1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
3 * added support for the ucs-2 external format (contributed by Ivan Boldyrev)
4 * minor incompatible change: pretty printing of objects of type
5 (cons symbol) is, in the default pprint-dispatch-table, now
6 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
8 * fixed bug: FILE-POSITION sometimes returned inconsistent results
9 for multibyte external-format streams (thanks to "vbzoli")
11 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
12 * feature: thread support on Solaris/x86, and experimental thread support
14 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
16 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
17 default method for SLOT-UNBOUND.
18 * minor incompatible change: prevent the user from specializing the
19 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
20 somewhat counter to the intent of the protocol, I (CSR) think, and
21 additionally it just doesn't work in SBCL as currently
22 implemented, thanks to optimizations (that are always valid for
23 the other three SLOT-VALUEish functions, but not for the setter).
24 * minor incompatible change: the :SB-LDB feature is now enabled by
25 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
26 the low-level debugger.
27 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
28 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
29 down to an SB-SYS:FD-STREAM, allowing direct communication in
30 more cases. Temporary files and pipes are still used as a fallback
32 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
33 on the non-reentrant C functions localtime(3) and gmtime(3).
34 * bug fix: global optimization policy was not visible in LOCALLY and
36 * bug fix: class objects can be used as specializers in methods.
37 (reported by Pascal Costanza)
38 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
39 with a type error. (reported by blitz_ on #lisp)
40 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
41 failed with a type error. (reported by Pascal Bourguignon)
42 * bug fix: merging pathnames against defaults with :DIRECTORY
43 starting with '(:RELATIVE :BACK) should preserve the :BACK.
44 (reported by James Y Knight)
45 * bug fix: saving large (>2GB) cores on x86-64 now works
46 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
47 constant index and a (simple-array (signed-byte 32)) array
48 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
49 error (patch by Robert J. Macomber)
50 * feature: TIME now displays the amount of run-time spent in GC
51 * bug fix: The debugger now does a better job of respecting
52 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
53 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
54 workaround for bug 403.)
55 * bug fix: floating point exception handling now works on FreeBSD
56 (thanks to NIIMI Satoshi)
57 * fixed some bugs revealed by Paul Dietz' test suite:
58 ** MISC.641: LET-conversion were not supposed to work in late
60 * improvements to the Win32/x86 port:
61 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
62 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
64 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
65 * new feature: source path information is generated for macro-expansion
66 errors for use in IDE's like Slime (thanks to Helmut Eller)
67 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
68 * bug fix: compiler does not loop forever on an invalid type in
70 * improvement: compilation of most CLOS applications is significantly
72 * optimization: added a limited bytecode compiler for simple toplevel
73 forms, speeding up compilation and FASL loading
75 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
76 * minor incompatible change: in sbcl-0.9.11 (but not earlier
77 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
78 into subclasses of the finalized class. Now user calls to
79 FINALIZE-INHERITANCE finalize just the one class, and calls by the
80 system return before any subclasses are finalized.
81 * minor incompatible change: The reader no longer ignores errors
82 regarding non-existent packages in #+ and #- feature tests.
83 * new feature: command line options --no-sysinit, --no-userinit to
84 inhibit loading the corresponding init files
85 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
86 for sending data through UDP sockets (thanks to François-René Rideau)
87 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
88 error is signalled instead (thanks to François-René Rideau)
89 * minor incompatible change: Error signalling behaviour of lexical
90 operations violating package locks has changed slightly. Refer to
91 documentation on package locks for details.
92 * enhancement: EVAL can process IF-expressions without resorting to the
94 * bug fix: Some operations involving SIN, COS, or TAN weren't being
95 constant-folded properly due to missing out-of-line functions.
96 (reported by Mika Pihlajamäki)
97 * bug fix: LISTEN sometimes returned T even in cases where no data was
98 immediately available from the stream
99 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
100 were reversed. (reported by Levente Mészáros)
101 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
102 when given a vector displaced to an adjustable array. (reported
104 * bug fix: derivation of float boundaries from numbers outside the
105 appropriate float range (reported by John Wiseman)
106 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
108 * bug fix: precomputing cache entries for generic functions with
109 some subclasses of specializers as yet invalid does not attempt to
110 fill a cache line with a negative offset. (reported by Levente
112 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
113 allow condition class objects as arguments to DOCUMENTATION and
114 (SETF DOCUMENTATION); only find and set documentation for
115 structure names for the STRUCTURE doc-type. (suggested by Gary
117 * improvements to the Win32/x86 port:
118 ** (user-homedir-pathname) and default initialization file
119 locations now know about the user's "Documents and Settings"
120 directory (thanks to Yaroslav Kavenchuk)
121 ** run-program is implemented (thanks to Mike Thomas)
122 ** sockets support (thanks to Timothy Ritchey)
123 ** better backtrace support (thanks to Alastair Bridgewater)
124 ** sb-grovel supported
125 ** asdf-install and sb-posix work somewhat
126 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
127 * improvements to the Solaris/x86 port:
128 ** works on Solaris 11/Solaris Express
129 ** floating-point exception handling support
130 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
131 * fixed some bugs revealed by Paul Dietz' test suite:
132 ** REMOVE-METHOD returns its generic function argument even when
133 the method is not one of the generic functions' methods.
134 ** objects declared as MEMBER types can be admissible arguments to
136 ** printing characters should simply be printed by the FORMAT ~:C
138 ** compiler failure when compiling functions with hairy constant
139 defaults for optional parameters.
140 ** compiler produces wrong code when MAYBE-INLINE-expanding a
141 function, which is already optimized.
143 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
144 * new platform: experimental support for SBCL x86/Darwin, including
145 MacOS X 10.4.5 on Intel.
146 * new feature: Unicode character names are now known to the system
147 (through CHAR-NAME and NAME-CHAR).
148 * new feature: the filesystem location of SBCL's core file is
149 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
150 * minor incompatible change: the contrib modules SB-POSIX and
151 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
152 this change is to make it easier to distribute
153 location-independent binaries.
154 * bug fix: as implied by AMOP, standardized classes no longer have
155 slots named by external symbols of public packages. (reported by
157 * optimization: calling functions via constant symbols -- as in
158 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
159 via the function object as in (FUNCALL #'FOO).
160 * optimization: on x86 and x86-64, the calling convention for the
161 case of dealing with an unknown number of values has been altered
162 to be friendlier to the prediction heuristics implemented,
163 particularly with reference to CALL and RET pairing. (thanks to
164 Alastair Bridgewater)
165 * optimization: on x86, the code for access to thread-local symbol
166 values and binding/unbinding of thread-local symbols is smaller.
167 (thanks to Alastair Bridgewater)
168 * enhancement: CONSTANTP is now able to determine constantness of
169 more complex forms, including calls to constant-foldable standardized
170 functions and some special forms beyond QUOTE.
171 * fixed bug: occasional GC crashes on Solaris/x86
172 * optimization: x86-64 supports stack allocation of results of simple
173 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
174 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
175 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
176 (thanks to James Knight)
177 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
178 by the standard (thanks to Stephen Compall)
180 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
181 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
182 be used for bundling the runtime and the core file into one
183 executable binary. This feature is not currently supported on all SBCL
184 platforms. (thanks to James Bielman and NIIMI Satoshi)
185 * new feature: a generational or ephemeral garbage collector is now
186 the default on the PowerPC platform (both Linux and Darwin). The
187 old Cheney (stop and copy) collector is a build-time option.
188 (thanks to Cyrus Harmon, after Raymond Toy)
189 * minor incompatible change: the method by which SBCL finds its
190 contributed modules has changed; it no longer relies on symbolic
191 links from an $SBCL_HOME/systems directory, but searches directly
192 in the subdirectories of $SBCL_HOME.
193 * enhancement: the dynamic heap size on the Linux/PPC platform is
194 markedly larger, even using the older Cheney garbage collector.
195 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
196 certain complicated string types. (reported by Gary King)
197 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
198 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
199 * fixed bug: DOCUMENTATION on structures no longer signals an
200 error if no structure type of the right name exists. (reported by
202 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
203 applicable non-standard primary or :AROUND method on
204 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
205 no longer cause unbound variable errors. (reported by Kevin Reid)
206 * fixed bug: metacircle resolution in cases where methods have slots
207 added before the slots from STANDARD-METHOD. (reported by Jean
209 * fixed bug: the Power PC assembly code for calling into lisp
210 sometimes computed its return address wrongly (depending on the
211 15th bit of the address, fixed at link time). This used to
212 manifest itself in a segmentation violation while building PCL.
213 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
214 many others over the years)
215 * fixed bug: ROOM no longer reports silly numbers for stack usage if
216 the stack pointer (treated as a signed integer) is negative.
217 (thanks to Peter van Eynde)
219 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
220 * new platform: experimental support for the Windows operating
221 system has been added. (thanks to Alastair Bridgewater)
222 * fixed several bugs in and robustified the PPC FFI (including
223 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
224 * bug fix: allow non-simple string symbol names (reported by Paul
226 * bug fix: interrupt handling on NetBSD (thanks to Richard M
228 * bug fix: saving a core corrupted callbacks on x86/x86-64
229 * bug fix: closed a loophole in metacircularity detection and
230 grounding in the PCL implementation of CLOS.
231 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
232 (reported by tomppa on #lisp)
233 * bug fix: clear the direction flag on Lisp -> C transitions, as
234 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
235 (reported by Andrej Grozin and Hendrik Maryns)
236 * optimization: major improvements to GC efficiency on GENCGC platforms
237 * optimization: faster implementation of EQUAL
238 * optimization: emit more efficient opcodes for some common
239 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
240 * optimization: several other minor code-generation improvements on x86-64
241 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
243 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
244 * minor incompatible change: (SETF CLASS-NAME) and (SETF
245 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
246 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
247 by ANSI as a generic function, and for consistency (SETF
248 GENERIC-FUNCTION-NAME) is treated likewise.
249 * fixed bug #233.b: make constraint propagation notice when a variable
250 value is changed after it is referenced but before it is used
251 * fixed bug #296: no more arbitrary behaviour from filenames with
252 odd characters as --load arguments. (reported by Adam Warner)
253 * fixed bug #390: :CHARACTER-SET pathname components now work as
254 expected. (reported by Tim Daly Jr)
255 * fixed bug #391: complicated :TYPE intersections in slot
256 definitions no longer cause an error in PCL internals.
257 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
258 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
259 returning the number of octets which would be written to the
260 file-stream. (thanks to Robert J. Macomber)
261 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
262 arguments to the internal stream functions. (thanks to David
264 * bug fix: stack exhaustion is now handled on the x86/SunOS
265 platform. (thanks to Daisuke Homma)
266 * bug fix: more accurate ROOM results on GENCGC platforms
267 * optimization: improved type inference for arithmetic-for
268 index variables in LOOP
269 * optimization: faster floating-point SQRT on x86-64
270 * fixed some bugs revealed by Paul Dietz' test suite:
271 ** DOCUMENTATION returns NIL instead of "" for method combinations
272 that don't have a docstring
274 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
275 * minor incompatible change: (SETF CLASS-NAME) and (SETF
276 GENERIC-FUNCTION-NAME) are no longer generic functions, and
277 therefore are not customizeable by user code (as seems to be at
278 least permitted and maybe required by AMOP). As a consolation,
279 however, the SBCL implementation of these functions now calls
280 REINITIALIZE-INSTANCE as specified by AMOP.
281 * bug fix: slot-definition documentation is propagated as per ANSI
282 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
283 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
284 argument, as do eslotd initialization methods. (from Pascal
285 Costanza's "Closer" project)
286 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
287 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
289 * bug fix: it is now possible to have more than one subclass of
290 STANDARD-GENERIC-FUNCTION without causing stack overflow.
291 (reported by Bruno Haible, Pascal Costanza and others)
292 * bug fix: the dependent update protocol now works for generic
293 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
295 * bug fix: condition-class instances corresponding to
296 DEFINE-CONDITION forms are now created eagerly. (reported by
297 Kalle Olavi Niemitalo on comp.lang.lisp)
298 * bug fix: floating point printing is more accurate in some
299 circumstances. (thanks to Simon Alexander)
300 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
301 merged with *DEFAULT-PATHNAME-DEFAULTS*.
302 * bug fix: callbacks on OS X now preserve stack-alignment.
303 (thanks to Cyrus Harmon)
304 * enhancement: the x86-64 disassembler is much better at
305 disassembling SSE instructions. (thanks to Lutz Euler)
306 * enhancement: CHECK-TYPE on a variable now allows type-inference
307 (thanks to Vincent Arkesteijn)
308 * optimization: improved performance of EUC-JP external format.
309 (thanks to NIIMI Satoshi)
310 * optimization: performance improvements to IO on file streams of
311 :ELEMENT-TYPE CHARACTER
312 * optimization: much faster memory allocation on x86-64
313 * optimization: faster garbage collections (latency and throughput)
315 * optimization: faster bignum subtraction and division on x86 and x86-64
316 (thanks to Lutz Euler)
317 * optimization: more accurate type-inference for loops that use a
318 floating point index variable or a negative step.
320 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
321 * bug fix: add a workaround to SBCL looping infinitely at startup on
322 Linux kernels with apparently buggy implementations of personality().
323 (thanks to Svein Ove Aas)
324 * bug fix: Unicode symbols are correctly printed in LDB backtraces
325 (thanks to David Lichteblau)
326 * bug fix: local bindings shadow global symbol macros for
327 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
328 * optimization: non-open coded uses of numeric comparison operators
329 (e.g. >) no longer cons when called with more than one parameter
330 on platforms supporting dynamic-extent allocation.
331 * enhancement: saving cores with foreign code loaded is now supported
332 on MIPS/Linux in addition to the previously supported platforms.
333 * bug fix: FIND-RESTART now tests for activity, not applicability when given
334 a restart object as identifier. (reported by Helmut Eller for CMUCL)
335 * bug fix: division by zero in sb-sprof when no samples were collected
336 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
337 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
338 (thanks to Alexander Kjeldaas)
339 * new feature: ignoring the return values of destructive functions
340 like NREVERSE signals a compile-time style-warning.
341 (thanks to Kevin Reid)
342 * new feature: the alignment of alien structure fields can be
343 explicitly specified. (thanks to Cyrus Harmon)
344 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
345 and dump core on SIGQUIT
347 ** incompatible change: threads do not inherit values of specials
348 from their parents (see manual)
349 ** bug fix: threads stacks belonging to dead threads are freed by the
350 next exiting thread, no need to gc to collect thread stacks anymore
351 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
352 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
353 ** bug fix: dynamic variable and thread start related gc lossage
354 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
356 * fixed some bugs revealed by Paul Dietz' test suite:
357 ** SUBTYPEP is slightly more accurate on heinously complicated
358 CONS types where some of the members have uncertain (in the
359 NIL, NIL sense) type relationships to each other.
361 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
362 no longer purified unless :PURIFY T is explicitly specified.
363 ** Non-purified cores are significantly smaller than before
365 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
366 * new feature: timers based on Zach Beane's excellent timer package
367 * added support for the following external formats: koi8-u,
368 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
369 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
370 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
371 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
372 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
373 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
374 * incompatible change: a threaded SBCL will no longer revert to
375 non-threaded mode on non-NPTL systems, but refuse to start entirely.
376 * bug fix: interrupts are disabled until startup is complete; no
377 more sigsegvs when receiving a signal to soon
378 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
380 * bug fix: add a workaround for the memory randomization features in
381 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
382 workaround will only be in effect on systems with the proc filesystem
384 * bug fix: printing objects of type HASH-TABLE signals a
385 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
387 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
388 have been running for over 50 days. (reported by Gilbert Baumann)
389 * bug fix: the logic for getting names of functions gets less
390 confused when confronded with alternate-metaclass
391 funcallable-instances. (reported by Cyrus Harmon)
392 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
393 consistent, even on internal alternate-metaclass objects.
394 * bug fix: flush closure information collected by physical
395 environment analysis prepass before the main pass. (bug reported
397 * bug fix: compiler pack phase does not modify a hashtable, which is
398 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
399 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
400 may be expanded. (reported by Alexander <malishev> on c.l.l)
401 * bug fix: no more style-warnings when compiling code using LAST
402 under high-SPEED low-SPACE settings. (reported by David Wallin
404 * fixed some incompatibilities between SBCL's MOP and the MOP
406 ** the METAOBJECT class is now implemented;
407 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
408 STANDARD-OBJECT, as required;
409 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
410 now compatible (as required by VALIDATE-SUPERCLASS); there
411 remains a consistency requirement over the presence or absence
412 of the FUNCTION class in the superclass list of finalized
413 classes; see the manual for more details;
414 ** the :AROUND method for COMPUTE-SLOTS on
415 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
416 requested slot ordering.
418 ** bug fix: parent thread now can be gc'ed even with a live
420 ** bug fix: binding a special with PROGV to no value is not
421 the same as that symbol not having been bound (thanks to
423 ** bug fix: binding specials is thread safe (thanks to
425 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
426 and SB-SPROF do not die with 'no handler for signal XX in
427 interrupt_handle_now(..)' anymore
428 ** bug fix: WITH-TIMEOUT works with multiple running threads
429 * fixed some bugs revealed by Paul Dietz' test suite:
430 ** ENSURE-GENERIC-FUNCTION should take a method class object for
431 the :method-class keyword argument.
433 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
434 * new port: the Solaris operating system on x86 processors is now
435 mostly supported, though some rough edges in the environment
436 remain. (thanks to Daisuke Homma)
437 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
438 space for its heap. (thanks to Thiemo Seufer)
439 * optimization: SBCL on MIPS platforms now supports dynamic-extent
440 closures. (thanks to Thiemo Seufer)
441 * minor incompatible change: eof selects abort in the debugger.
442 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
443 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
445 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
446 deprecated, and will go away in a future revision of SBCL.
447 * minor incompatible change: GC-ON and GC-OFF are no longer
448 implemented with a counter, it does not matter how many times gc
449 is switched on or off
450 * bug fix: discriminating functions for generic function classes
451 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
452 make invalid assumptions about method precedence order. (reported
454 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
455 platforms. (thanks to Thiemo Seufer)
456 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
457 * bug fix: correct pathname printing with printer escaping is on.
458 (thanks to Kevin Reid)
459 * bug fix: complex VOP definitions in "user-space" no longer trigger
460 package locks. (reported by Zach Beane)
461 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
462 for funcallable-instance functions, and is no different from
464 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
465 characters in the current readtable. (reported by Nicholas Neuss)
466 * bug fix: SUBTYPEP on various CONS types returns more a more
467 accurate acknowledgment of its certainty.
468 * optimizations: REMOVE-DUPLICATES now runs in linear time on
469 lists in some cases. This partially fixes bug 384.
470 * flush all standard streams before prompting in the REPL and the
472 * bug fix: signal handling and triggering gc do not conflict
473 directly anymore, in particular a high frequency sb-sprof does
474 not prevent gc from running
475 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
476 approximation for timezone and DST information between the
477 universal time epoch and the smallest negative 32-bit time_t.
478 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
479 year 1899 so long as the encoded time is non-negative.
480 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
481 the correct answer for non-constant rotate amounts when performing
482 an inline 32-bit rotation.
484 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
485 there is only one thread in the session
486 ** bug fix: memory leak for streams created in one thread and
487 written to in another
488 ** bug fix: lockup when compiled with gcc4
489 ** bug fix: race that allows the gc to be triggered when gc is
491 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
492 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
494 ** bug fix: finished threads are gc'ed properly
495 * fixed some bugs revealed by Paul Dietz' test suite:
496 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
497 has arguments with a different set of applicable methods from
498 the orignal arguments.
499 ** The type error thrown by MAP now has a correct expected-type
501 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
502 name a compiled function.
503 ** Three MISC tests where a large bignum was improperly coerced to
504 a float that couldn't represent that bignum during type
505 derivation were fixed.
506 ** SUBTYPEP can now handle types involving the negation of a
507 list-form FUNCTION type.
508 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
509 as (COMPLEX (AND RATIO (NOT FIXNUM))).
510 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
512 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
513 * New feature: Experimental support for bivalent streams: streams
514 opened with :element-type :default now allow character and binary
515 (unsigned-byte 8) I/O
516 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
517 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
519 * bug fix: no more highly sporadic "couldn't check whether ~S is
520 readable" when reading a stream and an interrupt hits in the middle
521 of a select system call
522 * compiler better recognizes complex arrays (reported by Eduardo
524 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
525 for variables on SBCL built with linkage-tables. (reported by Luis
527 * various error reporting improvements.
528 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
529 (Thanks to Thiemo Seufer)
530 * bug fix: nested reader invokations work correctly
531 * bug fix: it is possible to have simultaneous references to foreign
532 code and foreign data with the same name.
534 ** added x86-64 support
535 ** incompatible change: the threading api now works with thread
536 objects instead of thread ids
537 ** bug fix: threads are protected from signals and interruption when
538 starting up or going down
539 ** bug fix: a race where an exiting thread could lose its stack to gc
540 ** bug fix: don't halt on infinite error in threads if possible
541 ** fixed numerous gc deadlocks introduced in the pthread merge
542 ** bug fix: fixed thread safety issues in read and print
543 ** bug fix: debugger doesn't hang on session lock if interrupted at
544 an inappropriate moment
545 ** bug fix: run-program is now thread safe(r)
546 ** bug fix: inner with-recursive-lock no longer releases the mutex
547 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
548 * fixed some bugs revealed by Paul Dietz' test suite:
549 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
550 EXPECTED-TYPE slots filled.
551 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
553 ** MAKE-CONDITION accepts classes as type-designators.
554 ** COMPILE may never return NIL.
555 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
556 range before calling Unix time functions
558 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
559 * numerous signal handling fixes to increase stability
560 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
561 * minor incompatible change: we now correctly canonize default
562 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
563 as per the MOP, rather than the historical (INITARG INITFUNCTION
564 INITFORM). (reported by Bruno Haible)
565 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
566 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
567 for more information.
568 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
569 pathname is a directory pathname.
570 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
571 * dynamic space size on PPC has been increased to 768Mb. (thanks to
573 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
574 :METACLASS argument in addition to a class name. (reported by
575 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
576 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
577 (reported by Bruno Haible; patch from Gerd Moellmann)
578 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
580 * bug fix: more cleanups to the floating point exception handling on
581 x86-64 (thanks to James Knight)
582 * bug fix: the compiler does not try to constant fold calls of
583 COERCE and BIT-* functions when they return freshly constructed
584 objects. (reported by Nikodemus Siivola and Paul Dietz)
585 * optimization: improved the MIPS versions of generic (in the
586 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
587 * optimization: direct conversion of (unsigned-byte 32) to floats on
588 the PowerPC platform.
589 * optimization: structure instances with raw slots now use less
590 memory, and probably show better memory locality. (thanks to
592 * optimization: DYNAMIC-EXTENT declarations for lists and closures
593 are treated as requests for stack allocation on the x86-64,
594 Alpha-32, PPC and SPARC platforms.
595 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
596 the wrong file descriptor; implementation of SOCKET-OPEN-P.
597 (thanks to Tony Martinez)
599 ** gcing a dead thread can no longer lead to lockups
600 ** threads block signals until they are set up properly
601 ** errno is no longer shared by threads
602 ** interrupt-thread restores the eflags register on x86
603 ** fixed some lockups due to gc/thread interaction
604 * fixed some bugs revealed by Paul Dietz' test suite:
605 ** invalid dotted lists no longer raise a read error when
607 ** signal an error if a symbol that names a declaration is used
608 as the name of a type, or vice versa
609 ** allow using the (declare (typespec var*)) abbreviation for
610 (declare (type typespec var*)) with all type specifiers
611 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
612 ** always return NIL from PPRINT-POP when OBJECT is NIL
613 ** don't signal errors when pretty-printing malformed LABELS,
614 FLET or MACROLET forms
615 ** declarations in a DOLIST body are also in scope for the
617 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
619 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
622 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
623 * fixed cross-compiler leakages that prevented building a 32-bit
624 target with a 64-bit host compiler.
625 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
626 opened with :IF-EXISTS :OVERWRITE.
627 * fixed bug 281: error for an invalid qualifier in a short-form method
628 combination method is not signalled until the faulty method is called.
629 * bug fix: iteration variable type inferrer failed to deal with open
630 intervals. (reported by Alan Shields)
631 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
632 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
633 or more runtime options were provided to the sbcl binary.
634 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
635 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
636 * optimization: there's now a fast-path for fixnum arguments in the
637 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
638 * optimization: the code generated on x86-64 is more compact thanks
639 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
640 * bug fix: floating-point exception handling now partly works on
641 x86-64. (Thanks to James Knight)
642 * improvement to the MIPS backend from Thiemo Seufer: C-style
643 64-bit long long arguments and return values to alien functions
645 * fixed some bugs revealed by Paul Dietz' test suite:
646 ** the type-error signalled from WARN has a filled-in DATUM slot.
647 ** the type-error required when a stream is not associated with
648 a file has the stream as its datum.
649 ** type-errors on single-floats on x86-64 no longer have
650 :INVALID-OBJECT as the datum
651 ** the type-errors signalled for invalid function names now have
652 a correct expected type
653 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
654 for structures defined with a :TYPE. Documentation strings for
655 typed structures are no longer immediately discarded
656 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
657 broadcast streams and synonym streams. FILE-LENGTH now also works
658 on broadcast streams.
660 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
661 * incompatible change: the --noprogrammer option, deprecated since
662 version 0.7.5, has been removed. Please use the equivalent
663 --disable-debugger option instead.
664 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
666 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
667 inoperational for a while now) has been completely removed.
668 * null lexical environments are now printed as #<NULL-LEXENV>,
669 significantly reducing the amount of clutter in typical backtraces.
670 * documentation on weak pointers, finalization, and after GC hooks
671 has been added to the manual.
672 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
673 as other specialized array types, is much faster. SUBSEQ and
674 COPY-SEQ on such arrays have also been sped up.
675 * optimization: EQL is now more efficient when at least other argument
676 is known to be of type (OR FIXNUM (NOT NUMBER)).
677 * fixed bug: the runtime is now less vulnerable to changes in the
678 size of the SBCL object on OS X, and virtual memory is reserved for
679 all spaces that need to be at a fixed address.
680 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
681 * fixed bug: finalizers and after GC hooks that cause consing are now
683 * fixed bug: compiler error messages and summaries are now printed to
684 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
685 * fixed inference of the upper bound of an iteration variable.
686 (reported by Rajat Datta).
687 * fixed bug 211e: calling local functions with duplicated constant
688 keyword argument no longer causes a bogus style warning about an
690 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
691 ftype declaration to be disregarded. (reported by Dave Roberts)
692 * fixed bug 373: caused by erronous compilation of references to alien
693 variables in the runtime on ppc/darwin.
694 * fixed bug 376: CONJUGATE type deriver.
695 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
696 (reported by Baughn on #lisp)
697 * a cross-compiler bug on non-x86 platforms has been identified and
698 fixed. (thanks to Bruno Haible)
699 * improvements to the MIPS runtime code for increased stability.
700 (thanks to Thiemo Seufer)
701 * increased the maximimum compact environment size to allow
702 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
703 * partial workaround for bug 135: don't name the function we're
704 calling for hairy cases of EVAL, so as not to accumulate one environment
705 entry per call to EVAL. (reported by Kevin Reid)
706 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
707 * optimization: type testing for non-vector arrays should be faster.
708 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
710 * bug fix: setting 31st element of a bit vector to zero did not work
712 * bug fix: redefining a class definition which failed due to a
713 previous accessor / function clash now works (but see BUGS entry
714 #380 for more problems in this area). (thanks to Zach Beane)
715 * the long form of DEFINE-METHOD-COMBINATION disables method group
716 checking when given a single method group with pattern *. (thanks
718 * on x86 compiler supports stack allocation of results of simple
719 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
720 * support for building and running on Mac OS X 10.4 "Tiger" has been added
721 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
722 * fixed some bugs related to Unicode integration:
723 ** the restarts for recovering from input and output encoding
724 errors only appear when there is in fact such an error to
726 * fixed some bugs revealed by Paul Dietz' test suite:
727 ** MISC.549 and similar: late transformation of unsafe type
728 assertions into derived types caused unexpected code
730 ** SCALE-FLOAT type deriver is less wrong.
731 ** type derivers for EXP, LOG and similar functions compute result
732 types for complex arguments better.
733 ** (MISC.563) CONJUGATE type deriver works for very restricted
735 ** out-of-line type testers for character strings are available.
736 ** EQUAL compiler transform understands specialness of objects
738 ** accessing double-floats stored on the stack now works on x86-64.
739 ** debugger internals could sometimes create invalid lispobjs,
740 resulting in GC crashes.
741 ** MISC.548: type check weakening can convert required type into
743 ** initialization forms for bindings are not in scope of free special
746 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
747 * incompatible change: thread support for non-NPTL systems has
748 been removed - locking is buggy and unreliable. A threaded
749 SBCL build will now warn at startup and refuse to create
750 new threads, unless futex support is detected in the kernel
751 * incompatible change: the top level REPL now has only an ABORT
752 restart associated with it, not TOPLEVEL and ABORT as it used to.
753 TOP and TOPLEVEL are now available as debugger commands for
754 returning to the top level.
755 * incompatible change: forms evaluated in the REPL now use the
756 global optimization policy.
757 * incompatible change: user- and system-initialization files are
758 no longer processed with LOAD, but by READ and EVAL; hence the
759 global optimization policy, startup package, readtable, etc,
761 * The .fasl file format number has been incremented because of
762 various incompatible changes.
763 * internal entry point details and argument counts no longer appear
764 in backtraces unless explicitly requested by setting
765 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
766 * built-in and standard functions no longer have names like "top
767 level local call to FOO".
768 * fixed bug 32: functions defined in non-null lexical environments
769 now have more legible printed representation
770 * fixed bug 33: functions defined in non-null lexical environemnts
771 are now more amenable to inspection by INSPECT.
772 * workaround for bug 354: XEPs no longer appear in backtraces unless
773 explicitly requested.
774 * fixed bug: receiving the signal which results from attempting to
775 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
776 SIGBUS on other *BSDs) on architectures where the C stack is also
777 the Lisp stack (x86 and x86-64) from foreign code no longer leads
778 to debugger confusion or wild instability. (reported by Cheuksan
780 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
781 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
782 (reported by Lutz Euler)
783 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
784 compiled code. (reported by Paul Dietz)
785 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
786 specialized parameter to a method to something that is not TYPEP
787 the specializer is now possible.
788 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
789 face of package deletion.
790 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
791 pathname. (reported by David Tolpin on c.l.l)
792 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
793 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
794 STRING-STREAM and FILE-STREAM, respectively.
795 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
796 than previously; in addition, its test suite is now run on build.
797 (thanks to Luke Gorrie)
798 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
799 * optimization: added a immediate representation for single-floats
801 * fixed some bugs revealed by Paul Dietz' test suite:
802 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
803 ** attempting to create a package with a colliding nickname causes
804 correctable errors to be signalled.
805 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
806 ** several x86-64 backend bugs related to sign-extension of immediate
809 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
810 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
811 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
813 * fixed loading of multiply forward-referenced layouts.
814 (thanks to Cheuksan Wang)
815 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
816 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
817 (see COMPILE-FILE documentation for details of :PRINT option)
818 * fixed bugs 19 and 317: fixed-format floating point printing is
819 more accurate. This also fixes a bug reported by Adam Warner
820 related to the ~@F format directive.
821 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
823 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
824 dispatch table if the from-char is a dispatch macro character.
825 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
826 by the wordsize no longer ignore the last word. (reported by Lutz
828 * fixed bug in type checking of dynamic-extent variables. (reported
830 * optimization: sequence traversal functions use their freedom to
831 coerce function designators to functions.
832 * optimization: code with many calls to CLOS methods specialized on
833 CLOS classes has had redundant type checks removed.
834 * optimization: added declarations to speed up operations that access
835 the internal character database (for example STRING-UPCASE)
836 * optimization: comparison operations between floats and sufficiently small
837 fixnums no longer create extra rationals
838 * fixed some bugs related to Unicode integration:
839 ** portions of multibyte characters at the end of buffers for
840 character-based file input are correctly transferred to the
841 start of the buffer at the next read.
842 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
843 passing it through to OPEN.
844 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
845 argument given to internal calls to OPEN.
846 ** the built-in comment readers (introduced by character sequences
847 ";" and "#|") are more forgiving to encoding errors; they will
848 STYLE-WARN and then attempt to resync the stream at a character
849 boundary. (thanks to Teemu Kalvas)
850 * fixed some bugs in the x86-64 port:
851 ** Negative short int return values from c-calls are sign-extended
853 ** The stack is aligned to 16-bytes for c-calls, as required by
854 the ABI. (thanks to Cheuksan Wang)
855 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
856 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
857 for enums. (thanks to Vincent Arkesteijn)
858 ** Multiple small optimizations and bugfixes for floating point
860 * fixed some bugs revealed by Paul Dietz' test suite:
861 ** Space, Tab, Linefeed, Return and Page have the invalid
862 secondary constituent character trait.
863 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
865 ** WITH-INPUT-FROM-STRING should only update the index place on
867 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
869 ** Bit-array operations (BIT-AND and similar) worked incorrectly
870 with one-dimensional arrays with fill pointers.
871 ** TYPE-OF failed on a complex with an integer realpart and a
873 ** compiler failure during type inference for the code of form
874 (IF (EQL X (THE ...)) ...) (MISC.535).
876 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
877 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
878 platform. The port supports 61-bit fixnums, large memory spaces
879 and reloading shared object files.
880 * enhancement: saving cores with foreign code loaded is now
881 supported on ppc/Darwin in addition to the previously supported
883 * enhancement: the statistical profiler now walks deeper into the
884 call stack for more meaningful call-graphs and accrued time
885 reports (x86/x86-64 only). It also now reports time spent in
887 * enhancement: it is now possible to trace most individual methods
888 of a generic function in addition to tracing the generic function
890 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
892 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
893 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
894 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
895 produces an error. (thanks to Vincent Arkesteijn)
896 * bug fix: NAMESTRING on pathnames with :WILD components in their
897 directories works correctly. (thanks to Artem V. Andreev)
898 * fixed bug 125: compiler preserves identity of closures. (reported
900 * bug fixed: functions with &REST arguments sometimes failed with
901 "Undefined function" when compiled with (DEBUG 3). (reported by
903 * bug fix: overflow during compiling of setting element of a bit
904 vector with constant index and value. (reported by Timmy Douglas)
905 * build fix: fixed the dependence on *LOAD-PATHNAME* and
906 *COMPILE-FILE-PATHNAME* being absolute pathnames.
907 * on x86 compiler partially supports stack allocation of dynamic-extent
909 * GO and RETURN-FROM do not check the extent of their exit points
910 when compiled with SAFETY 0.
911 * fixed some bugs related to Unicode integration:
912 ** encoding and decoding errors are now much more robustly
913 handled; it should now be possible to recover even from invalid
914 input or output to the terminal. (thanks to Teemu Kalvas)
915 ** provided a first cut at implementing STRING-TO-OCTETS and
916 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
917 ** altered the SB-MD5 contributed module slightly, changing the
918 interface just enough for it to be supportable for builds where
919 lisp characters are not eight bits.
920 * fixed some bugs revealed by Paul Dietz' test suite:
921 ** the FORMATTER-generated functions for ~V[ conditionals require
922 the correct number of arguments.
923 ** READ-FROM-STRING returns the mandated second value when applied
924 to displaced strings.
925 ** the #\Rubout and #\Backspace characters are treated as invalid
926 constituent characters by the tokenizer.
928 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
929 * new feature: reloading changed shared object files with
930 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
931 * new feature: references to foreign variables and functions
932 can now be compiled and loaded before the corresponding shared
933 object file is loaded, as long as the foreign definitions are
934 available at runtime.
935 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
936 supported. (thanks to Dan Debertin)
937 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
938 just keywords. (thanks to Vincent Arkesteijn)
939 * fixed bug #331: structure-class instances corresponding to
940 DEFSTRUCT forms are now created eagerly.
941 * fixed bug #345: backtraces from calls to undefined functions work
942 on x86 as well. Related bug #61 is now also partially fixed on x86
943 (backtraces from throws to unknown catch tags.)
944 * bug fix: lambda-list parsing is now stricter vrt. order and number
945 of lambda-list keywords.
946 * bug fix: as specified by AMOP, an error is signalled if a
947 class-option appears multiple times in a DEFCLASS form. (reported
949 * bug fix: RANDOM can be compiled when the compiler derives the type
950 of its numeric argument as a disjoint set of small integers.
951 (reported by Paul Dietz)
952 * bug fix: starting a core saved with shared objects loaded when
953 those objects are not available no longer causes threaded SBCL to
954 hang. (reported by Sean Ross)
955 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
957 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
958 argument, which must always be NIL. (reported by Kalle Niemitalo)
959 * bug fix: printing 1.0d+23 no longer results in an error.
960 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
961 * bug fix: structure slot setters preserve evaluation order. (thanks
963 * fixed some bugs related to Unicode integration:
964 ** RUN-PROGRAM can allow its child to take input from a Lisp
965 stream. (reported by Stefan Scholl)
966 ** ASDF-INSTALL successfully downloads .tar.gz files in all
967 locales. (reported by Ken Causey)
968 * fixed some bugs revealed by Paul Dietz' test suite:
969 ** INCF, DECF and REMF evaluate their place form as specified in
971 ** FORMATTER expands ~{ ~} iteration directives with V or #
972 parameters correctly.
973 ** FORMATTER deals with the ~@[ ~] conditional directive where the
974 consequent uses no arguments correctly.
975 ** the system has a partial understanding of the (COMPLEX RATIO)
977 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
980 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
981 * new feature: a build-time option (controlled by the :SB-UNICODE
982 keyword feature, enabled by default) for building the system with
983 support for the entire 21-bit character space defined by the
985 * new feature: the system now has rudimentary external-format
986 support; the primary user-visible change at this time is that
987 characters with the high bit set (such as lower-case-e-acute) will
988 print correctly to a terminal in a UTF-8 environment.
989 * minor incompatible change: BASE-CHAR no longer names a class;
990 however, CHARACTER continues to do so, as required by ANSI.
991 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
992 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
993 has been moved to the SB-EXT package (temporarily re-exported from
995 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
996 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
998 * fixed bug #351: better error-handlind and reporting for malformed
1000 * fixed bug #350: bignum-printing is now more memory-efficient,
1001 allowing printing of very large bignums, eg. (expt 2 10000000).
1002 (reported by Bruno Haible)
1003 * fixed bug #302: better primitive-type selection for intersection
1005 * fixed bug #308: non-graphic characters now all have names, as
1006 required. (reported by Bruno Haible)
1007 * bug fix: redefining a class with different superclasses now correctly
1008 removes it from the direct-subclasses of its previous superclasses.
1009 (reported by David Morse)
1010 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1011 new value now works. (reported by Bruno Haible)
1012 * bug fix: correct canonicalization of multiple non-standard slot
1013 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1014 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1015 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1016 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1017 now exists, an signals an error.
1018 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1019 during macroexpansion for non-iteration variables as well. (reported
1020 by Bruno Haible for CMUCL)
1021 * bug fix: Cyclic structures and unprintable objects in compiler
1022 messages no longer cause errors. (reported by Bruno Haible)
1023 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1024 and READ-FROM-STRING all now return a primary value of NIL if
1025 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1026 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1027 (reported by Bruno Haible for CMUCL)
1028 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1029 do not designate a condition. (reported by Bruno Haible for
1031 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1032 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1033 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1034 permitting resolution in favour of any of the conflicting symbols.
1035 (reported by Bruno Haible for CMUCL)
1036 * bug fix: EQUAL compiler optimizations is less aggressive on
1037 strings which can potentially compare true despite having distinct
1038 specialized array element types.
1039 * bug fix: unit enumerations can be defined without dividing by
1040 zero. (thanks to Vincent Arkesteijn)
1041 * FORMAT compile-time argument count checking has been enhanced.
1042 (report from Bruno Haible for CMUCL)
1043 * a partial workaround for the bug 262: the compiler does not try to
1044 inline-expand a local function doing RETURN-FROM from a deleted
1045 BLOCK. (thanks to Peter Denno for the bug report and to David
1046 Wragg for the simple test case)
1047 * fixed some bugs revealed by Paul Dietz' test suite:
1048 ** RENAME-PACKAGE allows all package designators as new package
1050 ** constraint propagation and lambda variable substitution are
1051 more cautious in dealing with partially deleted code.
1052 ** compiler sometimes lost reoptimization passes.
1053 ** CERROR, given a condition as condition designator, passes its
1054 remaining arguments to the continue format control without
1056 ** Case-altering FORMAT directives work correctly on non-ASCII
1058 ** The REAL type specifier handles bounds outside the floating
1059 point ranges without signalling FLOATING-POINT-OVERFLOW.
1060 ** Functions with IR1-transformations can create intercomponent
1061 references to global functions.
1062 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1064 ** FORMAT ~R treats a nil value for its first parameter correctly.
1066 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1067 * enhancement: saving cores with foreign code loaded is now
1068 supported on x86/NetBSD and sparc/Linux in addition to the previously
1069 supported platforms.
1070 * bug fix: on some platforms repeated installations caused multiple
1071 copies of HTML documentation to be installed -- should not happen
1072 any more. (reported by Stefan Scholl)
1073 * bug fix: parsing self-recursive alien record types multiple times
1074 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1075 original patch by Helmut Eller for CMUCL)
1076 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1077 (thanks to Richard Kreuter)
1078 * bug fix: defining classes whose accessors are methods on existing
1079 generic functions in other (locked) packages no longer signals
1080 bogus package lock violations. (reported by François-René Rideau)
1081 * bug fix: special variables as DEFMETHOD parameters no longer have
1082 associated bogus type declarations. (reported by David Wragg and
1084 * bug fix: read-write consistency on streams of element-type
1085 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1086 * bug fix: redefiniton of the only method of a generic function with
1087 no DEFGENERIC no longer emits a full WARNING. In addition,
1088 redefinition of generic functions with no DEFGENERIC to an
1089 incompatible lambda list now signals an error. (thanks to Zach
1091 * bug fix: DEFGENERIC now works even when there's a function of the
1092 same name in an enclosing lexical environment. (thanks to Zach
1094 * fixed compiler failure, caused by instrumenting code during
1095 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1096 * optimization: added loop analysis and improved register allocation
1097 * fixed some bugs revealed by Paul Dietz' test suite:
1098 ** POSITION on displaced vectors with non-zero displacement
1099 returns the right answer.
1100 ** (SIMPLE-STRING) is a valid type specifier for sequence
1102 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1104 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1105 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1107 ** The justification version of the FORMAT ~< directive treats
1108 non-zero minpad parameter correctly.
1110 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1111 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1112 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1113 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1114 the supported interface.
1115 * new feature: Single-stepping of code compiled with DEBUG 2 or
1116 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1117 * new feature: saving cores with foreign code loaded is now
1118 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1119 Timothy Moore's work for CMUCL)
1120 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1121 parameters to * instead of NIL if no initform is supplied.
1122 (reported by Johan Bockgaard)
1123 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1124 correctly when FUNCALL appears as the car of the form. Note:
1125 despite this FUNCALL forms are not currently subject to
1126 compiler-macro expansion. (port of Raymond Toy's fix for the
1127 same from CMUCL, reported by Johan Bockgaard)
1128 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1129 (thanks for Teemu Kalvas)
1130 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1131 case instead of terminating the iteration (thanks for Julian
1132 Squires, Sean Champ and Raymond Toy)
1133 * bug fix: incorrect expansion of defgeneric that caused a style
1134 warning. (thanks for Zach Beane)
1135 * bug fix: slot accessor effective method computation works properly
1136 for classes with multiple non-standard applicable methods on
1137 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1138 * on x86 compiler supports stack allocation of results of LIST and
1139 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1140 CMUCL implementation by Gerd Moellmann)
1141 * fixed some bugs revealed by Paul Dietz' test suite:
1142 ** FORMAT strings with both the ~<~:;~> form of the justification
1143 directive and pretty-printing directives cause an error.
1145 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1146 * incompatible change: the internal functions
1147 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1148 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1149 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1150 instead of the old functions.
1151 * new feature: on platforms where "dladdr" is available foreign
1152 function names now appear in backtraces. (based on Helmut Eller's
1154 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1155 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1157 * documentation: ASDF manual is now distributed with SBCL.
1158 * bug fix: SBCL can now load its contributed modules with REQUIRE
1159 even if the system-provided entries have been removed from
1160 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1162 * bug fix: backtraces involving undefined functions or assembly
1163 routines are more informative. (thanks to Brian Downing)
1164 * bug fix: mutually referent alien structures now work correctly.
1165 (reported by Rick Taube)
1166 * bug fix: structures defined by WITH-ALIEN can be referred to
1167 within other definitions in the same WITH-ALIEN.
1168 * bug fix: division operators (MOD, TRUNCATE and the like) with
1169 constant zero divisors and integer dividends no longer generate
1171 * bug fix: provide default methods for INPUT-STREAM-P and
1172 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1173 * bug fix: improve the detection and resolution of MOP metacycles
1174 (where effective method computation depends on the generic function
1175 having its effective method computed). (reported by Bruno Haible)
1176 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1177 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1178 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1179 * bug fix: pathnames with relative directory components can be
1180 represented relative to default pathnames.
1181 * optimization: in taking the GCD of bignums, reduce the two bignums
1182 to approximately the same size (using Euclid's algorithm) before
1183 applying the more sophisticated binary GCD. (thanks to Juho
1185 * optimization: COUNT on bitvectors now operates word-at-a-time.
1186 * optimization: ASH with a positive, but not necessarily constant,
1187 (leftwards) shift, when in a modular context, is compiled to a
1189 * fixed some bugs revealed by Paul Dietz' test suite:
1190 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1191 if the corresponding argument is NIL.
1192 ** FORMAT directives accepting a minpad parameter treat negative
1194 ** literal commas as character parameters to FORMAT directives are
1196 ** literal spaces directly after ~<Newline> directives within a
1197 format-logical-block (~:< ~@:>) do not induce :FILL-style
1198 conditional newlines.
1199 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1200 as specified: it no longer includes conditional newlines).
1201 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1203 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1204 :PER-LINE-PREFIX argument does not evaluate to a string.
1206 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1207 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1208 the manual for details; for now, package locks can be disabled by
1209 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1210 if no major problems are found then it is likely that they will be
1211 compiled in unconditionally.
1212 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1213 unsupported operators on all platforms. To load a shared library
1214 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1215 object file, link it into a shared library outside of SBCL and
1216 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1218 * bug workaround: ROOM T and the GC were not working together
1219 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1220 reporting facility aren't preserved. That reporting has been
1221 disabled, so now until and unless someone figures out how to make it
1222 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1223 * minor incompatible change: as threatened around sbcl-0.8.0, the
1224 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1225 The symbols are also exported from SB-PCL for backwards
1226 compatibility, but more so than before SB-PCL should be treated as
1227 an implementation-internal package.
1228 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1230 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1231 It is known as of this release not to work on the Alpha, however.
1232 * fixed bug #167: errors signalled due to illegal syntax in method
1233 bodies are now more legible.
1234 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1235 designators and can hence be used with TYPEP.
1236 * fixed bug #333: CHECK-TYPE now ensures that the type error
1237 signalled, if any, has the right object to be accessed by
1238 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1239 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1240 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1242 * fixed bug #334: programmatic addition of slots using specialized
1243 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1244 and :ALLOCATION :CLASS slots.
1245 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1246 float exponents. (rereported by Peter Seibel)
1247 * fixed a bug: #\Space (and other whitespace characters) are no
1248 longer considered to be macro characters in standard syntax by
1249 GET-MACRO-CHARACTER.
1250 * fixed bug: initialization of condition class metaobjects no longer
1251 causes an instance of the condition to be created. (reported by
1253 * fixed bug: it is now possible to have slots such that
1254 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1255 description is neither :INSTANCE nor :CLASS.
1256 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1257 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1258 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1259 non-standard SLOT-DEFINITION-ALLOCATION values.
1260 * fixed bugs: various race conditions handling exiting threads.
1261 CL-PPCRE's thread torture test now passes.
1262 * fixed arguably-a-bug: GC time proportional to number of threads in
1263 system even when most of them are idle
1264 * optimization: improved performance of BIT and SBIT on bit-vectors.
1265 * .fasl file incompatibility: The fasl file version number has
1266 been incremented because of changes associated with package locks.
1268 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1269 * minor incompatible change: the system no longer provides
1270 optimization hints (conditions which are TYPEP
1271 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1272 mode; these hints are still emitted when higher SPEED optimization
1274 * new contrib module: a sampling profiler (profiling by statistical
1275 sampling, rather than by instrumenting functions) is available as
1276 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1277 * the behaviour of the standard function ED is now customizeable by
1278 third parties through a hook variable: see ED's documentation
1279 string for information on the protocol.
1280 * the compiler no longer emits efficiency notes for (FUNCALL X)
1281 when the type of X is uncertain under default optimization
1283 * fixed bug 276: mutating a binding of a specialized parameter to a
1284 method to something that is not TYPEP the specializer is now
1286 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1287 exhibit the required behaviour on the x86 platform. (thanks to
1288 Peter van Eynde, Eric Marsden and Bruno Haible)
1289 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1290 even for difficult arguments. (reported by Peter Graves)
1291 * fixed bug 141a: the backquote printer now descends quoted
1293 * fixed another bug in backquote printing: no more destructive
1294 modification of the form's list structure. (reported by Brian
1296 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1297 move between its address being taken and the call to
1298 interrupt_thread, fixing a crashing race condition.
1299 * the SB-POSIX contrib implementation has been adjusted so that it
1300 no longer exhibits ridiculously poor performance when constructing
1301 instances corresponding to C structs.
1303 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1304 * minor incompatible change: the sb-grovel contrib now treats C
1305 structures as alien (in the sense of SB-ALIEN) objects rather than
1306 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1307 has implications for memory management of client code
1308 (sb-grovel-returned objects must now be manually managed) and for
1309 type safety (alien objects now have full types).
1310 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1311 used to control emission of compiler diagnostics, rather than the
1312 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1313 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1314 quality should be considered deprecated.
1315 * install.sh now installs the user manual as well
1316 * (not quite a new documentable feature, but worth considering in
1317 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1318 beginnings of a semantically meaningful condition hierarchy is
1319 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1321 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1322 designator as the defaults argument.
1323 * fixed bug: Displaced arrays whose displaced-to array has become
1324 too small now cause ARRAY-DIMENSION to signal an error, providing
1325 for safer bounds-checking. (reported by Bruno Haible)
1326 * fixed bug: DEFCLASS slot definitions with identical :READER and
1327 :WRITER names now signal a reasonable error. (reported by Thomas
1329 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1331 * fixed bug: Invalid slot specification errors now print correctly.
1332 (thanks to Zach Beane)
1333 * fixed bug 320: Shared to local slot value transfers in class
1334 redefinitions now happen corrently from superclasses as
1335 well. (reported by Bruno Haible)
1336 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1338 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1339 improper lists correctly. (reported by Bruno Haible)
1340 * fixed bug 313: source-transform for <fun-name> was erroneously
1341 applied to a call of a value of a variable with name <fun-name>.
1342 (reported by Antonio Menezes Leitao)
1343 * fixed bug 307: The obsolete instance protocol ensures that
1344 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1345 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1346 remove dead unknown-values globs from the middle of the stack.
1347 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1348 easier to resume long computations after using *BREAK-ON-SIGNALS*
1349 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1350 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1351 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1352 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1353 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1354 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1356 * fixed a (fairly theoretical) bug in string printing: if
1357 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1358 does not have array-element-type equal to the most general string
1360 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1361 function name. (thanks to Nikodemus Siivola)
1362 * fixed bug: compile-time format string checker failed on
1363 non-closed ~{. (reported by Thomas F Burdick)
1364 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1365 don't warn on legal loop code involving more than one aggregate
1366 boolean. (thanks to Nikodemus Siivola)
1367 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1368 in keyword position to MAKE-PACKAGE was not being reported as
1370 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1371 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1373 * fixed bugs 280 and 312: the checking for multiple definitions in a
1374 file is less likely to become confused by uses of inline
1376 * fixed bug: the #S reader macro performs the keyword coercion
1377 specified for slot names. (reported by Kalle Niemitalo)
1378 * fixed bug: lambda lists may contain symbols whose names start with
1379 & but are not lambda-list-keywords; their occurrence triggers a
1381 * fixed bug 321: define-method-combination argument lambda lists do
1382 not cause invalid code to be generated when &optional or &aux
1383 variables have default values. (reported by Bruno Haible)
1384 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1385 have CLOS classes; this appears also to have fixed reported
1386 instability in stack exhaustion detection.
1387 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1388 MUFFLE-WARNING when no associated restart is present is now
1390 * optimization: rearranged the expansion of various defining macros
1391 so that each expands into only one top-level form in a
1392 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1394 * optimization: used a previously little-used slot in symbols to
1395 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1397 * fixed some bugs revealed by Paul Dietz' test suite:
1398 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1400 ** Fixed an optimization bug related to inheritance of initforms
1401 from local to shared slots.
1402 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1403 ** CAST optimizer forgot to flush argument derived type.
1404 ** print/read consistency on floats is now orders of magnitude
1405 more likely. (thanks also to Bruno Haible for a similar report
1407 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1408 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1409 for objects of type REAL. Make it so.
1410 ** SXHASH is specified (infelicitously) to respect similarity,
1411 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1412 it so. (thanks to Markus Ziegler)
1413 ** on the Alpha, the compiler succeeds in compiling functions
1414 returning a known number of arguments greater than 63.
1415 ** fixed handling of invalid NIL arguments in keyword position
1417 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1419 ** provided a readably-printable representation for RANDOM-STATE
1421 ** ensured that pathnames, where they have a namestring, always
1422 print using #P"..." syntax.
1424 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1425 * Support for the forthcoming 2.0 version of the NetBSD kernel
1426 running on x86 hardware has been added. (thanks to Perry
1427 E. Metzger most immediately, and others for their past work)
1428 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1429 current release version; SBCL's OpenBSD support had been broken
1430 since about the time of OpenBSD's switch to ELF binary format.)
1431 * [placeholder for DX summary]
1432 ** user code with &REST lists declared dynamic-extent, under high
1433 speed or space and low safety and debug optimization policy.
1434 * The manual has been converted to Texinfo format and the debugger
1435 chapter from the cmucl manual has been added.
1436 * A facility has been added to extract documentation strings from
1437 sbcl and store them as Texinfo-formatted snippets for inclusion in
1438 the manual (via Texinfo's @include directive)
1439 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1440 the test case to Dave Roberts)
1441 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1442 pointers. (reported by Sean Ross)
1443 * bug fix: PROFILE output is printed nicely even for large numerical
1444 values. (thanks to Zach Beane)
1445 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1446 greater than 32 handle EOF correctly.
1447 * bug fix: on X86 an immediate argument of the IMUL instruction is
1448 correctly printed in disassembly. (thanks to Lutz Euler)
1449 * bug fix: class slots in redefined classes preserve their old
1450 values. (thanks to Bruno Haible and Nikodemus Siivola)
1451 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1452 succeeds. (reported by Marco Baringer)
1453 * fixed some bugs revealed by Paul Dietz' test suite:
1454 ** READ-SEQUENCE now works on ECHO-STREAMs.
1455 ** RATIONALIZE works more according to its specification. (thanks
1458 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1459 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1460 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1461 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1462 should matter to you only if you rebind the printer control
1463 variables and then find you want different bindings in the
1464 debugger than in the ordinary execution of your program.)
1465 * The runtime build system has been tweaked to support building
1466 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1467 assembler and linker. (thanks to Nikodemus Siivola)
1468 * Unbound, undefined, undeclared variables now trigger full
1469 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1470 more often programmer error than deliberate exploitation of undefined
1472 * optimization: the hash algorithm for strings has changed to one
1473 that is less vulnerable to spurious collisions. (thanks to Juho
1475 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1476 needless bounds checking. (thanks to Juho Snellman)
1477 * optimization: implemented multiplication as a modular
1478 (UNSIGNED-BYTE 32) operation on the PPC backend.
1479 * fixed some bugs revealed by Paul Dietz' test suite:
1480 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1482 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1483 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1484 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1486 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1488 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1490 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1491 their output stream on EOF from read.
1492 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1493 have been read to end-of-file.
1494 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1496 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1497 description of determination of which consecutive characters
1499 ** Printing the "Space" character with escaping on now yields "#\\ ",
1500 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1501 ** Reading floating-point numbers with *READ-BASE* set to a number
1502 less than 10 works correctly.
1503 ** Reading floating-point numbers with *READ-BASE* set to a number
1504 more than 10 works correctly.
1505 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1506 the readtable currently in effect.
1508 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1509 * minor incompatible change: parsing of namestrings on a physical
1510 (Unix) host has changed; numbers after the final #\. in a
1511 namestring are no longer interpreted as a version field. This is
1512 intented to be largely invisible to the user, except that the
1513 meaning of the namestring "*.*.*" has changed: it now refers to a
1514 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1515 should usually be replaced by
1516 (make-pathname :name :wild :type :wild :version :wild)
1517 with the added benefit that this is more likely to be portable.
1518 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1519 option to OPEN now signals an error if the file being opened
1520 exists; this may have an impact on existing code.
1521 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1522 signals received, on the PowerPC platforms (both Linux and
1523 Darwin). (thanks to Pierre Mai for pointing out the location of
1525 * several fixes on OS X: The system now builds and runs cleanly on
1526 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1527 (thanks to Brian Mastenbrook)
1528 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1529 second-resolution: integer multiples of 1/3600 between -24 and 24.
1530 (thanks to Vincent Arkesteijn)
1531 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1532 type when called with 1 argument; PEEK-CHAR checked type of
1533 PEEK-TYPE only after having read first character from a
1534 stream. (reported by Peter Graves)
1535 * bug fix: the garbage collector now has much better locality
1536 behaviour, and in particular no longer treats all memory as being
1537 exhausted when many small objects point to each other in a deeply
1539 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1540 recognized as being TYPEP their class.
1541 * bug fix: the PUSHNEW documentation string has been corrected.
1542 (thanks to Vincent Arkesteijn)
1543 * bug fix: defaulting of the value for the last of an atypically
1544 large number of multiple values being bound was not being
1545 performed correctly on the Alpha or PPC platforms
1546 * optimization: implemented multiplication as a modular
1547 (UNSIGNED-BYTE 32) operation on the x86 backend.
1548 * optimization: SEARCH on simple-base-strings can now be open-coded.
1549 (see also contrib/compiler-extras.lisp for inspiration for
1550 teaching the compiler about the Boyer-Moore algorithm).
1551 * value, returned by MAX (and MIN) called with several EQUALP, but
1552 not EQL, arguments now does not depend on compiler settings.
1553 * fixed some bugs revealed by Paul Dietz' test suite:
1554 ** in stack analysis liveness information is propagated from
1555 non-local entry points.
1556 ** pathwise CAST removing failed when the CAST node did not start
1558 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1559 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1561 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1562 host is already defined.
1563 ** RENAME-FILE works on streams instead of signalling an internal
1565 ** PEEK-CHAR uses the current readtable when determining whether
1566 or not a character is whitespace.
1567 ** MERGE-PATHNAMES handles the case when the pathname does not
1568 specify a name while the default-pathname specifies a version.
1569 ** Pathnames now stand a better chance of respecting print/read
1571 ** Attempting to use standardized file system operators with a
1572 pathname with invalid :DIRECTORY components signals a
1574 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1576 ** DIRECTORY on logical pathnames is more correct.
1577 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1578 signal a TYPE-ERROR if their argument is not a stream.
1579 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1580 designator argument does not designate a stream.
1581 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1582 examining the synonym.
1583 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1585 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1586 element-type larger than ([UN]SIGNED-BYTE 32).
1588 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1589 * When built with the :SB-FUTEX feature, threaded builds now take
1590 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1591 for faster/more reliable mutex and condition variable support.
1592 * Incompatible change (but one you probably shouldn't have been using
1593 anyway): the interface and code for arbitrating between multiple
1594 threads in the same user session has been redesigned.
1595 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1596 error when called without an explicit environment argument.
1597 (thanks to Nikodemus Siivola)
1598 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1599 confused about their position. (thanks to Adam Warner and Gerd
1601 * bug fix: Pretty printing backquoted forms with unquotations in the
1602 argument list position of various code constructs such as LAMBDA
1603 now works correctly. (reported by Paul Dietz)
1604 * bug fix: Pretty printing unquotations no longer loses all
1605 stream position information.
1606 * optimization: performance of string output streams is now less
1607 poor for multiple small sequence writes.
1608 * optimization: performance of CSUBTYPEP in the presence of complex
1609 expressions involving CONS and NOT many times has been improved.
1610 (reported by Paul Dietz)
1611 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1613 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1614 streams for the REPL, for the debugger, and so on. (thanks to
1616 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1617 * fixed some bugs revealed by Paul Dietz' test suite:
1618 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1619 is ignored if it is too small, rather than propagating through
1620 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1621 ** extremely complex negations of CONS types were not being
1622 sufficiently canonicalized, leading to inconsistencies in
1624 ** VALUES tranformer lost derived type.
1626 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1627 * fixed a bootstrapping bug: the build process no longer assumes
1628 that the various BOOLE-related constants have the same value in
1629 host and target lisps. (noted by Paul Dietz' test suite on an
1630 SBCL binary built from CLISP)
1631 * The system can now be dynamically linked on the MIPS platform,
1632 which enables dynamic loading of foreign code from Lisp. (thanks
1633 to Ralf Baechle for discussions on the MIPS ABI)
1634 * The system now records debugging information for its own source
1635 files in a filesystem-position-independent manner, relative to
1636 the "SYS" logical host.
1637 * fixed a compiler bug: MV-LET convertion did not check references
1638 to the "max args" entry point. (reported by Brian Downing)
1639 * tweaked disassembly notes to be less confident about proclaiming
1640 some instruction as an LRA. (thanks to Brian Downing)
1641 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1642 now each have their own history, command character, and other
1643 characteristics. (thanks to David Lichteblau)
1644 * fixed some bugs revealed by Paul Dietz' test suite:
1645 ** compiler failure in compiling LOGAND expressions including a
1647 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1648 shift greater than 32.
1649 ** FLUSH-DEST did not mark blocks for type check regeneration.
1650 ** HANDLER-CASE failed to accept declarations in handler clauses
1651 in some circumstances.
1653 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1654 * New code in contrib/sb-introspect (still probably not entirely
1655 stable yet) provides some support for smart Lisp development
1656 environments like SLIME.
1657 * The conditions signalled for errors occurring when loading .fasl
1658 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1659 in a way which should help ASDF recover gracefully.
1660 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1661 stuff has been cleaned up. If you code contrib/ stuff, this might
1662 affect you, and you can look at contrib/README, contrib/STANDARDS,
1663 and/or the 0.8.4.27 diff to check.
1664 * In full calls the compiler now does not generate checks for declared
1665 argument types for all arguments.
1666 * various threading fixes
1667 ** and some experimental patches which didn't make it into
1668 the main tree for this release, but which are shipped in
1669 contrib/experimental-thread.patch as a possible fix for some
1670 failures (deadlock, spinning...) in GC-intensive multithreaded
1672 * fixed PPC build problem (source code incompatibility of different
1673 library versions): added offsetof() hackery which attempts to divine
1674 where glibc maintainers put uc_mcontext today
1675 * fixed bug 282: compiler does not trust type assertions while passing
1676 arguments to a full call.
1677 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1678 &OPTIONAL VALUES type parameter.
1679 * fix bug 214: algorithm for noting rejected templates is now more
1680 similar to that of template seletion. (also reported by rydis on
1682 * fixed bug 141b: printing backquoted information readably and prettily
1683 inserts a space where necessary.
1684 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1685 (CHARACTER 10) are now reported as errors, rather than propagated
1686 as unknown types. (reported by piso on #lisp)
1687 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1688 with values NIL and :ERROR. (thanks to Milan Zamazal)
1689 * fixed bug 191c: CLOS now does proper keyword argument checking as
1690 described in CLHS 7.6.5 and 7.6.5.1.
1691 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1692 longer raise an error; further, using a list as a for-as-arithmetic
1693 counter now raises a meaningful error.
1694 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1695 now understood by sequence creation functions such as MAKE-SEQUENCE
1697 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1698 a STRING-INPUT-STREAM.
1699 * compiler enhancement: SIGNUM is now better able to derive the type
1701 * type declarations inside WITH-SLOTS are checked. (reported by
1703 * fixed some bugs revealed by Paul Dietz' test suite:
1704 ** incorrect optimization of TRUNCATE for a positive first
1705 argument and negative second.
1706 ** compiler failure in let-convertion during flushing dead code.
1707 ** compiler failure while deriving type of TRUNCATE on an
1708 interval, containing 0.
1709 ** ASH of a negative bignum by a negative bignum count now returns
1711 ** intersection of CONS types now canonicalizes properly, fixing
1712 inconsistencies in SUBTYPEP.
1714 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1715 * incompatible change: The --disable-debugger command line
1716 option now clobbers the debugger at a more fundamental
1717 level, by redefining #'INVOKE-DEBUGGER instead of by
1718 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1719 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1720 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1721 enter the debugger REPL and then suffer recursive errors
1722 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1723 while under the new system, BREAK in --disable-debugger mode
1724 terminates the system just as an unhandled error would.
1725 * fixed compiler performance when processing loops with a step >1;
1726 * bug fix: DOCUMENTATION now retrieves generic function
1727 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1728 support has been systematized, and now supports the methods
1729 specified by ANSI, along with a default method and a method for
1730 slot documentation. (reported by Nathan Froyd)
1731 * bug fix: effective methods associated with a generic function are
1732 no longer cached over a change of that generic function's method
1733 combination. (reported by Andreas Fuchs)
1734 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1735 names a non-existent directory. (thanks to Andreas Fuchs)
1736 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1737 the PPC platform, lead to stack corruption; this has been fixed.
1738 (reported by Rainer Joswig)
1739 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1740 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1741 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1742 arithmetic in general, is now much more reliable on the PPC
1744 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1745 platform now returns the right answer.
1746 * optimization: restored some effective method precomputation in
1747 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1748 precomputation is now tunable.
1749 * optimization: compiler-internal data structure use has been
1750 reviewed, and changes have been made that should improve the
1751 performance of the compiler by about 20%.
1752 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1753 simple-base-strings and simple-bit-vectors is improved.
1754 * optimization: the optimization of 32-bit logical and arithmetic
1755 functions introduced in version 0.8.3 on the x86 has been
1756 implemented on the mips, ppc and sparc platforms; an
1757 implementation of the same facility, but for 64-bit arithmetic,
1758 has been added for the alpha.
1759 * microoptimization: the compiler is better able to make use of the
1760 x86 LEA instruction for multiplication by constants.
1761 * bug fix: in some situations compiler did not report usage of
1762 generic arithmetic in (SPEED 3) policy.
1763 * bug 145b fix: compiler used wrong type specifier while converting
1764 MEMBER-types to numeric.
1765 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1767 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1768 index argument now works properly on the Alpha platform.
1769 * bug fix: floating point exception treatment on the Alpha platform
1771 * bug fix: FILE-POSITION works much better on string input and
1772 output streams. (thanks to Nikodemus Siivola)
1773 * bug fix: many threading/garbage collection symptoms sorted.
1774 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1775 might be pseudo-atomic.
1776 * internal change: Stopping for GC is now done with signals not
1777 ptrace. GC is now done in whichever thread wanted it, instead of
1779 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1781 * bug fix: result form in DO is not contained in the implicit
1783 * incompatible change: ICR structure is changed; the value part of
1784 CONTINUATION is now called LVAR; corresponding functions are
1785 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1786 * added type deriver for ISQRT (thanks to Robert E. Brown).
1787 * bug fix: better support for loading from the command line when an
1788 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1790 * fixed some bugs revealed by Paul Dietz' test suite:
1791 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1792 ** ROUND and FROUND now give the right answer when given very
1793 small float arguments.
1794 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1796 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1797 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1798 ** LCM should return a non-negative integer.
1799 ** PARSE-INTEGER returned the index of a terminator instead of the
1800 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1801 ** PARSE-INTEGER returned an incorrect index being applied to a
1803 ** LCM with two arguments of 0 returns 0 rather than signalling
1805 ** unsigned addition of a 32-bit constant with the high bit set no
1806 longer causes an internal compiler error on the x86.
1807 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1808 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1809 with negative last argument.
1810 ** byte specifiers with zero size and position no longer cause
1811 an error during type derivation.
1812 ** bignum multiplication on the Alpha platform now returns the
1814 * porting: The system now builds on SuSE AMD64, although it still
1815 generates a 32-bit binary.
1816 * .fasl file incompatibility: The fasl file version number has
1817 been incremented (because of the changes to internal compiler
1818 data structures referred to above).
1820 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1821 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1822 more accurately, on the Darwin kernel running on PowerPC hardware.
1823 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1824 * Compiler code deletion notes now signal a condition of type
1825 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1826 an associated MUFFLE-WARNING restart.
1827 * The compiler now performs limited argument count validation of
1828 constant format strings in FORMAT, and where appropriate in ERROR,
1829 CERROR and WARN. (thanks to Gerd Moellmann)
1830 * New ASDF-INSTALL contrib can be used for automatic download and
1831 installation of third-party Lisp code from CCLAN or other sites
1833 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1834 which forces another thread to execute a function supplied by the
1836 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1837 now accept and act upon their :ELEMENT-TYPE keyword argument.
1838 (reported by Martin Atzmueller, Edi Weitz)
1839 * bug fix: FILE-POSITION now accepts position designators up to
1840 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1841 is the greater. (thanks to Patrik Nordebo)
1842 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1843 Kalle Olavi Niemitalo)
1844 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1845 printable. (reported by Eric Marsden)
1846 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1847 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1848 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1849 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1850 (thanks to Stig Sandoe)
1851 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1852 UNREAD-CHAR now correctly includes the unread character in the
1853 target sequence. (thanks to Gerd Moellmann)
1854 * bug fix in threaded builds: the system can now be suspended and
1855 resumed by shell job control with minimal disruption.
1856 * bug fixes in times and timezones >2038AD
1857 * better handling of "where is GNU make?" problem in build scripts
1858 (thanks to Nikodemus Siivola)
1859 * new optimization: inside a named function any reference to a
1860 function with the same name is considered to be a self-reference;
1861 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1862 optimization quality.
1863 * new optimization on x86: logical functions and + now have
1864 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1865 used when the result is truncated to 32 bits.
1866 * VALUES declaration is partially enabled.
1867 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1868 * fixed some bugs revealed by Paul Dietz' test suite:
1869 ** The system now obeys the constraint imposed by
1870 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1871 types form a lattice under type intersection.
1872 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1873 ** ASSOC now ignores NIL elements in an alist.
1874 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1875 and (1+ MOST-POSITIVE-FIXNUM) answers.
1876 ** The addition of a method with invalid qualifiers to a generic
1877 function does not cause an error to be signalled immediately;
1878 a warning is signalled, and the error is generated only on
1879 calling the generic function.
1880 * changed .fasl file version number, in part to add type codes for
1881 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1882 obscure ANSI requirements
1884 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1885 * fixed bug 148: failure to inline-expand a local function left
1886 garbage, confusing the compiler.
1887 * fixed bugs 3cd: structure slot readers perform type check if the
1888 slot can have an invalid value (i.e. it is either not initialized
1889 or can be written with a less specific slot writer).
1890 * bug fix: the compiler now traps array references to elements off
1891 the end of an array; previously, the bounds checking in some
1892 circumstances could go off-by-one.
1893 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1895 * type declarations for array element types now obey the description
1896 on the CLHS page "Declaration TYPE", as per discussions on
1897 sbcl-help around 2003-05-08. This means that a declaration
1898 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1899 declaration, all references to BAR will be asserted or assumed
1900 (with THE, so dependent on compiler policy) to involve objects of
1901 type FOO. Note that no such declaration is implied in
1902 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1903 * declared types of functions from the "Conditions"
1904 chapter. (reported by Paul Dietz)
1905 * bug fix: CERROR accepts a function as its first argument.
1906 * bug fix: NTH an NTHCDR accept a bignum as index
1907 arguments. (reported by Adam Warner)
1908 * optimization: character compare routines now optimize comparing
1909 against a constant character. (reported by Gilbert Baumann)
1910 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1911 argument now works properly on the MIPS platform.
1912 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1913 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1914 declared to have a functional type, does not cause an error
1916 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1917 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1918 * fixed bug: if last continuation of a deleted block has a
1919 destination, this destination should be deleted too. (reported by
1921 * fixed a bug in the bootstrap process: the host compiler's values
1922 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1923 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1924 test case from Patrik Nordebo)
1925 * improved the ability of the disassembler on the PPC platform to
1926 provide helpful disassembly notes.
1927 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1928 the class in more cases than previously.
1929 * bug fix: FILE-POSITION now understands :START and :END for
1930 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1931 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1932 8) data. (thanks to David Lichteblau)
1933 * bug fix: it is possible to add a method to a generic function
1934 without lambda list.
1935 * bug fix: reader failed to signal END-OF-FILE inside an
1936 object representation. (reported by Nikodemus Siivola)
1937 * fixed some bugs revealed by Paul Dietz' test suite:
1938 ** LAST and [N]BUTLAST should accept a bignum.
1939 ** condition slot accessors are methods.
1940 ** (VECTOR NIL) is a subtype of STRING.
1942 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1943 * minor incompatible change: some nonsensical specialized lambda
1944 lists (used in DEFMETHOD) which were previously ignored now signal
1946 * minor incompatible change: the system is now aware of the types of
1947 variables in the COMMON-LISP package, and will signal errors for
1948 most violations of these type constraints (where previously they
1949 were silently accepted).
1950 * minor incompatible change: COMPILE-FILE now uses the freedom
1951 afforded (ANSI 3.2.2.3) to use derived function types for
1952 functions defined in the same file. This also permits the system
1953 to warn on static type mismatches and function
1954 redefinition. (Currently it does not work with high DEBUG level.)
1955 * minor incompatible change: VALUES declaration is disabled.
1956 * When issuing notes, the compiler now signals a condition of type
1957 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1958 restart for use in user handlers. It is expected that the
1959 COMPILER-NOTE condition will eventually become a condition
1960 supertype to a hierarchy of note types, which will then be
1961 handleable in a similar fashion. However, at the moment, no such
1962 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1963 but it's an implementation detail, not a classification for the
1965 * Changes in type checking closed the following bugs:
1966 ** type checking of unused values (192b, 194d, 203);
1967 ** template selection based on unsafe type assertions (192c, 236);
1968 ** type checking in branches (194bc).
1969 * A short form of VALUES type specifier has ANSI meaning (it has
1970 increased the number of situations when SBCL cannot perform type
1972 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1973 keywords or constants is permissible.
1974 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1975 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1976 classes. (thanks to Antonio Martinez)
1977 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1978 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1979 argument to a pretty-print stream. Instead, it leaves any such
1980 operations to DESCRIBE-OBJECT methods.
1981 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1983 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1984 lambda lists. (thanks to David Lichteblau)
1985 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1986 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1987 (thanks to Gerd Moellmann)
1988 * fixed bug 246: increased compilation speed of long
1989 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1991 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1993 * DEFCONSTANT now throws a condition of type
1994 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1995 constant to a non-EQL value; CONTINUE and ABORT restarts
1996 respectively change and preserve the value.
1997 * fixed bug 63: The code walker, part of the implementation of CLOS,
1998 is now better at handling symbol macros.
1999 * bug fix: There is no longer an internal implementation type named
2000 CL:LENGTH. (reported by Raymond Toy)
2001 * bug fix: In macro-like defining macros/special operators the
2002 implicit block does not enclose the lambda list.
2003 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2004 enclosed in parentheses) are not suitable as type specifiers, and
2005 their use properly signals an error now.
2006 * bug fix: An argument count mismatch for a type specifier in code
2007 being compiled no longer causes an unhandled error at compile
2008 time, but signals a compile-time warning.
2009 * fixed simple vector readable printing
2010 * bug fix: DESCRIBE takes more care over whether the class
2011 precedence list slot of a class is bound before accessing it.
2012 (reported by Markus Krummenacker)
2013 * bug fix: FORMATTER can successfully compile pretty-printer format
2014 strings which use variants of the ~* directive inside.
2015 * bug fix: SEARCH now applies its TEST predicate to the elements of
2016 the arguments in the correct order. (thanks to Wolfhard Buss)
2017 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2018 in type inference. (thanks to Robert E. Brown)
2019 * bug fix: Reading in symbols with an explicit package name of ""
2020 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2021 * fixed some bugs revealed by Paul Dietz' test suite:
2022 ** NIL is now allowed as a structure slot name.
2023 ** Arbitrary numbers, not just REALs, are allowed in certain
2024 circumstances in LOOP for-as-arithmetic clauses.
2025 ** Multiple class redefinitions before slot access no longer
2026 causes a type error.
2027 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2028 association between the name and a class.
2029 ** Generic functions with non-standard method-combination and over
2030 five methods all of which return constants no longer return NIL
2031 after the first few invocations. (thanks to Gerd Moellmann)
2032 ** CALL-NEXT-METHOD with no arguments now passes the original
2033 values of the arguments, even in the presence of assignment.
2034 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2036 ** DISASSEMBLE works with closures and funcallable instances.
2037 ** ADD-METHOD now returns the generic function, not the new method.
2038 ** FIND-METHOD signals an error if the lengths of the specializers
2039 is incompatible with the generic function, even if the ERRORP
2041 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2042 which its argument is a member.
2043 ** DEFCLASS only redefines the class named by its class-name
2044 argument if that name is the proper name of the class;
2045 otherwise, it creates a new class.
2046 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2047 of the UNBOUND-SLOT condition to the name of the slot.
2048 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2049 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2050 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2051 SLOT-MAKUNBOUND in the specified fashion.
2053 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2054 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2055 cross-compilation host. As a consequence, we can now bootstrap our
2056 way up to SBCL starting with a bare gcc toolchain and human-readable
2057 source code (first the source to CLISP, then the source to SBCL).
2058 * A contributed module containing a partial implementation of the
2059 simple-streams interface has been included. (thanks to Rudi
2061 * A contributed module implementing the RFC1321 Message Digest
2062 Algorithm, known as MD5, has been included.
2063 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2064 no longer has any effect, as the code controlled by this feature
2065 has been deleted. (As far as we know, no-one has ever built using
2066 this feature, and its semantics were confused in any case).
2067 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2068 work on conditions (as required by the ANSI specification),
2069 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2070 expected behaviour on conditions. Users should note, however,
2071 that such behaviour is not required by the ANSI specification,
2072 and so use of this behaviour may render their code unportable.
2073 * fixed some bugs revealed by Paul Dietz' test suite:
2074 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2076 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2077 ** SXHASH on condition objects no longer returns NIL.
2078 ** :ALLOCATION :CLASS slots are better treated; their values are
2079 updated on class redefinition, and initforms inherited from
2080 superclasses are applied.
2081 ** REMOVE-METHOD returns its generic function argument even when
2082 no method was removed.
2083 ** SHARED-INITIALIZE now initializes the values of the requested
2084 slots, including those with :ALLOCATION :CLASS.
2085 ** ALLOCATE-INSTANCE now works on structure classes defined via
2086 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2088 ** SLOT-EXISTS-P now works on conditions, as well as structures
2090 ** MAKE-LOAD-FORM now has the required methods on
2091 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2092 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2093 keyword, and now implements the SLOT-NAMES argument.
2094 ** methods with &OPTIONAL arguments no longer allow too many
2095 arguments to be passed in the call without error.
2096 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2097 option is consistent with the required arguments of the generic
2098 function lambda list.
2099 * bug fix: REQUIRE accepts a string designator. (Thanks to
2101 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2102 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2103 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2104 * bug fix: The compiler now checks for duplicated variables in macro
2106 * bug fix: SETQ on globals returns the correct value.
2107 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2108 not a COMPILER-ERROR (followed by some other strange error on
2109 choosing the CONTINUE restart).
2110 * bug fix: make.sh and friends are now more consistent in the way that
2111 they look for GNU "make".
2113 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2114 * experimental native threads support (on x86 Linux >=2.4 only).
2115 This is not compiled in by default: you need to add :SB-THREAD to
2116 the target features. See the "Beyond ANSI" chapter of the manual
2118 * fix for longstanding nonANSIism: The old distinction between
2119 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2120 The return value from CL:FIND-CLASS is now a CLOS class, and
2121 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2122 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2123 * An interface to the MetaObject Protocol, as described in Kiczales,
2124 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2125 MIT Press, 1991, is available from the SB-MOP package.
2126 * incompatible change: the SB-PCL package should now be considered
2127 a private implementation detail, and no longer a semi-private MOP
2129 * minor incompatible change: due to rearrangement for threads, the
2130 control stack and binding stack are now allocated at arbitrary
2131 addresses instead of being hardcoded per-port. Users affected by
2132 this probably have to be doing advanced things with shared
2133 libraries, and will know who they are.
2134 * minor incompatible change: Previously, all --eval forms used were
2135 processed with READ before any of them were processed with EVAL.
2136 Now each --eval form is processed with both READ and EVAL before
2137 the next --eval form is processed. (Thus package operations like
2138 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2139 work as the user might reasonably expect.)
2140 * minor incompatible change: *STANDARD-INPUT* is now only an
2141 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2143 * minor incompatible change: Y-OR-N-P is now character-oriented,
2144 not line oriented. Also, YES-OR-NO-P now works without errors.
2145 (thanks to Antonio Martinez)
2146 * sb-aclrepl module improvements: an integrated inspector, added
2147 repl features, and a bug fix to :trace command.
2148 * Known functions, which cannot be open coded by the backend, are
2149 considered to be able to check types of their arguments. (fixing
2150 a bug report by Nathan J. Froyd)
2151 * fixed a bug in computing method discriminating functions: It is
2152 now possible to define methods specialized on classes which have
2153 forward-referenced superclasses. (thanks to Gerd Moellmann)
2154 * fixed evaluation order in optional entries (reported by Gilbert
2156 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2157 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2158 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2159 required-by-AMOP NAME argument, as well as CLASS and
2160 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2161 * fixed bug 20: DEFMETHOD can define methods using names that are
2162 not the proper names of classes to designate class specializers.
2163 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2164 Unix file descriptors, instead of blowing up. (thanks to Antonio
2166 * Garbage collection refactoring: user-visible change is that a
2167 call to the GC function during WITHOUT-GCING will not do garbage
2168 collection until the end of the WITHOUT-GCING. If you were doing
2169 this you were probably losing anyway.
2170 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2171 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2172 TYPEP the latter but not the former.
2173 * The compiler issues a full WARNING for calls to undefined functions
2174 with names from the CL package.
2175 * MAP-INTO for a vector destination is open coded. (reported by
2176 Brian Downing on c.l.l)
2177 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2178 documentation string.
2179 * fixed some bugs revealed by Paul Dietz' test suite:
2180 ** COPY-ALIST now signals an error if its argument is a dotted
2182 ** Condition slots are now accessed more correctly in the presence
2183 of multiple initargs for a given slot.
2184 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2185 exclude restarts of the same name associated with a different
2187 ** DEFCLASS of forward-referenced classes with another
2188 forward-referenced class in the superclasses list no longer
2190 ** Condition slots are now initialized once each, not multiple
2191 times. (thanks to Gerd Moellmann)
2192 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2194 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2195 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2196 arguments contain duplicated elements.
2197 ** RESTART-CASE understands local macros.
2198 ** RESTART-CASE associates exactly its own restarts with a condition.
2199 ** ENDP in safe mode checks its argument to be of type LIST.
2200 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2201 in question is unbound.
2202 ** Optimized MAKE-INSTANCE functions no longer cause internal
2203 assertion failures in the presence of duplicate initargs.
2204 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2205 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2206 propagated even in the ctor optimized implementation of
2208 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2210 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2211 * a better implementation of SXHASH on (simple) bit vectors,
2212 measured both in execution speed and in distribution of results
2213 over the positive fixnums, has been installed. Likewise, a better
2214 implementation of EQUAL for simple bit vectors is now available.
2215 * fixed CEILING optimization for a divisor of form 2^k.
2216 * fixed bug 240 (emitting extra style warnings "using the lexical
2217 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2218 by Antonio Martinez)
2219 * fixed SXHASH, giving different results for NIL depending on type
2220 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2221 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2222 variable. (found by Rolf Wester)
2223 * SBCL does not ignore type declarations for special
2224 variables. (reported by rif on c.l.l 2003-03-05)
2225 * some bug fixes in contrib/sb-aclrepl/
2226 * fixed some bugs revealed by Paul Dietz' test suite:
2227 ** a bug in the CONS type specifier, whereby the CAR and CDR
2228 types got intertwined, has been fixed;
2229 ** the type system is now able to reason about the interaction
2230 between INTEGER and RATIO types more completely;
2231 ** APPEND, [N]REVERSE and NRECONC check that those their
2232 arguments, which must be proper lists, are really so;
2233 ** An array specialized to be unable to hold elements has been
2234 implemented, as required -- yes, really -- by ANSI;
2235 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2236 on malformed property lists;
2238 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2239 * incompatible packaging change: in line with Unix convention,
2240 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2241 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2242 * REQUIRE and PROVIDE are now optionally capable of doing something
2243 useful. See the documentation string for REQUIRE.
2244 * infrastructure for a managed SBCL contrib system: contributed
2245 modules in this release include:
2246 ** the ASDF system definition facility;
2247 ** an interface to the BSD Sockets API;
2248 ** an ACL-like convenience interface to the repl;
2249 (thanks to Kevin Rosenberg)
2250 ** an implementation of ROTATE-BYTE, with efficient implementation
2252 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2253 gives a closer approximation to the right answer than previously.
2254 (thanks to Raymond Toy)
2255 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2256 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2257 optional environment arguments, as required by ANSI.
2258 * fixed bugs in other functions taking environment objects, allowing
2259 calls with an explicit NIL environment argument to be compiled
2261 * fixed bug 228: primary return values from
2262 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2263 COMPILE or FUNCTION.
2264 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2265 :TYPE will no longer signal errors on innocuous objects.
2266 * fixed bug 231b: SETQ is better at respecting type declarations in
2267 the lexical environment.
2268 * fixed a bug in DEFCLASS: classes named by symbols with no or
2269 unprintable packages can now be defined.
2270 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2271 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2272 * fixed compiler failure related to checking types of functions.
2273 (reported by Robert E. Brown)
2274 * the compiler is now much more consistent in its error-checking
2275 treatment of bounding index arguments to sequence functions: in
2276 (SAFETY 3) code, errors will be signalled in almost all cases if
2277 invalid sequence bounding indices are passed to functions defined
2278 by ANSI to operate on sequences.
2279 * fixed a bug in the build procedure: documentation of SBCL-specific
2280 packages is now preserved and available in the final Lisp image.
2281 * lifted FDEFINITION lookup out of loops in the implementation of
2282 many list operations. (thanks to Robert E. Brown)
2283 * fixed a bug in the reader: the #n# reader macro now works for
2284 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2285 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2286 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2287 E. Brown and Tony Martinez respectively)
2288 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2289 SLOT-BOUNDP: in particular, it is now able to optimize them much
2290 better, and is now not vulnerable to having packages renamed.
2291 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2292 always called when a slot is not present in an instance. (thanks
2294 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2295 longer signal an error on unreading a character following EOF on
2296 the previous constituent stream. (thanks to Tony Martinez)
2297 * fixed some bugs revealed by Paul Dietz' test suite:
2298 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2299 not just nonnegative fixnums;
2300 ** the logical bit-array operators such as BIT-AND now accept an
2301 explicit NIL for their "opt-arg" argument (to indicate a
2302 freshly-consed result bit-array);
2303 ** ELT now signals an error on an invalid sequence index in safe
2305 ** the type system is now cleverer about negations of numeric
2306 types, and consequently understands the BIGNUM and RATIO types
2308 ** the type system is now cleverer about the interaction between
2309 INTEGER and RATIO types: while bugs still remain, many more
2310 cases are accurately computed;
2311 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2312 if it is in the last clause;
2313 ** CONSTANTLY now correctly returns a side-effect-free function in
2315 ** DECLARE is no longer treated as a special-operator; in
2316 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2317 * incremented fasl file version number due to the change in the
2318 DEFSTRUCT-SLOT-DESCRIPTION structure.
2320 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2321 * minor incompatible change: code processed by the "interpreter" or
2322 EVAL now has a compilation optimization policy of (DEBUG 2)
2323 (changed from (DEBUG 1)) to improve debuggability of interactive
2324 development, and to allow the use of the debug RETURN command in
2326 * an experimental implementation of the RETURN command for the
2327 debugger has been included. (thanks to Frederik Kuivinen)
2328 * fixed bug 62: constraints were not propagated into a loop.
2329 * fixed bug in embedded calls of SORT (reported and investigated by
2331 * fixed some bugs revealed by Paul Dietz' test suite:
2332 ** printing and reading of arrays with some dimensions having
2333 length 0 (thanks to Gerd Moellmann);
2334 ** BOA constructor with &AUX argument without a default value does
2335 not cause a type error;
2336 ** CONSTANTP now returns true for all self-evaluating objects.
2338 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2339 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2340 accessors that are related by inheritance, as specified in the
2341 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2342 Valtteri Vuorikoski)
2343 * The compiler is now able to inline functions that were defined in
2344 a complex lexical environment (e.g. inside a MACROLET).
2345 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2347 * Support for the upcoming FreeBSD-5.0 release has been included.
2348 (thanks to Dag-Erling Smorgrav)
2349 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2350 effect when it is not in a toplevel context.
2351 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2352 stand a better chance of being correct. (thanks to Gerd
2354 * fixed bug in COERCE, which now signals an error on coercing a
2355 rational to a bounded real type which excludes the expected
2357 * The compiler is now able to derive types more accurately from the
2358 COERCE and COMPILE functions.
2359 * fixed bug 223: functional binding is considered to be constant
2360 only for symbols in the CL package.
2361 * fixed bug 231: SETQ did not check the type of a variable being set
2362 (reported by Robert E. Brown)
2363 * A new optimization for MAKE-INSTANCE has been included, fixing
2364 various bugs (including relating to :ALLOCATION :CLASS slots and
2365 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2367 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2368 ** As required by ANSI, LOOP now disallows anonymous collection
2369 clauses such as COLLECT I in conjunction with aggregate boolean
2370 clauses such as THEREIS (= I 1);
2371 ** LOOP now signals an error when any variable is reused in the
2372 same loop (including the potentially useful construct analogous
2373 to WITH A = 1 WITH A = (1+ A);
2374 ** IT is only a special loop symbol within the first clause of a
2375 conditional loop clause;
2376 ** LOOP with a typed iteration variable over a hashtable now
2377 signals a type error iff it should.
2378 * fixed some other bugs revealed by Paul Dietz' test suite:
2379 ** FILE-STREAM now names the class previously known as FD-STREAM;
2380 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2381 argument) no longer signals an error;
2382 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2383 of the slot symbol, rather than using the current package
2384 ((:CONC-NAME "") continues to intern the slot's name in the
2386 * incremented fasl file version number, because of the incompatible
2387 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2388 of the new implementation of DEFINE-COMPILER-MACRO.
2390 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2391 * Support for building SBCL for MIPS platforms running in
2392 little-endian mode has now been checked in, and basic
2393 functionality on said platforms verified.
2394 * minor incompatible change: PCL now records the pathname of a file
2395 in which methods and the like are defined, rather than its
2397 * minor incompatible change: TRUENAME now considers the truename of
2398 a file naming a directory to be the pathname with :DIRECTORY
2399 component indicating that directory.
2400 * minor incompatible change: a NAMED clause in the extended form of
2401 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2402 reason for the previous behaviour is unclear.
2403 * more systematization and improvement of CLOS and MOP conformance
2404 in PCL (thanks to Gerd Moellman and Pierre Mai):
2405 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2407 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2408 primary methods with no specializers;
2409 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2411 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2412 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2413 FORWARD-REFERENCED-CLASSes; error reporting on
2414 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2416 ** SXHASH on CLOS instances now uses a slot internal to the
2417 instance to return different numbers on distinct instances,
2418 while preserving the same return value through invocations of
2420 ** DEFMETHOD signals errors when methods with longer incongruent
2421 lambda lists are added to generic functions;
2422 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2423 CLASS, as specified in AMOP;
2424 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2425 on the order returned by the primary method for classes of
2426 class STANDARD-CLASS;
2427 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2428 * fixed some bugs shown by Paul Dietz' test suite:
2429 ** DOLIST puts its body in TAGBODY;
2430 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2432 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2433 value producing form;
2434 ** if more variables are given to PROGV than values, extra
2435 variables are bound and made to have no value;
2436 ** NSUBSTITUTE on list arguments gets the right answer with
2438 ** ELT signals an error of type TYPE-ERROR when the index argument
2439 is not a valid sequence index;
2440 ** LOOP signals (at macroexpansion time) an error of type
2441 PROGRAM-ERROR when duplicate variable names are found;
2442 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2443 ** FUNCALL of special-operators now cause an error of type
2445 ** PSETQ now works as required in the presence of side-effecting
2446 symbol-macro places;
2447 ** NCONC accepts any object as its last argument;
2448 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2450 ** loop-for-as-package does not require a package to be explicitely
2452 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2454 * fixed bug 166: compiler preserves "there is a way to go"
2455 invariant when deleting code.
2456 * fixed bug 172: macro lambda lists with required arguments after
2457 &REST arguments now cause an error to be signalled. (thanks to
2459 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2460 bug. (thanks to Matthew Danish)
2461 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2463 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2464 arguments are lexically rebound. (thanks to Gerd Moellmann and
2466 * fixed bug 194: error messages are now more informative when there
2467 is no primary method applicable in a call to a generic
2468 function. (thanks to Gerd Moellmann)
2469 * fixed bug in command line argument checking (thanks to Julian
2471 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2472 (thanks to Matthew Danish)
2473 * incremented fasl file version number, because of the
2474 SXHASH-related changes in the layout of CLOS data structures
2476 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2477 * minor incompatible change: The runtime (the Unix executable named
2478 "sbcl") is now much pickier about the .core files it will load.
2479 Essentially it now requires .core files to descend from the same
2480 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2481 as the runtime does. (The intent is to prevent the crashes which
2482 can occur, and which can even be reported as mysterious failures,
2483 when people patch the sources or change the build parameters
2484 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2485 sbcl and .core files.)
2486 * fixed bug: VALUES-LIST is no longer optimized away.
2487 * fixed bug 142: The FFI conversion of C string values to Lisp
2488 string values no longer conses excessively. (thanks to Nathan
2489 Froyd porting Raymond Toy's fix to CMU CL)
2490 * began to systematize and improve MOP conformance in PCL (thanks to
2491 Nathan Froyd, Gerd Moellman and Pierre Mai):
2492 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2494 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2495 ** FINALIZE-INHERITANCE is now called on class finalization;
2496 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2497 argument precedence order.
2498 * fixed bug 202: The compiler no longer fails on functions whose
2499 derived types contradict their declared type.
2500 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2501 so it can be non-toplevel.
2502 * The fasl file version number has changed (because of the new
2503 implementation of DEFMACRO).
2504 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2505 safe code, that any length requirement by their type-specifier
2506 argument is valid. The exceptions to this are described in bug
2508 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2509 functions now check the types of their inputs as required by ANSI.
2510 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2511 introduced symbol is DECLAREd to be SPECIAL.
2512 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2513 * fixed bug: :COUNT argument to sequence functions may be negative
2514 * fixed bug: body of DO-SYMBOLS may contain declarations
2515 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2516 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2517 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2518 symbol macro only once
2519 * fixed printing of call frame when argument list is unavailable
2520 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2521 * compiler no longer signals WARNING on unknown keyword
2524 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2525 * A beta-quality port to the mips architecture running Linux,
2526 based on the old CMUCL backend, has been made. It has been tested
2527 on a big-endian kernel, and works sufficiently well to be able to
2528 rebuild itself; it has not been tested in little-endian mode.
2529 * fixed an inconsistency between gencgc.c and purify.c which made
2530 dumping/loading .core files unreliable
2531 * fixed bug 120a: The compiler now deals correctly with IFs where
2532 the consequent is the same as the alternative, instead of
2533 misderiving the return type. (thanks to Alexey Dejneka)
2534 * fixed bug 113: Logical pathnames are now dumpable (the logical
2535 host is resolved at load-time, throwing an error if it is not
2537 * fixed bug 174: FORMAT's error message is slightly clearer when a
2538 non-printing character is used in a format directive.
2539 * fixed several bugs in compiler checking of type declarations, i.e.
2540 violations of the Python "declarations are assertions" principle
2541 (thanks to Alexey Dejneka)
2542 * fixed several bugs in PCL's error checking (thanks to Gerd
2544 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2546 * fixed bug in compilation of functions as first class values
2547 (thanks to Antonio Martinez-Shotton)
2548 * The compiler's handling TYPE-ERRORs which it can prove will
2549 inevitably happen at runtime has been cleaned up and corrected
2550 in several ways. (thanks to Alexey Dejneka)
2551 * improved argument type checking for various basic arithmetic
2552 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2553 lavished on them in the past that they can be compiled in many
2554 ways in different special cases
2555 * fixed bug 181: compiler checks validity of user supplied type
2557 * cleaned up code flushing in optimization: Function calls which
2558 should signal errors for safety purposes (e.g. which ANSI says
2559 should signal errors when their arguments are of incorrect type)
2560 are no longer optimized away.
2561 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2562 * incremented fasl file version number, because changes in the
2563 implementation of sequence functions like COERCE caused
2564 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2565 in old inline expansions) to become undefined. (Actually these
2566 changes were later undone, so we might very well be binary
2567 compatible with 0.7.7 after all, but leaving the version number
2568 incremented seemed like the simplest and most conservative
2571 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2572 * An alpha-quality port to the parisc architecture running Linux,
2573 based on the old CMUCL backend, has been made. This, even more so
2574 than the other backends, should be considered still a work in
2575 progress; known problems include that the Linux kernel in 64-bit
2576 mode does not propagate the correct sigcontext structure to
2577 userspace, and consequently SBCL on a parisc64 kernel will not
2579 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2580 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2581 suppressed.) Also now that the compiler is looking at declarations
2582 in the environment, it checks optimization declarations as well,
2583 and suppresses inlining when (> DEBUG SPEED).
2584 * More fixes have been made to treatment of floating point exception
2585 treatment and other Unix signals. In particular, floating point
2586 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2587 * The detection and handling of control stack exhaustion (infinite
2588 or very deeply nested recursion) has changed. Stack exhaustion
2589 detection is now done by write-protecting pages at the OS level
2590 and applies at all optimization settings; when found, a
2591 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2592 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2593 be caught using IGNORE-ERRORS.
2594 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2595 correctly when transforming recursive LABELS functions to LETs.
2596 (thanks to Alexey Dejneka)
2597 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2598 that are names of constants or global variables.
2599 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2600 alien routines with docstrings.
2601 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2602 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2604 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2605 (thanks to Eric Marsden)
2606 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2607 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2608 * Bug fix: The SPARC backend can now compile functions involving
2609 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2610 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2611 to a functions expecting a PATHNAME-DESIGNATOR.
2612 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2613 lambda lists. (thanks to Alexey Dejneka)
2614 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2615 functions so that EXPORTing the name of the class doesn't cause
2616 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2617 to Antonio Martinez for reporting this)
2618 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2619 constant-folded without failing an assertion. (thanks to Einar
2620 Floystad Dorum for reporting this)
2621 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2622 types are now dealt with more correctly by the compiler.
2623 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2624 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2625 behaviour analogous to RENAME-FILE. This puts its behaviour more
2626 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2628 * The fasl file version number has changed again. (because of the
2629 bug fix involving the names of PCL MAKE-INSTANCE functions)
2631 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2632 * bug fix: Floating point exceptions are treated much more
2633 consistently on the x86/Linux and PPC/Linux platforms.
2634 * Array initialization with :INITIAL-ELEMENT is now much faster for
2635 cases when the compiler cannot open code the array creation, but
2636 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2637 array accesses have also seen a speed increase.
2638 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2639 specified. (This was at the root of some bad interactions between
2640 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2642 * bug fix: Internal error arguments for undefined functions are now
2643 computed correctly on the PPC/Linux platform.
2644 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2645 Raymond Toy's patch for CMU CL)
2646 * Support for the Solaris 9 operating environment has been included
2647 (thanks to Daniel Merritt)
2648 * A very ugly but hopefully complete draft of the missing FFI chapter
2649 of the manual has been created by reformatting the corresponding
2650 CMU CL manual chapter into (currently very ugly and incoherent)
2651 DocBook and bringing it up to date for SBCL behavior. Thus, the
2652 manual is now essentially complete, at least by my extreme
2653 once-and-only-once standards, whereby it's acceptable to refer to
2654 the doc strings of SB-EXT functions as the primary documentation.
2655 * The fasl file version number has changed again, due to cleanup of
2656 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2657 is no longer a static symbol.)
2659 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2660 * SBCL now builds with OpenMCL (version 0.12) as the
2661 cross-compilation host; also, more progress has been made toward
2662 bootstrapping under CLISP.
2663 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2665 * bug 158 fixed: The compiler can now deal with integer loop
2666 increments different from 1; fixing this turned out also to fix
2668 * bug 169 fixed: no more bogus warnings about using lexical bindings
2669 despite the presence of perfectly good SPECIAL declarations (thanks
2670 to David Lichteblau)
2671 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2672 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2673 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2675 * bug 180 fixed: Method combination specifications no longer ignore
2676 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2677 * bug fix: Structure type predicate functions now check their argument
2678 count as they should.
2679 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2680 correctly. (thanks to Pierre Mai)
2681 * minor incompatible change: The --noprogrammer option is deprecated
2682 in favor of the new --disable-debugger option, which is very similar.
2683 (The major difference is that it takes effect at a slightly different
2684 time at startup, causing handling of errors in --sysinit and
2685 --userinit files will be affected differently.) The
2686 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2687 been added to allow this functionality to be controlled from ordinary
2688 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2689 maintainers, who might want to run non-interactive scripts to
2690 build SBCL cores which will later be used interactively.)
2691 * minor incompatible change: The LOAD function no longer, when given
2692 a wild pathname to load, loads all files matching that pathname.
2693 Instead, an error of type FILE-ERROR is signalled.
2695 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2696 * bug 147 fixed: The compiler preserves its block link/count
2697 invariants more correctly now so that it doesn't crash. (thanks
2699 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2701 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2702 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2703 no-value-for-this-character as NIL (as specified by ANSI).
2704 * HOST-NAMESTRING on physical pathnames now returns a string that is
2705 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2706 (thanks to Christophe Rhodes)
2707 * The Alpha port handles icache flushing more correctly. (thanks to
2709 * More progress has been made toward bootstrapping under CLISP. (thanks
2710 to Christophe Rhodes)
2711 * The fasl file format has changed again, because dynamic loading
2712 on OpenBSD (which has non-ELF object files) motivated some cleanups
2713 in the way that foreign symbols are transformed and passed around.
2714 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2715 is no longer treated as whitespace by the reader, but instead as
2716 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2717 |A
\7fB|, instead of A as it used to.
2719 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2720 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2721 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2722 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2723 this as of 0.7.1.45, but was left out of the previous news section
2724 (thanks to Dan Barlow)
2725 * SBCL now runs on the Solaris operating system on SPARC architectures
2726 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2727 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2728 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2729 * SPARC backend cleanups, allowing builds of cores optimized for V8
2730 and V9 SPARCS, and also emission of code targeted to a particular
2731 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2733 * SBCL is closer to bootstrapping under CLISP, thanks to various
2734 fixes by Christophe Rhodes.
2735 * The fasl file format has changed again, to allow the compiler's
2736 INFO database to support symbol macros.
2737 * The user manual (in doc/) is formatted into HTML more nicely.
2738 (thanks to coreythomas)
2739 * The system is smarter about SUBTYPEP relationships, especially
2740 those involving NOT types (including types such as ATOM which are
2741 represented internally using NOT types). Thus SUBTYPEP is less
2742 likely to return (VALUES NIL NIL) in general, and in particular
2743 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2745 * The fasl file format has changed again, because the internal
2746 representation of types now includes a new slot to support the new
2747 SUBTYPEP-of-NOT-types logic.
2748 * (not a change in the main branch of SBCL, but a related prototype
2749 which can hopefully be merged into the main branch of SBCL in the
2750 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2751 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2752 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2753 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2754 dispatch macro character combinations. (thanks to Alexey Dejneka)
2755 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2756 that superficially look like logical namestrings correctly.
2757 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2759 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2760 default supplied-p) for &optional and &key arguments. (thanks to
2763 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2764 * incompatible change: The compiler is now less aggressive about
2765 tail call optimization, doing it only when (> SPACE DEBUG) or
2766 (> SPEED DEBUG). (This is an incompatible change because there are
2767 programs which relied on the old CMU-CL-style behavior to optimize
2768 away their unbounded recursion which will now die of stack overflow.)
2769 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2770 for non-GENCGC systems has been increased to 20M (since that
2771 seems much closer to the likely performance optimum for modern
2772 systems than the old 4M value was)
2773 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2774 and *DEBUG-PRINT-LEVEL*
2775 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2776 of CMU CL's support for SPARC, and various endianness and other
2777 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2778 * new syntactic sugar for the Unix command line: --load foo.bar is now
2779 an alternate notation for --eval '(load "foo.bar")'.
2781 ** The system now detects stack overflow and handles it gracefully,
2782 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2783 optimization settings. (This is a good thing in general, and
2784 its introduction in this version should be particularly timely
2785 for anyone whose code fails because of suppression of tail
2787 ** The system now hunts for the C variable "environ" in a more
2788 devious way, to avoid segfaults when the C library version differs
2789 between compile time and run time. (thanks to Christophe Rhodes)
2790 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2791 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2792 ** The compiler no longer issues bogus style warnings for undefined
2793 classes in the same source file as the DEFCLASSes which defined
2794 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2796 ** fixes in CONDITION class precedence list for undefined function
2797 errors (thanks to Alexei Dejneka)
2798 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2799 correctly. (thanks to Dan Barlow)
2800 ** portability fixes aiming at bootstrapping under CLISP (thanks
2801 to Dave McDonald and Christophe Rhodes)
2802 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2803 ** fixes in type translation and and type inference (thanks to
2805 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2806 ** various fixes in the new ports (thanks to Dan Barlow)
2807 * several changes related to debugging:
2808 ** suppression of tail recursion, as noted above
2809 ** stack overflow detection, as noted above
2810 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2811 is now the default. (For some time encapsulation has been more
2812 reliable than the breakpoint-based :ENCAPSULATE NIL
2813 implementation, at least on X86 systems; and I just noticed that
2814 encapsulation also seems closer to the spirit of the ANSI
2817 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2819 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2820 up properly again. (There was a packaging bug in 0.7.0 which
2821 left their definitions in SB-SYS::LOAD-FOREIGN and
2822 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2823 vital for most things which interface to C-level interfaces,
2824 like extensions working with sockets or databases or
2825 Perl-compatible regexes or whatever, and the need to fix
2826 this bug is the main reason that 0.7.1 was released so
2828 ** DEFGENERIC is now choosier about the methods it redefines, so that
2829 reLOADing a previously-LOADed file containing DEFGENERICs does
2830 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2831 cycle now works as it should. (thanks to Alexey Dejneka)
2832 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2833 fixed by Christophe Rhodes. (He actually submitted this patch
2834 months ago, and I delayed until after 0.7.0.)
2835 ** Bug 111 (internal compiler confusion about runtime checks on
2836 FUNCTION types) was fixed by Alexey Dejneka.
2837 * Some internal cleanups (getting rid of variables which aren't
2838 needed now that the byte interpreter is gone) caused the fasl
2839 file format number to change again.
2841 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2842 * major incompatible change: The default fasl file extension, i.e. the
2843 default extension for files produced by COMPILE-FILE, has changed
2844 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2846 ** There are many changes in the implementation of the compiler.
2847 SBCL is now essentially a compiler-only implementation of ANSI
2848 Common Lisp. EVAL still "interprets" a few special cases, but
2849 almost all the interesting cases are handled by creating
2850 a LAMBDA expression, calling COMPILE on it, then calling
2851 FUNCALL on the result.
2852 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2853 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2854 Since the code is newer, there might still be some new bugs
2855 (though not as many as before Martin Atzmueller's fixes:-). But
2856 the new code is substantially simpler and clearer, and hopefully
2857 any remaining bugs will be simpler, less fundamental, and more
2858 fixable then the bugs in the old code.
2859 ** The revised compiler is still a little unsteady on its feet.
2861 *** The debugging information it produces (particularly the names
2862 of FUNCTION objects) is sometimes much less useful than what
2863 the old compiler produced.
2864 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2865 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2866 within a MACROLET) has been temporarily weakened.
2867 ** There are new compiler optimizations for various functions:
2868 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2869 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2870 *** the math functions TRUNCATE, FLOOR, and CEILING
2871 *** the function-of-all-trades COERCE
2872 Mostly these should be transparent, but there's one
2873 potentially-annoying problem (bug 117): when the compiler
2874 inline-expands a function and does type analysis on the result,
2875 it can create control paths which have type mismatches, and
2876 when it can't prove that those control paths aren't taken,
2877 it will issue WARNINGs about the type mismatches. This is
2878 a particular problem in practice for the new sequence functions.
2879 It's not clear how this should be fixed, and for now, a
2880 workaround is given in the entry for 117 in the BUGS file.
2881 ** (Because of the interaction between the two previous items --
2882 occasional inlining problems and new inline expansions -- some
2883 of the new sequence function optimizations won't really kick in
2884 completely until debugging information, and then inlining, are
2885 straightened out in some future version.)
2886 * minor incompatible changes:
2887 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2888 DIRECTORY no longer implicitly promotes NIL slots of its
2889 pathname argument to :WILD. In particular, when you ask for the
2890 contents of a directory (which you used to be able to do without
2891 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2892 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2893 ** changes in behavior that ANSI explicitly defines to be
2894 implementation dependent:
2895 *** The new compiler-only implementation still conforms with ANSI,
2896 but acts a little different than before. Besides the obvious
2897 changes in performance tradeoffs (that the cost per form passed
2898 to EVAL has gone up, and the cost per form executed by EVAL
2899 has gone down), the behavior of the system changes a little
2900 because there are no longer any interpreted function objects.
2901 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2902 e.g. doing COMPILE on the output of interactive DEFUN is
2904 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2906 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2907 has changed from (:CL) to NIL.
2908 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2909 e.g. #\Newline and #\Tab, don't have names specified in the
2910 ANSI Common Lisp standard, is now based on their ASCII symbolic
2911 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2912 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2913 are no longer used for output.
2914 ** changes in internal implementation constants:
2915 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2916 4 million. (If your application spends a lot of time GCing and
2917 you have a lot of RAM, you might want to experiment with
2918 increasing it even more.)
2919 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2920 However, almost all old code should still continue to work without
2921 immediate update, as SB-C-CALL is now a (deprecated) nickname
2923 ** Old operator names in the style DEF-FOO are now deprecated in
2924 favor of new corresponding names DEFINE-FOO, for consistency with
2925 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2926 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2927 internal symbols, but a few supported extensions like
2928 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2929 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2930 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2931 etc. as you get deeper into recursive calls to the debugger
2932 command loop, instead of the old "5]", "5]]", "5]]]"
2933 sequence. (I was motivated to do this when squabbles between
2934 ILISP and SBCL left me very deeply nested in the debugger. In the
2935 short term, this change will probably provoke more ILISP/SBCL
2936 squabbles, but hopefully it will be an improvement in the long run.)
2937 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2938 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2939 function, since declaiming return types involving aliens is
2940 (1) annoyingly messy to do by hand and (2) vital to efficient
2941 compilation of code which calls such functions.
2942 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2943 longer reexported by the SB-EXT package. They're solely useful
2944 for alien code, so it seems more logical that you should get
2945 them from the SB-ALIEN package, not in SB-EXT.
2946 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2947 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2948 features. Instead, the code that they used to control is always
2949 built into the system.
2950 * many other bug fixes
2951 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2952 advantage of the new EVAL-WHEN stuff and to clean them up in
2953 general, and they are now more ANSI-compliant in a number of
2954 ways. Martin Atzmueller is responsible for a lot of this.
2955 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2957 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2958 *** correct ERROR type for various file operations
2959 *** some fixes for Lisp streams
2960 *** DEFMETHOD syntax checking
2961 *** changing old weird representation of debug information as
2962 strings (which, among their other deficiencies, don't transform
2963 correctly when you rename packages, and don't change their
2964 print representation when you change things like *PACKAGE*
2965 and *PRINT-LENGTH*) to symbols and lists of symbols
2966 He also made several improvements and fixed several bugs in DESCRIBE.
2967 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2969 *** misbehavior of WRITE-STRING/WRITE-LINE
2970 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2971 and several other LOOP problems as well
2972 *** DIRECTORY when similar filenames are present
2973 *** DEFGENERIC with :METHOD options
2974 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2975 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2976 *** argument ordering in FIND with :TEST option
2977 *** mishandled package designator argument in APROPOS-LIST
2978 *** various problems in the backquote readmacro
2979 *** a bug in APROPOS
2980 *** probably some others that I'm not describing very well here,
2981 since the CVS log documents them by reference to sbcl-devel
2982 messages, and the SourceForge archives aren't working well.:-(
2983 ** Dan Barlow improved the Alpha port (and is making progress on the
2984 PPC port, for those of you who think different).
2985 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2986 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2987 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2988 continues to work on the SPARC port (for those of you in a position
2989 to look down upon our little PC-compatible boxes from a great height).
2990 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2991 and uses that copy, rather than the current dynamic value, when
2992 it's trying to decide whether to truncate output. Thus e.g.
2993 (let ((*print-lines* 50))
2994 (pprint-logical-block (stream nil)
2996 (let ((*print-lines* 8))
2997 (print (aref possiblybigthings i) stream)))))
2998 should now truncate the logical block only at 50 lines, instead of
2999 often truncating it at 8 lines, as it did before.
3000 * The doc/cmucl/ directory, containing old CMU CL documentation
3001 from the time of the fork, is no longer part of the base system.
3002 SourceForge has shut down its anonymous FTP service, and with it
3003 my original plan for distributing the old CMU CL documentation
3004 there. For now, if you need these files you can download an old
3005 SBCL source release and extract them from it.
3006 * The fasl file version number changed again, for dozens of reasons,
3007 some of which are apparent above.
3009 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3010 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3011 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3012 patch, so that DISASSEMBLE output is much nicer.
3013 * The code in the SB-PROFILE package now seems reasonably stable.
3014 I still haven't decided what the final interface should look like
3015 (I'd like PROFILE to interact cleanly with TRACE, since both
3016 facilities use function encapsulation) but if you have a need
3017 for profiling now, you can probably use it successfully with
3018 the current CMU-CL-style interface.
3019 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3020 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3021 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3022 link now returns the dangling link itself, and for similar
3023 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3024 link itself. (In these cases the old code signalled an error and
3025 looped endlessly, respectively.) Thus, DIRECTORY now works even
3026 in the presence of dangling and cyclic symbolic links.
3027 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3028 is now a supported extension again, since the consensus on
3029 sbcl-devel was that it can be useful for ordinary development
3030 work, not just for debugging SBCL itself.
3031 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3032 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3033 that currently-defined functions might be redefined later with
3034 different return types.
3035 * Hash tables can be printed readably, as inspired by CMU CL code
3036 of Eric Marsden and SBCL code of Martin Atzmueller.
3037 * better error handling in CLOS method combination, thanks to
3038 Martin Atzmueller porting Pierre Mai's CMU CL patches
3039 * more overflow fixes for >16Mbyte I/O buffers
3040 * A bug in READ has been fixed, so that now a single Ctrl-D
3041 character suffices to cause end-of-file on character streams.
3042 In particular, now you only need one Ctrl-D at the command
3043 line (not two) to exit SBCL.
3044 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3046 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3047 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3048 does the right thing.
3049 * The implementation of some type tests, especially for CONDITION
3050 types, is now tidier and maybe faster, due to CMU CL code
3051 originally by Douglas Crosher, ported by Martin Atzmueller.
3052 * Some math functions have been fixed, and there are new
3053 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3054 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3055 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3056 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3057 likely to be built into sbcl-0.7.0.)
3058 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3059 CL, ported to SBCL by Martin Atzmueller)
3060 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3061 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3062 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3063 * There's a new slam.sh hack to shorten the edit/compile/debug
3064 cycle for low-level changes to SBCL itself, and a new
3065 :SB-AFTER-XC-CORE target feature to control the generation of
3066 the after-xc.core file needed by slam.sh.
3067 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3068 COMPILE-FILE is no longer supported, so that now every function
3069 gets an entry point, so that block compilation looks a little
3070 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3071 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3072 returns the number of bytes consed since the system started,
3073 rather than the number consed since the first time the function
3074 was called. (The new definition parallels ANSI functions like
3075 CL:GET-INTERNAL-RUN-TIME.)
3076 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3077 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3078 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3079 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3080 * Fasl file version numbers are now independent of the target CPU,
3081 since historically most system changes which required version
3082 number changes have affected all CPUs equally. Similarly,
3083 the byte fasl file version is now equal to the ordinary
3086 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3087 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3088 is no longer recognized. I apologize for this, because it was
3089 listed in SB-EXT as a supported extension, but I found that
3090 its existing behavior was poorly specified, as well as incorrectly
3091 specified, and it looked like too much of a mess to straighten it
3092 out. I have enough on my hands trying to get ANSI stuff to work..
3093 * many patches ported from CMU CL by Martin Atzmueller, with
3094 half a dozen bug fixes in pretty-printing and the debugger, and
3095 half a dozen others elsewhere
3096 * fixed bug 13: Floating point infinities are now supported again.
3097 They might still be a little bit flaky, but thanks to bug reports
3098 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3099 as flaky as they were.
3100 * The --noprogrammer command line option is now supported. (Its
3101 behavior is slightly different in detail from what the old man
3102 page claimed it would do, but it's still appropriate under the
3103 same circumstances that the man page talks about.)
3104 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3105 are now supported, and enabled by default. Thus, the compiler can
3106 handle many floating point and complex operations much less
3107 inefficiently. (Thus e.g. you can implement a complex FFT
3109 * The compiler now detects type mismatches between DECLAIM FTYPE
3110 and DEFUN better, and implements CHECK-TYPE more correctly, and
3111 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3112 thanks to patches from Martin Atzmueller.
3113 * various fixes to make the cross-compiler more portable to
3114 ANSI-conforming-but-different cross-compilation hosts (notably
3115 Lispworks for Windows, following bug reports from Arthur Lemmens)
3116 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3117 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3118 fixed thanks to Pierre Mai's CMU CL patches.
3119 * fixes to keep the system from overflowing internal counters when
3120 it tries to use i/o buffers larger than 16M bytes
3121 * fixed bug 45a: Various internal functions required to support
3122 complex special functions have been merged from CMU CL sources.
3123 (When I was first setting up SBCL, I misunderstood a compile-time
3124 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3125 * improved support for type intersection and union, fixing bug 12
3126 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3127 more obscure bugs as well
3128 * some steps toward byte-compiling non-performance-critical
3129 parts of the system, courtesy of patches from Martin Atzmueller
3130 * Christophe Rhodes has made some debian packages of sbcl at
3131 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3132 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3133 stable, but are nonetheless usable. When he's ready, I'd be happy
3134 to add them to the SourceForge "File Releases" section. (And if
3135 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3136 * new fasl file format version number (because of changes in
3137 internal representation of (OR ..) types to accommodate the new
3138 support for (AND ..) types, among other things)
3140 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3141 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3143 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3145 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3146 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3147 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3148 * DESCRIBE now works on structure objects again.
3149 * Most function call argument type mismatches are now handled as
3150 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3151 know whether the function will be redefined before the call is
3152 executed. (The compiler could flag local calls with full WARNINGs,
3153 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3154 it doesn't keep track of enough information to know whether calls
3155 are local in this sense.)
3156 * Compiler output is now more verbose, with messages truncated
3157 later than before. (There should be some supported way for users
3158 to override the default verbosity, but I haven't decided how to
3159 provide it yet, so this behavior is still controlled by the internal
3160 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3161 src/compiler/ir1util.lisp.)
3162 * Fasl file format version numbers have increased again, because
3163 support for the Gray streams extension changes the layout of the
3164 system's STREAM objects.
3165 * The Gray subclassable streams extension now works, thanks to a
3166 patch from Martin Atzmueller.
3167 * The full LOAD-FOREIGN extension (not just the primitive
3168 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3169 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3170 but like most other programs, it defaults to copying the Unix
3171 environment from the original process instead of starting the
3172 new process in an empty environment.
3173 * Extensions which manipulate the Unix environment now support
3174 an :ENVIRONMENT keyword option which doesn't smash case or
3175 do other bad things. The CMU-CL-style :ENV option is retained
3176 for porting convenience.
3177 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3178 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3180 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3182 * A patch from Martin Atzmueller seems to have solved the SIGINT
3183 problem, and as far as we know, signal-handling now works cleanly.
3184 (If you find any new bugs, please report them!)
3185 * The system no longer defaults Lisp source file names to types
3186 ".l", ".cl", or ".lsp", but only to ".lisp".
3187 * The compiler no longer uses special default file extensions for
3188 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3189 seems to expect a single default extension for all compiled code,
3190 and there's no compelling reason to try to stretch the standard
3191 to allow two different extensions.) Instead, byte-compiled files
3192 default to the same extension as native-compiled files.
3193 * Fasl file format version numbers have increased again, because
3194 a rearrangement of internal implementation packages made some
3195 dumped symbols in old fasl files unreadable in new cores.
3196 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3197 many fewer weird special cases.
3198 * Bug #17 (differing COMPILE-FILE behavior between logical and
3199 physical pathnames) has been fixed, and some related misbehavior too,
3200 thanks to a patch from Martin Atzmueller.
3201 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3202 by Tim Moore, ported to SBCL by Martin Atzmueller.
3203 * Martin Atzmueller fixed several filesystem-related problems,
3204 including bug #36, in part by porting CMU CL patches, which were
3205 written in part by Paul Werkowski.
3206 * More compiler warnings in src/runtime/ are gone, thanks to
3207 more patches from Martin Atzmueller.
3208 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3211 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3213 * DESCRIBE now works on CONDITION objects.
3214 * The debugger now handles errors which arise when trying to print
3215 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3217 * The build system now uses an additional file, customize-target-features.lisp,
3218 to allow local modifications to the target *FEATURES* list. (The point of
3219 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3220 debugging features enabled, without having to worry about propagating it
3221 into everyone's system when I do a "cvs update".) When no
3222 customize-target-features.lisp file exists, the target *FEATURES* list
3223 should be constructed the same way as before.
3224 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3225 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3226 warn about behavior which is undefined under the ANSI spec. Note: This
3227 is specified by ANSI, but it's not very popular with programmers.
3228 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3229 macro in the SBCL sources for an example of a workaround which you
3230 might use to make portable ANSI-standard code which does what you want.
3231 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3232 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3233 when it's used as a non-toplevel form. (This is required in order
3234 to implement the DEFCONSTANT-EQX macro.)
3235 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3236 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3237 funny things, due to bugs in the implementation of EVAL-WHEN
3238 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3239 non-toplevel DEFCONSTANTs are uncommon.)
3240 * The core file version number and fasl file version number have been
3241 incremented, because the old noncompliant DEFCONSTANT behavior involved
3242 calling functions which no longer exist, and because I also took the
3243 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3244 * fixed bug 1 (error handling before read-eval-print loop starts), and
3245 redid debugger restarts and related debugger commands somewhat while
3247 ** The QUIT debugger command is gone, since it did something
3248 rather different than the SB-EXT:QUIT command, and since it never
3249 worked properly outside the main toplevel read/eval/print loop.
3250 Invoking the new TOPLEVEL restart provides the same functionality.
3251 ** The GO debugger command is also gone, since you can just invoke
3252 the CONTINUE restart directly instead.
3253 ** The TOP debugger command is also gone, since it's redundant with the
3254 FRAME 0 command, and since it interfered with abbreviations for the
3256 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3258 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3259 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3260 some porting work by Martin Atzmueller)
3261 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3262 with a boatload of patches which he ported to SBCL. Now that those
3264 ** The system tries to make sure that its low-priority messages
3265 are prefixed by semicolons, to help people who like to use
3266 syntax highlighting in their ILISP buffer. (This patch
3267 was originally due to Raymond Toy.)
3268 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3269 patches originally written by Raymond Toy.
3270 ** The compiler understands coercion between single-value and
3271 multiple-VALUES type expressions better, getting rid of some very
3272 weird behavior, thanks to patches originally by Robert MacLachlan
3273 and Douglas Crosher.
3274 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3275 lambda lists, thanks to a patch originally by Pierre Mai.
3276 ** The system no longer bogusly warns about "abbreviated type
3278 ** The compiler gets less confused by inlining and RETURN-FROM,
3279 thanks to some patches originally by Tim Moore.
3280 ** The system no longer hangs when dumping circular lists to fasl
3281 files, thanks to a patch originally from Douglas Crosher.
3282 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3283 undefined function error.
3284 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3285 in structure slot initforms) for now, documented workaround instead:-|
3286 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3287 * fixed bug 5: added stubs for various Gray stream functions called
3288 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3289 installed, at least appropriate type errors are generated
3290 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3291 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3292 consistently in DEFMETHOD forms.
3293 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3294 it doesn't seem to affect SBCL after all
3295 * The C runtime system now builds with better optimization and many
3296 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3298 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3300 * The system is now under CVS at SourceForge (instead of the
3301 CVS repository on my home machine).
3302 * The new signal handling code has been tweaked to treat register
3303 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3304 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3305 downstream, e.g. in debug-int.lisp, has implicit dependencies
3306 on the unsignedness of integer representation of machine words,
3307 and that caused the system to bomb out with infinite regress
3308 when trying to recover from type errors involving signed values,
3309 e.g. (BUTLAST '(1 2 3) -1).)
3310 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3311 (This was one of the bugs Peter Van Eynde reported back in July.)
3312 * The system now uses code inspired by Colin Walters' O(N)
3313 implementation of MAP (from the cmucl-imp@cons.org mailing
3314 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3315 inline the MAP operation, and there is more than one
3316 sequence argument to the MAP call (so that it can't just
3317 do ETYPECASE once and for all based on the type of the
3318 single sequence argument). (The old non-inline implementation
3319 of the general M-argument sequence-of-length-N case required
3320 O(M*N*N) time when any of the sequence arguments were LISTs.)
3321 * The QUIT :UNIX-CODE keyword argument has been renamed to
3322 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3323 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3325 * Raymond Wiker's patches to port dynamic loading from Linux to
3326 FreeBSD have been added.
3327 * The BUGS file is now more nearly up to date, thanks in large part
3328 to Martin Atzmueller's review of it.
3329 * The debugger now flushes standard output streams before it begins
3330 its output ("debugger invoked" and so forth).
3331 * The core version number and fasl file version number have both
3332 been incremented, because of incompatible changes in the layout
3334 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3335 used to not be called for a saved Lisp image.)
3336 * Martin Atzmueller's version of a patch to fix a compiler crash,
3337 as posted on sbcl-devel 13 September 2000, has been installed.
3338 * Instead of installing Martin Atzmueller's patch for the
3339 compiler transform for SUBSEQ, I deleted the compiler transform,
3340 and transforms for some similar consing operations.
3341 * A bug in signal handling which kept TRACE from working on OpenBSD
3343 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3344 away by constant folding
3345 * The system now defines its address space constants in one place
3346 (in the Lisp sources), and propagates them automatically elsewhere
3347 (through GENESIS and the sbcl.h file). Therefore, patching the
3348 address map is less unnecessarily tedious and error-prone. The
3349 Lisp names of address space constants have also been systematized.
3350 * CVS tags like dollar-Header-dollar have been removed from
3351 the sources, because they have never saved me trouble and
3352 they've been source of trouble working with patches and other
3353 diff-related operations.
3354 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3355 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3357 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3359 * The system has been ported to OpenBSD.
3360 * The system now compiles with a simple "sh make.sh" on the systems
3361 that it's supported on. I.e., now you no longer need to tweak
3362 text in the target-features.lisp-expr and symlinks in src/runtime/
3363 by hand, the make.sh takes care of it for you.
3364 * The system is no longer so grossly inefficient when compiling code
3365 involving vectors implemented as general (not simple) vectors (VECTOR T),
3366 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3367 which dares to use the various sequence functions on non-simple
3368 vectors, takes less of a performance hit.
3369 * There is now a primitive type predicate VECTOR-T-P
3370 to test for the (VECTOR T) type, so that e.g.
3371 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3372 can now be compiled with some semblance of efficiency. (The old code
3373 turned the type declaration into a full call to %TYPEP at runtime!)
3374 * AREF on (VECTOR T) is still not fast, since it's still compiled
3375 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3376 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3377 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3378 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3379 without making full calls to SUBTYPEP at runtime.
3380 (Some analogous efficiency issues for non-simple vectors specialized to
3381 element types other than T, or for non-simple multidimensional arrays,
3382 have not been addressed. They could almost certainly be handled the
3383 same way if anyone is motivated to do so.)
3384 * The changes in array handling break binary compatibility, so
3385 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3386 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3387 instead of (VALUES T T).
3388 * By following the instructions that Dan Barlow posted to sbcl-devel
3389 on 2 July 2000, I was able to enable primitive dynamic object
3390 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3391 functionality is not implemented (since it calls ld to resolve
3392 library references automatically, requiring RUN-PROGRAM for its
3393 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3394 try to resolve library references) is now supported.
3395 * The system now flushes the standard output streams when it terminates,
3396 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3397 them at several other probably-convenient times, e.g. in each pass of
3398 the toplevel read-eval-print loop, and after evaluating a form given
3399 as an "--eval" command-line option. (These changes were motivated by a
3400 discussion of stream flushing issues on cmucl-imp in August 2000.)
3401 * The source transform for TYPEP of array types no longer assumes
3402 that an array whose element type is a not-yet-defined type
3403 is implemented as an array of T, but instead punts, so that the
3404 type will be interpreted at runtime.
3405 * There is now some support for cross-compiling in make.sh: each of
3406 the phases of make.sh has its own script. (This should be transparent
3407 to people doing ordinary, non-cross-compile builds.)
3408 * Since my laptop doesn't have hundreds of megabytes of memory like
3409 my desktop machine, I became more motivated to do some items on
3410 my to-do list in order to reduce the size of the system a little:
3411 ** Arrange for various needed-only-at-cold-init things to be
3412 uninterned after cold init. To support this, those things have
3413 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3414 symbols with such names are now uninterned after cold init).
3415 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3416 fasl files for cold load.
3417 ** Remove the old compiler structure pooling code (which used to
3418 be conditional on the target feature :SB-ALLOC) completely.
3419 ** Redo the representation of some data in cold init to be more compact.
3420 (I also looked into supporting byte compiled code at bootstrap time,
3421 which would probably reduce the size of the system a lot, but that
3422 looked too complicated, so I punted for now.)
3423 * The maximum signal nesting depth in the src/runtime/ support code has
3424 been reduced from 4096 to 256. (I don't know any reason for the very
3425 large old value. If the new smaller value turns out to break something,
3426 I'll probably just bump it back up.)
3427 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3429 * Many, many bugs reported by Peter Van Eynde have been added to
3430 the BUGS list; some have even been fixed.
3431 * While enabling dynamic object file loading, I tried to make the
3432 code easier to understand, renaming various functions and variables
3433 with less ambiguous names, and changing some function calling
3434 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3435 * While trying to figure out how to do the OpenBSD port, I tried to
3436 clean up some of the code in src/runtime/. In particular, I dropped
3437 support for non-POSIX signal handling, added various comments,
3438 tweaked the code to reduce the number of compilation warnings, and
3439 renamed some files to increase consistency.
3440 * To support the new automatic configuration functionality in make.sh,
3441 the source file target-features.lisp-expr has been replaced with the
3442 source file base-target-features.lisp-expr and the machine-generated
3443 file local-target-features.lisp-expr.
3444 * fixed a stupid quoting error in make.sh so that using CMU CL
3445 "lisp -batch" as cross-compilation host works again
3447 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3449 * DESCRIBE no longer tries to call itself recursively to describe
3450 bound/fbound values, so that it no longer fails on symbols which are
3451 bound to themselves (like keywords, T, and NIL).
3452 * DESCRIBE now works on generic functions.
3453 * The printer now prints less-screwed-up representations of closures
3454 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3455 * A private symbol is used instead of the :EMPTY keyword previously
3456 used to mark empty slots in hash tables. Thus
3457 (DEFVAR *HT* (MAKE-HASH-TABLE))
3458 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3459 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3460 now does what ANSI says that it should. (You can still get
3461 similar noncompliant behavior if bang on the hash table
3462 implementation with all the symbols you get back from
3463 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3464 This breaks binary compatibility, since tests for equality to
3465 :EMPTY are wired into things like the macroexpansion of
3466 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3468 * There's now a minimal placeholder implementation for CL:STEP,
3469 as required by ANSI.
3470 * An obscure bug in the interaction of the normal compiler, the byte
3471 compiler, inlining, and structure predicates has been patched
3472 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3473 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3475 * Missing ordinary arguments in a macro call are now detected even
3476 when the macro lambda list contains &KEY or &REST.
3477 * The debugger no longer complains about encountering the top of the
3478 stack when you type "FRAME 0" to explicitly instruct it to go to
3479 the top of the stack. And it now prints the frame you request even
3480 if it's the current frame (instead of saying "You are here.").
3481 * As specified by ANSI, the system now always prints keywords
3482 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3484 * The default initial SIZE of HASH-TABLEs is now smaller.
3485 * Type information from CLOS class dispatch is now propagated
3486 into DEFMETHOD bodies, so that e.g.
3487 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3489 is now basically equivalent to
3490 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3491 (DECLARE (TYPE SINGLE-FLOAT X))
3493 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3494 operation, without having to do run-time type dispatch.
3495 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3496 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3497 * Since I don't seem to be making big file reorganizations very often
3498 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3499 is overflowing my ability to conveniently back them up), I've finally
3500 checked the system into CVS. (The CVS repository is on my home system,
3501 not at SourceForge -- putting it on SourceForge might come later.)
3502 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3503 high-level GC-NOTIFY-FOO functions send their output. (There's
3504 still very little control of where low-level verbose GC functions
3505 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3506 less than it used to -- the GC-NOTIFY-FOO functions are now under
3507 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3508 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3509 in only one place in the source code, and propagates it automatically
3510 everywhere that it's needed. Thus e.g. when I bump the version from
3511 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3512 * The C source files now include boilerplate legalese and documentation
3513 at the head of each file (just as the Lisp source files already did).
3514 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3515 to his page will be replaced with a link to his new CLiki service.
3517 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3519 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3520 * The build process now looks for GNU make under the default name "gmake",
3521 instead of "make" as it used to. If GNU make is not available as "gmake"
3522 on your system, you can change this default behavior by setting the
3523 GNUMAKE environment variable.
3524 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3525 can build without error under CMU CL.
3527 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3529 * There is now a partial SBCL user manual (with some new text and some
3530 text cribbed from the CMU CL manual).
3531 * The beginnings of a profiler have been added (starting with the
3532 CMU CL profiler and simplifying and cleaning up). Eventually the
3533 main interface should be through the TRACE macro, but for now,
3534 it's still accessed through vaguely CMU-CL-style functions and macros
3535 exported from the package SB-PROFILE.
3536 * Some problems left over from porting CMU CL to the new
3537 cross-compilation bootstrap process have been cleaned up:
3538 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3539 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3540 part when getting rid of the compiler *BACKEND* stuff.)
3541 ** The value of *NULL-TYPE* was screwed up, because it was
3542 being initialized before the type system knew the final
3543 definition of the 'NULL type. This screwed up several key
3544 optimizations in the compiler, causing inefficiency in all sorts
3545 of places. (I found it because I wanted to understand why
3546 GET-INTERNAL-RUN-TIME was consing.)
3547 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3548 PROCLAIM FTYPE information. Unfortunately this broke binary
3549 compatibility again, since now the forms output by DEFGENERIC
3550 to refer to functions which didn't exist in 0.6.3.
3551 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3552 can use the new (as of 0.6.3) transform for SOME into MAP into
3554 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3555 values to (INTEGER 0 1000000), so that the time code will no longer
3556 occasionally get blown up by Linux returning 1000000 microseconds
3557 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3558 its output conform to the ANSI spec. (Alas, this makes its output
3559 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3560 * A full call to MAP NIL with a single sequence argument no longer conses.
3561 * fixes to problems pointed out by Martin Atzmueller:
3562 * The manual page no longer talks about multiprocessing as though
3563 it were currently supported.
3564 * The ILISP support patches have been removed from the distribution,
3565 because as of version 5.10.1, ILISP now supports SBCL without us
3566 having to maintain patches.
3567 * added a modified version of Raymond Toy's recent CMU CL patch for
3568 EQUALP comparison of HASH-TABLE
3570 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3572 * The system still can't cross-compile itself with
3573 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3574 entails), but at least it can compile more of itself that way
3575 than it used to be able to, and various buglets which were uncovered
3576 by trying to cross-compile itself that way have now been fixed.
3577 * This release breaks binary compatibility again. This time
3578 at least I've incremented the FASL file format version to 2, so that the
3579 problem can be detected reliably instead of just causing weird errors.
3580 * various new style warnings:
3581 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3582 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3583 ** using the lexical binding of a variable named in the *FOO* style
3584 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3585 as specified by ANSI.
3586 * *RANDOM-STATE* is no longer automatically initialized from
3587 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3588 default behavior of the system is to repeat its behavior every time
3589 it's run. If you'd like to change this behavior, you can always
3590 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3591 old convention there was no comparably easy way to get the system to
3592 repeat its behavior every time it was run.
3593 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3594 FTYPE declarations has been removed, in favor of their ANSI
3595 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3596 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3597 the types of their sequence arguments can be determined at compile time.
3598 This is done through a new open code expansion for MAP which eliminates
3599 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3600 argument types can be determined at compile time.
3601 * The optimizer now transforms COERCE into an identity operation when it
3602 can prove that the coerced object is already of the correct type. (This
3603 can be a win for machine generated code, including the output of other
3604 optimization transforms, such as the MAP transform above.)
3605 * Credit information has been moved from source file headers into CREDITS.
3606 * Source file headers have been made more standard.
3607 * The CASE macro now compiles without complaining even when it has
3610 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3612 * (Note that the way that the PCL macroexpansions were rewritten
3613 to accommodate the change in DEFGENERIC below breaks binary
3614 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3615 not run under sbcl-0.6.2. Once we get out of alpha releases,
3616 i.e. hit release 1.0.0, we'll probably try to maintain binary
3617 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3618 and sbcl-1.4.4. Until then, however, it might be fairly common
3619 for maintenance releases to break binary compatibility.)
3620 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3622 * The reporting of unhandled signals has been changed to print some
3623 explanatory text as well as the report form. (Previously only
3624 the report form was printed.)
3625 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3626 it defines, so that the compiler no longer issues undefined function
3627 warnings for compiled-but-not-yet-loaded generic functions.
3628 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3629 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3630 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3631 section of the ANSI spec are supported.
3632 * The "" nickname for the "KEYWORD" package has been removed.
3633 The reader still handles symbol tokens which begin with a package marker
3634 as keywords, but it doesn't expose its mechanism for doing so in the
3635 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3636 * The system now issues STYLE-WARNINGs for contradictory TYPE
3637 proclamations. (Warnings for contradictory FTYPE proclamations would
3638 be nice too, but those can't be done usefully unless the type system
3639 is made smarter about FUNCTION types.)
3640 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3641 systematized, so that "*target-*.lisp is supposed to exist only on the
3642 target and imply that there's a related file which exists on the
3643 host, and *host-*.lisp is supposed to exist only on the host and imply
3644 that there's a related file which exists on the target. This involves a
3645 lot of renaming. Hopefully the acute confusion caused by the renaming
3646 will be justified by the reduction in chronic confusion..
3647 ** runtime-type.lisp -> early-target-type.lisp
3648 ** target-type.lisp -> late-target-type.lisp
3649 ** early-host-format.lisp -> early-format.lisp
3650 ** late-host-format.lisp -> late-format.lisp
3651 ** host-error.lisp -> misc-error.lisp
3652 ** early-error.lisp -> early-target-error.lisp
3653 ** late-error.lisp -> late-target-error.lisp
3654 ** host-defboot.lisp -> early-defboot.lisp
3655 ** code/misc.lisp -> code/target-misc.lisp
3656 ** code/host-misc.lisp -> code/misc.lisp
3657 ** code/numbers.lisp -> code/target-numbers.lisp
3658 ** code/early-numbers.lisp -> numbers.lisp
3659 ** early-host-type.lisp -> early-type.lisp
3660 ** late-host-type.lisp -> late-type.lisp
3661 ** host-typep.lisp -> typep.lisp
3662 ** load.lisp -> target-load.lisp
3663 ** host-load.lisp -> load.lisp
3664 ** host-disassem.lisp -> disassem.lisp
3665 ** host-insts.lisp -> insts.lisp
3666 ** byte-comp.lisp -> target-byte-comp.lisp
3667 ** host-byte-comp.lisp -> byte-comp.lisp
3668 ** host-signal.lisp -> signal.lisp
3669 ** host-defstruct.lisp -> defstruct.lisp
3670 ** late-target-type.lisp -> deftypes-for-target.lisp
3671 Furthermore, several other previously target-only files foo.lisp (e.g.
3672 hash-table.lisp and random.lisp) have been split into a target-and-host
3673 foo.lisp file and a target-only target-foo.lisp file, with their key type
3674 definitions in the target-and-host part, so that the cross-compiler will
3675 know more about target types.
3676 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3677 gone away. In their place are various *BACKEND-FOO* variables
3678 corresponding to the slots of the old structure.
3679 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3680 into the target SBCL has been fixed.
3681 * The chill.lisp system for loading cold code into a running SBCL
3683 * Support for the CMU CL "scavenger hook" extension has been removed.
3684 (It was undocumented and unused in the CMU CL sources that SBCL was
3685 derived from, and stale in sbcl-0.6.1.)
3686 * Various errors in the cross-compiler type system were detected
3687 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3688 (enabling various consistency checks). Many of them were fixed,
3689 but some hard problems remain, so the compiler is back to
3690 running without *TYPE-SYSTEM-INITIALIZED* for now.
3691 * As part of the cross-compiler type system cleanup, I implemented
3692 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3693 * I have started adding UNCROSS calls throughout the type system
3694 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3695 on cross-compiler input files will be able to go away, and various
3697 * CONSTANTP now returns true for quoted forms (as explicitly required
3700 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3702 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3703 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3704 (MAKE-LIST -1). In the longer term, it ought to become true
3705 that declarations are assertions even at SAFETY 1. For now, it's not
3706 quite true even at SAFETY 3, but it's at least more nearly true..
3707 (Note that this change seems to increases the size of the system by
3708 O(5%) and to decrease the speed of the compiler by 20% or more.)
3709 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3710 for the problem of printing dozens of lines of distracting information
3711 about low-level system machinery as part of the top stack frame
3712 on entry to the debugger when an undefined function was called.
3713 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3714 is not reset to COMMON-LISP-USER.
3715 * Compilation of stuff related to dyncount.lisp has been made conditional
3716 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3717 smaller. The various dyncount-related symbols have been moved into
3718 a new "SB-DYNCOUNT" package.
3719 * tty-inspect.lisp has been renamed to inspect.lisp.
3720 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3721 feature has gone away. (When we eventually port to other flavors of
3722 libc and/or Unix, we'll try to make the differences between flavors
3723 invisible at the user level.)
3724 * Various other *FEATURES* tags, and/or their associated conditionals,
3725 have been removed if obsolescent, or given better documentation, or
3726 sometimes given more-mnemonic names.
3728 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3730 * tidied up "make.sh" script
3731 * tidied up system directory structure
3732 * better "clean.sh" behavior
3733 * added doc/FOR-CMUCL-DEVELOPERS
3734 * many many small tweaks to output format, e.g. removing possibly-confusing
3735 trailing #\. character in DESCRIBE-INSTANCE
3736 * (EQUALP #\A 'A) no longer signals an error.
3737 * new hashing code, including EQUALP hashing
3738 * tidied up Lisp initialization and toplevel
3739 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3740 * command line argument processing
3741 * added POSIX-GETENV function to deal with Unix-ish environment variables
3742 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3743 terminating SBCL on EOF
3744 * non-verbose GC by default
3745 * There is no more "sbcl" shell script; the sbcl file is now the C
3746 runtime executable (just like CMU CL).
3747 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3749 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3750 symbol and package information as Lisp native symbol and package objects
3751 instead of strings naming symbols and strings naming packages. This way,
3752 whenever packages are renamed (as in warm init), debug information is
3753 transformed along with everything else.
3754 * tweaked the optimization policy declarations which control the building
3755 of SBCL itself. Now, among other things, the system no longer saves
3756 source location debugging information. (This helps two problems at once
3757 by reducing SBCL size and by keeping SBCL from trying to look for its
3758 sources -- which may not exist -- when reporting errors.)
3759 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3760 debugging and testing purposes
3761 * cleaned up printing, making the printer call PRINT-OBJECT for
3762 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3763 methods, giving nearly-ANSI behavior
3764 * converted almost all special variables to use *FOO* naming convention
3765 * deleted PARSE-TIME functionality, since it can be done portably
3766 * moved some files out of cold init into warm init
3767 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3769 * regularized formatting of source files
3770 * added an install.sh script
3771 * fixed ridiculous memory usage of cross-compiler by making
3772 compiler/alloc.lisp not try to do pooling unless it can hook
3773 itself into the GC of the cross-compilation host. Now the system
3774 builds nicely on my old laptop.
3775 * added :SB-ALLOC in target-features.lisp-expr
3776 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3777 was not implemented)
3778 * re-did condition handling and note reporting in the compiler. Notes
3779 are no longer handled by signalling conditions. Style warnings
3780 and warnings are handled more correctly and reported in such a way
3781 that it's easy to find one or the other in your output (so that you
3782 can e.g. figure out which of many problems caused COMPILE-FILE to
3784 * changed the severity of several compiler warnings from full WARNING
3785 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3786 compiler note reporting so that it doesn't use the condition system
3787 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3788 COMPILE-FILE command)
3789 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3790 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3791 no longer supported, since their implementation was deeply intertwingled
3792 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3793 * removed bogus "support" for compiler macros named (SETF FOO), and
3794 removed the compiler macro for SETF INFO (but only after making a fool
3795 of myself on the cmucl-imp mailing list by posting a bogus patch for
3796 DEFINE-COMPILER-MACRO..)
3797 * Compiled files containing forms which have side effects on the Lisp
3798 reader (such as DEFPACKAGE forms) are now handled more correctly.
3799 (Compiler queuing of top level lambdas has been suppressed by setting
3800 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3801 * deleted various currently-unused source files, e.g. gengc.lisp. They
3802 may be added back at some point e.g. when porting to other architectures,
3803 but until they are it's distracting to distribute them and to try to
3805 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3806 there were so many of them they're just distractions, and UNCROSS is
3807 known to be able to handle the current sources
3808 * moved PROFILE functionality into TRACE, so that it will be clear
3809 how the wrapping and unwrapping of functions when you profile them
3810 interacts with the wrapping and unwrapping of functions when you
3811 trace them. (Actually, the functionality isn't there yet, but at least
3812 the interface specification is there. Hopefully, the functionality will
3813 arrive with some future maintenance release.)
3814 * removed host-oops.lisp
3815 * changed signature of QUIT function to allow UNIX-CODE argument
3816 * fixed READ-SEQUENCE bug
3817 * tweaked verbose GC output so that it looks more like the progress
3818 output that ANSI specifies for functions like LOAD
3819 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3820 * added <http://sbcl.sourceforge.com> to the banner information printed by