary numbers, not just REALs, are allowed in certain
1962 circumstances in LOOP for-as-arithmetic clauses.
1963 ** Multiple class redefinitions before slot access no longer
1964 causes a type error.
1965 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1966 association between the name and a class.
1967 ** Generic functions with non-standard method-combination and over
1968 five methods all of which return constants no longer return NIL
1969 after the first few invocations. (thanks to Gerd Moellmann)
1970 ** CALL-NEXT-METHOD with no arguments now passes the original
1971 values of the arguments, even in the presence of assignment.
1972 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1974 ** DISASSEMBLE works with closures and funcallable instances.
1975 ** ADD-METHOD now returns the generic function, not the new method.
1976 ** FIND-METHOD signals an error if the lengths of the specializers
1977 is incompatible with the generic function, even if the ERRORP
1979 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1980 which its argument is a member.
1981 ** DEFCLASS only redefines the class named by its class-name
1982 argument if that name is the proper name of the class;
1983 otherwise, it creates a new class.
1984 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1985 of the UNBOUND-SLOT condition to the name of the slot.
1986 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1987 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1988 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1989 SLOT-MAKUNBOUND in the specified fashion.
1991 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1992 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1993 cross-compilation host. As a consequence, we can now bootstrap our
1994 way up to SBCL starting with a bare gcc toolchain and human-readable
1995 source code (first the source to CLISP, then the source to SBCL).
1996 * A contributed module containing a partial implementation of the
1997 simple-streams interface has been included. (thanks to Rudi
1999 * A contributed module implementing the RFC1321 Message Digest
2000 Algorithm, known as MD5, has been included.
2001 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2002 no longer has any effect, as the code controlled by this feature
2003 has been deleted. (As far as we know, no-one has ever built using
2004 this feature, and its semantics were confused in any case).
2005 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2006 work on conditions (as required by the ANSI specification),
2007 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2008 expected behaviour on conditions. Users should note, however,
2009 that such behaviour is not required by the ANSI specification,
2010 and so use of this behaviour may render their code unportable.
2011 * fixed some bugs revealed by Paul Dietz' test suite:
2012 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2014 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2015 ** SXHASH on condition objects no longer returns NIL.
2016 ** :ALLOCATION :CLASS slots are better treated; their values are
2017 updated on class redefinition, and initforms inherited from
2018 superclasses are applied.
2019 ** REMOVE-METHOD returns its generic function argument even when
2020 no method was removed.
2021 ** SHARED-INITIALIZE now initializes the values of the requested
2022 slots, including those with :ALLOCATION :CLASS.
2023 ** ALLOCATE-INSTANCE now works on structure classes defined via
2024 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2026 ** SLOT-EXISTS-P now works on conditions, as well as structures
2028 ** MAKE-LOAD-FORM now has the required methods on
2029 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2030 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2031 keyword, and now implements the SLOT-NAMES argument.
2032 ** methods with &OPTIONAL arguments no longer allow too many
2033 arguments to be passed in the call without error.
2034 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2035 option is consistent with the required arguments of the generic
2036 function lambda list.
2037 * bug fix: REQUIRE accepts a string designator. (Thanks to
2039 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2040 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2041 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2042 * bug fix: The compiler now checks for duplicated variables in macro
2044 * bug fix: SETQ on globals returns the correct value.
2045 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2046 not a COMPILER-ERROR (followed by some other strange error on
2047 choosing the CONTINUE restart).
2048 * bug fix: make.sh and friends are now more consistent in the way that
2049 they look for GNU "make".
2051 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2052 * experimental native threads support (on x86 Linux >=2.4 only).
2053 This is not compiled in by default: you need to add :SB-THREAD to
2054 the target features. See the "Beyond ANSI" chapter of the manual
2056 * fix for longstanding nonANSIism: The old distinction between
2057 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2058 The return value from CL:FIND-CLASS is now a CLOS class, and
2059 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2060 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2061 * An interface to the MetaObject Protocol, as described in Kiczales,
2062 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2063 MIT Press, 1991, is available from the SB-MOP package.
2064 * incompatible change: the SB-PCL package should now be considered
2065 a private implementation detail, and no longer a semi-private MOP
2067 * minor incompatible change: due to rearrangement for threads, the
2068 control stack and binding stack are now allocated at arbitrary
2069 addresses instead of being hardcoded per-port. Users affected by
2070 this probably have to be doing advanced things with shared
2071 libraries, and will know who they are.
2072 * minor incompatible change: Previously, all --eval forms used were
2073 processed with READ before any of them were processed with EVAL.
2074 Now each --eval form is processed with both READ and EVAL before
2075 the next --eval form is processed. (Thus package operations like
2076 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2077 work as the user might reasonably expect.)
2078 * minor incompatible change: *STANDARD-INPUT* is now only an
2079 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2081 * minor incompatible change: Y-OR-N-P is now character-oriented,
2082 not line oriented. Also, YES-OR-NO-P now works without errors.
2083 (thanks to Antonio Martinez)
2084 * sb-aclrepl module improvements: an integrated inspector, added
2085 repl features, and a bug fix to :trace command.
2086 * Known functions, which cannot be open coded by the backend, are
2087 considered to be able to check types of their arguments. (fixing
2088 a bug report by Nathan J. Froyd)
2089 * fixed a bug in computing method discriminating functions: It is
2090 now possible to define methods specialized on classes which have
2091 forward-referenced superclasses. (thanks to Gerd Moellmann)
2092 * fixed evaluation order in optional entries (reported by Gilbert
2094 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2095 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2096 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2097 required-by-AMOP NAME argument, as well as CLASS and
2098 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2099 * fixed bug 20: DEFMETHOD can define methods using names that are
2100 not the proper names of classes to designate class specializers.
2101 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2102 Unix file descriptors, instead of blowing up. (thanks to Antonio
2104 * Garbage collection refactoring: user-visible change is that a
2105 call to the GC function during WITHOUT-GCING will not do garbage
2106 collection until the end of the WITHOUT-GCING. If you were doing
2107 this you were probably losing anyway.
2108 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2109 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2110 TYPEP the latter but not the former.
2111 * The compiler issues a full WARNING for calls to undefined functions
2112 with names from the CL package.
2113 * MAP-INTO for a vector destination is open coded. (reported by
2114 Brian Downing on c.l.l)
2115 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2116 documentation string.
2117 * fixed some bugs revealed by Paul Dietz' test suite:
2118 ** COPY-ALIST now signals an error if its argument is a dotted
2120 ** Condition slots are now accessed more correctly in the presence
2121 of multiple initargs for a given slot.
2122 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2123 exclude restarts of the same name associated with a different
2125 ** DEFCLASS of forward-referenced classes with another
2126 forward-referenced class in the superclasses list no longer
2128 ** Condition slots are now initialized once each, not multiple
2129 times. (thanks to Gerd Moellmann)
2130 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2132 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2133 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2134 arguments contain duplicated elements.
2135 ** RESTART-CASE understands local macros.
2136 ** RESTART-CASE associates exactly its own restarts with a condition.
2137 ** ENDP in safe mode checks its argument to be of type LIST.
2138 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2139 in question is unbound.
2140 ** Optimized MAKE-INSTANCE functions no longer cause internal
2141 assertion failures in the presence of duplicate initargs.
2142 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2143 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2144 propagated even in the ctor optimized implementation of
2146 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2148 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2149 * a better implementation of SXHASH on (simple) bit vectors,
2150 measured both in execution speed and in distribution of results
2151 over the positive fixnums, has been installed. Likewise, a better
2152 implementation of EQUAL for simple bit vectors is now available.
2153 * fixed CEILING optimization for a divisor of form 2^k.
2154 * fixed bug 240 (emitting extra style warnings "using the lexical
2155 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2156 by Antonio Martinez)
2157 * fixed SXHASH, giving different results for NIL depending on type
2158 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2159 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2160 variable. (found by Rolf Wester)
2161 * SBCL does not ignore type declarations for special
2162 variables. (reported by rif on c.l.l 2003-03-05)
2163 * some bug fixes in contrib/sb-aclrepl/
2164 * fixed some bugs revealed by Paul Dietz' test suite:
2165 ** a bug in the CONS type specifier, whereby the CAR and CDR
2166 types got intertwined, has been fixed;
2167 ** the type system is now able to reason about the interaction
2168 between INTEGER and RATIO types more completely;
2169 ** APPEND, [N]REVERSE and NRECONC check that those their
2170 arguments, which must be proper lists, are really so;
2171 ** An array specialized to be unable to hold elements has been
2172 implemented, as required -- yes, really -- by ANSI;
2173 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2174 on malformed property lists;
2176 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2177 * incompatible packaging change: in line with Unix convention,
2178 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2179 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2180 * REQUIRE and PROVIDE are now optionally capable of doing something
2181 useful. See the documentation string for REQUIRE.
2182 * infrastructure for a managed SBCL contrib system: contributed
2183 modules in this release include:
2184 ** the ASDF system definition facility;
2185 ** an interface to the BSD Sockets API;
2186 ** an ACL-like convenience interface to the repl;
2187 (thanks to Kevin Rosenberg)
2188 ** an implementation of ROTATE-BYTE, with efficient implementation
2190 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2191 gives a closer approximation to the right answer than previously.
2192 (thanks to Raymond Toy)
2193 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2194 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2195 optional environment arguments, as required by ANSI.
2196 * fixed bugs in other functions taking environment objects, allowing
2197 calls with an explicit NIL environment argument to be compiled
2199 * fixed bug 228: primary return values from
2200 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2201 COMPILE or FUNCTION.
2202 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2203 :TYPE will no longer signal errors on innocuous objects.
2204 * fixed bug 231b: SETQ is better at respecting type declarations in
2205 the lexical environment.
2206 * fixed a bug in DEFCLASS: classes named by symbols with no or
2207 unprintable packages can now be defined.
2208 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2209 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2210 * fixed compiler failure related to checking types of functions.
2211 (reported by Robert E. Brown)
2212 * the compiler is now much more consistent in its error-checking
2213 treatment of bounding index arguments to sequence functions: in
2214 (SAFETY 3) code, errors will be signalled in almost all cases if
2215 invalid sequence bounding indices are passed to functions defined
2216 by ANSI to operate on sequences.
2217 * fixed a bug in the build procedure: documentation of SBCL-specific
2218 packages is now preserved and available in the final Lisp image.
2219 * lifted FDEFINITION lookup out of loops in the implementation of
2220 many list operations. (thanks to Robert E. Brown)
2221 * fixed a bug in the reader: the #n# reader macro now works for
2222 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2223 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2224 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2225 E. Brown and Tony Martinez respectively)
2226 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2227 SLOT-BOUNDP: in particular, it is now able to optimize them much
2228 better, and is now not vulnerable to having packages renamed.
2229 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2230 always called when a slot is not present in an instance. (thanks
2232 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2233 longer signal an error on unreading a character following EOF on
2234 the previous constituent stream. (thanks to Tony Martinez)
2235 * fixed some bugs revealed by Paul Dietz' test suite:
2236 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2237 not just nonnegative fixnums;
2238 ** the logical bit-array operators such as BIT-AND now accept an
2239 explicit NIL for their "opt-arg" argument (to indicate a
2240 freshly-consed result bit-array);
2241 ** ELT now signals an error on an invalid sequence index in safe
2243 ** the type system is now cleverer about negations of numeric
2244 types, and consequently understands the BIGNUM and RATIO types
2246 ** the type system is now cleverer about the interaction between
2247 INTEGER and RATIO types: while bugs still remain, many more
2248 cases are accurately computed;
2249 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2250 if it is in the last clause;
2251 ** CONSTANTLY now correctly returns a side-effect-free function in
2253 ** DECLARE is no longer treated as a special-operator; in
2254 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2255 * incremented fasl file version number due to the change in the
2256 DEFSTRUCT-SLOT-DESCRIPTION structure.
2258 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2259 * minor incompatible change: code processed by the "interpreter" or
2260 EVAL now has a compilation optimization policy of (DEBUG 2)
2261 (changed from (DEBUG 1)) to improve debuggability of interactive
2262 development, and to allow the use of the debug RETURN command in
2264 * an experimental implementation of the RETURN command for the
2265 debugger has been included. (thanks to Frederik Kuivinen)
2266 * fixed bug 62: constraints were not propagated into a loop.
2267 * fixed bug in embedded calls of SORT (reported and investigated by
2269 * fixed some bugs revealed by Paul Dietz' test suite:
2270 ** printing and reading of arrays with some dimensions having
2271 length 0 (thanks to Gerd Moellmann);
2272 ** BOA constructor with &AUX argument without a default value does
2273 not cause a type error;
2274 ** CONSTANTP now returns true for all self-evaluating objects.
2276 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2277 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2278 accessors that are related by inheritance, as specified in the
2279 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2280 Valtteri Vuorikoski)
2281 * The compiler is now able to inline functions that were defined in
2282 a complex lexical environment (e.g. inside a MACROLET).
2283 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2285 * Support for the upcoming FreeBSD-5.0 release has been included.
2286 (thanks to Dag-Erling Smorgrav)
2287 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2288 effect when it is not in a toplevel context.
2289 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2290 stand a better chance of being correct. (thanks to Gerd
2292 * fixed bug in COERCE, which now signals an error on coercing a
2293 rational to a bounded real type which excludes the expected
2295 * The compiler is now able to derive types more accurately from the
2296 COERCE and COMPILE functions.
2297 * fixed bug 223: functional binding is considered to be constant
2298 only for symbols in the CL package.
2299 * fixed bug 231: SETQ did not check the type of a variable being set
2300 (reported by Robert E. Brown)
2301 * A new optimization for MAKE-INSTANCE has been included, fixing
2302 various bugs (including relating to :ALLOCATION :CLASS slots and
2303 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2305 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2306 ** As required by ANSI, LOOP now disallows anonymous collection
2307 clauses such as COLLECT I in conjunction with aggregate boolean
2308 clauses such as THEREIS (= I 1);
2309 ** LOOP now signals an error when any variable is reused in the
2310 same loop (including the potentially useful construct analogous
2311 to WITH A = 1 WITH A = (1+ A);
2312 ** IT is only a special loop symbol within the first clause of a
2313 conditional loop clause;
2314 ** LOOP with a typed iteration variable over a hashtable now
2315 signals a type error iff it should.
2316 * fixed some other bugs revealed by Paul Dietz' test suite:
2317 ** FILE-STREAM now names the class previously known as FD-STREAM;
2318 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2319 argument) no longer signals an error;
2320 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2321 of the slot symbol, rather than using the current package
2322 ((:CONC-NAME "") continues to intern the slot's name in the
2324 * incremented fasl file version number, because of the incompatible
2325 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2326 of the new implementation of DEFINE-COMPILER-MACRO.
2328 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2329 * Support for building SBCL for MIPS platforms running in
2330 little-endian mode has now been checked in, and basic
2331 functionality on said platforms verified.
2332 * minor incompatible change: PCL now records the pathname of a file
2333 in which methods and the like are defined, rather than its
2335 * minor incompatible change: TRUENAME now considers the truename of
2336 a file naming a directory to be the pathname with :DIRECTORY
2337 component indicating that directory.
2338 * minor incompatible change: a NAMED clause in the extended form of
2339 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2340 reason for the previous behaviour is unclear.
2341 * more systematization and improvement of CLOS and MOP conformance
2342 in PCL (thanks to Gerd Moellman and Pierre Mai):
2343 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2345 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2346 primary methods with no specializers;
2347 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2349 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2350 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2351 FORWARD-REFERENCED-CLASSes; error reporting on
2352 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2354 ** SXHASH on CLOS instances now uses a slot internal to the
2355 instance to return different numbers on distinct instances,
2356 while preserving the same return value through invocations of
2358 ** DEFMETHOD signals errors when methods with longer incongruent
2359 lambda lists are added to generic functions;
2360 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2361 CLASS, as specified in AMOP;
2362 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2363 on the order returned by the primary method for classes of
2364 class STANDARD-CLASS;
2365 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2366 * fixed some bugs shown by Paul Dietz' test suite:
2367 ** DOLIST puts its body in TAGBODY;
2368 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2370 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2371 value producing form;
2372 ** if more variables are given to PROGV than values, extra
2373 variables are bound and made to have no value;
2374 ** NSUBSTITUTE on list arguments gets the right answer with
2376 ** ELT signals an error of type TYPE-ERROR when the index argument
2377 is not a valid sequence index;
2378 ** LOOP signals (at macroexpansion time) an error of type
2379 PROGRAM-ERROR when duplicate variable names are found;
2380 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2381 ** FUNCALL of special-operators now cause an error of type
2383 ** PSETQ now works as required in the presence of side-effecting
2384 symbol-macro places;
2385 ** NCONC accepts any object as its last argument;
2386 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2388 ** loop-for-as-package does not require a package to be explicitely
2390 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2392 * fixed bug 166: compiler preserves "there is a way to go"
2393 invariant when deleting code.
2394 * fixed bug 172: macro lambda lists with required arguments after
2395 &REST arguments now cause an error to be signalled. (thanks to
2397 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2398 bug. (thanks to Matthew Danish)
2399 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2401 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2402 arguments are lexically rebound. (thanks to Gerd Moellmann and
2404 * fixed bug 194: error messages are now more informative when there
2405 is no primary method applicable in a call to a generic
2406 function. (thanks to Gerd Moellmann)
2407 * fixed bug in command line argument checking (thanks to Julian
2409 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2410 (thanks to Matthew Danish)
2411 * incremented fasl file version number, because of the
2412 SXHASH-related changes in the layout of CLOS data structures
2414 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2415 * minor incompatible change: The runtime (the Unix executable named
2416 "sbcl") is now much pickier about the .core files it will load.
2417 Essentially it now requires .core files to descend from the same
2418 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2419 as the runtime does. (The intent is to prevent the crashes which
2420 can occur, and which can even be reported as mysterious failures,
2421 when people patch the sources or change the build parameters
2422 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2423 sbcl and .core files.)
2424 * fixed bug: VALUES-LIST is no longer optimized away.
2425 * fixed bug 142: The FFI conversion of C string values to Lisp
2426 string values no longer conses excessively. (thanks to Nathan
2427 Froyd porting Raymond Toy's fix to CMU CL)
2428 * began to systematize and improve MOP conformance in PCL (thanks to
2429 Nathan Froyd, Gerd Moellman and Pierre Mai):
2430 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2432 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2433 ** FINALIZE-INHERITANCE is now called on class finalization;
2434 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2435 argument precedence order.
2436 * fixed bug 202: The compiler no longer fails on functions whose
2437 derived types contradict their declared type.
2438 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2439 so it can be non-toplevel.
2440 * The fasl file version number has changed (because of the new
2441 implementation of DEFMACRO).
2442 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2443 safe code, that any length requirement by their type-specifier
2444 argument is valid. The exceptions to this are described in bug
2446 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2447 functions now check the types of their inputs as required by ANSI.
2448 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2449 introduced symbol is DECLAREd to be SPECIAL.
2450 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2451 * fixed bug: :COUNT argument to sequence functions may be negative
2452 * fixed bug: body of DO-SYMBOLS may contain declarations
2453 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2454 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2455 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2456 symbol macro only once
2457 * fixed printing of call frame when argument list is unavailable
2458 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2459 * compiler no longer signals WARNING on unknown keyword
2462 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2463 * A beta-quality port to the mips architecture running Linux,
2464 based on the old CMUCL backend, has been made. It has been tested
2465 on a big-endian kernel, and works sufficiently well to be able to
2466 rebuild itself; it has not been tested in little-endian mode.
2467 * fixed an inconsistency between gencgc.c and purify.c which made
2468 dumping/loading .core files unreliable
2469 * fixed bug 120a: The compiler now deals correctly with IFs where
2470 the consequent is the same as the alternative, instead of
2471 misderiving the return type. (thanks to Alexey Dejneka)
2472 * fixed bug 113: Logical pathnames are now dumpable (the logical
2473 host is resolved at load-time, throwing an error if it is not
2475 * fixed bug 174: FORMAT's error message is slightly clearer when a
2476 non-printing character is used in a format directive.
2477 * fixed several bugs in compiler checking of type declarations, i.e.
2478 violations of the Python "declarations are assertions" principle
2479 (thanks to Alexey Dejneka)
2480 * fixed several bugs in PCL's error checking (thanks to Gerd
2482 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2484 * fixed bug in compilation of functions as first class values
2485 (thanks to Antonio Martinez-Shotton)
2486 * The compiler's handling TYPE-ERRORs which it can prove will
2487 inevitably happen at runtime has been cleaned up and corrected
2488 in several ways. (thanks to Alexey Dejneka)
2489 * improved argument type checking for various basic arithmetic
2490 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2491 lavished on them in the past that they can be compiled in many
2492 ways in different special cases
2493 * fixed bug 181: compiler checks validity of user supplied type
2495 * cleaned up code flushing in optimization: Function calls which
2496 should signal errors for safety purposes (e.g. which ANSI says
2497 should signal errors when their arguments are of incorrect type)
2498 are no longer optimized away.
2499 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2500 * incremented fasl file version number, because changes in the
2501 implementation of sequence functions like COERCE caused
2502 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2503 in old inline expansions) to become undefined. (Actually these
2504 changes were later undone, so we might very well be binary
2505 compatible with 0.7.7 after all, but leaving the version number
2506 incremented seemed like the simplest and most conservative
2509 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2510 * An alpha-quality port to the parisc architecture running Linux,
2511 based on the old CMUCL backend, has been made. This, even more so
2512 than the other backends, should be considered still a work in
2513 progress; known problems include that the Linux kernel in 64-bit
2514 mode does not propagate the correct sigcontext structure to
2515 userspace, and consequently SBCL on a parisc64 kernel will not
2517 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2518 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2519 suppressed.) Also now that the compiler is looking at declarations
2520 in the environment, it checks optimization declarations as well,
2521 and suppresses inlining when (> DEBUG SPEED).
2522 * More fixes have been made to treatment of floating point exception
2523 treatment and other Unix signals. In particular, floating point
2524 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2525 * The detection and handling of control stack exhaustion (infinite
2526 or very deeply nested recursion) has changed. Stack exhaustion
2527 detection is now done by write-protecting pages at the OS level
2528 and applies at all optimization settings; when found, a
2529 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2530 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2531 be caught using IGNORE-ERRORS.
2532 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2533 correctly when transforming recursive LABELS functions to LETs.
2534 (thanks to Alexey Dejneka)
2535 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2536 that are names of constants or global variables.
2537 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2538 alien routines with docstrings.
2539 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2540 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2542 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2543 (thanks to Eric Marsden)
2544 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2545 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2546 * Bug fix: The SPARC backend can now compile functions involving
2547 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2548 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2549 to a functions expecting a PATHNAME-DESIGNATOR.
2550 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2551 lambda lists. (thanks to Alexey Dejneka)
2552 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2553 functions so that EXPORTing the name of the class doesn't cause
2554 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2555 to Antonio Martinez for reporting this)
2556 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2557 constant-folded without failing an assertion. (thanks to Einar
2558 Floystad Dorum for reporting this)
2559 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2560 types are now dealt with more correctly by the compiler.
2561 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2562 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2563 behaviour analogous to RENAME-FILE. This puts its behaviour more
2564 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2566 * The fasl file version number has changed again. (because of the
2567 bug fix involving the names of PCL MAKE-INSTANCE functions)
2569 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2570 * bug fix: Floating point exceptions are treated much more
2571 consistently on the x86/Linux and PPC/Linux platforms.
2572 * Array initialization with :INITIAL-ELEMENT is now much faster for
2573 cases when the compiler cannot open code the array creation, but
2574 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2575 array accesses have also seen a speed increase.
2576 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2577 specified. (This was at the root of some bad interactions between
2578 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2580 * bug fix: Internal error arguments for undefined functions are now
2581 computed correctly on the PPC/Linux platform.
2582 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2583 Raymond Toy's patch for CMU CL)
2584 * Support for the Solaris 9 operating environment has been included
2585 (thanks to Daniel Merritt)
2586 * A very ugly but hopefully complete draft of the missing FFI chapter
2587 of the manual has been created by reformatting the corresponding
2588 CMU CL manual chapter into (currently very ugly and incoherent)
2589 DocBook and bringing it up to date for SBCL behavior. Thus, the
2590 manual is now essentially complete, at least by my extreme
2591 once-and-only-once standards, whereby it's acceptable to refer to
2592 the doc strings of SB-EXT functions as the primary documentation.
2593 * The fasl file version number has changed again, due to cleanup of
2594 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2595 is no longer a static symbol.)
2597 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2598 * SBCL now builds with OpenMCL (version 0.12) as the
2599 cross-compilation host; also, more progress has been made toward
2600 bootstrapping under CLISP.
2601 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2603 * bug 158 fixed: The compiler can now deal with integer loop
2604 increments different from 1; fixing this turned out also to fix
2606 * bug 169 fixed: no more bogus warnings about using lexical bindings
2607 despite the presence of perfectly good SPECIAL declarations (thanks
2608 to David Lichteblau)
2609 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2610 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2611 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2613 * bug 180 fixed: Method combination specifications no longer ignore
2614 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2615 * bug fix: Structure type predicate functions now check their argument
2616 count as they should.
2617 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2618 correctly. (thanks to Pierre Mai)
2619 * minor incompatible change: The --noprogrammer option is deprecated
2620 in favor of the new --disable-debugger option, which is very similar.
2621 (The major difference is that it takes effect at a slightly different
2622 time at startup, causing handling of errors in --sysinit and
2623 --userinit files will be affected differently.) The
2624 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2625 been added to allow this functionality to be controlled from ordinary
2626 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2627 maintainers, who might want to run non-interactive scripts to
2628 build SBCL cores which will later be used interactively.)
2629 * minor incompatible change: The LOAD function no longer, when given
2630 a wild pathname to load, loads all files matching that pathname.
2631 Instead, an error of type FILE-ERROR is signalled.
2633 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2634 * bug 147 fixed: The compiler preserves its block link/count
2635 invariants more correctly now so that it doesn't crash. (thanks
2637 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2639 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2640 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2641 no-value-for-this-character as NIL (as specified by ANSI).
2642 * HOST-NAMESTRING on physical pathnames now returns a string that is
2643 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2644 (thanks to Christophe Rhodes)
2645 * The Alpha port handles icache flushing more correctly. (thanks to
2647 * More progress has been made toward bootstrapping under CLISP. (thanks
2648 to Christophe Rhodes)
2649 * The fasl file format has changed again, because dynamic loading
2650 on OpenBSD (which has non-ELF object files) motivated some cleanups
2651 in the way that foreign symbols are transformed and passed around.
2652 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2653 is no longer treated as whitespace by the reader, but instead as
2654 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
2655 |A
\x7fB|, instead of A as it used to.
2657 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2658 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2659 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2660 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2661 this as of 0.7.1.45, but was left out of the previous news section
2662 (thanks to Dan Barlow)
2663 * SBCL now runs on the Solaris operating system on SPARC architectures
2664 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2665 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2666 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2667 * SPARC backend cleanups, allowing builds of cores optimized for V8
2668 and V9 SPARCS, and also emission of code targeted to a particular
2669 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2671 * SBCL is closer to bootstrapping under CLISP, thanks to various
2672 fixes by Christophe Rhodes.
2673 * The fasl file format has changed again, to allow the compiler's
2674 INFO database to support symbol macros.
2675 * The user manual (in doc/) is formatted into HTML more nicely.
2676 (thanks to coreythomas)
2677 * The system is smarter about SUBTYPEP relationships, especially
2678 those involving NOT types (including types such as ATOM which are
2679 represented internally using NOT types). Thus SUBTYPEP is less
2680 likely to return (VALUES NIL NIL) in general, and in particular
2681 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2683 * The fasl file format has changed again, because the internal
2684 representation of types now includes a new slot to support the new
2685 SUBTYPEP-of-NOT-types logic.
2686 * (not a change in the main branch of SBCL, but a related prototype
2687 which can hopefully be merged into the main branch of SBCL in the
2688 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2689 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2690 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2691 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2692 dispatch macro character combinations. (thanks to Alexey Dejneka)
2693 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2694 that superficially look like logical namestrings correctly.
2695 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2697 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2698 default supplied-p) for &optional and &key arguments. (thanks to
2701 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2702 * incompatible change: The compiler is now less aggressive about
2703 tail call optimization, doing it only when (> SPACE DEBUG) or
2704 (> SPEED DEBUG). (This is an incompatible change because there are
2705 programs which relied on the old CMU-CL-style behavior to optimize
2706 away their unbounded recursion which will now die of stack overflow.)
2707 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2708 for non-GENCGC systems has been increased to 20M (since that
2709 seems much closer to the likely performance optimum for modern
2710 systems than the old 4M value was)
2711 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2712 and *DEBUG-PRINT-LEVEL*
2713 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2714 of CMU CL's support for SPARC, and various endianness and other
2715 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2716 * new syntactic sugar for the Unix command line: --load foo.bar is now
2717 an alternate notation for --eval '(load "foo.bar")'.
2719 ** The system now detects stack overflow and handles it gracefully,
2720 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2721 optimization settings. (This is a good thing in general, and
2722 its introduction in this version should be particularly timely
2723 for anyone whose code fails because of suppression of tail
2725 ** The system now hunts for the C variable "environ" in a more
2726 devious way, to avoid segfaults when the C library version differs
2727 between compile time and run time. (thanks to Christophe Rhodes)
2728 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2729 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2730 ** The compiler no longer issues bogus style warnings for undefined
2731 classes in the same source file as the DEFCLASSes which defined
2732 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2734 ** fixes in CONDITION class precedence list for undefined function
2735 errors (thanks to Alexei Dejneka)
2736 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2737 correctly. (thanks to Dan Barlow)
2738 ** portability fixes aiming at bootstrapping under CLISP (thanks
2739 to Dave McDonald and Christophe Rhodes)
2740 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2741 ** fixes in type translation and and type inference (thanks to
2743 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2744 ** various fixes in the new ports (thanks to Dan Barlow)
2745 * several changes related to debugging:
2746 ** suppression of tail recursion, as noted above
2747 ** stack overflow detection, as noted above
2748 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2749 is now the default. (For some time encapsulation has been more
2750 reliable than the breakpoint-based :ENCAPSULATE NIL
2751 implementation, at least on X86 systems; and I just noticed that
2752 encapsulation also seems closer to the spirit of the ANSI
2755 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2757 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2758 up properly again. (There was a packaging bug in 0.7.0 which
2759 left their definitions in SB-SYS::LOAD-FOREIGN and
2760 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2761 vital for most things which interface to C-level interfaces,
2762 like extensions working with sockets or databases or
2763 Perl-compatible regexes or whatever, and the need to fix
2764 this bug is the main reason that 0.7.1 was released so
2766 ** DEFGENERIC is now choosier about the methods it redefines, so that
2767 reLOADing a previously-LOADed file containing DEFGENERICs does
2768 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2769 cycle now works as it should. (thanks to Alexey Dejneka)
2770 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2771 fixed by Christophe Rhodes. (He actually submitted this patch
2772 months ago, and I delayed until after 0.7.0.)
2773 ** Bug 111 (internal compiler confusion about runtime checks on
2774 FUNCTION types) was fixed by Alexey Dejneka.
2775 * Some internal cleanups (getting rid of variables which aren't
2776 needed now that the byte interpreter is gone) caused the fasl
2777 file format number to change again.
2779 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2780 * major incompatible change: The default fasl file extension, i.e. the
2781 default extension for files produced by COMPILE-FILE, has changed
2782 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2784 ** There are many changes in the implementation of the compiler.
2785 SBCL is now essentially a compiler-only implementation of ANSI
2786 Common Lisp. EVAL still "interprets" a few special cases, but
2787 almost all the interesting cases are handled by creating
2788 a LAMBDA expression, calling COMPILE on it, then calling
2789 FUNCALL on the result.
2790 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2791 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2792 Since the code is newer, there might still be some new bugs
2793 (though not as many as before Martin Atzmueller's fixes:-). But
2794 the new code is substantially simpler and clearer, and hopefully
2795 any remaining bugs will be simpler, less fundamental, and more
2796 fixable then the bugs in the old code.
2797 ** The revised compiler is still a little unsteady on its feet.
2799 *** The debugging information it produces (particularly the names
2800 of FUNCTION objects) is sometimes much less useful than what
2801 the old compiler produced.
2802 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2803 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2804 within a MACROLET) has been temporarily weakened.
2805 ** There are new compiler optimizations for various functions:
2806 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2807 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2808 *** the math functions TRUNCATE, FLOOR, and CEILING
2809 *** the function-of-all-trades COERCE
2810 Mostly these should be transparent, but there's one
2811 potentially-annoying problem (bug 117): when the compiler
2812 inline-expands a function and does type analysis on the result,
2813 it can create control paths which have type mismatches, and
2814 when it can't prove that those control paths aren't taken,
2815 it will issue WARNINGs about the type mismatches. This is
2816 a particular problem in practice for the new sequence functions.
2817 It's not clear how this should be fixed, and for now, a
2818 workaround is given in the entry for 117 in the BUGS file.
2819 ** (Because of the interaction between the two previous items --
2820 occasional inlining problems and new inline expansions -- some
2821 of the new sequence function optimizations won't really kick in
2822 completely until debugging information, and then inlining, are
2823 straightened out in some future version.)
2824 * minor incompatible changes:
2825 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2826 DIRECTORY no longer implicitly promotes NIL slots of its
2827 pathname argument to :WILD. In particular, when you ask for the
2828 contents of a directory (which you used to be able to do without
2829 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2830 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2831 ** changes in behavior that ANSI explicitly defines to be
2832 implementation dependent:
2833 *** The new compiler-only implementation still conforms with ANSI,
2834 but acts a little different than before. Besides the obvious
2835 changes in performance tradeoffs (that the cost per form passed
2836 to EVAL has gone up, and the cost per form executed by EVAL
2837 has gone down), the behavior of the system changes a little
2838 because there are no longer any interpreted function objects.
2839 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2840 e.g. doing COMPILE on the output of interactive DEFUN is
2842 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2844 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2845 has changed from (:CL) to NIL.
2846 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2847 e.g. #\Newline and #\Tab, don't have names specified in the
2848 ANSI Common Lisp standard, is now based on their ASCII symbolic
2849 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2850 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2851 are no longer used for output.
2852 ** changes in internal implementation constants:
2853 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2854 4 million. (If your application spends a lot of time GCing and
2855 you have a lot of RAM, you might want to experiment with
2856 increasing it even more.)
2857 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2858 However, almost all old code should still continue to work without
2859 immediate update, as SB-C-CALL is now a (deprecated) nickname
2861 ** Old operator names in the style DEF-FOO are now deprecated in
2862 favor of new corresponding names DEFINE-FOO, for consistency with
2863 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2864 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2865 internal symbols, but a few supported extensions like
2866 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2867 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2868 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2869 etc. as you get deeper into recursive calls to the debugger
2870 command loop, instead of the old "5]", "5]]", "5]]]"
2871 sequence. (I was motivated to do this when squabbles between
2872 ILISP and SBCL left me very deeply nested in the debugger. In the
2873 short term, this change will probably provoke more ILISP/SBCL
2874 squabbles, but hopefully it will be an improvement in the long run.)
2875 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2876 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2877 function, since declaiming return types involving aliens is
2878 (1) annoyingly messy to do by hand and (2) vital to efficient
2879 compilation of code which calls such functions.
2880 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2881 longer reexported by the SB-EXT package. They're solely useful
2882 for alien code, so it seems more logical that you should get
2883 them from the SB-ALIEN package, not in SB-EXT.
2884 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2885 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2886 features. Instead, the code that they used to control is always
2887 built into the system.
2888 * many other bug fixes
2889 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2890 advantage of the new EVAL-WHEN stuff and to clean them up in
2891 general, and they are now more ANSI-compliant in a number of
2892 ways. Martin Atzmueller is responsible for a lot of this.
2893 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2895 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2896 *** correct ERROR type for various file operations
2897 *** some fixes for Lisp streams
2898 *** DEFMETHOD syntax checking
2899 *** changing old weird representation of debug information as
2900 strings (which, among their other deficiencies, don't transform
2901 correctly when you rename packages, and don't change their
2902 print representation when you change things like *PACKAGE*
2903 and *PRINT-LENGTH*) to symbols and lists of symbols
2904 He also made several improvements and fixed several bugs in DESCRIBE.
2905 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2907 *** misbehavior of WRITE-STRING/WRITE-LINE
2908 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2909 and several other LOOP problems as well
2910 *** DIRECTORY when similar filenames are present
2911 *** DEFGENERIC with :METHOD options
2912 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2913 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2914 *** argument ordering in FIND with :TEST option
2915 *** mishandled package designator argument in APROPOS-LIST
2916 *** various problems in the backquote readmacro
2917 *** a bug in APROPOS
2918 *** probably some others that I'm not describing very well here,
2919 since the CVS log documents them by reference to sbcl-devel
2920 messages, and the SourceForge archives aren't working well.:-(
2921 ** Dan Barlow improved the Alpha port (and is making progress on the
2922 PPC port, for those of you who think different).
2923 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2924 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2925 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2926 continues to work on the SPARC port (for those of you in a position
2927 to look down upon our little PC-compatible boxes from a great height).
2928 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2929 and uses that copy, rather than the current dynamic value, when
2930 it's trying to decide whether to truncate output. Thus e.g.
2931 (let ((*print-lines* 50))
2932 (pprint-logical-block (stream nil)
2934 (let ((*print-lines* 8))
2935 (print (aref possiblybigthings i) stream)))))
2936 should now truncate the logical block only at 50 lines, instead of
2937 often truncating it at 8 lines, as it did before.
2938 * The doc/cmucl/ directory, containing old CMU CL documentation
2939 from the time of the fork, is no longer part of the base system.
2940 SourceForge has shut down its anonymous FTP service, and with it
2941 my original plan for distributing the old CMU CL documentation
2942 there. For now, if you need these files you can download an old
2943 SBCL source release and extract them from it.
2944 * The fasl file version number changed again, for dozens of reasons,
2945 some of which are apparent above.
2947 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2948 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2949 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2950 patch, so that DISASSEMBLE output is much nicer.
2951 * The code in the SB-PROFILE package now seems reasonably stable.
2952 I still haven't decided what the final interface should look like
2953 (I'd like PROFILE to interact cleanly with TRACE, since both
2954 facilities use function encapsulation) but if you have a need
2955 for profiling now, you can probably use it successfully with
2956 the current CMU-CL-style interface.
2957 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2958 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2959 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2960 link now returns the dangling link itself, and for similar
2961 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2962 link itself. (In these cases the old code signalled an error and
2963 looped endlessly, respectively.) Thus, DIRECTORY now works even
2964 in the presence of dangling and cyclic symbolic links.
2965 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2966 is now a supported extension again, since the consensus on
2967 sbcl-devel was that it can be useful for ordinary development
2968 work, not just for debugging SBCL itself.
2969 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2970 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2971 that currently-defined functions might be redefined later with
2972 different return types.
2973 * Hash tables can be printed readably, as inspired by CMU CL code
2974 of Eric Marsden and SBCL code of Martin Atzmueller.
2975 * better error handling in CLOS method combination, thanks to
2976 Martin Atzmueller porting Pierre Mai's CMU CL patches
2977 * more overflow fixes for >16Mbyte I/O buffers
2978 * A bug in READ has been fixed, so that now a single Ctrl-D
2979 character suffices to cause end-of-file on character streams.
2980 In particular, now you only need one Ctrl-D at the command
2981 line (not two) to exit SBCL.
2982 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2984 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2985 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2986 does the right thing.
2987 * The implementation of some type tests, especially for CONDITION
2988 types, is now tidier and maybe faster, due to CMU CL code
2989 originally by Douglas Crosher, ported by Martin Atzmueller.
2990 * Some math functions have been fixed, and there are new
2991 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2992 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2993 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2994 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2995 likely to be built into sbcl-0.7.0.)
2996 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2997 CL, ported to SBCL by Martin Atzmueller)
2998 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2999 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3000 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3001 * There's a new slam.sh hack to shorten the edit/compile/debug
3002 cycle for low-level changes to SBCL itself, and a new
3003 :SB-AFTER-XC-CORE target feature to control the generation of
3004 the after-xc.core file needed by slam.sh.
3005 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3006 COMPILE-FILE is no longer supported, so that now every function
3007 gets an entry point, so that block compilation looks a little
3008 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3009 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3010 returns the number of bytes consed since the system started,
3011 rather than the number consed since the first time the function
3012 was called. (The new definition parallels ANSI functions like
3013 CL:GET-INTERNAL-RUN-TIME.)
3014 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3015 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3016 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3017 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3018 * Fasl file version numbers are now independent of the target CPU,
3019 since historically most system changes which required version
3020 number changes have affected all CPUs equally. Similarly,
3021 the byte fasl file version is now equal to the ordinary
3024 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3025 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3026 is no longer recognized. I apologize for this, because it was
3027 listed in SB-EXT as a supported extension, but I found that
3028 its existing behavior was poorly specified, as well as incorrectly
3029 specified, and it looked like too much of a mess to straighten it
3030 out. I have enough on my hands trying to get ANSI stuff to work..
3031 * many patches ported from CMU CL by Martin Atzmueller, with
3032 half a dozen bug fixes in pretty-printing and the debugger, and
3033 half a dozen others elsewhere
3034 * fixed bug 13: Floating point infinities are now supported again.
3035 They might still be a little bit flaky, but thanks to bug reports
3036 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3037 as flaky as they were.
3038 * The --noprogrammer command line option is now supported. (Its
3039 behavior is slightly different in detail from what the old man
3040 page claimed it would do, but it's still appropriate under the
3041 same circumstances that the man page talks about.)
3042 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3043 are now supported, and enabled by default. Thus, the compiler can
3044 handle many floating point and complex operations much less
3045 inefficiently. (Thus e.g. you can implement a complex FFT
3047 * The compiler now detects type mismatches between DECLAIM FTYPE
3048 and DEFUN better, and implements CHECK-TYPE more correctly, and
3049 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3050 thanks to patches from Martin Atzmueller.
3051 * various fixes to make the cross-compiler more portable to
3052 ANSI-conforming-but-different cross-compilation hosts (notably
3053 Lispworks for Windows, following bug reports from Arthur Lemmens)
3054 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3055 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3056 fixed thanks to Pierre Mai's CMU CL patches.
3057 * fixes to keep the system from overflowing internal counters when
3058 it tries to use i/o buffers larger than 16M bytes
3059 * fixed bug 45a: Various internal functions required to support
3060 complex special functions have been merged from CMU CL sources.
3061 (When I was first setting up SBCL, I misunderstood a compile-time
3062 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3063 * improved support for type intersection and union, fixing bug 12
3064 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3065 more obscure bugs as well
3066 * some steps toward byte-compiling non-performance-critical
3067 parts of the system, courtesy of patches from Martin Atzmueller
3068 * Christophe Rhodes has made some debian packages of sbcl at
3069 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3070 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3071 stable, but are nonetheless usable. When he's ready, I'd be happy
3072 to add them to the SourceForge "File Releases" section. (And if
3073 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3074 * new fasl file format version number (because of changes in
3075 internal representation of (OR ..) types to accommodate the new
3076 support for (AND ..) types, among other things)
3078 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3079 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3081 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3083 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3084 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3085 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3086 * DESCRIBE now works on structure objects again.
3087 * Most function call argument type mismatches are now handled as
3088 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3089 know whether the function will be redefined before the call is
3090 executed. (The compiler could flag local calls with full WARNINGs,
3091 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3092 it doesn't keep track of enough information to know whether calls
3093 are local in this sense.)
3094 * Compiler output is now more verbose, with messages truncated
3095 later than before. (There should be some supported way for users
3096 to override the default verbosity, but I haven't decided how to
3097 provide it yet, so this behavior is still controlled by the internal
3098 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3099 src/compiler/ir1util.lisp.)
3100 * Fasl file format version numbers have increased again, because
3101 support for the Gray streams extension changes the layout of the
3102 system's STREAM objects.
3103 * The Gray subclassable streams extension now works, thanks to a
3104 patch from Martin Atzmueller.
3105 * The full LOAD-FOREIGN extension (not just the primitive
3106 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3107 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3108 but like most other programs, it defaults to copying the Unix
3109 environment from the original process instead of starting the
3110 new process in an empty environment.
3111 * Extensions which manipulate the Unix environment now support
3112 an :ENVIRONMENT keyword option which doesn't smash case or
3113 do other bad things. The CMU-CL-style :ENV option is retained
3114 for porting convenience.
3115 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3116 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3118 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3120 * A patch from Martin Atzmueller seems to have solved the SIGINT
3121 problem, and as far as we know, signal-handling now works cleanly.
3122 (If you find any new bugs, please report them!)
3123 * The system no longer defaults Lisp source file names to types
3124 ".l", ".cl", or ".lsp", but only to ".lisp".
3125 * The compiler no longer uses special default file extensions for
3126 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3127 seems to expect a single default extension for all compiled code,
3128 and there's no compelling reason to try to stretch the standard
3129 to allow two different extensions.) Instead, byte-compiled files
3130 default to the same extension as native-compiled files.
3131 * Fasl file format version numbers have increased again, because
3132 a rearrangement of internal implementation packages made some
3133 dumped symbols in old fasl files unreadable in new cores.
3134 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3135 many fewer weird special cases.
3136 * Bug #17 (differing COMPILE-FILE behavior between logical and
3137 physical pathnames) has been fixed, and some related misbehavior too,
3138 thanks to a patch from Martin Atzmueller.
3139 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3140 by Tim Moore, ported to SBCL by Martin Atzmueller.
3141 * Martin Atzmueller fixed several filesystem-related problems,
3142 including bug #36, in part by porting CMU CL patches, which were
3143 written in part by Paul Werkowski.
3144 * More compiler warnings in src/runtime/ are gone, thanks to
3145 more patches from Martin Atzmueller.
3146 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3149 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3151 * DESCRIBE now works on CONDITION objects.
3152 * The debugger now handles errors which arise when trying to print
3153 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3155 * The build system now uses an additional file, customize-target-features.lisp,
3156 to allow local modifications to the target *FEATURES* list. (The point of
3157 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3158 debugging features enabled, without having to worry about propagating it
3159 into everyone's system when I do a "cvs update".) When no
3160 customize-target-features.lisp file exists, the target *FEATURES* list
3161 should be constructed the same way as before.
3162 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3163 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3164 warn about behavior which is undefined under the ANSI spec. Note: This
3165 is specified by ANSI, but it's not very popular with programmers.
3166 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3167 macro in the SBCL sources for an example of a workaround which you
3168 might use to make portable ANSI-standard code which does what you want.
3169 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3170 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3171 when it's used as a non-toplevel form. (This is required in order
3172 to implement the DEFCONSTANT-EQX macro.)
3173 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3174 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3175 funny things, due to bugs in the implementation of EVAL-WHEN
3176 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3177 non-toplevel DEFCONSTANTs are uncommon.)
3178 * The core file version number and fasl file version number have been
3179 incremented, because the old noncompliant DEFCONSTANT behavior involved
3180 calling functions which no longer exist, and because I also took the
3181 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3182 * fixed bug 1 (error handling before read-eval-print loop starts), and
3183 redid debugger restarts and related debugger commands somewhat while
3185 ** The QUIT debugger command is gone, since it did something
3186 rather different than the SB-EXT:QUIT command, and since it never
3187 worked properly outside the main toplevel read/eval/print loop.
3188 Invoking the new TOPLEVEL restart provides the same functionality.
3189 ** The GO debugger command is also gone, since you can just invoke
3190 the CONTINUE restart directly instead.
3191 ** The TOP debugger command is also gone, since it's redundant with the
3192 FRAME 0 command, and since it interfered with abbreviations for the
3194 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3196 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3197 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3198 some porting work by Martin Atzmueller)
3199 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3200 with a boatload of patches which he ported to SBCL. Now that those
3202 ** The system tries to make sure that its low-priority messages
3203 are prefixed by semicolons, to help people who like to use
3204 syntax highlighting in their ILISP buffer. (This patch
3205 was originally due to Raymond Toy.)
3206 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3207 patches originally written by Raymond Toy.
3208 ** The compiler understands coercion between single-value and
3209 multiple-VALUES type expressions better, getting rid of some very
3210 weird behavior, thanks to patches originally by Robert MacLachlan
3211 and Douglas Crosher.
3212 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3213 lambda lists, thanks to a patch originally by Pierre Mai.
3214 ** The system no longer bogusly warns about "abbreviated type
3216 ** The compiler gets less confused by inlining and RETURN-FROM,
3217 thanks to some patches originally by Tim Moore.
3218 ** The system no longer hangs when dumping circular lists to fasl
3219 files, thanks to a patch originally from Douglas Crosher.
3220 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3221 undefined function error.
3222 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3223 in structure slot initforms) for now, documented workaround instead:-|
3224 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3225 * fixed bug 5: added stubs for various Gray stream functions called
3226 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3227 installed, at least appropriate type errors are generated
3228 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3229 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3230 consistently in DEFMETHOD forms.
3231 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3232 it doesn't seem to affect SBCL after all
3233 * The C runtime system now builds with better optimization and many
3234 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3236 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3238 * The system is now under CVS at SourceForge (instead of the
3239 CVS repository on my home machine).
3240 * The new signal handling code has been tweaked to treat register
3241 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3242 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3243 downstream, e.g. in debug-int.lisp, has implicit dependencies
3244 on the unsignedness of integer representation of machine words,
3245 and that caused the system to bomb out with infinite regress
3246 when trying to recover from type errors involving signed values,
3247 e.g. (BUTLAST '(1 2 3) -1).)
3248 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3249 (This was one of the bugs Peter Van Eynde reported back in July.)
3250 * The system now uses code inspired by Colin Walters' O(N)
3251 implementation of MAP (from the cmucl-imp@cons.org mailing
3252 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3253 inline the MAP operation, and there is more than one
3254 sequence argument to the MAP call (so that it can't just
3255 do ETYPECASE once and for all based on the type of the
3256 single sequence argument). (The old non-inline implementation
3257 of the general M-argument sequence-of-length-N case required
3258 O(M*N*N) time when any of the sequence arguments were LISTs.)
3259 * The QUIT :UNIX-CODE keyword argument has been renamed to
3260 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3261 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3263 * Raymond Wiker's patches to port dynamic loading from Linux to
3264 FreeBSD have been added.
3265 * The BUGS file is now more nearly up to date, thanks in large part
3266 to Martin Atzmueller's review of it.
3267 * The debugger now flushes standard output streams before it begins
3268 its output ("debugger invoked" and so forth).
3269 * The core version number and fasl file version number have both
3270 been incremented, because of incompatible changes in the layout
3272 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3273 used to not be called for a saved Lisp image.)
3274 * Martin Atzmueller's version of a patch to fix a compiler crash,
3275 as posted on sbcl-devel 13 September 2000, has been installed.
3276 * Instead of installing Martin Atzmueller's patch for the
3277 compiler transform for SUBSEQ, I deleted the compiler transform,
3278 and transforms for some similar consing operations.
3279 * A bug in signal handling which kept TRACE from working on OpenBSD
3281 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3282 away by constant folding
3283 * The system now defines its address space constants in one place
3284 (in the Lisp sources), and propagates them automatically elsewhere
3285 (through GENESIS and the sbcl.h file). Therefore, patching the
3286 address map is less unnecessarily tedious and error-prone. The
3287 Lisp names of address space constants have also been systematized.
3288 * CVS tags like dollar-Header-dollar have been removed from
3289 the sources, because they have never saved me trouble and
3290 they've been source of trouble working with patches and other
3291 diff-related operations.
3292 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3293 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3295 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3297 * The system has been ported to OpenBSD.
3298 * The system now compiles with a simple "sh make.sh" on the systems
3299 that it's supported on. I.e., now you no longer need to tweak
3300 text in the target-features.lisp-expr and symlinks in src/runtime/
3301 by hand, the make.sh takes care of it for you.
3302 * The system is no longer so grossly inefficient when compiling code
3303 involving vectors implemented as general (not simple) vectors (VECTOR T),
3304 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3305 which dares to use the various sequence functions on non-simple
3306 vectors, takes less of a performance hit.
3307 * There is now a primitive type predicate VECTOR-T-P
3308 to test for the (VECTOR T) type, so that e.g.
3309 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3310 can now be compiled with some semblance of efficiency. (The old code
3311 turned the type declaration into a full call to %TYPEP at runtime!)
3312 * AREF on (VECTOR T) is still not fast, since it's still compiled
3313 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3314 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3315 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3316 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3317 without making full calls to SUBTYPEP at runtime.
3318 (Some analogous efficiency issues for non-simple vectors specialized to
3319 element types other than T, or for non-simple multidimensional arrays,
3320 have not been addressed. They could almost certainly be handled the
3321 same way if anyone is motivated to do so.)
3322 * The changes in array handling break binary compatibility, so
3323 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3324 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3325 instead of (VALUES T T).
3326 * By following the instructions that Dan Barlow posted to sbcl-devel
3327 on 2 July 2000, I was able to enable primitive dynamic object
3328 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3329 functionality is not implemented (since it calls ld to resolve
3330 library references automatically, requiring RUN-PROGRAM for its
3331 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3332 try to resolve library references) is now supported.
3333 * The system now flushes the standard output streams when it terminates,
3334 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3335 them at several other probably-convenient times, e.g. in each pass of
3336 the toplevel read-eval-print loop, and after evaluating a form given
3337 as an "--eval" command-line option. (These changes were motivated by a
3338 discussion of stream flushing issues on cmucl-imp in August 2000.)
3339 * The source transform for TYPEP of array types no longer assumes
3340 that an array whose element type is a not-yet-defined type
3341 is implemented as an array of T, but instead punts, so that the
3342 type will be interpreted at runtime.
3343 * There is now some support for cross-compiling in make.sh: each of
3344 the phases of make.sh has its own script. (This should be transparent
3345 to people doing ordinary, non-cross-compile builds.)
3346 * Since my laptop doesn't have hundreds of megabytes of memory like
3347 my desktop machine, I became more motivated to do some items on
3348 my to-do list in order to reduce the size of the system a little:
3349 ** Arrange for various needed-only-at-cold-init things to be
3350 uninterned after cold init. To support this, those things have
3351 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3352 symbols with such names are now uninterned after cold init).
3353 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3354 fasl files for cold load.
3355 ** Remove the old compiler structure pooling code (which used to
3356 be conditional on the target feature :SB-ALLOC) completely.
3357 ** Redo the representation of some data in cold init to be more compact.
3358 (I also looked into supporting byte compiled code at bootstrap time,
3359 which would probably reduce the size of the system a lot, but that
3360 looked too complicated, so I punted for now.)
3361 * The maximum signal nesting depth in the src/runtime/ support code has
3362 been reduced from 4096 to 256. (I don't know any reason for the very
3363 large old value. If the new smaller value turns out to break something,
3364 I'll probably just bump it back up.)
3365 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3367 * Many, many bugs reported by Peter Van Eynde have been added to
3368 the BUGS list; some have even been fixed.
3369 * While enabling dynamic object file loading, I tried to make the
3370 code easier to understand, renaming various functions and variables
3371 with less ambiguous names, and changing some function calling
3372 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3373 * While trying to figure out how to do the OpenBSD port, I tried to
3374 clean up some of the code in src/runtime/. In particular, I dropped
3375 support for non-POSIX signal handling, added various comments,
3376 tweaked the code to reduce the number of compilation warnings, and
3377 renamed some files to increase consistency.
3378 * To support the new automatic configuration functionality in make.sh,
3379 the source file target-features.lisp-expr has been replaced with the
3380 source file base-target-features.lisp-expr and the machine-generated
3381 file local-target-features.lisp-expr.
3382 * fixed a stupid quoting error in make.sh so that using CMU CL
3383 "lisp -batch" as cross-compilation host works again
3385 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3387 * DESCRIBE no longer tries to call itself recursively to describe
3388 bound/fbound values, so that it no longer fails on symbols which are
3389 bound to themselves (like keywords, T, and NIL).
3390 * DESCRIBE now works on generic functions.
3391 * The printer now prints less-screwed-up representations of closures
3392 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3393 * A private symbol is used instead of the :EMPTY keyword previously
3394 used to mark empty slots in hash tables. Thus
3395 (DEFVAR *HT* (MAKE-HASH-TABLE))
3396 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3397 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3398 now does what ANSI says that it should. (You can still get
3399 similar noncompliant behavior if bang on the hash table
3400 implementation with all the symbols you get back from
3401 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3402 This breaks binary compatibility, since tests for equality to
3403 :EMPTY are wired into things like the macroexpansion of
3404 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3406 * There's now a minimal placeholder implementation for CL:STEP,
3407 as required by ANSI.
3408 * An obscure bug in the interaction of the normal compiler, the byte
3409 compiler, inlining, and structure predicates has been patched
3410 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3411 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3413 * Missing ordinary arguments in a macro call are now detected even
3414 when the macro lambda list contains &KEY or &REST.
3415 * The debugger no longer complains about encountering the top of the
3416 stack when you type "FRAME 0" to explicitly instruct it to go to
3417 the top of the stack. And it now prints the frame you request even
3418 if it's the current frame (instead of saying "You are here.").
3419 * As specified by ANSI, the system now always prints keywords
3420 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3422 * The default initial SIZE of HASH-TABLEs is now smaller.
3423 * Type information from CLOS class dispatch is now propagated
3424 into DEFMETHOD bodies, so that e.g.
3425 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3427 is now basically equivalent to
3428 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3429 (DECLARE (TYPE SINGLE-FLOAT X))
3431 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3432 operation, without having to do run-time type dispatch.
3433 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3434 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3435 * Since I don't seem to be making big file reorganizations very often
3436 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3437 is overflowing my ability to conveniently back them up), I've finally
3438 checked the system into CVS. (The CVS repository is on my home system,
3439 not at SourceForge -- putting it on SourceForge might come later.)
3440 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3441 high-level GC-NOTIFY-FOO functions send their output. (There's
3442 still very little control of where low-level verbose GC functions
3443 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3444 less than it used to -- the GC-NOTIFY-FOO functions are now under
3445 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3446 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3447 in only one place in the source code, and propagates it automatically
3448 everywhere that it's needed. Thus e.g. when I bump the version from
3449 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3450 * The C source files now include boilerplate legalese and documentation
3451 at the head of each file (just as the Lisp source files already did).
3452 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3453 to his page will be replaced with a link to his new CLiki service.
3455 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3457 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3458 * The build process now looks for GNU make under the default name "gmake",
3459 instead of "make" as it used to. If GNU make is not available as "gmake"
3460 on your system, you can change this default behavior by setting the
3461 GNUMAKE environment variable.
3462 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3463 can build without error under CMU CL.
3465 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3467 * There is now a partial SBCL user manual (with some new text and some
3468 text cribbed from the CMU CL manual).
3469 * The beginnings of a profiler have been added (starting with the
3470 CMU CL profiler and simplifying and cleaning up). Eventually the
3471 main interface should be through the TRACE macro, but for now,
3472 it's still accessed through vaguely CMU-CL-style functions and macros
3473 exported from the package SB-PROFILE.
3474 * Some problems left over from porting CMU CL to the new
3475 cross-compilation bootstrap process have been cleaned up:
3476 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3477 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3478 part when getting rid of the compiler *BACKEND* stuff.)
3479 ** The value of *NULL-TYPE* was screwed up, because it was
3480 being initialized before the type system knew the final
3481 definition of the 'NULL type. This screwed up several key
3482 optimizations in the compiler, causing inefficiency in all sorts
3483 of places. (I found it because I wanted to understand why
3484 GET-INTERNAL-RUN-TIME was consing.)
3485 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3486 PROCLAIM FTYPE information. Unfortunately this broke binary
3487 compatibility again, since now the forms output by DEFGENERIC
3488 to refer to functions which didn't exist in 0.6.3.
3489 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3490 can use the new (as of 0.6.3) transform for SOME into MAP into
3492 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3493 values to (INTEGER 0 1000000), so that the time code will no longer
3494 occasionally get blown up by Linux returning 1000000 microseconds
3495 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3496 its output conform to the ANSI spec. (Alas, this makes its output
3497 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3498 * A full call to MAP NIL with a single sequence argument no longer conses.
3499 * fixes to problems pointed out by Martin Atzmueller:
3500 * The manual page no longer talks about multiprocessing as though
3501 it were currently supported.
3502 * The ILISP support patches have been removed from the distribution,
3503 because as of version 5.10.1, ILISP now supports SBCL without us
3504 having to maintain patches.
3505 * added a modified version of Raymond Toy's recent CMU CL patch for
3506 EQUALP comparison of HASH-TABLE
3508 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3510 * The system still can't cross-compile itself with
3511 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3512 entails), but at least it can compile more of itself that way
3513 than it used to be able to, and various buglets which were uncovered
3514 by trying to cross-compile itself that way have now been fixed.
3515 * This release breaks binary compatibility again. This time
3516 at least I've incremented the FASL file format version to 2, so that the
3517 problem can be detected reliably instead of just causing weird errors.
3518 * various new style warnings:
3519 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3520 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3521 ** using the lexical binding of a variable named in the *FOO* style
3522 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3523 as specified by ANSI.
3524 * *RANDOM-STATE* is no longer automatically initialized from
3525 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3526 default behavior of the system is to repeat its behavior every time
3527 it's run. If you'd like to change this behavior, you can always
3528 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3529 old convention there was no comparably easy way to get the system to
3530 repeat its behavior every time it was run.
3531 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3532 FTYPE declarations has been removed, in favor of their ANSI
3533 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3534 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3535 the types of their sequence arguments can be determined at compile time.
3536 This is done through a new open code expansion for MAP which eliminates
3537 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3538 argument types can be determined at compile time.
3539 * The optimizer now transforms COERCE into an identity operation when it
3540 can prove that the coerced object is already of the correct type. (This
3541 can be a win for machine generated code, including the output of other
3542 optimization transforms, such as the MAP transform above.)
3543 * Credit information has been moved from source file headers into CREDITS.
3544 * Source file headers have been made more standard.
3545 * The CASE macro now compiles without complaining even when it has
3548 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3550 * (Note that the way that the PCL macroexpansions were rewritten
3551 to accommodate the change in DEFGENERIC below breaks binary
3552 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3553 not run under sbcl-0.6.2. Once we get out of alpha releases,
3554 i.e. hit release 1.0.0, we'll probably try to maintain binary
3555 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3556 and sbcl-1.4.4. Until then, however, it might be fairly common
3557 for maintenance releases to break binary compatibility.)
3558 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3560 * The reporting of unhandled signals has been changed to print some
3561 explanatory text as well as the report form. (Previously only
3562 the report form was printed.)
3563 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3564 it defines, so that the compiler no longer issues undefined function
3565 warnings for compiled-but-not-yet-loaded generic functions.
3566 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3567 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3568 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3569 section of the ANSI spec are supported.
3570 * The "" nickname for the "KEYWORD" package has been removed.
3571 The reader still handles symbol tokens which begin with a package marker
3572 as keywords, but it doesn't expose its mechanism for doing so in the
3573 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3574 * The system now issues STYLE-WARNINGs for contradictory TYPE
3575 proclamations. (Warnings for contradictory FTYPE proclamations would
3576 be nice too, but those can't be done usefully unless the type system
3577 is made smarter about FUNCTION types.)
3578 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3579 systematized, so that "*target-*.lisp is supposed to exist only on the
3580 target and imply that there's a related file which exists on the
3581 host, and *host-*.lisp is supposed to exist only on the host and imply
3582 that there's a related file which exists on the target. This involves a
3583 lot of renaming. Hopefully the acute confusion caused by the renaming
3584 will be justified by the reduction in chronic confusion..
3585 ** runtime-type.lisp -> early-target-type.lisp
3586 ** target-type.lisp -> late-target-type.lisp
3587 ** early-host-format.lisp -> early-format.lisp
3588 ** late-host-format.lisp -> late-format.lisp
3589 ** host-error.lisp -> misc-error.lisp
3590 ** early-error.lisp -> early-target-error.lisp
3591 ** late-error.lisp -> late-target-error.lisp
3592 ** host-defboot.lisp -> early-defboot.lisp
3593 ** code/misc.lisp -> code/target-misc.lisp
3594 ** code/host-misc.lisp -> code/misc.lisp
3595 ** code/numbers.lisp -> code/target-numbers.lisp
3596 ** code/early-numbers.lisp -> numbers.lisp
3597 ** early-host-type.lisp -> early-type.lisp
3598 ** late-host-type.lisp -> late-type.lisp
3599 ** host-typep.lisp -> typep.lisp
3600 ** load.lisp -> target-load.lisp
3601 ** host-load.lisp -> load.lisp
3602 ** host-disassem.lisp -> disassem.lisp
3603 ** host-insts.lisp -> insts.lisp
3604 ** byte-comp.lisp -> target-byte-comp.lisp
3605 ** host-byte-comp.lisp -> byte-comp.lisp
3606 ** host-signal.lisp -> signal.lisp
3607 ** host-defstruct.lisp -> defstruct.lisp
3608 ** late-target-type.lisp -> deftypes-for-target.lisp
3609 Furthermore, several other previously target-only files foo.lisp (e.g.
3610 hash-table.lisp and random.lisp) have been split into a target-and-host
3611 foo.lisp file and a target-only target-foo.lisp file, with their key type
3612 definitions in the target-and-host part, so that the cross-compiler will
3613 know more about target types.
3614 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3615 gone away. In their place are various *BACKEND-FOO* variables
3616 corresponding to the slots of the old structure.
3617 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3618 into the target SBCL has been fixed.
3619 * The chill.lisp system for loading cold code into a running SBCL
3621 * Support for the CMU CL "scavenger hook" extension has been removed.
3622 (It was undocumented and unused in the CMU CL sources that SBCL was
3623 derived from, and stale in sbcl-0.6.1.)
3624 * Various errors in the cross-compiler type system were detected
3625 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3626 (enabling various consistency checks). Many of them were fixed,
3627 but some hard problems remain, so the compiler is back to
3628 running without *TYPE-SYSTEM-INITIALIZED* for now.
3629 * As part of the cross-compiler type system cleanup, I implemented
3630 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3631 * I have started adding UNCROSS calls throughout the type system
3632 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3633 on cross-compiler input files will be able to go away, and various
3635 * CONSTANTP now returns true for quoted forms (as explicitly required
3638 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3640 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3641 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3642 (MAKE-LIST -1). In the longer term, it ought to become true
3643 that declarations are assertions even at SAFETY 1. For now, it's not
3644 quite true even at SAFETY 3, but it's at least more nearly true..
3645 (Note that this change seems to increases the size of the system by
3646 O(5%) and to decrease the speed of the compiler by 20% or more.)
3647 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3648 for the problem of printing dozens of lines of distracting information
3649 about low-level system machinery as part of the top stack frame
3650 on entry to the debugger when an undefined function was called.
3651 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3652 is not reset to COMMON-LISP-USER.
3653 * Compilation of stuff related to dyncount.lisp has been made conditional
3654 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3655 smaller. The various dyncount-related symbols have been moved into
3656 a new "SB-DYNCOUNT" package.
3657 * tty-inspect.lisp has been renamed to inspect.lisp.
3658 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3659 feature has gone away. (When we eventually port to other flavors of
3660 libc and/or Unix, we'll try to make the differences between flavors
3661 invisible at the user level.)
3662 * Various other *FEATURES* tags, and/or their associated conditionals,
3663 have been removed if obsolescent, or given better documentation, or
3664 sometimes given more-mnemonic names.
3666 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3668 * tidied up "make.sh" script
3669 * tidied up system directory structure
3670 * better "clean.sh" behavior
3671 * added doc/FOR-CMUCL-DEVELOPERS
3672 * many many small tweaks to output format, e.g. removing possibly-confusing
3673 trailing #\. character in DESCRIBE-INSTANCE
3674 * (EQUALP #\A 'A) no longer signals an error.
3675 * new hashing code, including EQUALP hashing
3676 * tidied up Lisp initialization and toplevel
3677 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3678 * command line argument processing
3679 * added POSIX-GETENV function to deal with Unix-ish environment variables
3680 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3681 terminating SBCL on EOF
3682 * non-verbose GC by default
3683 * There is no more "sbcl" shell script; the sbcl file is now the C
3684 runtime executable (just like CMU CL).
3685 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3687 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3688 symbol and package information as Lisp native symbol and package objects
3689 instead of strings naming symbols and strings naming packages. This way,
3690 whenever packages are renamed (as in warm init), debug information is
3691 transformed along with everything else.
3692 * tweaked the optimization policy declarations which control the building
3693 of SBCL itself. Now, among other things, the system no longer saves
3694 source location debugging information. (This helps two problems at once
3695 by reducing SBCL size and by keeping SBCL from trying to look for its
3696 sources -- which may not exist -- when reporting errors.)
3697 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3698 debugging and testing purposes
3699 * cleaned up printing, making the printer call PRINT-OBJECT for
3700 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3701 methods, giving nearly-ANSI behavior
3702 * converted almost all special variables to use *FOO* naming convention
3703 * deleted PARSE-TIME functionality, since it can be done portably
3704 * moved some files out of cold init into warm init
3705 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3707 * regularized formatting of source files
3708 * added an install.sh script
3709 * fixed ridiculous memory usage of cross-compiler by making
3710 compiler/alloc.lisp not try to do pooling unless it can hook
3711 itself into the GC of the cross-compilation host. Now the system
3712 builds nicely on my old laptop.
3713 * added :SB-ALLOC in target-features.lisp-expr
3714 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3715 was not implemented)
3716 * re-did condition handling and note reporting in the compiler. Notes
3717 are no longer handled by signalling conditions. Style warnings
3718 and warnings are handled more correctly and reported in such a way
3719 that it's easy to find one or the other in your output (so that you
3720 can e.g. figure out which of many problems caused COMPILE-FILE to
3722 * changed the severity of several compiler warnings from full WARNING
3723 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3724 compiler note reporting so that it doesn't use the condition system
3725 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3726 COMPILE-FILE command)
3727 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3728 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3729 no longer supported, since their implementation was deeply intertwingled
3730 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3731 * removed bogus "support" for compiler macros named (SETF FOO), and
3732 removed the compiler macro for SETF INFO (but only after making a fool
3733 of myself on the cmucl-imp mailing list by posting a bogus patch for
3734 DEFINE-COMPILER-MACRO..)
3735 * Compiled files containing forms which have side effects on the Lisp
3736 reader (such as DEFPACKAGE forms) are now handled more correctly.
3737 (Compiler queuing of top level lambdas has been suppressed by setting
3738 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3739 * deleted various currently-unused source files, e.g. gengc.lisp. They
3740 may be added back at some point e.g. when porting to other architectures,
3741 but until they are it's distracting to distribute them and to try to
3743 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3744 there were so many of them they're just distractions, and UNCROSS is
3745 known to be able to handle the current sources
3746 * moved PROFILE functionality into TRACE, so that it will be clear
3747 how the wrapping and unwrapping of functions when you profile them
3748 interacts with the wrapping and unwrapping of functions when you
3749 trace them. (Actually, the functionality isn't there yet, but at least
3750 the interface specification is there. Hopefully, the functionality will
3751 arrive with some future maintenance release.)
3752 * removed host-oops.lisp
3753 * changed signature of QUIT function to allow UNIX-CODE argument
3754 * fixed READ-SEQUENCE bug
3755 * tweaked verbose GC output so that it looks more like the progress
3756 output that ANSI specifies for functions like LOAD
3757 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3758 * added <http://sbcl.sourceforge.com> to the banner information printed by