1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
3 * feature: thread support on Solaris/x86, and experimental thread support
5 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
7 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
8 default method for SLOT-UNBOUND.
9 * minor incompatible change: prevent the user from specializing the
10 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
11 somewhat counter to the intent of the protocol, I (CSR) think, and
12 additionally it just doesn't work in SBCL as currently
13 implemented, thanks to optimizations (that are always valid for
14 the other three SLOT-VALUEish functions, but not for the setter).
15 * minor incompatible change: the :SB-LDB feature is now enabled by
16 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
17 the low-level debugger.
18 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
19 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
20 down to an SB-SYS:FD-STREAM, allowing direct communication in
21 more cases. Temporary files and pipes are still used as a fallback
23 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
24 on the non-reentrant C functions localtime(3) and gmtime(3).
25 * bug fix: global optimization policy was not visible in LOCALLY and
27 * bug fix: class objects can be used as specializers in methods.
28 (reported by Pascal Costanza)
29 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
30 with a type error. (reported by blitz_ on #lisp)
31 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
32 failed with a type error. (reported by Pascal Bourguignon)
33 * bug fix: merging pathnames against defaults with :DIRECTORY
34 starting with '(:RELATIVE :BACK) should preserve the :BACK.
35 (reported by James Y Knight)
36 * bug fix: saving large (>2GB) cores on x86-64 now works
37 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
38 constant index and a (simple-array (signed-byte 32)) array
39 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
40 error (patch by Robert J. Macomber)
41 * feature: TIME now displays the amount of run-time spent in GC
42 * bug fix: The debugger now does a better job of respecting
43 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
44 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
45 workaround for bug 403.)
46 * bug fix: floating point exception handling now works on FreeBSD
47 (thanks to NIIMI Satoshi)
48 * fixed some bugs revealed by Paul Dietz' test suite:
49 ** MISC.641: LET-conversion were not supposed to work in late
51 * improvements to the Win32/x86 port:
52 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
53 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
55 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
56 * new feature: source path information is generated for macro-expansion
57 errors for use in IDE's like Slime (thanks to Helmut Eller)
58 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
59 * bug fix: compiler does not loop forever on an invalid type in
61 * improvement: compilation of most CLOS applications is significantly
63 * optimization: added a limited bytecode compiler for simple toplevel
64 forms, speeding up compilation and FASL loading
66 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
67 * minor incompatible change: in sbcl-0.9.11 (but not earlier
68 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
69 into subclasses of the finalized class. Now user calls to
70 FINALIZE-INHERITANCE finalize just the one class, and calls by the
71 system return before any subclasses are finalized.
72 * minor incompatible change: The reader no longer ignores errors
73 regarding non-existent packages in #+ and #- feature tests.
74 * new feature: command line options --no-sysinit, --no-userinit to
75 inhibit loading the corresponding init files
76 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
77 for sending data through UDP sockets (thanks to François-René Rideau)
78 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
79 error is signalled instead (thanks to François-René Rideau)
80 * minor incompatible change: Error signalling behaviour of lexical
81 operations violating package locks has changed slightly. Refer to
82 documentation on package locks for details.
83 * enhancement: EVAL can process IF-expressions without resorting to the
85 * bug fix: Some operations involving SIN, COS, or TAN weren't being
86 constant-folded properly due to missing out-of-line functions.
87 (reported by Mika Pihlajamäki)
88 * bug fix: LISTEN sometimes returned T even in cases where no data was
89 immediately available from the stream
90 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
91 were reversed. (reported by Levente Mészáros)
92 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
93 when given a vector displaced to an adjustable array. (reported
95 * bug fix: derivation of float boundaries from numbers outside the
96 appropriate float range (reported by John Wiseman)
97 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
99 * bug fix: precomputing cache entries for generic functions with
100 some subclasses of specializers as yet invalid does not attempt to
101 fill a cache line with a negative offset. (reported by Levente
103 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
104 allow condition class objects as arguments to DOCUMENTATION and
105 (SETF DOCUMENTATION); only find and set documentation for
106 structure names for the STRUCTURE doc-type. (suggested by Gary
108 * improvements to the Win32/x86 port:
109 ** (user-homedir-pathname) and default initialization file
110 locations now know about the user's "Documents and Settings"
111 directory (thanks to Yaroslav Kavenchuk)
112 ** run-program is implemented (thanks to Mike Thomas)
113 ** sockets support (thanks to Timothy Ritchey)
114 ** better backtrace support (thanks to Alastair Bridgewater)
115 ** sb-grovel supported
116 ** asdf-install and sb-posix work somewhat
117 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
118 * improvements to the Solaris/x86 port:
119 ** works on Solaris 11/Solaris Express
120 ** floating-point exception handling support
121 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
122 * fixed some bugs revealed by Paul Dietz' test suite:
123 ** REMOVE-METHOD returns its generic function argument even when
124 the method is not one of the generic functions' methods.
125 ** objects declared as MEMBER types can be admissible arguments to
127 ** printing characters should simply be printed by the FORMAT ~:C
129 ** compiler failure when compiling functions with hairy constant
130 defaults for optional parameters.
131 ** compiler produces wrong code when MAYBE-INLINE-expanding a
132 function, which is already optimized.
134 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
135 * new platform: experimental support for SBCL x86/Darwin, including
136 MacOS X 10.4.5 on Intel.
137 * new feature: Unicode character names are now known to the system
138 (through CHAR-NAME and NAME-CHAR).
139 * new feature: the filesystem location of SBCL's core file is
140 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
141 * minor incompatible change: the contrib modules SB-POSIX and
142 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
143 this change is to make it easier to distribute
144 location-independent binaries.
145 * bug fix: as implied by AMOP, standardized classes no longer have
146 slots named by external symbols of public packages. (reported by
148 * optimization: calling functions via constant symbols -- as in
149 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
150 via the function object as in (FUNCALL #'FOO).
151 * optimization: on x86 and x86-64, the calling convention for the
152 case of dealing with an unknown number of values has been altered
153 to be friendlier to the prediction heuristics implemented,
154 particularly with reference to CALL and RET pairing. (thanks to
155 Alastair Bridgewater)
156 * optimization: on x86, the code for access to thread-local symbol
157 values and binding/unbinding of thread-local symbols is smaller.
158 (thanks to Alastair Bridgewater)
159 * enhancement: CONSTANTP is now able to determine constantness of
160 more complex forms, including calls to constant-foldable standardized
161 functions and some special forms beyond QUOTE.
162 * fixed bug: occasional GC crashes on Solaris/x86
163 * optimization: x86-64 supports stack allocation of results of simple
164 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
165 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
166 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
167 (thanks to James Knight)
168 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
169 by the standard (thanks to Stephen Compall)
171 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
172 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
173 be used for bundling the runtime and the core file into one
174 executable binary. This feature is not currently supported on all SBCL
175 platforms. (thanks to James Bielman and NIIMI Satoshi)
176 * new feature: a generational or ephemeral garbage collector is now
177 the default on the PowerPC platform (both Linux and Darwin). The
178 old Cheney (stop and copy) collector is a build-time option.
179 (thanks to Cyrus Harmon, after Raymond Toy)
180 * minor incompatible change: the method by which SBCL finds its
181 contributed modules has changed; it no longer relies on symbolic
182 links from an $SBCL_HOME/systems directory, but searches directly
183 in the subdirectories of $SBCL_HOME.
184 * enhancement: the dynamic heap size on the Linux/PPC platform is
185 markedly larger, even using the older Cheney garbage collector.
186 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
187 certain complicated string types. (reported by Gary King)
188 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
189 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
190 * fixed bug: DOCUMENTATION on structures no longer signals an
191 error if no structure type of the right name exists. (reported by
193 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
194 applicable non-standard primary or :AROUND method on
195 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
196 no longer cause unbound variable errors. (reported by Kevin Reid)
197 * fixed bug: metacircle resolution in cases where methods have slots
198 added before the slots from STANDARD-METHOD. (reported by Jean
200 * fixed bug: the Power PC assembly code for calling into lisp
201 sometimes computed its return address wrongly (depending on the
202 15th bit of the address, fixed at link time). This used to
203 manifest itself in a segmentation violation while building PCL.
204 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
205 many others over the years)
206 * fixed bug: ROOM no longer reports silly numbers for stack usage if
207 the stack pointer (treated as a signed integer) is negative.
208 (thanks to Peter van Eynde)
210 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
211 * new platform: experimental support for the Windows operating
212 system has been added. (thanks to Alastair Bridgewater)
213 * fixed several bugs in and robustified the PPC FFI (including
214 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
215 * bug fix: allow non-simple string symbol names (reported by Paul
217 * bug fix: interrupt handling on NetBSD (thanks to Richard M
219 * bug fix: saving a core corrupted callbacks on x86/x86-64
220 * bug fix: closed a loophole in metacircularity detection and
221 grounding in the PCL implementation of CLOS.
222 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
223 (reported by tomppa on #lisp)
224 * bug fix: clear the direction flag on Lisp -> C transitions, as
225 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
226 (reported by Andrej Grozin and Hendrik Maryns)
227 * optimization: major improvements to GC efficiency on GENCGC platforms
228 * optimization: faster implementation of EQUAL
229 * optimization: emit more efficient opcodes for some common
230 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
231 * optimization: several other minor code-generation improvements on x86-64
232 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
234 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
235 * minor incompatible change: (SETF CLASS-NAME) and (SETF
236 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
237 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
238 by ANSI as a generic function, and for consistency (SETF
239 GENERIC-FUNCTION-NAME) is treated likewise.
240 * fixed bug #233.b: make constraint propagation notice when a variable
241 value is changed after it is referenced but before it is used
242 * fixed bug #296: no more arbitrary behaviour from filenames with
243 odd characters as --load arguments. (reported by Adam Warner)
244 * fixed bug #390: :CHARACTER-SET pathname components now work as
245 expected. (reported by Tim Daly Jr)
246 * fixed bug #391: complicated :TYPE intersections in slot
247 definitions no longer cause an error in PCL internals.
248 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
249 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
250 returning the number of octets which would be written to the
251 file-stream. (thanks to Robert J. Macomber)
252 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
253 arguments to the internal stream functions. (thanks to David
255 * bug fix: stack exhaustion is now handled on the x86/SunOS
256 platform. (thanks to Daisuke Homma)
257 * bug fix: more accurate ROOM results on GENCGC platforms
258 * optimization: improved type inference for arithmetic-for
259 index variables in LOOP
260 * optimization: faster floating-point SQRT on x86-64
261 * fixed some bugs revealed by Paul Dietz' test suite:
262 ** DOCUMENTATION returns NIL instead of "" for method combinations
263 that don't have a docstring
265 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
266 * minor incompatible change: (SETF CLASS-NAME) and (SETF
267 GENERIC-FUNCTION-NAME) are no longer generic functions, and
268 therefore are not customizeable by user code (as seems to be at
269 least permitted and maybe required by AMOP). As a consolation,
270 however, the SBCL implementation of these functions now calls
271 REINITIALIZE-INSTANCE as specified by AMOP.
272 * bug fix: slot-definition documentation is propagated as per ANSI
273 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
274 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
275 argument, as do eslotd initialization methods. (from Pascal
276 Costanza's "Closer" project)
277 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
278 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
280 * bug fix: it is now possible to have more than one subclass of
281 STANDARD-GENERIC-FUNCTION without causing stack overflow.
282 (reported by Bruno Haible, Pascal Costanza and others)
283 * bug fix: the dependent update protocol now works for generic
284 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
286 * bug fix: condition-class instances corresponding to
287 DEFINE-CONDITION forms are now created eagerly. (reported by
288 Kalle Olavi Niemitalo on comp.lang.lisp)
289 * bug fix: floating point printing is more accurate in some
290 circumstances. (thanks to Simon Alexander)
291 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
292 merged with *DEFAULT-PATHNAME-DEFAULTS*.
293 * bug fix: callbacks on OS X now preserve stack-alignment.
294 (thanks to Cyrus Harmon)
295 * enhancement: the x86-64 disassembler is much better at
296 disassembling SSE instructions. (thanks to Lutz Euler)
297 * enhancement: CHECK-TYPE on a variable now allows type-inference
298 (thanks to Vincent Arkesteijn)
299 * optimization: improved performance of EUC-JP external format.
300 (thanks to NIIMI Satoshi)
301 * optimization: performance improvements to IO on file streams of
302 :ELEMENT-TYPE CHARACTER
303 * optimization: much faster memory allocation on x86-64
304 * optimization: faster garbage collections (latency and throughput)
306 * optimization: faster bignum subtraction and division on x86 and x86-64
307 (thanks to Lutz Euler)
308 * optimization: more accurate type-inference for loops that use a
309 floating point index variable or a negative step.
311 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
312 * bug fix: add a workaround to SBCL looping infinitely at startup on
313 Linux kernels with apparently buggy implementations of personality().
314 (thanks to Svein Ove Aas)
315 * bug fix: Unicode symbols are correctly printed in LDB backtraces
316 (thanks to David Lichteblau)
317 * bug fix: local bindings shadow global symbol macros for
318 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
319 * optimization: non-open coded uses of numeric comparison operators
320 (e.g. >) no longer cons when called with more than one parameter
321 on platforms supporting dynamic-extent allocation.
322 * enhancement: saving cores with foreign code loaded is now supported
323 on MIPS/Linux in addition to the previously supported platforms.
324 * bug fix: FIND-RESTART now tests for activity, not applicability when given
325 a restart object as identifier. (reported by Helmut Eller for CMUCL)
326 * bug fix: division by zero in sb-sprof when no samples were collected
327 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
328 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
329 (thanks to Alexander Kjeldaas)
330 * new feature: ignoring the return values of destructive functions
331 like NREVERSE signals a compile-time style-warning.
332 (thanks to Kevin Reid)
333 * new feature: the alignment of alien structure fields can be
334 explicitly specified. (thanks to Cyrus Harmon)
335 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
336 and dump core on SIGQUIT
338 ** incompatible change: threads do not inherit values of specials
339 from their parents (see manual)
340 ** bug fix: threads stacks belonging to dead threads are freed by the
341 next exiting thread, no need to gc to collect thread stacks anymore
342 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
343 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
344 ** bug fix: dynamic variable and thread start related gc lossage
345 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
347 * fixed some bugs revealed by Paul Dietz' test suite:
348 ** SUBTYPEP is slightly more accurate on heinously complicated
349 CONS types where some of the members have uncertain (in the
350 NIL, NIL sense) type relationships to each other.
352 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
353 no longer purified unless :PURIFY T is explicitly specified.
354 ** Non-purified cores are significantly smaller than before
356 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
357 * new feature: timers based on Zach Beane's excellent timer package
358 * added support for the following external formats: koi8-u,
359 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
360 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
361 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
362 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
363 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
364 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
365 * incompatible change: a threaded SBCL will no longer revert to
366 non-threaded mode on non-NPTL systems, but refuse to start entirely.
367 * bug fix: interrupts are disabled until startup is complete; no
368 more sigsegvs when receiving a signal to soon
369 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
371 * bug fix: add a workaround for the memory randomization features in
372 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
373 workaround will only be in effect on systems with the proc filesystem
375 * bug fix: printing objects of type HASH-TABLE signals a
376 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
378 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
379 have been running for over 50 days. (reported by Gilbert Baumann)
380 * bug fix: the logic for getting names of functions gets less
381 confused when confronded with alternate-metaclass
382 funcallable-instances. (reported by Cyrus Harmon)
383 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
384 consistent, even on internal alternate-metaclass objects.
385 * bug fix: flush closure information collected by physical
386 environment analysis prepass before the main pass. (bug reported
388 * bug fix: compiler pack phase does not modify a hashtable, which is
389 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
390 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
391 may be expanded. (reported by Alexander <malishev> on c.l.l)
392 * bug fix: no more style-warnings when compiling code using LAST
393 under high-SPEED low-SPACE settings. (reported by David Wallin
395 * fixed some incompatibilities between SBCL's MOP and the MOP
397 ** the METAOBJECT class is now implemented;
398 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
399 STANDARD-OBJECT, as required;
400 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
401 now compatible (as required by VALIDATE-SUPERCLASS); there
402 remains a consistency requirement over the presence or absence
403 of the FUNCTION class in the superclass list of finalized
404 classes; see the manual for more details;
405 ** the :AROUND method for COMPUTE-SLOTS on
406 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
407 requested slot ordering.
409 ** bug fix: parent thread now can be gc'ed even with a live
411 ** bug fix: binding a special with PROGV to no value is not
412 the same as that symbol not having been bound (thanks to
414 ** bug fix: binding specials is thread safe (thanks to
416 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
417 and SB-SPROF do not die with 'no handler for signal XX in
418 interrupt_handle_now(..)' anymore
419 ** bug fix: WITH-TIMEOUT works with multiple running threads
420 * fixed some bugs revealed by Paul Dietz' test suite:
421 ** ENSURE-GENERIC-FUNCTION should take a method class object for
422 the :method-class keyword argument.
424 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
425 * new port: the Solaris operating system on x86 processors is now
426 mostly supported, though some rough edges in the environment
427 remain. (thanks to Daisuke Homma)
428 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
429 space for its heap. (thanks to Thiemo Seufer)
430 * optimization: SBCL on MIPS platforms now supports dynamic-extent
431 closures. (thanks to Thiemo Seufer)
432 * minor incompatible change: eof selects abort in the debugger.
433 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
434 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
436 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
437 deprecated, and will go away in a future revision of SBCL.
438 * minor incompatible change: GC-ON and GC-OFF are no longer
439 implemented with a counter, it does not matter how many times gc
440 is switched on or off
441 * bug fix: discriminating functions for generic function classes
442 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
443 make invalid assumptions about method precedence order. (reported
445 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
446 platforms. (thanks to Thiemo Seufer)
447 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
448 * bug fix: correct pathname printing with printer escaping is on.
449 (thanks to Kevin Reid)
450 * bug fix: complex VOP definitions in "user-space" no longer trigger
451 package locks. (reported by Zach Beane)
452 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
453 for funcallable-instance functions, and is no different from
455 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
456 characters in the current readtable. (reported by Nicholas Neuss)
457 * bug fix: SUBTYPEP on various CONS types returns more a more
458 accurate acknowledgment of its certainty.
459 * optimizations: REMOVE-DUPLICATES now runs in linear time on
460 lists in some cases. This partially fixes bug 384.
461 * flush all standard streams before prompting in the REPL and the
463 * bug fix: signal handling and triggering gc do not conflict
464 directly anymore, in particular a high frequency sb-sprof does
465 not prevent gc from running
466 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
467 approximation for timezone and DST information between the
468 universal time epoch and the smallest negative 32-bit time_t.
469 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
470 year 1899 so long as the encoded time is non-negative.
471 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
472 the correct answer for non-constant rotate amounts when performing
473 an inline 32-bit rotation.
475 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
476 there is only one thread in the session
477 ** bug fix: memory leak for streams created in one thread and
478 written to in another
479 ** bug fix: lockup when compiled with gcc4
480 ** bug fix: race that allows the gc to be triggered when gc is
482 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
483 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
485 ** bug fix: finished threads are gc'ed properly
486 * fixed some bugs revealed by Paul Dietz' test suite:
487 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
488 has arguments with a different set of applicable methods from
489 the orignal arguments.
490 ** The type error thrown by MAP now has a correct expected-type
492 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
493 name a compiled function.
494 ** Three MISC tests where a large bignum was improperly coerced to
495 a float that couldn't represent that bignum during type
496 derivation were fixed.
497 ** SUBTYPEP can now handle types involving the negation of a
498 list-form FUNCTION type.
499 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
500 as (COMPLEX (AND RATIO (NOT FIXNUM))).
501 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
503 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
504 * New feature: Experimental support for bivalent streams: streams
505 opened with :element-type :default now allow character and binary
506 (unsigned-byte 8) I/O
507 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
508 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
510 * bug fix: no more highly sporadic "couldn't check whether ~S is
511 readable" when reading a stream and an interrupt hits in the middle
512 of a select system call
513 * compiler better recognizes complex arrays (reported by Eduardo
515 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
516 for variables on SBCL built with linkage-tables. (reported by Luis
518 * various error reporting improvements.
519 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
520 (Thanks to Thiemo Seufer)
521 * bug fix: nested reader invokations work correctly
522 * bug fix: it is possible to have simultaneous references to foreign
523 code and foreign data with the same name.
525 ** added x86-64 support
526 ** incompatible change: the threading api now works with thread
527 objects instead of thread ids
528 ** bug fix: threads are protected from signals and interruption when
529 starting up or going down
530 ** bug fix: a race where an exiting thread could lose its stack to gc
531 ** bug fix: don't halt on infinite error in threads if possible
532 ** fixed numerous gc deadlocks introduced in the pthread merge
533 ** bug fix: fixed thread safety issues in read and print
534 ** bug fix: debugger doesn't hang on session lock if interrupted at
535 an inappropriate moment
536 ** bug fix: run-program is now thread safe(r)
537 ** bug fix: inner with-recursive-lock no longer releases the mutex
538 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
539 * fixed some bugs revealed by Paul Dietz' test suite:
540 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
541 EXPECTED-TYPE slots filled.
542 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
544 ** MAKE-CONDITION accepts classes as type-designators.
545 ** COMPILE may never return NIL.
546 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
547 range before calling Unix time functions
549 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
550 * numerous signal handling fixes to increase stability
551 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
552 * minor incompatible change: we now correctly canonize default
553 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
554 as per the MOP, rather than the historical (INITARG INITFUNCTION
555 INITFORM). (reported by Bruno Haible)
556 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
557 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
558 for more information.
559 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
560 pathname is a directory pathname.
561 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
562 * dynamic space size on PPC has been increased to 768Mb. (thanks to
564 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
565 :METACLASS argument in addition to a class name. (reported by
566 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
567 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
568 (reported by Bruno Haible; patch from Gerd Moellmann)
569 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
571 * bug fix: more cleanups to the floating point exception handling on
572 x86-64 (thanks to James Knight)
573 * bug fix: the compiler does not try to constant fold calls of
574 COERCE and BIT-* functions when they return freshly constructed
575 objects. (reported by Nikodemus Siivola and Paul Dietz)
576 * optimization: improved the MIPS versions of generic (in the
577 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
578 * optimization: direct conversion of (unsigned-byte 32) to floats on
579 the PowerPC platform.
580 * optimization: structure instances with raw slots now use less
581 memory, and probably show better memory locality. (thanks to
583 * optimization: DYNAMIC-EXTENT declarations for lists and closures
584 are treated as requests for stack allocation on the x86-64,
585 Alpha-32, PPC and SPARC platforms.
586 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
587 the wrong file descriptor; implementation of SOCKET-OPEN-P.
588 (thanks to Tony Martinez)
590 ** gcing a dead thread can no longer lead to lockups
591 ** threads block signals until they are set up properly
592 ** errno is no longer shared by threads
593 ** interrupt-thread restores the eflags register on x86
594 ** fixed some lockups due to gc/thread interaction
595 * fixed some bugs revealed by Paul Dietz' test suite:
596 ** invalid dotted lists no longer raise a read error when
598 ** signal an error if a symbol that names a declaration is used
599 as the name of a type, or vice versa
600 ** allow using the (declare (typespec var*)) abbreviation for
601 (declare (type typespec var*)) with all type specifiers
602 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
603 ** always return NIL from PPRINT-POP when OBJECT is NIL
604 ** don't signal errors when pretty-printing malformed LABELS,
605 FLET or MACROLET forms
606 ** declarations in a DOLIST body are also in scope for the
608 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
610 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
613 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
614 * fixed cross-compiler leakages that prevented building a 32-bit
615 target with a 64-bit host compiler.
616 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
617 opened with :IF-EXISTS :OVERWRITE.
618 * fixed bug 281: error for an invalid qualifier in a short-form method
619 combination method is not signalled until the faulty method is called.
620 * bug fix: iteration variable type inferrer failed to deal with open
621 intervals. (reported by Alan Shields)
622 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
623 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
624 or more runtime options were provided to the sbcl binary.
625 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
626 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
627 * optimization: there's now a fast-path for fixnum arguments in the
628 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
629 * optimization: the code generated on x86-64 is more compact thanks
630 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
631 * bug fix: floating-point exception handling now partly works on
632 x86-64. (Thanks to James Knight)
633 * improvement to the MIPS backend from Thiemo Seufer: C-style
634 64-bit long long arguments and return values to alien functions
636 * fixed some bugs revealed by Paul Dietz' test suite:
637 ** the type-error signalled from WARN has a filled-in DATUM slot.
638 ** the type-error required when a stream is not associated with
639 a file has the stream as its datum.
640 ** type-errors on single-floats on x86-64 no longer have
641 :INVALID-OBJECT as the datum
642 ** the type-errors signalled for invalid function names now have
643 a correct expected type
644 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
645 for structures defined with a :TYPE. Documentation strings for
646 typed structures are no longer immediately discarded
647 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
648 broadcast streams and synonym streams. FILE-LENGTH now also works
649 on broadcast streams.
651 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
652 * incompatible change: the --noprogrammer option, deprecated since
653 version 0.7.5, has been removed. Please use the equivalent
654 --disable-debugger option instead.
655 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
657 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
658 inoperational for a while now) has been completely removed.
659 * null lexical environments are now printed as #<NULL-LEXENV>,
660 significantly reducing the amount of clutter in typical backtraces.
661 * documentation on weak pointers, finalization, and after GC hooks
662 has been added to the manual.
663 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
664 as other specialized array types, is much faster. SUBSEQ and
665 COPY-SEQ on such arrays have also been sped up.
666 * optimization: EQL is now more efficient when at least other argument
667 is known to be of type (OR FIXNUM (NOT NUMBER)).
668 * fixed bug: the runtime is now less vulnerable to changes in the
669 size of the SBCL object on OS X, and virtual memory is reserved for
670 all spaces that need to be at a fixed address.
671 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
672 * fixed bug: finalizers and after GC hooks that cause consing are now
674 * fixed bug: compiler error messages and summaries are now printed to
675 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
676 * fixed inference of the upper bound of an iteration variable.
677 (reported by Rajat Datta).
678 * fixed bug 211e: calling local functions with duplicated constant
679 keyword argument no longer causes a bogus style warning about an
681 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
682 ftype declaration to be disregarded. (reported by Dave Roberts)
683 * fixed bug 373: caused by erronous compilation of references to alien
684 variables in the runtime on ppc/darwin.
685 * fixed bug 376: CONJUGATE type deriver.
686 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
687 (reported by Baughn on #lisp)
688 * a cross-compiler bug on non-x86 platforms has been identified and
689 fixed. (thanks to Bruno Haible)
690 * improvements to the MIPS runtime code for increased stability.
691 (thanks to Thiemo Seufer)
692 * increased the maximimum compact environment size to allow
693 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
694 * partial workaround for bug 135: don't name the function we're
695 calling for hairy cases of EVAL, so as not to accumulate one environment
696 entry per call to EVAL. (reported by Kevin Reid)
697 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
698 * optimization: type testing for non-vector arrays should be faster.
699 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
701 * bug fix: setting 31st element of a bit vector to zero did not work
703 * bug fix: redefining a class definition which failed due to a
704 previous accessor / function clash now works (but see BUGS entry
705 #380 for more problems in this area). (thanks to Zach Beane)
706 * the long form of DEFINE-METHOD-COMBINATION disables method group
707 checking when given a single method group with pattern *. (thanks
709 * on x86 compiler supports stack allocation of results of simple
710 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
711 * support for building and running on Mac OS X 10.4 "Tiger" has been added
712 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
713 * fixed some bugs related to Unicode integration:
714 ** the restarts for recovering from input and output encoding
715 errors only appear when there is in fact such an error to
717 * fixed some bugs revealed by Paul Dietz' test suite:
718 ** MISC.549 and similar: late transformation of unsafe type
719 assertions into derived types caused unexpected code
721 ** SCALE-FLOAT type deriver is less wrong.
722 ** type derivers for EXP, LOG and similar functions compute result
723 types for complex arguments better.
724 ** (MISC.563) CONJUGATE type deriver works for very restricted
726 ** out-of-line type testers for character strings are available.
727 ** EQUAL compiler transform understands specialness of objects
729 ** accessing double-floats stored on the stack now works on x86-64.
730 ** debugger internals could sometimes create invalid lispobjs,
731 resulting in GC crashes.
732 ** MISC.548: type check weakening can convert required type into
734 ** initialization forms for bindings are not in scope of free special
737 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
738 * incompatible change: thread support for non-NPTL systems has
739 been removed - locking is buggy and unreliable. A threaded
740 SBCL build will now warn at startup and refuse to create
741 new threads, unless futex support is detected in the kernel
742 * incompatible change: the top level REPL now has only an ABORT
743 restart associated with it, not TOPLEVEL and ABORT as it used to.
744 TOP and TOPLEVEL are now available as debugger commands for
745 returning to the top level.
746 * incompatible change: forms evaluated in the REPL now use the
747 global optimization policy.
748 * incompatible change: user- and system-initialization files are
749 no longer processed with LOAD, but by READ and EVAL; hence the
750 global optimization policy, startup package, readtable, etc,
752 * The .fasl file format number has been incremented because of
753 various incompatible changes.
754 * internal entry point details and argument counts no longer appear
755 in backtraces unless explicitly requested by setting
756 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
757 * built-in and standard functions no longer have names like "top
758 level local call to FOO".
759 * fixed bug 32: functions defined in non-null lexical environments
760 now have more legible printed representation
761 * fixed bug 33: functions defined in non-null lexical environemnts
762 are now more amenable to inspection by INSPECT.
763 * workaround for bug 354: XEPs no longer appear in backtraces unless
764 explicitly requested.
765 * fixed bug: receiving the signal which results from attempting to
766 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
767 SIGBUS on other *BSDs) on architectures where the C stack is also
768 the Lisp stack (x86 and x86-64) from foreign code no longer leads
769 to debugger confusion or wild instability. (reported by Cheuksan
771 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
772 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
773 (reported by Lutz Euler)
774 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
775 compiled code. (reported by Paul Dietz)
776 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
777 specialized parameter to a method to something that is not TYPEP
778 the specializer is now possible.
779 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
780 face of package deletion.
781 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
782 pathname. (reported by David Tolpin on c.l.l)
783 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
784 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
785 STRING-STREAM and FILE-STREAM, respectively.
786 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
787 than previously; in addition, its test suite is now run on build.
788 (thanks to Luke Gorrie)
789 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
790 * optimization: added a immediate representation for single-floats
792 * fixed some bugs revealed by Paul Dietz' test suite:
793 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
794 ** attempting to create a package with a colliding nickname causes
795 correctable errors to be signalled.
796 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
797 ** several x86-64 backend bugs related to sign-extension of immediate
800 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
801 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
802 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
804 * fixed loading of multiply forward-referenced layouts.
805 (thanks to Cheuksan Wang)
806 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
807 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
808 (see COMPILE-FILE documentation for details of :PRINT option)
809 * fixed bugs 19 and 317: fixed-format floating point printing is
810 more accurate. This also fixes a bug reported by Adam Warner
811 related to the ~@F format directive.
812 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
814 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
815 dispatch table if the from-char is a dispatch macro character.
816 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
817 by the wordsize no longer ignore the last word. (reported by Lutz
819 * fixed bug in type checking of dynamic-extent variables. (reported
821 * optimization: sequence traversal functions use their freedom to
822 coerce function designators to functions.
823 * optimization: code with many calls to CLOS methods specialized on
824 CLOS classes has had redundant type checks removed.
825 * optimization: added declarations to speed up operations that access
826 the internal character database (for example STRING-UPCASE)
827 * optimization: comparison operations between floats and sufficiently small
828 fixnums no longer create extra rationals
829 * fixed some bugs related to Unicode integration:
830 ** portions of multibyte characters at the end of buffers for
831 character-based file input are correctly transferred to the
832 start of the buffer at the next read.
833 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
834 passing it through to OPEN.
835 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
836 argument given to internal calls to OPEN.
837 ** the built-in comment readers (introduced by character sequences
838 ";" and "#|") are more forgiving to encoding errors; they will
839 STYLE-WARN and then attempt to resync the stream at a character
840 boundary. (thanks to Teemu Kalvas)
841 * fixed some bugs in the x86-64 port:
842 ** Negative short int return values from c-calls are sign-extended
844 ** The stack is aligned to 16-bytes for c-calls, as required by
845 the ABI. (thanks to Cheuksan Wang)
846 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
847 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
848 for enums. (thanks to Vincent Arkesteijn)
849 ** Multiple small optimizations and bugfixes for floating point
851 * fixed some bugs revealed by Paul Dietz' test suite:
852 ** Space, Tab, Linefeed, Return and Page have the invalid
853 secondary constituent character trait.
854 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
856 ** WITH-INPUT-FROM-STRING should only update the index place on
858 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
860 ** Bit-array operations (BIT-AND and similar) worked incorrectly
861 with one-dimensional arrays with fill pointers.
862 ** TYPE-OF failed on a complex with an integer realpart and a
864 ** compiler failure during type inference for the code of form
865 (IF (EQL X (THE ...)) ...) (MISC.535).
867 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
868 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
869 platform. The port supports 61-bit fixnums, large memory spaces
870 and reloading shared object files.
871 * enhancement: saving cores with foreign code loaded is now
872 supported on ppc/Darwin in addition to the previously supported
874 * enhancement: the statistical profiler now walks deeper into the
875 call stack for more meaningful call-graphs and accrued time
876 reports (x86/x86-64 only). It also now reports time spent in
878 * enhancement: it is now possible to trace most individual methods
879 of a generic function in addition to tracing the generic function
881 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
883 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
884 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
885 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
886 produces an error. (thanks to Vincent Arkesteijn)
887 * bug fix: NAMESTRING on pathnames with :WILD components in their
888 directories works correctly. (thanks to Artem V. Andreev)
889 * fixed bug 125: compiler preserves identity of closures. (reported
891 * bug fixed: functions with &REST arguments sometimes failed with
892 "Undefined function" when compiled with (DEBUG 3). (reported by
894 * bug fix: overflow during compiling of setting element of a bit
895 vector with constant index and value. (reported by Timmy Douglas)
896 * build fix: fixed the dependence on *LOAD-PATHNAME* and
897 *COMPILE-FILE-PATHNAME* being absolute pathnames.
898 * on x86 compiler partially supports stack allocation of dynamic-extent
900 * GO and RETURN-FROM do not check the extent of their exit points
901 when compiled with SAFETY 0.
902 * fixed some bugs related to Unicode integration:
903 ** encoding and decoding errors are now much more robustly
904 handled; it should now be possible to recover even from invalid
905 input or output to the terminal. (thanks to Teemu Kalvas)
906 ** provided a first cut at implementing STRING-TO-OCTETS and
907 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
908 ** altered the SB-MD5 contributed module slightly, changing the
909 interface just enough for it to be supportable for builds where
910 lisp characters are not eight bits.
911 * fixed some bugs revealed by Paul Dietz' test suite:
912 ** the FORMATTER-generated functions for ~V[ conditionals require
913 the correct number of arguments.
914 ** READ-FROM-STRING returns the mandated second value when applied
915 to displaced strings.
916 ** the #\Rubout and #\Backspace characters are treated as invalid
917 constituent characters by the tokenizer.
919 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
920 * new feature: reloading changed shared object files with
921 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
922 * new feature: references to foreign variables and functions
923 can now be compiled and loaded before the corresponding shared
924 object file is loaded, as long as the foreign definitions are
925 available at runtime.
926 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
927 supported. (thanks to Dan Debertin)
928 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
929 just keywords. (thanks to Vincent Arkesteijn)
930 * fixed bug #331: structure-class instances corresponding to
931 DEFSTRUCT forms are now created eagerly.
932 * fixed bug #345: backtraces from calls to undefined functions work
933 on x86 as well. Related bug #61 is now also partially fixed on x86
934 (backtraces from throws to unknown catch tags.)
935 * bug fix: lambda-list parsing is now stricter vrt. order and number
936 of lambda-list keywords.
937 * bug fix: as specified by AMOP, an error is signalled if a
938 class-option appears multiple times in a DEFCLASS form. (reported
940 * bug fix: RANDOM can be compiled when the compiler derives the type
941 of its numeric argument as a disjoint set of small integers.
942 (reported by Paul Dietz)
943 * bug fix: starting a core saved with shared objects loaded when
944 those objects are not available no longer causes threaded SBCL to
945 hang. (reported by Sean Ross)
946 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
948 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
949 argument, which must always be NIL. (reported by Kalle Niemitalo)
950 * bug fix: printing 1.0d+23 no longer results in an error.
951 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
952 * bug fix: structure slot setters preserve evaluation order. (thanks
954 * fixed some bugs related to Unicode integration:
955 ** RUN-PROGRAM can allow its child to take input from a Lisp
956 stream. (reported by Stefan Scholl)
957 ** ASDF-INSTALL successfully downloads .tar.gz files in all
958 locales. (reported by Ken Causey)
959 * fixed some bugs revealed by Paul Dietz' test suite:
960 ** INCF, DECF and REMF evaluate their place form as specified in
962 ** FORMATTER expands ~{ ~} iteration directives with V or #
963 parameters correctly.
964 ** FORMATTER deals with the ~@[ ~] conditional directive where the
965 consequent uses no arguments correctly.
966 ** the system has a partial understanding of the (COMPLEX RATIO)
968 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
971 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
972 * new feature: a build-time option (controlled by the :SB-UNICODE
973 keyword feature, enabled by default) for building the system with
974 support for the entire 21-bit character space defined by the
976 * new feature: the system now has rudimentary external-format
977 support; the primary user-visible change at this time is that
978 characters with the high bit set (such as lower-case-e-acute) will
979 print correctly to a terminal in a UTF-8 environment.
980 * minor incompatible change: BASE-CHAR no longer names a class;
981 however, CHARACTER continues to do so, as required by ANSI.
982 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
983 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
984 has been moved to the SB-EXT package (temporarily re-exported from
986 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
987 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
989 * fixed bug #351: better error-handlind and reporting for malformed
991 * fixed bug #350: bignum-printing is now more memory-efficient,
992 allowing printing of very large bignums, eg. (expt 2 10000000).
993 (reported by Bruno Haible)
994 * fixed bug #302: better primitive-type selection for intersection
996 * fixed bug #308: non-graphic characters now all have names, as
997 required. (reported by Bruno Haible)
998 * bug fix: redefining a class with different superclasses now correctly
999 removes it from the direct-subclasses of its previous superclasses.
1000 (reported by David Morse)
1001 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1002 new value now works. (reported by Bruno Haible)
1003 * bug fix: correct canonicalization of multiple non-standard slot
1004 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1005 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1006 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1007 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1008 now exists, an signals an error.
1009 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1010 during macroexpansion for non-iteration variables as well. (reported
1011 by Bruno Haible for CMUCL)
1012 * bug fix: Cyclic structures and unprintable objects in compiler
1013 messages no longer cause errors. (reported by Bruno Haible)
1014 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1015 and READ-FROM-STRING all now return a primary value of NIL if
1016 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1017 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1018 (reported by Bruno Haible for CMUCL)
1019 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1020 do not designate a condition. (reported by Bruno Haible for
1022 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1023 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1024 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1025 permitting resolution in favour of any of the conflicting symbols.
1026 (reported by Bruno Haible for CMUCL)
1027 * bug fix: EQUAL compiler optimizations is less aggressive on
1028 strings which can potentially compare true despite having distinct
1029 specialized array element types.
1030 * bug fix: unit enumerations can be defined without dividing by
1031 zero. (thanks to Vincent Arkesteijn)
1032 * FORMAT compile-time argument count checking has been enhanced.
1033 (report from Bruno Haible for CMUCL)
1034 * a partial workaround for the bug 262: the compiler does not try to
1035 inline-expand a local function doing RETURN-FROM from a deleted
1036 BLOCK. (thanks to Peter Denno for the bug report and to David
1037 Wragg for the simple test case)
1038 * fixed some bugs revealed by Paul Dietz' test suite:
1039 ** RENAME-PACKAGE allows all package designators as new package
1041 ** constraint propagation and lambda variable substitution are
1042 more cautious in dealing with partially deleted code.
1043 ** compiler sometimes lost reoptimization passes.
1044 ** CERROR, given a condition as condition designator, passes its
1045 remaining arguments to the continue format control without
1047 ** Case-altering FORMAT directives work correctly on non-ASCII
1049 ** The REAL type specifier handles bounds outside the floating
1050 point ranges without signalling FLOATING-POINT-OVERFLOW.
1051 ** Functions with IR1-transformations can create intercomponent
1052 references to global functions.
1053 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1055 ** FORMAT ~R treats a nil value for its first parameter correctly.
1057 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1058 * enhancement: saving cores with foreign code loaded is now
1059 supported on x86/NetBSD and sparc/Linux in addition to the previously
1060 supported platforms.
1061 * bug fix: on some platforms repeated installations caused multiple
1062 copies of HTML documentation to be installed -- should not happen
1063 any more. (reported by Stefan Scholl)
1064 * bug fix: parsing self-recursive alien record types multiple times
1065 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1066 original patch by Helmut Eller for CMUCL)
1067 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1068 (thanks to Richard Kreuter)
1069 * bug fix: defining classes whose accessors are methods on existing
1070 generic functions in other (locked) packages no longer signals
1071 bogus package lock violations. (reported by François-René Rideau)
1072 * bug fix: special variables as DEFMETHOD parameters no longer have
1073 associated bogus type declarations. (reported by David Wragg and
1075 * bug fix: read-write consistency on streams of element-type
1076 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1077 * bug fix: redefiniton of the only method of a generic function with
1078 no DEFGENERIC no longer emits a full WARNING. In addition,
1079 redefinition of generic functions with no DEFGENERIC to an
1080 incompatible lambda list now signals an error. (thanks to Zach
1082 * bug fix: DEFGENERIC now works even when there's a function of the
1083 same name in an enclosing lexical environment. (thanks to Zach
1085 * fixed compiler failure, caused by instrumenting code during
1086 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1087 * optimization: added loop analysis and improved register allocation
1088 * fixed some bugs revealed by Paul Dietz' test suite:
1089 ** POSITION on displaced vectors with non-zero displacement
1090 returns the right answer.
1091 ** (SIMPLE-STRING) is a valid type specifier for sequence
1093 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1095 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1096 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1098 ** The justification version of the FORMAT ~< directive treats
1099 non-zero minpad parameter correctly.
1101 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1102 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1103 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1104 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1105 the supported interface.
1106 * new feature: Single-stepping of code compiled with DEBUG 2 or
1107 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1108 * new feature: saving cores with foreign code loaded is now
1109 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1110 Timothy Moore's work for CMUCL)
1111 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1112 parameters to * instead of NIL if no initform is supplied.
1113 (reported by Johan Bockgaard)
1114 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1115 correctly when FUNCALL appears as the car of the form. Note:
1116 despite this FUNCALL forms are not currently subject to
1117 compiler-macro expansion. (port of Raymond Toy's fix for the
1118 same from CMUCL, reported by Johan Bockgaard)
1119 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1120 (thanks for Teemu Kalvas)
1121 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1122 case instead of terminating the iteration (thanks for Julian
1123 Squires, Sean Champ and Raymond Toy)
1124 * bug fix: incorrect expansion of defgeneric that caused a style
1125 warning. (thanks for Zach Beane)
1126 * bug fix: slot accessor effective method computation works properly
1127 for classes with multiple non-standard applicable methods on
1128 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1129 * on x86 compiler supports stack allocation of results of LIST and
1130 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1131 CMUCL implementation by Gerd Moellmann)
1132 * fixed some bugs revealed by Paul Dietz' test suite:
1133 ** FORMAT strings with both the ~<~:;~> form of the justification
1134 directive and pretty-printing directives cause an error.
1136 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1137 * incompatible change: the internal functions
1138 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1139 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1140 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1141 instead of the old functions.
1142 * new feature: on platforms where "dladdr" is available foreign
1143 function names now appear in backtraces. (based on Helmut Eller's
1145 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1146 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1148 * documentation: ASDF manual is now distributed with SBCL.
1149 * bug fix: SBCL can now load its contributed modules with REQUIRE
1150 even if the system-provided entries have been removed from
1151 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1153 * bug fix: backtraces involving undefined functions or assembly
1154 routines are more informative. (thanks to Brian Downing)
1155 * bug fix: mutually referent alien structures now work correctly.
1156 (reported by Rick Taube)
1157 * bug fix: structures defined by WITH-ALIEN can be referred to
1158 within other definitions in the same WITH-ALIEN.
1159 * bug fix: division operators (MOD, TRUNCATE and the like) with
1160 constant zero divisors and integer dividends no longer generate
1162 * bug fix: provide default methods for INPUT-STREAM-P and
1163 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1164 * bug fix: improve the detection and resolution of MOP metacycles
1165 (where effective method computation depends on the generic function
1166 having its effective method computed). (reported by Bruno Haible)
1167 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1168 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1169 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1170 * bug fix: pathnames with relative directory components can be
1171 represented relative to default pathnames.
1172 * optimization: in taking the GCD of bignums, reduce the two bignums
1173 to approximately the same size (using Euclid's algorithm) before
1174 applying the more sophisticated binary GCD. (thanks to Juho
1176 * optimization: COUNT on bitvectors now operates word-at-a-time.
1177 * optimization: ASH with a positive, but not necessarily constant,
1178 (leftwards) shift, when in a modular context, is compiled to a
1180 * fixed some bugs revealed by Paul Dietz' test suite:
1181 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1182 if the corresponding argument is NIL.
1183 ** FORMAT directives accepting a minpad parameter treat negative
1185 ** literal commas as character parameters to FORMAT directives are
1187 ** literal spaces directly after ~<Newline> directives within a
1188 format-logical-block (~:< ~@:>) do not induce :FILL-style
1189 conditional newlines.
1190 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1191 as specified: it no longer includes conditional newlines).
1192 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1194 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1195 :PER-LINE-PREFIX argument does not evaluate to a string.
1197 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1198 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1199 the manual for details; for now, package locks can be disabled by
1200 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1201 if no major problems are found then it is likely that they will be
1202 compiled in unconditionally.
1203 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1204 unsupported operators on all platforms. To load a shared library
1205 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1206 object file, link it into a shared library outside of SBCL and
1207 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1209 * bug workaround: ROOM T and the GC were not working together
1210 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1211 reporting facility aren't preserved. That reporting has been
1212 disabled, so now until and unless someone figures out how to make it
1213 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1214 * minor incompatible change: as threatened around sbcl-0.8.0, the
1215 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1216 The symbols are also exported from SB-PCL for backwards
1217 compatibility, but more so than before SB-PCL should be treated as
1218 an implementation-internal package.
1219 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1221 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1222 It is known as of this release not to work on the Alpha, however.
1223 * fixed bug #167: errors signalled due to illegal syntax in method
1224 bodies are now more legible.
1225 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1226 designators and can hence be used with TYPEP.
1227 * fixed bug #333: CHECK-TYPE now ensures that the type error
1228 signalled, if any, has the right object to be accessed by
1229 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1230 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1231 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1233 * fixed bug #334: programmatic addition of slots using specialized
1234 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1235 and :ALLOCATION :CLASS slots.
1236 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1237 float exponents. (rereported by Peter Seibel)
1238 * fixed a bug: #\Space (and other whitespace characters) are no
1239 longer considered to be macro characters in standard syntax by
1240 GET-MACRO-CHARACTER.
1241 * fixed bug: initialization of condition class metaobjects no longer
1242 causes an instance of the condition to be created. (reported by
1244 * fixed bug: it is now possible to have slots such that
1245 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1246 description is neither :INSTANCE nor :CLASS.
1247 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1248 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1249 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1250 non-standard SLOT-DEFINITION-ALLOCATION values.
1251 * fixed bugs: various race conditions handling exiting threads.
1252 CL-PPCRE's thread torture test now passes.
1253 * fixed arguably-a-bug: GC time proportional to number of threads in
1254 system even when most of them are idle
1255 * optimization: improved performance of BIT and SBIT on bit-vectors.
1256 * .fasl file incompatibility: The fasl file version number has
1257 been incremented because of changes associated with package locks.
1259 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1260 * minor incompatible change: the system no longer provides
1261 optimization hints (conditions which are TYPEP
1262 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1263 mode; these hints are still emitted when higher SPEED optimization
1265 * new contrib module: a sampling profiler (profiling by statistical
1266 sampling, rather than by instrumenting functions) is available as
1267 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1268 * the behaviour of the standard function ED is now customizeable by
1269 third parties through a hook variable: see ED's documentation
1270 string for information on the protocol.
1271 * the compiler no longer emits efficiency notes for (FUNCALL X)
1272 when the type of X is uncertain under default optimization
1274 * fixed bug 276: mutating a binding of a specialized parameter to a
1275 method to something that is not TYPEP the specializer is now
1277 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1278 exhibit the required behaviour on the x86 platform. (thanks to
1279 Peter van Eynde, Eric Marsden and Bruno Haible)
1280 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1281 even for difficult arguments. (reported by Peter Graves)
1282 * fixed bug 141a: the backquote printer now descends quoted
1284 * fixed another bug in backquote printing: no more destructive
1285 modification of the form's list structure. (reported by Brian
1287 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1288 move between its address being taken and the call to
1289 interrupt_thread, fixing a crashing race condition.
1290 * the SB-POSIX contrib implementation has been adjusted so that it
1291 no longer exhibits ridiculously poor performance when constructing
1292 instances corresponding to C structs.
1294 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1295 * minor incompatible change: the sb-grovel contrib now treats C
1296 structures as alien (in the sense of SB-ALIEN) objects rather than
1297 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1298 has implications for memory management of client code
1299 (sb-grovel-returned objects must now be manually managed) and for
1300 type safety (alien objects now have full types).
1301 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1302 used to control emission of compiler diagnostics, rather than the
1303 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1304 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1305 quality should be considered deprecated.
1306 * install.sh now installs the user manual as well
1307 * (not quite a new documentable feature, but worth considering in
1308 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1309 beginnings of a semantically meaningful condition hierarchy is
1310 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1312 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1313 designator as the defaults argument.
1314 * fixed bug: Displaced arrays whose displaced-to array has become
1315 too small now cause ARRAY-DIMENSION to signal an error, providing
1316 for safer bounds-checking. (reported by Bruno Haible)
1317 * fixed bug: DEFCLASS slot definitions with identical :READER and
1318 :WRITER names now signal a reasonable error. (reported by Thomas
1320 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1322 * fixed bug: Invalid slot specification errors now print correctly.
1323 (thanks to Zach Beane)
1324 * fixed bug 320: Shared to local slot value transfers in class
1325 redefinitions now happen corrently from superclasses as
1326 well. (reported by Bruno Haible)
1327 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1329 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1330 improper lists correctly. (reported by Bruno Haible)
1331 * fixed bug 313: source-transform for <fun-name> was erroneously
1332 applied to a call of a value of a variable with name <fun-name>.
1333 (reported by Antonio Menezes Leitao)
1334 * fixed bug 307: The obsolete instance protocol ensures that
1335 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1336 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1337 remove dead unknown-values globs from the middle of the stack.
1338 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1339 easier to resume long computations after using *BREAK-ON-SIGNALS*
1340 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1341 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1342 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1343 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1344 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1345 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1347 * fixed a (fairly theoretical) bug in string printing: if
1348 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1349 does not have array-element-type equal to the most general string
1351 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1352 function name. (thanks to Nikodemus Siivola)
1353 * fixed bug: compile-time format string checker failed on
1354 non-closed ~{. (reported by Thomas F Burdick)
1355 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1356 don't warn on legal loop code involving more than one aggregate
1357 boolean. (thanks to Nikodemus Siivola)
1358 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1359 in keyword position to MAKE-PACKAGE was not being reported as
1361 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1362 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1364 * fixed bugs 280 and 312: the checking for multiple definitions in a
1365 file is less likely to become confused by uses of inline
1367 * fixed bug: the #S reader macro performs the keyword coercion
1368 specified for slot names. (reported by Kalle Niemitalo)
1369 * fixed bug: lambda lists may contain symbols whose names start with
1370 & but are not lambda-list-keywords; their occurrence triggers a
1372 * fixed bug 321: define-method-combination argument lambda lists do
1373 not cause invalid code to be generated when &optional or &aux
1374 variables have default values. (reported by Bruno Haible)
1375 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1376 have CLOS classes; this appears also to have fixed reported
1377 instability in stack exhaustion detection.
1378 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1379 MUFFLE-WARNING when no associated restart is present is now
1381 * optimization: rearranged the expansion of various defining macros
1382 so that each expands into only one top-level form in a
1383 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1385 * optimization: used a previously little-used slot in symbols to
1386 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1388 * fixed some bugs revealed by Paul Dietz' test suite:
1389 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1391 ** Fixed an optimization bug related to inheritance of initforms
1392 from local to shared slots.
1393 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1394 ** CAST optimizer forgot to flush argument derived type.
1395 ** print/read consistency on floats is now orders of magnitude
1396 more likely. (thanks also to Bruno Haible for a similar report
1398 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1399 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1400 for objects of type REAL. Make it so.
1401 ** SXHASH is specified (infelicitously) to respect similarity,
1402 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1403 it so. (thanks to Markus Ziegler)
1404 ** on the Alpha, the compiler succeeds in compiling functions
1405 returning a known number of arguments greater than 63.
1406 ** fixed handling of invalid NIL arguments in keyword position
1408 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1410 ** provided a readably-printable representation for RANDOM-STATE
1412 ** ensured that pathnames, where they have a namestring, always
1413 print using #P"..." syntax.
1415 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1416 * Support for the forthcoming 2.0 version of the NetBSD kernel
1417 running on x86 hardware has been added. (thanks to Perry
1418 E. Metzger most immediately, and others for their past work)
1419 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1420 current release version; SBCL's OpenBSD support had been broken
1421 since about the time of OpenBSD's switch to ELF binary format.)
1422 * [placeholder for DX summary]
1423 ** user code with &REST lists declared dynamic-extent, under high
1424 speed or space and low safety and debug optimization policy.
1425 * The manual has been converted to Texinfo format and the debugger
1426 chapter from the cmucl manual has been added.
1427 * A facility has been added to extract documentation strings from
1428 sbcl and store them as Texinfo-formatted snippets for inclusion in
1429 the manual (via Texinfo's @include directive)
1430 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1431 the test case to Dave Roberts)
1432 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1433 pointers. (reported by Sean Ross)
1434 * bug fix: PROFILE output is printed nicely even for large numerical
1435 values. (thanks to Zach Beane)
1436 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1437 greater than 32 handle EOF correctly.
1438 * bug fix: on X86 an immediate argument of the IMUL instruction is
1439 correctly printed in disassembly. (thanks to Lutz Euler)
1440 * bug fix: class slots in redefined classes preserve their old
1441 values. (thanks to Bruno Haible and Nikodemus Siivola)
1442 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1443 succeeds. (reported by Marco Baringer)
1444 * fixed some bugs revealed by Paul Dietz' test suite:
1445 ** READ-SEQUENCE now works on ECHO-STREAMs.
1446 ** RATIONALIZE works more according to its specification. (thanks
1449 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1450 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1451 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1452 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1453 should matter to you only if you rebind the printer control
1454 variables and then find you want different bindings in the
1455 debugger than in the ordinary execution of your program.)
1456 * The runtime build system has been tweaked to support building
1457 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1458 assembler and linker. (thanks to Nikodemus Siivola)
1459 * Unbound, undefined, undeclared variables now trigger full
1460 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1461 more often programmer error than deliberate exploitation of undefined
1463 * optimization: the hash algorithm for strings has changed to one
1464 that is less vulnerable to spurious collisions. (thanks to Juho
1466 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1467 needless bounds checking. (thanks to Juho Snellman)
1468 * optimization: implemented multiplication as a modular
1469 (UNSIGNED-BYTE 32) operation on the PPC backend.
1470 * fixed some bugs revealed by Paul Dietz' test suite:
1471 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1473 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1474 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1475 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1477 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1479 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1481 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1482 their output stream on EOF from read.
1483 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1484 have been read to end-of-file.
1485 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1487 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1488 description of determination of which consecutive characters
1490 ** Printing the "Space" character with escaping on now yields "#\\ ",
1491 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1492 ** Reading floating-point numbers with *READ-BASE* set to a number
1493 less than 10 works correctly.
1494 ** Reading floating-point numbers with *READ-BASE* set to a number
1495 more than 10 works correctly.
1496 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1497 the readtable currently in effect.
1499 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1500 * minor incompatible change: parsing of namestrings on a physical
1501 (Unix) host has changed; numbers after the final #\. in a
1502 namestring are no longer interpreted as a version field. This is
1503 intented to be largely invisible to the user, except that the
1504 meaning of the namestring "*.*.*" has changed: it now refers to a
1505 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1506 should usually be replaced by
1507 (make-pathname :name :wild :type :wild :version :wild)
1508 with the added benefit that this is more likely to be portable.
1509 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1510 option to OPEN now signals an error if the file being opened
1511 exists; this may have an impact on existing code.
1512 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1513 signals received, on the PowerPC platforms (both Linux and
1514 Darwin). (thanks to Pierre Mai for pointing out the location of
1516 * several fixes on OS X: The system now builds and runs cleanly on
1517 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1518 (thanks to Brian Mastenbrook)
1519 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1520 second-resolution: integer multiples of 1/3600 between -24 and 24.
1521 (thanks to Vincent Arkesteijn)
1522 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1523 type when called with 1 argument; PEEK-CHAR checked type of
1524 PEEK-TYPE only after having read first character from a
1525 stream. (reported by Peter Graves)
1526 * bug fix: the garbage collector now has much better locality
1527 behaviour, and in particular no longer treats all memory as being
1528 exhausted when many small objects point to each other in a deeply
1530 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1531 recognized as being TYPEP their class.
1532 * bug fix: the PUSHNEW documentation string has been corrected.
1533 (thanks to Vincent Arkesteijn)
1534 * bug fix: defaulting of the value for the last of an atypically
1535 large number of multiple values being bound was not being
1536 performed correctly on the Alpha or PPC platforms
1537 * optimization: implemented multiplication as a modular
1538 (UNSIGNED-BYTE 32) operation on the x86 backend.
1539 * optimization: SEARCH on simple-base-strings can now be open-coded.
1540 (see also contrib/compiler-extras.lisp for inspiration for
1541 teaching the compiler about the Boyer-Moore algorithm).
1542 * value, returned by MAX (and MIN) called with several EQUALP, but
1543 not EQL, arguments now does not depend on compiler settings.
1544 * fixed some bugs revealed by Paul Dietz' test suite:
1545 ** in stack analysis liveness information is propagated from
1546 non-local entry points.
1547 ** pathwise CAST removing failed when the CAST node did not start
1549 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1550 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1552 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1553 host is already defined.
1554 ** RENAME-FILE works on streams instead of signalling an internal
1556 ** PEEK-CHAR uses the current readtable when determining whether
1557 or not a character is whitespace.
1558 ** MERGE-PATHNAMES handles the case when the pathname does not
1559 specify a name while the default-pathname specifies a version.
1560 ** Pathnames now stand a better chance of respecting print/read
1562 ** Attempting to use standardized file system operators with a
1563 pathname with invalid :DIRECTORY components signals a
1565 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1567 ** DIRECTORY on logical pathnames is more correct.
1568 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1569 signal a TYPE-ERROR if their argument is not a stream.
1570 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1571 designator argument does not designate a stream.
1572 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1573 examining the synonym.
1574 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1576 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1577 element-type larger than ([UN]SIGNED-BYTE 32).
1579 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1580 * When built with the :SB-FUTEX feature, threaded builds now take
1581 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1582 for faster/more reliable mutex and condition variable support.
1583 * Incompatible change (but one you probably shouldn't have been using
1584 anyway): the interface and code for arbitrating between multiple
1585 threads in the same user session has been redesigned.
1586 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1587 error when called without an explicit environment argument.
1588 (thanks to Nikodemus Siivola)
1589 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1590 confused about their position. (thanks to Adam Warner and Gerd
1592 * bug fix: Pretty printing backquoted forms with unquotations in the
1593 argument list position of various code constructs such as LAMBDA
1594 now works correctly. (reported by Paul Dietz)
1595 * bug fix: Pretty printing unquotations no longer loses all
1596 stream position information.
1597 * optimization: performance of string output streams is now less
1598 poor for multiple small sequence writes.
1599 * optimization: performance of CSUBTYPEP in the presence of complex
1600 expressions involving CONS and NOT many times has been improved.
1601 (reported by Paul Dietz)
1602 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1604 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1605 streams for the REPL, for the debugger, and so on. (thanks to
1607 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1608 * fixed some bugs revealed by Paul Dietz' test suite:
1609 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1610 is ignored if it is too small, rather than propagating through
1611 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1612 ** extremely complex negations of CONS types were not being
1613 sufficiently canonicalized, leading to inconsistencies in
1615 ** VALUES tranformer lost derived type.
1617 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1618 * fixed a bootstrapping bug: the build process no longer assumes
1619 that the various BOOLE-related constants have the same value in
1620 host and target lisps. (noted by Paul Dietz' test suite on an
1621 SBCL binary built from CLISP)
1622 * The system can now be dynamically linked on the MIPS platform,
1623 which enables dynamic loading of foreign code from Lisp. (thanks
1624 to Ralf Baechle for discussions on the MIPS ABI)
1625 * The system now records debugging information for its own source
1626 files in a filesystem-position-independent manner, relative to
1627 the "SYS" logical host.
1628 * fixed a compiler bug: MV-LET convertion did not check references
1629 to the "max args" entry point. (reported by Brian Downing)
1630 * tweaked disassembly notes to be less confident about proclaiming
1631 some instruction as an LRA. (thanks to Brian Downing)
1632 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1633 now each have their own history, command character, and other
1634 characteristics. (thanks to David Lichteblau)
1635 * fixed some bugs revealed by Paul Dietz' test suite:
1636 ** compiler failure in compiling LOGAND expressions including a
1638 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1639 shift greater than 32.
1640 ** FLUSH-DEST did not mark blocks for type check regeneration.
1641 ** HANDLER-CASE failed to accept declarations in handler clauses
1642 in some circumstances.
1644 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1645 * New code in contrib/sb-introspect (still probably not entirely
1646 stable yet) provides some support for smart Lisp development
1647 environments like SLIME.
1648 * The conditions signalled for errors occurring when loading .fasl
1649 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1650 in a way which should help ASDF recover gracefully.
1651 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1652 stuff has been cleaned up. If you code contrib/ stuff, this might
1653 affect you, and you can look at contrib/README, contrib/STANDARDS,
1654 and/or the 0.8.4.27 diff to check.
1655 * In full calls the compiler now does not generate checks for declared
1656 argument types for all arguments.
1657 * various threading fixes
1658 ** and some experimental patches which didn't make it into
1659 the main tree for this release, but which are shipped in
1660 contrib/experimental-thread.patch as a possible fix for some
1661 failures (deadlock, spinning...) in GC-intensive multithreaded
1663 * fixed PPC build problem (source code incompatibility of different
1664 library versions): added offsetof() hackery which attempts to divine
1665 where glibc maintainers put uc_mcontext today
1666 * fixed bug 282: compiler does not trust type assertions while passing
1667 arguments to a full call.
1668 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1669 &OPTIONAL VALUES type parameter.
1670 * fix bug 214: algorithm for noting rejected templates is now more
1671 similar to that of template seletion. (also reported by rydis on
1673 * fixed bug 141b: printing backquoted information readably and prettily
1674 inserts a space where necessary.
1675 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1676 (CHARACTER 10) are now reported as errors, rather than propagated
1677 as unknown types. (reported by piso on #lisp)
1678 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1679 with values NIL and :ERROR. (thanks to Milan Zamazal)
1680 * fixed bug 191c: CLOS now does proper keyword argument checking as
1681 described in CLHS 7.6.5 and 7.6.5.1.
1682 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1683 longer raise an error; further, using a list as a for-as-arithmetic
1684 counter now raises a meaningful error.
1685 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1686 now understood by sequence creation functions such as MAKE-SEQUENCE
1688 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1689 a STRING-INPUT-STREAM.
1690 * compiler enhancement: SIGNUM is now better able to derive the type
1692 * type declarations inside WITH-SLOTS are checked. (reported by
1694 * fixed some bugs revealed by Paul Dietz' test suite:
1695 ** incorrect optimization of TRUNCATE for a positive first
1696 argument and negative second.
1697 ** compiler failure in let-convertion during flushing dead code.
1698 ** compiler failure while deriving type of TRUNCATE on an
1699 interval, containing 0.
1700 ** ASH of a negative bignum by a negative bignum count now returns
1702 ** intersection of CONS types now canonicalizes properly, fixing
1703 inconsistencies in SUBTYPEP.
1705 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1706 * incompatible change: The --disable-debugger command line
1707 option now clobbers the debugger at a more fundamental
1708 level, by redefining #'INVOKE-DEBUGGER instead of by
1709 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1710 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1711 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1712 enter the debugger REPL and then suffer recursive errors
1713 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1714 while under the new system, BREAK in --disable-debugger mode
1715 terminates the system just as an unhandled error would.
1716 * fixed compiler performance when processing loops with a step >1;
1717 * bug fix: DOCUMENTATION now retrieves generic function
1718 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1719 support has been systematized, and now supports the methods
1720 specified by ANSI, along with a default method and a method for
1721 slot documentation. (reported by Nathan Froyd)
1722 * bug fix: effective methods associated with a generic function are
1723 no longer cached over a change of that generic function's method
1724 combination. (reported by Andreas Fuchs)
1725 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1726 names a non-existent directory. (thanks to Andreas Fuchs)
1727 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1728 the PPC platform, lead to stack corruption; this has been fixed.
1729 (reported by Rainer Joswig)
1730 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1731 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1732 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1733 arithmetic in general, is now much more reliable on the PPC
1735 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1736 platform now returns the right answer.
1737 * optimization: restored some effective method precomputation in
1738 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1739 precomputation is now tunable.
1740 * optimization: compiler-internal data structure use has been
1741 reviewed, and changes have been made that should improve the
1742 performance of the compiler by about 20%.
1743 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1744 simple-base-strings and simple-bit-vectors is improved.
1745 * optimization: the optimization of 32-bit logical and arithmetic
1746 functions introduced in version 0.8.3 on the x86 has been
1747 implemented on the mips, ppc and sparc platforms; an
1748 implementation of the same facility, but for 64-bit arithmetic,
1749 has been added for the alpha.
1750 * microoptimization: the compiler is better able to make use of the
1751 x86 LEA instruction for multiplication by constants.
1752 * bug fix: in some situations compiler did not report usage of
1753 generic arithmetic in (SPEED 3) policy.
1754 * bug 145b fix: compiler used wrong type specifier while converting
1755 MEMBER-types to numeric.
1756 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1758 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1759 index argument now works properly on the Alpha platform.
1760 * bug fix: floating point exception treatment on the Alpha platform
1762 * bug fix: FILE-POSITION works much better on string input and
1763 output streams. (thanks to Nikodemus Siivola)
1764 * bug fix: many threading/garbage collection symptoms sorted.
1765 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1766 might be pseudo-atomic.
1767 * internal change: Stopping for GC is now done with signals not
1768 ptrace. GC is now done in whichever thread wanted it, instead of
1770 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1772 * bug fix: result form in DO is not contained in the implicit
1774 * incompatible change: ICR structure is changed; the value part of
1775 CONTINUATION is now called LVAR; corresponding functions are
1776 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1777 * added type deriver for ISQRT (thanks to Robert E. Brown).
1778 * bug fix: better support for loading from the command line when an
1779 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1781 * fixed some bugs revealed by Paul Dietz' test suite:
1782 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1783 ** ROUND and FROUND now give the right answer when given very
1784 small float arguments.
1785 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1787 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1788 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1789 ** LCM should return a non-negative integer.
1790 ** PARSE-INTEGER returned the index of a terminator instead of the
1791 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1792 ** PARSE-INTEGER returned an incorrect index being applied to a
1794 ** LCM with two arguments of 0 returns 0 rather than signalling
1796 ** unsigned addition of a 32-bit constant with the high bit set no
1797 longer causes an internal compiler error on the x86.
1798 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1799 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1800 with negative last argument.
1801 ** byte specifiers with zero size and position no longer cause
1802 an error during type derivation.
1803 ** bignum multiplication on the Alpha platform now returns the
1805 * porting: The system now builds on SuSE AMD64, although it still
1806 generates a 32-bit binary.
1807 * .fasl file incompatibility: The fasl file version number has
1808 been incremented (because of the changes to internal compiler
1809 data structures referred to above).
1811 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1812 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1813 more accurately, on the Darwin kernel running on PowerPC hardware.
1814 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1815 * Compiler code deletion notes now signal a condition of type
1816 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1817 an associated MUFFLE-WARNING restart.
1818 * The compiler now performs limited argument count validation of
1819 constant format strings in FORMAT, and where appropriate in ERROR,
1820 CERROR and WARN. (thanks to Gerd Moellmann)
1821 * New ASDF-INSTALL contrib can be used for automatic download and
1822 installation of third-party Lisp code from CCLAN or other sites
1824 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1825 which forces another thread to execute a function supplied by the
1827 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1828 now accept and act upon their :ELEMENT-TYPE keyword argument.
1829 (reported by Martin Atzmueller, Edi Weitz)
1830 * bug fix: FILE-POSITION now accepts position designators up to
1831 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1832 is the greater. (thanks to Patrik Nordebo)
1833 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1834 Kalle Olavi Niemitalo)
1835 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1836 printable. (reported by Eric Marsden)
1837 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1838 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1839 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1840 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1841 (thanks to Stig Sandoe)
1842 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1843 UNREAD-CHAR now correctly includes the unread character in the
1844 target sequence. (thanks to Gerd Moellmann)
1845 * bug fix in threaded builds: the system can now be suspended and
1846 resumed by shell job control with minimal disruption.
1847 * bug fixes in times and timezones >2038AD
1848 * better handling of "where is GNU make?" problem in build scripts
1849 (thanks to Nikodemus Siivola)
1850 * new optimization: inside a named function any reference to a
1851 function with the same name is considered to be a self-reference;
1852 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1853 optimization quality.
1854 * new optimization on x86: logical functions and + now have
1855 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1856 used when the result is truncated to 32 bits.
1857 * VALUES declaration is partially enabled.
1858 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1859 * fixed some bugs revealed by Paul Dietz' test suite:
1860 ** The system now obeys the constraint imposed by
1861 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1862 types form a lattice under type intersection.
1863 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1864 ** ASSOC now ignores NIL elements in an alist.
1865 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1866 and (1+ MOST-POSITIVE-FIXNUM) answers.
1867 ** The addition of a method with invalid qualifiers to a generic
1868 function does not cause an error to be signalled immediately;
1869 a warning is signalled, and the error is generated only on
1870 calling the generic function.
1871 * changed .fasl file version number, in part to add type codes for
1872 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1873 obscure ANSI requirements
1875 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1876 * fixed bug 148: failure to inline-expand a local function left
1877 garbage, confusing the compiler.
1878 * fixed bugs 3cd: structure slot readers perform type check if the
1879 slot can have an invalid value (i.e. it is either not initialized
1880 or can be written with a less specific slot writer).
1881 * bug fix: the compiler now traps array references to elements off
1882 the end of an array; previously, the bounds checking in some
1883 circumstances could go off-by-one.
1884 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1886 * type declarations for array element types now obey the description
1887 on the CLHS page "Declaration TYPE", as per discussions on
1888 sbcl-help around 2003-05-08. This means that a declaration
1889 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1890 declaration, all references to BAR will be asserted or assumed
1891 (with THE, so dependent on compiler policy) to involve objects of
1892 type FOO. Note that no such declaration is implied in
1893 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1894 * declared types of functions from the "Conditions"
1895 chapter. (reported by Paul Dietz)
1896 * bug fix: CERROR accepts a function as its first argument.
1897 * bug fix: NTH an NTHCDR accept a bignum as index
1898 arguments. (reported by Adam Warner)
1899 * optimization: character compare routines now optimize comparing
1900 against a constant character. (reported by Gilbert Baumann)
1901 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1902 argument now works properly on the MIPS platform.
1903 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1904 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1905 declared to have a functional type, does not cause an error
1907 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1908 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1909 * fixed bug: if last continuation of a deleted block has a
1910 destination, this destination should be deleted too. (reported by
1912 * fixed a bug in the bootstrap process: the host compiler's values
1913 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1914 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1915 test case from Patrik Nordebo)
1916 * improved the ability of the disassembler on the PPC platform to
1917 provide helpful disassembly notes.
1918 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1919 the class in more cases than previously.
1920 * bug fix: FILE-POSITION now understands :START and :END for
1921 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1922 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1923 8) data. (thanks to David Lichteblau)
1924 * bug fix: it is possible to add a method to a generic function
1925 without lambda list.
1926 * bug fix: reader failed to signal END-OF-FILE inside an
1927 object representation. (reported by Nikodemus Siivola)
1928 * fixed some bugs revealed by Paul Dietz' test suite:
1929 ** LAST and [N]BUTLAST should accept a bignum.
1930 ** condition slot accessors are methods.
1931 ** (VECTOR NIL) is a subtype of STRING.
1933 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1934 * minor incompatible change: some nonsensical specialized lambda
1935 lists (used in DEFMETHOD) which were previously ignored now signal
1937 * minor incompatible change: the system is now aware of the types of
1938 variables in the COMMON-LISP package, and will signal errors for
1939 most violations of these type constraints (where previously they
1940 were silently accepted).
1941 * minor incompatible change: COMPILE-FILE now uses the freedom
1942 afforded (ANSI 3.2.2.3) to use derived function types for
1943 functions defined in the same file. This also permits the system
1944 to warn on static type mismatches and function
1945 redefinition. (Currently it does not work with high DEBUG level.)
1946 * minor incompatible change: VALUES declaration is disabled.
1947 * When issuing notes, the compiler now signals a condition of type
1948 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1949 restart for use in user handlers. It is expected that the
1950 COMPILER-NOTE condition will eventually become a condition
1951 supertype to a hierarchy of note types, which will then be
1952 handleable in a similar fashion. However, at the moment, no such
1953 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1954 but it's an implementation detail, not a classification for the
1956 * Changes in type checking closed the following bugs:
1957 ** type checking of unused values (192b, 194d, 203);
1958 ** template selection based on unsafe type assertions (192c, 236);
1959 ** type checking in branches (194bc).
1960 * A short form of VALUES type specifier has ANSI meaning (it has
1961 increased the number of situations when SBCL cannot perform type
1963 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1964 keywords or constants is permissible.
1965 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1966 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1967 classes. (thanks to Antonio Martinez)
1968 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1969 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1970 argument to a pretty-print stream. Instead, it leaves any such
1971 operations to DESCRIBE-OBJECT methods.
1972 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1974 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1975 lambda lists. (thanks to David Lichteblau)
1976 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1977 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1978 (thanks to Gerd Moellmann)
1979 * fixed bug 246: increased compilation speed of long
1980 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1982 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1984 * DEFCONSTANT now throws a condition of type
1985 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1986 constant to a non-EQL value; CONTINUE and ABORT restarts
1987 respectively change and preserve the value.
1988 * fixed bug 63: The code walker, part of the implementation of CLOS,
1989 is now better at handling symbol macros.
1990 * bug fix: There is no longer an internal implementation type named
1991 CL:LENGTH. (reported by Raymond Toy)
1992 * bug fix: In macro-like defining macros/special operators the
1993 implicit block does not enclose the lambda list.
1994 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1995 enclosed in parentheses) are not suitable as type specifiers, and
1996 their use properly signals an error now.
1997 * bug fix: An argument count mismatch for a type specifier in code
1998 being compiled no longer causes an unhandled error at compile
1999 time, but signals a compile-time warning.
2000 * fixed simple vector readable printing
2001 * bug fix: DESCRIBE takes more care over whether the class
2002 precedence list slot of a class is bound before accessing it.
2003 (reported by Markus Krummenacker)
2004 * bug fix: FORMATTER can successfully compile pretty-printer format
2005 strings which use variants of the ~* directive inside.
2006 * bug fix: SEARCH now applies its TEST predicate to the elements of
2007 the arguments in the correct order. (thanks to Wolfhard Buss)
2008 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2009 in type inference. (thanks to Robert E. Brown)
2010 * bug fix: Reading in symbols with an explicit package name of ""
2011 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2012 * fixed some bugs revealed by Paul Dietz' test suite:
2013 ** NIL is now allowed as a structure slot name.
2014 ** Arbitrary numbers, not just REALs, are allowed in certain
2015 circumstances in LOOP for-as-arithmetic clauses.
2016 ** Multiple class redefinitions before slot access no longer
2017 causes a type error.
2018 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2019 association between the name and a class.
2020 ** Generic functions with non-standard method-combination and over
2021 five methods all of which return constants no longer return NIL
2022 after the first few invocations. (thanks to Gerd Moellmann)
2023 ** CALL-NEXT-METHOD with no arguments now passes the original
2024 values of the arguments, even in the presence of assignment.
2025 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2027 ** DISASSEMBLE works with closures and funcallable instances.
2028 ** ADD-METHOD now returns the generic function, not the new method.
2029 ** FIND-METHOD signals an error if the lengths of the specializers
2030 is incompatible with the generic function, even if the ERRORP
2032 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2033 which its argument is a member.
2034 ** DEFCLASS only redefines the class named by its class-name
2035 argument if that name is the proper name of the class;
2036 otherwise, it creates a new class.
2037 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2038 of the UNBOUND-SLOT condition to the name of the slot.
2039 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2040 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2041 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2042 SLOT-MAKUNBOUND in the specified fashion.
2044 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2045 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2046 cross-compilation host. As a consequence, we can now bootstrap our
2047 way up to SBCL starting with a bare gcc toolchain and human-readable
2048 source code (first the source to CLISP, then the source to SBCL).
2049 * A contributed module containing a partial implementation of the
2050 simple-streams interface has been included. (thanks to Rudi
2052 * A contributed module implementing the RFC1321 Message Digest
2053 Algorithm, known as MD5, has been included.
2054 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2055 no longer has any effect, as the code controlled by this feature
2056 has been deleted. (As far as we know, no-one has ever built using
2057 this feature, and its semantics were confused in any case).
2058 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2059 work on conditions (as required by the ANSI specification),
2060 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2061 expected behaviour on conditions. Users should note, however,
2062 that such behaviour is not required by the ANSI specification,
2063 and so use of this behaviour may render their code unportable.
2064 * fixed some bugs revealed by Paul Dietz' test suite:
2065 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2067 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2068 ** SXHASH on condition objects no longer returns NIL.
2069 ** :ALLOCATION :CLASS slots are better treated; their values are
2070 updated on class redefinition, and initforms inherited from
2071 superclasses are applied.
2072 ** REMOVE-METHOD returns its generic function argument even when
2073 no method was removed.
2074 ** SHARED-INITIALIZE now initializes the values of the requested
2075 slots, including those with :ALLOCATION :CLASS.
2076 ** ALLOCATE-INSTANCE now works on structure classes defined via
2077 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2079 ** SLOT-EXISTS-P now works on conditions, as well as structures
2081 ** MAKE-LOAD-FORM now has the required methods on
2082 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2083 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2084 keyword, and now implements the SLOT-NAMES argument.
2085 ** methods with &OPTIONAL arguments no longer allow too many
2086 arguments to be passed in the call without error.
2087 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2088 option is consistent with the required arguments of the generic
2089 function lambda list.
2090 * bug fix: REQUIRE accepts a string designator. (Thanks to
2092 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2093 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2094 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2095 * bug fix: The compiler now checks for duplicated variables in macro
2097 * bug fix: SETQ on globals returns the correct value.
2098 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2099 not a COMPILER-ERROR (followed by some other strange error on
2100 choosing the CONTINUE restart).
2101 * bug fix: make.sh and friends are now more consistent in the way that
2102 they look for GNU "make".
2104 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2105 * experimental native threads support (on x86 Linux >=2.4 only).
2106 This is not compiled in by default: you need to add :SB-THREAD to
2107 the target features. See the "Beyond ANSI" chapter of the manual
2109 * fix for longstanding nonANSIism: The old distinction between
2110 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2111 The return value from CL:FIND-CLASS is now a CLOS class, and
2112 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2113 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2114 * An interface to the MetaObject Protocol, as described in Kiczales,
2115 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2116 MIT Press, 1991, is available from the SB-MOP package.
2117 * incompatible change: the SB-PCL package should now be considered
2118 a private implementation detail, and no longer a semi-private MOP
2120 * minor incompatible change: due to rearrangement for threads, the
2121 control stack and binding stack are now allocated at arbitrary
2122 addresses instead of being hardcoded per-port. Users affected by
2123 this probably have to be doing advanced things with shared
2124 libraries, and will know who they are.
2125 * minor incompatible change: Previously, all --eval forms used were
2126 processed with READ before any of them were processed with EVAL.
2127 Now each --eval form is processed with both READ and EVAL before
2128 the next --eval form is processed. (Thus package operations like
2129 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2130 work as the user might reasonably expect.)
2131 * minor incompatible change: *STANDARD-INPUT* is now only an
2132 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2134 * minor incompatible change: Y-OR-N-P is now character-oriented,
2135 not line oriented. Also, YES-OR-NO-P now works without errors.
2136 (thanks to Antonio Martinez)
2137 * sb-aclrepl module improvements: an integrated inspector, added
2138 repl features, and a bug fix to :trace command.
2139 * Known functions, which cannot be open coded by the backend, are
2140 considered to be able to check types of their arguments. (fixing
2141 a bug report by Nathan J. Froyd)
2142 * fixed a bug in computing method discriminating functions: It is
2143 now possible to define methods specialized on classes which have
2144 forward-referenced superclasses. (thanks to Gerd Moellmann)
2145 * fixed evaluation order in optional entries (reported by Gilbert
2147 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2148 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2149 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2150 required-by-AMOP NAME argument, as well as CLASS and
2151 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2152 * fixed bug 20: DEFMETHOD can define methods using names that are
2153 not the proper names of classes to designate class specializers.
2154 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2155 Unix file descriptors, instead of blowing up. (thanks to Antonio
2157 * Garbage collection refactoring: user-visible change is that a
2158 call to the GC function during WITHOUT-GCING will not do garbage
2159 collection until the end of the WITHOUT-GCING. If you were doing
2160 this you were probably losing anyway.
2161 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2162 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2163 TYPEP the latter but not the former.
2164 * The compiler issues a full WARNING for calls to undefined functions
2165 with names from the CL package.
2166 * MAP-INTO for a vector destination is open coded. (reported by
2167 Brian Downing on c.l.l)
2168 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2169 documentation string.
2170 * fixed some bugs revealed by Paul Dietz' test suite:
2171 ** COPY-ALIST now signals an error if its argument is a dotted
2173 ** Condition slots are now accessed more correctly in the presence
2174 of multiple initargs for a given slot.
2175 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2176 exclude restarts of the same name associated with a different
2178 ** DEFCLASS of forward-referenced classes with another
2179 forward-referenced class in the superclasses list no longer
2181 ** Condition slots are now initialized once each, not multiple
2182 times. (thanks to Gerd Moellmann)
2183 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2185 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2186 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2187 arguments contain duplicated elements.
2188 ** RESTART-CASE understands local macros.
2189 ** RESTART-CASE associates exactly its own restarts with a condition.
2190 ** ENDP in safe mode checks its argument to be of type LIST.
2191 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2192 in question is unbound.
2193 ** Optimized MAKE-INSTANCE functions no longer cause internal
2194 assertion failures in the presence of duplicate initargs.
2195 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2196 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2197 propagated even in the ctor optimized implementation of
2199 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2201 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2202 * a better implementation of SXHASH on (simple) bit vectors,
2203 measured both in execution speed and in distribution of results
2204 over the positive fixnums, has been installed. Likewise, a better
2205 implementation of EQUAL for simple bit vectors is now available.
2206 * fixed CEILING optimization for a divisor of form 2^k.
2207 * fixed bug 240 (emitting extra style warnings "using the lexical
2208 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2209 by Antonio Martinez)
2210 * fixed SXHASH, giving different results for NIL depending on type
2211 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2212 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2213 variable. (found by Rolf Wester)
2214 * SBCL does not ignore type declarations for special
2215 variables. (reported by rif on c.l.l 2003-03-05)
2216 * some bug fixes in contrib/sb-aclrepl/
2217 * fixed some bugs revealed by Paul Dietz' test suite:
2218 ** a bug in the CONS type specifier, whereby the CAR and CDR
2219 types got intertwined, has been fixed;
2220 ** the type system is now able to reason about the interaction
2221 between INTEGER and RATIO types more completely;
2222 ** APPEND, [N]REVERSE and NRECONC check that those their
2223 arguments, which must be proper lists, are really so;
2224 ** An array specialized to be unable to hold elements has been
2225 implemented, as required -- yes, really -- by ANSI;
2226 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2227 on malformed property lists;
2229 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2230 * incompatible packaging change: in line with Unix convention,
2231 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2232 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2233 * REQUIRE and PROVIDE are now optionally capable of doing something
2234 useful. See the documentation string for REQUIRE.
2235 * infrastructure for a managed SBCL contrib system: contributed
2236 modules in this release include:
2237 ** the ASDF system definition facility;
2238 ** an interface to the BSD Sockets API;
2239 ** an ACL-like convenience interface to the repl;
2240 (thanks to Kevin Rosenberg)
2241 ** an implementation of ROTATE-BYTE, with efficient implementation
2243 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2244 gives a closer approximation to the right answer than previously.
2245 (thanks to Raymond Toy)
2246 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2247 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2248 optional environment arguments, as required by ANSI.
2249 * fixed bugs in other functions taking environment objects, allowing
2250 calls with an explicit NIL environment argument to be compiled
2252 * fixed bug 228: primary return values from
2253 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2254 COMPILE or FUNCTION.
2255 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2256 :TYPE will no longer signal errors on innocuous objects.
2257 * fixed bug 231b: SETQ is better at respecting type declarations in
2258 the lexical environment.
2259 * fixed a bug in DEFCLASS: classes named by symbols with no or
2260 unprintable packages can now be defined.
2261 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2262 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2263 * fixed compiler failure related to checking types of functions.
2264 (reported by Robert E. Brown)
2265 * the compiler is now much more consistent in its error-checking
2266 treatment of bounding index arguments to sequence functions: in
2267 (SAFETY 3) code, errors will be signalled in almost all cases if
2268 invalid sequence bounding indices are passed to functions defined
2269 by ANSI to operate on sequences.
2270 * fixed a bug in the build procedure: documentation of SBCL-specific
2271 packages is now preserved and available in the final Lisp image.
2272 * lifted FDEFINITION lookup out of loops in the implementation of
2273 many list operations. (thanks to Robert E. Brown)
2274 * fixed a bug in the reader: the #n# reader macro now works for
2275 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2276 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2277 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2278 E. Brown and Tony Martinez respectively)
2279 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2280 SLOT-BOUNDP: in particular, it is now able to optimize them much
2281 better, and is now not vulnerable to having packages renamed.
2282 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2283 always called when a slot is not present in an instance. (thanks
2285 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2286 longer signal an error on unreading a character following EOF on
2287 the previous constituent stream. (thanks to Tony Martinez)
2288 * fixed some bugs revealed by Paul Dietz' test suite:
2289 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2290 not just nonnegative fixnums;
2291 ** the logical bit-array operators such as BIT-AND now accept an
2292 explicit NIL for their "opt-arg" argument (to indicate a
2293 freshly-consed result bit-array);
2294 ** ELT now signals an error on an invalid sequence index in safe
2296 ** the type system is now cleverer about negations of numeric
2297 types, and consequently understands the BIGNUM and RATIO types
2299 ** the type system is now cleverer about the interaction between
2300 INTEGER and RATIO types: while bugs still remain, many more
2301 cases are accurately computed;
2302 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2303 if it is in the last clause;
2304 ** CONSTANTLY now correctly returns a side-effect-free function in
2306 ** DECLARE is no longer treated as a special-operator; in
2307 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2308 * incremented fasl file version number due to the change in the
2309 DEFSTRUCT-SLOT-DESCRIPTION structure.
2311 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2312 * minor incompatible change: code processed by the "interpreter" or
2313 EVAL now has a compilation optimization policy of (DEBUG 2)
2314 (changed from (DEBUG 1)) to improve debuggability of interactive
2315 development, and to allow the use of the debug RETURN command in
2317 * an experimental implementation of the RETURN command for the
2318 debugger has been included. (thanks to Frederik Kuivinen)
2319 * fixed bug 62: constraints were not propagated into a loop.
2320 * fixed bug in embedded calls of SORT (reported and investigated by
2322 * fixed some bugs revealed by Paul Dietz' test suite:
2323 ** printing and reading of arrays with some dimensions having
2324 length 0 (thanks to Gerd Moellmann);
2325 ** BOA constructor with &AUX argument without a default value does
2326 not cause a type error;
2327 ** CONSTANTP now returns true for all self-evaluating objects.
2329 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2330 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2331 accessors that are related by inheritance, as specified in the
2332 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2333 Valtteri Vuorikoski)
2334 * The compiler is now able to inline functions that were defined in
2335 a complex lexical environment (e.g. inside a MACROLET).
2336 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2338 * Support for the upcoming FreeBSD-5.0 release has been included.
2339 (thanks to Dag-Erling Smorgrav)
2340 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2341 effect when it is not in a toplevel context.
2342 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2343 stand a better chance of being correct. (thanks to Gerd
2345 * fixed bug in COERCE, which now signals an error on coercing a
2346 rational to a bounded real type which excludes the expected
2348 * The compiler is now able to derive types more accurately from the
2349 COERCE and COMPILE functions.
2350 * fixed bug 223: functional binding is considered to be constant
2351 only for symbols in the CL package.
2352 * fixed bug 231: SETQ did not check the type of a variable being set
2353 (reported by Robert E. Brown)
2354 * A new optimization for MAKE-INSTANCE has been included, fixing
2355 various bugs (including relating to :ALLOCATION :CLASS slots and
2356 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2358 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2359 ** As required by ANSI, LOOP now disallows anonymous collection
2360 clauses such as COLLECT I in conjunction with aggregate boolean
2361 clauses such as THEREIS (= I 1);
2362 ** LOOP now signals an error when any variable is reused in the
2363 same loop (including the potentially useful construct analogous
2364 to WITH A = 1 WITH A = (1+ A);
2365 ** IT is only a special loop symbol within the first clause of a
2366 conditional loop clause;
2367 ** LOOP with a typed iteration variable over a hashtable now
2368 signals a type error iff it should.
2369 * fixed some other bugs revealed by Paul Dietz' test suite:
2370 ** FILE-STREAM now names the class previously known as FD-STREAM;
2371 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2372 argument) no longer signals an error;
2373 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2374 of the slot symbol, rather than using the current package
2375 ((:CONC-NAME "") continues to intern the slot's name in the
2377 * incremented fasl file version number, because of the incompatible
2378 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2379 of the new implementation of DEFINE-COMPILER-MACRO.
2381 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2382 * Support for building SBCL for MIPS platforms running in
2383 little-endian mode has now been checked in, and basic
2384 functionality on said platforms verified.
2385 * minor incompatible change: PCL now records the pathname of a file
2386 in which methods and the like are defined, rather than its
2388 * minor incompatible change: TRUENAME now considers the truename of
2389 a file naming a directory to be the pathname with :DIRECTORY
2390 component indicating that directory.
2391 * minor incompatible change: a NAMED clause in the extended form of
2392 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2393 reason for the previous behaviour is unclear.
2394 * more systematization and improvement of CLOS and MOP conformance
2395 in PCL (thanks to Gerd Moellman and Pierre Mai):
2396 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2398 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2399 primary methods with no specializers;
2400 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2402 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2403 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2404 FORWARD-REFERENCED-CLASSes; error reporting on
2405 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2407 ** SXHASH on CLOS instances now uses a slot internal to the
2408 instance to return different numbers on distinct instances,
2409 while preserving the same return value through invocations of
2411 ** DEFMETHOD signals errors when methods with longer incongruent
2412 lambda lists are added to generic functions;
2413 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2414 CLASS, as specified in AMOP;
2415 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2416 on the order returned by the primary method for classes of
2417 class STANDARD-CLASS;
2418 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2419 * fixed some bugs shown by Paul Dietz' test suite:
2420 ** DOLIST puts its body in TAGBODY;
2421 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2423 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2424 value producing form;
2425 ** if more variables are given to PROGV than values, extra
2426 variables are bound and made to have no value;
2427 ** NSUBSTITUTE on list arguments gets the right answer with
2429 ** ELT signals an error of type TYPE-ERROR when the index argument
2430 is not a valid sequence index;
2431 ** LOOP signals (at macroexpansion time) an error of type
2432 PROGRAM-ERROR when duplicate variable names are found;
2433 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2434 ** FUNCALL of special-operators now cause an error of type
2436 ** PSETQ now works as required in the presence of side-effecting
2437 symbol-macro places;
2438 ** NCONC accepts any object as its last argument;
2439 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2441 ** loop-for-as-package does not require a package to be explicitely
2443 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2445 * fixed bug 166: compiler preserves "there is a way to go"
2446 invariant when deleting code.
2447 * fixed bug 172: macro lambda lists with required arguments after
2448 &REST arguments now cause an error to be signalled. (thanks to
2450 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2451 bug. (thanks to Matthew Danish)
2452 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2454 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2455 arguments are lexically rebound. (thanks to Gerd Moellmann and
2457 * fixed bug 194: error messages are now more informative when there
2458 is no primary method applicable in a call to a generic
2459 function. (thanks to Gerd Moellmann)
2460 * fixed bug in command line argument checking (thanks to Julian
2462 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2463 (thanks to Matthew Danish)
2464 * incremented fasl file version number, because of the
2465 SXHASH-related changes in the layout of CLOS data structures
2467 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2468 * minor incompatible change: The runtime (the Unix executable named
2469 "sbcl") is now much pickier about the .core files it will load.
2470 Essentially it now requires .core files to descend from the same
2471 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2472 as the runtime does. (The intent is to prevent the crashes which
2473 can occur, and which can even be reported as mysterious failures,
2474 when people patch the sources or change the build parameters
2475 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2476 sbcl and .core files.)
2477 * fixed bug: VALUES-LIST is no longer optimized away.
2478 * fixed bug 142: The FFI conversion of C string values to Lisp
2479 string values no longer conses excessively. (thanks to Nathan
2480 Froyd porting Raymond Toy's fix to CMU CL)
2481 * began to systematize and improve MOP conformance in PCL (thanks to
2482 Nathan Froyd, Gerd Moellman and Pierre Mai):
2483 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2485 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2486 ** FINALIZE-INHERITANCE is now called on class finalization;
2487 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2488 argument precedence order.
2489 * fixed bug 202: The compiler no longer fails on functions whose
2490 derived types contradict their declared type.
2491 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2492 so it can be non-toplevel.
2493 * The fasl file version number has changed (because of the new
2494 implementation of DEFMACRO).
2495 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2496 safe code, that any length requirement by their type-specifier
2497 argument is valid. The exceptions to this are described in bug
2499 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2500 functions now check the types of their inputs as required by ANSI.
2501 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2502 introduced symbol is DECLAREd to be SPECIAL.
2503 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2504 * fixed bug: :COUNT argument to sequence functions may be negative
2505 * fixed bug: body of DO-SYMBOLS may contain declarations
2506 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2507 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2508 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2509 symbol macro only once
2510 * fixed printing of call frame when argument list is unavailable
2511 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2512 * compiler no longer signals WARNING on unknown keyword
2515 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2516 * A beta-quality port to the mips architecture running Linux,
2517 based on the old CMUCL backend, has been made. It has been tested
2518 on a big-endian kernel, and works sufficiently well to be able to
2519 rebuild itself; it has not been tested in little-endian mode.
2520 * fixed an inconsistency between gencgc.c and purify.c which made
2521 dumping/loading .core files unreliable
2522 * fixed bug 120a: The compiler now deals correctly with IFs where
2523 the consequent is the same as the alternative, instead of
2524 misderiving the return type. (thanks to Alexey Dejneka)
2525 * fixed bug 113: Logical pathnames are now dumpable (the logical
2526 host is resolved at load-time, throwing an error if it is not
2528 * fixed bug 174: FORMAT's error message is slightly clearer when a
2529 non-printing character is used in a format directive.
2530 * fixed several bugs in compiler checking of type declarations, i.e.
2531 violations of the Python "declarations are assertions" principle
2532 (thanks to Alexey Dejneka)
2533 * fixed several bugs in PCL's error checking (thanks to Gerd
2535 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2537 * fixed bug in compilation of functions as first class values
2538 (thanks to Antonio Martinez-Shotton)
2539 * The compiler's handling TYPE-ERRORs which it can prove will
2540 inevitably happen at runtime has been cleaned up and corrected
2541 in several ways. (thanks to Alexey Dejneka)
2542 * improved argument type checking for various basic arithmetic
2543 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2544 lavished on them in the past that they can be compiled in many
2545 ways in different special cases
2546 * fixed bug 181: compiler checks validity of user supplied type
2548 * cleaned up code flushing in optimization: Function calls which
2549 should signal errors for safety purposes (e.g. which ANSI says
2550 should signal errors when their arguments are of incorrect type)
2551 are no longer optimized away.
2552 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2553 * incremented fasl file version number, because changes in the
2554 implementation of sequence functions like COERCE caused
2555 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2556 in old inline expansions) to become undefined. (Actually these
2557 changes were later undone, so we might very well be binary
2558 compatible with 0.7.7 after all, but leaving the version number
2559 incremented seemed like the simplest and most conservative
2562 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2563 * An alpha-quality port to the parisc architecture running Linux,
2564 based on the old CMUCL backend, has been made. This, even more so
2565 than the other backends, should be considered still a work in
2566 progress; known problems include that the Linux kernel in 64-bit
2567 mode does not propagate the correct sigcontext structure to
2568 userspace, and consequently SBCL on a parisc64 kernel will not
2570 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2571 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2572 suppressed.) Also now that the compiler is looking at declarations
2573 in the environment, it checks optimization declarations as well,
2574 and suppresses inlining when (> DEBUG SPEED).
2575 * More fixes have been made to treatment of floating point exception
2576 treatment and other Unix signals. In particular, floating point
2577 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2578 * The detection and handling of control stack exhaustion (infinite
2579 or very deeply nested recursion) has changed. Stack exhaustion
2580 detection is now done by write-protecting pages at the OS level
2581 and applies at all optimization settings; when found, a
2582 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2583 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2584 be caught using IGNORE-ERRORS.
2585 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2586 correctly when transforming recursive LABELS functions to LETs.
2587 (thanks to Alexey Dejneka)
2588 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2589 that are names of constants or global variables.
2590 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2591 alien routines with docstrings.
2592 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2593 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2595 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2596 (thanks to Eric Marsden)
2597 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2598 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2599 * Bug fix: The SPARC backend can now compile functions involving
2600 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2601 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2602 to a functions expecting a PATHNAME-DESIGNATOR.
2603 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2604 lambda lists. (thanks to Alexey Dejneka)
2605 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2606 functions so that EXPORTing the name of the class doesn't cause
2607 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2608 to Antonio Martinez for reporting this)
2609 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2610 constant-folded without failing an assertion. (thanks to Einar
2611 Floystad Dorum for reporting this)
2612 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2613 types are now dealt with more correctly by the compiler.
2614 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2615 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2616 behaviour analogous to RENAME-FILE. This puts its behaviour more
2617 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2619 * The fasl file version number has changed again. (because of the
2620 bug fix involving the names of PCL MAKE-INSTANCE functions)
2622 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2623 * bug fix: Floating point exceptions are treated much more
2624 consistently on the x86/Linux and PPC/Linux platforms.
2625 * Array initialization with :INITIAL-ELEMENT is now much faster for
2626 cases when the compiler cannot open code the array creation, but
2627 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2628 array accesses have also seen a speed increase.
2629 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2630 specified. (This was at the root of some bad interactions between
2631 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2633 * bug fix: Internal error arguments for undefined functions are now
2634 computed correctly on the PPC/Linux platform.
2635 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2636 Raymond Toy's patch for CMU CL)
2637 * Support for the Solaris 9 operating environment has been included
2638 (thanks to Daniel Merritt)
2639 * A very ugly but hopefully complete draft of the missing FFI chapter
2640 of the manual has been created by reformatting the corresponding
2641 CMU CL manual chapter into (currently very ugly and incoherent)
2642 DocBook and bringing it up to date for SBCL behavior. Thus, the
2643 manual is now essentially complete, at least by my extreme
2644 once-and-only-once standards, whereby it's acceptable to refer to
2645 the doc strings of SB-EXT functions as the primary documentation.
2646 * The fasl file version number has changed again, due to cleanup of
2647 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2648 is no longer a static symbol.)
2650 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2651 * SBCL now builds with OpenMCL (version 0.12) as the
2652 cross-compilation host; also, more progress has been made toward
2653 bootstrapping under CLISP.
2654 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2656 * bug 158 fixed: The compiler can now deal with integer loop
2657 increments different from 1; fixing this turned out also to fix
2659 * bug 169 fixed: no more bogus warnings about using lexical bindings
2660 despite the presence of perfectly good SPECIAL declarations (thanks
2661 to David Lichteblau)
2662 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2663 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2664 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2666 * bug 180 fixed: Method combination specifications no longer ignore
2667 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2668 * bug fix: Structure type predicate functions now check their argument
2669 count as they should.
2670 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2671 correctly. (thanks to Pierre Mai)
2672 * minor incompatible change: The --noprogrammer option is deprecated
2673 in favor of the new --disable-debugger option, which is very similar.
2674 (The major difference is that it takes effect at a slightly different
2675 time at startup, causing handling of errors in --sysinit and
2676 --userinit files will be affected differently.) The
2677 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2678 been added to allow this functionality to be controlled from ordinary
2679 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2680 maintainers, who might want to run non-interactive scripts to
2681 build SBCL cores which will later be used interactively.)
2682 * minor incompatible change: The LOAD function no longer, when given
2683 a wild pathname to load, loads all files matching that pathname.
2684 Instead, an error of type FILE-ERROR is signalled.
2686 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2687 * bug 147 fixed: The compiler preserves its block link/count
2688 invariants more correctly now so that it doesn't crash. (thanks
2690 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2692 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2693 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2694 no-value-for-this-character as NIL (as specified by ANSI).
2695 * HOST-NAMESTRING on physical pathnames now returns a string that is
2696 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2697 (thanks to Christophe Rhodes)
2698 * The Alpha port handles icache flushing more correctly. (thanks to
2700 * More progress has been made toward bootstrapping under CLISP. (thanks
2701 to Christophe Rhodes)
2702 * The fasl file format has changed again, because dynamic loading
2703 on OpenBSD (which has non-ELF object files) motivated some cleanups
2704 in the way that foreign symbols are transformed and passed around.
2705 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2706 is no longer treated as whitespace by the reader, but instead as
2707 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2708 |A
\7fB|, instead of A as it used to.
2710 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2711 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2712 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2713 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2714 this as of 0.7.1.45, but was left out of the previous news section
2715 (thanks to Dan Barlow)
2716 * SBCL now runs on the Solaris operating system on SPARC architectures
2717 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2718 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2719 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2720 * SPARC backend cleanups, allowing builds of cores optimized for V8
2721 and V9 SPARCS, and also emission of code targeted to a particular
2722 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2724 * SBCL is closer to bootstrapping under CLISP, thanks to various
2725 fixes by Christophe Rhodes.
2726 * The fasl file format has changed again, to allow the compiler's
2727 INFO database to support symbol macros.
2728 * The user manual (in doc/) is formatted into HTML more nicely.
2729 (thanks to coreythomas)
2730 * The system is smarter about SUBTYPEP relationships, especially
2731 those involving NOT types (including types such as ATOM which are
2732 represented internally using NOT types). Thus SUBTYPEP is less
2733 likely to return (VALUES NIL NIL) in general, and in particular
2734 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2736 * The fasl file format has changed again, because the internal
2737 representation of types now includes a new slot to support the new
2738 SUBTYPEP-of-NOT-types logic.
2739 * (not a change in the main branch of SBCL, but a related prototype
2740 which can hopefully be merged into the main branch of SBCL in the
2741 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2742 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2743 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2744 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2745 dispatch macro character combinations. (thanks to Alexey Dejneka)
2746 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2747 that superficially look like logical namestrings correctly.
2748 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2750 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2751 default supplied-p) for &optional and &key arguments. (thanks to
2754 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2755 * incompatible change: The compiler is now less aggressive about
2756 tail call optimization, doing it only when (> SPACE DEBUG) or
2757 (> SPEED DEBUG). (This is an incompatible change because there are
2758 programs which relied on the old CMU-CL-style behavior to optimize
2759 away their unbounded recursion which will now die of stack overflow.)
2760 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2761 for non-GENCGC systems has been increased to 20M (since that
2762 seems much closer to the likely performance optimum for modern
2763 systems than the old 4M value was)
2764 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2765 and *DEBUG-PRINT-LEVEL*
2766 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2767 of CMU CL's support for SPARC, and various endianness and other
2768 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2769 * new syntactic sugar for the Unix command line: --load foo.bar is now
2770 an alternate notation for --eval '(load "foo.bar")'.
2772 ** The system now detects stack overflow and handles it gracefully,
2773 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2774 optimization settings. (This is a good thing in general, and
2775 its introduction in this version should be particularly timely
2776 for anyone whose code fails because of suppression of tail
2778 ** The system now hunts for the C variable "environ" in a more
2779 devious way, to avoid segfaults when the C library version differs
2780 between compile time and run time. (thanks to Christophe Rhodes)
2781 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2782 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2783 ** The compiler no longer issues bogus style warnings for undefined
2784 classes in the same source file as the DEFCLASSes which defined
2785 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2787 ** fixes in CONDITION class precedence list for undefined function
2788 errors (thanks to Alexei Dejneka)
2789 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2790 correctly. (thanks to Dan Barlow)
2791 ** portability fixes aiming at bootstrapping under CLISP (thanks
2792 to Dave McDonald and Christophe Rhodes)
2793 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2794 ** fixes in type translation and and type inference (thanks to
2796 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2797 ** various fixes in the new ports (thanks to Dan Barlow)
2798 * several changes related to debugging:
2799 ** suppression of tail recursion, as noted above
2800 ** stack overflow detection, as noted above
2801 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2802 is now the default. (For some time encapsulation has been more
2803 reliable than the breakpoint-based :ENCAPSULATE NIL
2804 implementation, at least on X86 systems; and I just noticed that
2805 encapsulation also seems closer to the spirit of the ANSI
2808 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2810 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2811 up properly again. (There was a packaging bug in 0.7.0 which
2812 left their definitions in SB-SYS::LOAD-FOREIGN and
2813 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2814 vital for most things which interface to C-level interfaces,
2815 like extensions working with sockets or databases or
2816 Perl-compatible regexes or whatever, and the need to fix
2817 this bug is the main reason that 0.7.1 was released so
2819 ** DEFGENERIC is now choosier about the methods it redefines, so that
2820 reLOADing a previously-LOADed file containing DEFGENERICs does
2821 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2822 cycle now works as it should. (thanks to Alexey Dejneka)
2823 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2824 fixed by Christophe Rhodes. (He actually submitted this patch
2825 months ago, and I delayed until after 0.7.0.)
2826 ** Bug 111 (internal compiler confusion about runtime checks on
2827 FUNCTION types) was fixed by Alexey Dejneka.
2828 * Some internal cleanups (getting rid of variables which aren't
2829 needed now that the byte interpreter is gone) caused the fasl
2830 file format number to change again.
2832 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2833 * major incompatible change: The default fasl file extension, i.e. the
2834 default extension for files produced by COMPILE-FILE, has changed
2835 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2837 ** There are many changes in the implementation of the compiler.
2838 SBCL is now essentially a compiler-only implementation of ANSI
2839 Common Lisp. EVAL still "interprets" a few special cases, but
2840 almost all the interesting cases are handled by creating
2841 a LAMBDA expression, calling COMPILE on it, then calling
2842 FUNCALL on the result.
2843 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2844 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2845 Since the code is newer, there might still be some new bugs
2846 (though not as many as before Martin Atzmueller's fixes:-). But
2847 the new code is substantially simpler and clearer, and hopefully
2848 any remaining bugs will be simpler, less fundamental, and more
2849 fixable then the bugs in the old code.
2850 ** The revised compiler is still a little unsteady on its feet.
2852 *** The debugging information it produces (particularly the names
2853 of FUNCTION objects) is sometimes much less useful than what
2854 the old compiler produced.
2855 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2856 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2857 within a MACROLET) has been temporarily weakened.
2858 ** There are new compiler optimizations for various functions:
2859 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2860 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2861 *** the math functions TRUNCATE, FLOOR, and CEILING
2862 *** the function-of-all-trades COERCE
2863 Mostly these should be transparent, but there's one
2864 potentially-annoying problem (bug 117): when the compiler
2865 inline-expands a function and does type analysis on the result,
2866 it can create control paths which have type mismatches, and
2867 when it can't prove that those control paths aren't taken,
2868 it will issue WARNINGs about the type mismatches. This is
2869 a particular problem in practice for the new sequence functions.
2870 It's not clear how this should be fixed, and for now, a
2871 workaround is given in the entry for 117 in the BUGS file.
2872 ** (Because of the interaction between the two previous items --
2873 occasional inlining problems and new inline expansions -- some
2874 of the new sequence function optimizations won't really kick in
2875 completely until debugging information, and then inlining, are
2876 straightened out in some future version.)
2877 * minor incompatible changes:
2878 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2879 DIRECTORY no longer implicitly promotes NIL slots of its
2880 pathname argument to :WILD. In particular, when you ask for the
2881 contents of a directory (which you used to be able to do without
2882 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2883 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2884 ** changes in behavior that ANSI explicitly defines to be
2885 implementation dependent:
2886 *** The new compiler-only implementation still conforms with ANSI,
2887 but acts a little different than before. Besides the obvious
2888 changes in performance tradeoffs (that the cost per form passed
2889 to EVAL has gone up, and the cost per form executed by EVAL
2890 has gone down), the behavior of the system changes a little
2891 because there are no longer any interpreted function objects.
2892 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2893 e.g. doing COMPILE on the output of interactive DEFUN is
2895 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2897 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2898 has changed from (:CL) to NIL.
2899 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2900 e.g. #\Newline and #\Tab, don't have names specified in the
2901 ANSI Common Lisp standard, is now based on their ASCII symbolic
2902 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2903 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2904 are no longer used for output.
2905 ** changes in internal implementation constants:
2906 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2907 4 million. (If your application spends a lot of time GCing and
2908 you have a lot of RAM, you might want to experiment with
2909 increasing it even more.)
2910 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2911 However, almost all old code should still continue to work without
2912 immediate update, as SB-C-CALL is now a (deprecated) nickname
2914 ** Old operator names in the style DEF-FOO are now deprecated in
2915 favor of new corresponding names DEFINE-FOO, for consistency with
2916 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2917 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2918 internal symbols, but a few supported extensions like
2919 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2920 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2921 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2922 etc. as you get deeper into recursive calls to the debugger
2923 command loop, instead of the old "5]", "5]]", "5]]]"
2924 sequence. (I was motivated to do this when squabbles between
2925 ILISP and SBCL left me very deeply nested in the debugger. In the
2926 short term, this change will probably provoke more ILISP/SBCL
2927 squabbles, but hopefully it will be an improvement in the long run.)
2928 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2929 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2930 function, since declaiming return types involving aliens is
2931 (1) annoyingly messy to do by hand and (2) vital to efficient
2932 compilation of code which calls such functions.
2933 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2934 longer reexported by the SB-EXT package. They're solely useful
2935 for alien code, so it seems more logical that you should get
2936 them from the SB-ALIEN package, not in SB-EXT.
2937 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2938 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2939 features. Instead, the code that they used to control is always
2940 built into the system.
2941 * many other bug fixes
2942 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2943 advantage of the new EVAL-WHEN stuff and to clean them up in
2944 general, and they are now more ANSI-compliant in a number of
2945 ways. Martin Atzmueller is responsible for a lot of this.
2946 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2948 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2949 *** correct ERROR type for various file operations
2950 *** some fixes for Lisp streams
2951 *** DEFMETHOD syntax checking
2952 *** changing old weird representation of debug information as
2953 strings (which, among their other deficiencies, don't transform
2954 correctly when you rename packages, and don't change their
2955 print representation when you change things like *PACKAGE*
2956 and *PRINT-LENGTH*) to symbols and lists of symbols
2957 He also made several improvements and fixed several bugs in DESCRIBE.
2958 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2960 *** misbehavior of WRITE-STRING/WRITE-LINE
2961 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2962 and several other LOOP problems as well
2963 *** DIRECTORY when similar filenames are present
2964 *** DEFGENERIC with :METHOD options
2965 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2966 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2967 *** argument ordering in FIND with :TEST option
2968 *** mishandled package designator argument in APROPOS-LIST
2969 *** various problems in the backquote readmacro
2970 *** a bug in APROPOS
2971 *** probably some others that I'm not describing very well here,
2972 since the CVS log documents them by reference to sbcl-devel
2973 messages, and the SourceForge archives aren't working well.:-(
2974 ** Dan Barlow improved the Alpha port (and is making progress on the
2975 PPC port, for those of you who think different).
2976 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2977 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2978 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2979 continues to work on the SPARC port (for those of you in a position
2980 to look down upon our little PC-compatible boxes from a great height).
2981 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2982 and uses that copy, rather than the current dynamic value, when
2983 it's trying to decide whether to truncate output. Thus e.g.
2984 (let ((*print-lines* 50))
2985 (pprint-logical-block (stream nil)
2987 (let ((*print-lines* 8))
2988 (print (aref possiblybigthings i) stream)))))
2989 should now truncate the logical block only at 50 lines, instead of
2990 often truncating it at 8 lines, as it did before.
2991 * The doc/cmucl/ directory, containing old CMU CL documentation
2992 from the time of the fork, is no longer part of the base system.
2993 SourceForge has shut down its anonymous FTP service, and with it
2994 my original plan for distributing the old CMU CL documentation
2995 there. For now, if you need these files you can download an old
2996 SBCL source release and extract them from it.
2997 * The fasl file version number changed again, for dozens of reasons,
2998 some of which are apparent above.
3000 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3001 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3002 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3003 patch, so that DISASSEMBLE output is much nicer.
3004 * The code in the SB-PROFILE package now seems reasonably stable.
3005 I still haven't decided what the final interface should look like
3006 (I'd like PROFILE to interact cleanly with TRACE, since both
3007 facilities use function encapsulation) but if you have a need
3008 for profiling now, you can probably use it successfully with
3009 the current CMU-CL-style interface.
3010 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3011 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3012 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3013 link now returns the dangling link itself, and for similar
3014 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3015 link itself. (In these cases the old code signalled an error and
3016 looped endlessly, respectively.) Thus, DIRECTORY now works even
3017 in the presence of dangling and cyclic symbolic links.
3018 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3019 is now a supported extension again, since the consensus on
3020 sbcl-devel was that it can be useful for ordinary development
3021 work, not just for debugging SBCL itself.
3022 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3023 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3024 that currently-defined functions might be redefined later with
3025 different return types.
3026 * Hash tables can be printed readably, as inspired by CMU CL code
3027 of Eric Marsden and SBCL code of Martin Atzmueller.
3028 * better error handling in CLOS method combination, thanks to
3029 Martin Atzmueller porting Pierre Mai's CMU CL patches
3030 * more overflow fixes for >16Mbyte I/O buffers
3031 * A bug in READ has been fixed, so that now a single Ctrl-D
3032 character suffices to cause end-of-file on character streams.
3033 In particular, now you only need one Ctrl-D at the command
3034 line (not two) to exit SBCL.
3035 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3037 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3038 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3039 does the right thing.
3040 * The implementation of some type tests, especially for CONDITION
3041 types, is now tidier and maybe faster, due to CMU CL code
3042 originally by Douglas Crosher, ported by Martin Atzmueller.
3043 * Some math functions have been fixed, and there are new
3044 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3045 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3046 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3047 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3048 likely to be built into sbcl-0.7.0.)
3049 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3050 CL, ported to SBCL by Martin Atzmueller)
3051 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3052 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3053 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3054 * There's a new slam.sh hack to shorten the edit/compile/debug
3055 cycle for low-level changes to SBCL itself, and a new
3056 :SB-AFTER-XC-CORE target feature to control the generation of
3057 the after-xc.core file needed by slam.sh.
3058 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3059 COMPILE-FILE is no longer supported, so that now every function
3060 gets an entry point, so that block compilation looks a little
3061 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3062 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3063 returns the number of bytes consed since the system started,
3064 rather than the number consed since the first time the function
3065 was called. (The new definition parallels ANSI functions like
3066 CL:GET-INTERNAL-RUN-TIME.)
3067 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3068 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3069 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3070 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3071 * Fasl file version numbers are now independent of the target CPU,
3072 since historically most system changes which required version
3073 number changes have affected all CPUs equally. Similarly,
3074 the byte fasl file version is now equal to the ordinary
3077 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3078 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3079 is no longer recognized. I apologize for this, because it was
3080 listed in SB-EXT as a supported extension, but I found that
3081 its existing behavior was poorly specified, as well as incorrectly
3082 specified, and it looked like too much of a mess to straighten it
3083 out. I have enough on my hands trying to get ANSI stuff to work..
3084 * many patches ported from CMU CL by Martin Atzmueller, with
3085 half a dozen bug fixes in pretty-printing and the debugger, and
3086 half a dozen others elsewhere
3087 * fixed bug 13: Floating point infinities are now supported again.
3088 They might still be a little bit flaky, but thanks to bug reports
3089 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3090 as flaky as they were.
3091 * The --noprogrammer command line option is now supported. (Its
3092 behavior is slightly different in detail from what the old man
3093 page claimed it would do, but it's still appropriate under the
3094 same circumstances that the man page talks about.)
3095 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3096 are now supported, and enabled by default. Thus, the compiler can
3097 handle many floating point and complex operations much less
3098 inefficiently. (Thus e.g. you can implement a complex FFT
3100 * The compiler now detects type mismatches between DECLAIM FTYPE
3101 and DEFUN better, and implements CHECK-TYPE more correctly, and
3102 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3103 thanks to patches from Martin Atzmueller.
3104 * various fixes to make the cross-compiler more portable to
3105 ANSI-conforming-but-different cross-compilation hosts (notably
3106 Lispworks for Windows, following bug reports from Arthur Lemmens)
3107 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3108 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3109 fixed thanks to Pierre Mai's CMU CL patches.
3110 * fixes to keep the system from overflowing internal counters when
3111 it tries to use i/o buffers larger than 16M bytes
3112 * fixed bug 45a: Various internal functions required to support
3113 complex special functions have been merged from CMU CL sources.
3114 (When I was first setting up SBCL, I misunderstood a compile-time
3115 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3116 * improved support for type intersection and union, fixing bug 12
3117 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3118 more obscure bugs as well
3119 * some steps toward byte-compiling non-performance-critical
3120 parts of the system, courtesy of patches from Martin Atzmueller
3121 * Christophe Rhodes has made some debian packages of sbcl at
3122 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3123 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3124 stable, but are nonetheless usable. When he's ready, I'd be happy
3125 to add them to the SourceForge "File Releases" section. (And if
3126 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3127 * new fasl file format version number (because of changes in
3128 internal representation of (OR ..) types to accommodate the new
3129 support for (AND ..) types, among other things)
3131 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3132 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3134 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3136 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3137 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3138 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3139 * DESCRIBE now works on structure objects again.
3140 * Most function call argument type mismatches are now handled as
3141 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3142 know whether the function will be redefined before the call is
3143 executed. (The compiler could flag local calls with full WARNINGs,
3144 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3145 it doesn't keep track of enough information to know whether calls
3146 are local in this sense.)
3147 * Compiler output is now more verbose, with messages truncated
3148 later than before. (There should be some supported way for users
3149 to override the default verbosity, but I haven't decided how to
3150 provide it yet, so this behavior is still controlled by the internal
3151 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3152 src/compiler/ir1util.lisp.)
3153 * Fasl file format version numbers have increased again, because
3154 support for the Gray streams extension changes the layout of the
3155 system's STREAM objects.
3156 * The Gray subclassable streams extension now works, thanks to a
3157 patch from Martin Atzmueller.
3158 * The full LOAD-FOREIGN extension (not just the primitive
3159 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3160 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3161 but like most other programs, it defaults to copying the Unix
3162 environment from the original process instead of starting the
3163 new process in an empty environment.
3164 * Extensions which manipulate the Unix environment now support
3165 an :ENVIRONMENT keyword option which doesn't smash case or
3166 do other bad things. The CMU-CL-style :ENV option is retained
3167 for porting convenience.
3168 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3169 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3171 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3173 * A patch from Martin Atzmueller seems to have solved the SIGINT
3174 problem, and as far as we know, signal-handling now works cleanly.
3175 (If you find any new bugs, please report them!)
3176 * The system no longer defaults Lisp source file names to types
3177 ".l", ".cl", or ".lsp", but only to ".lisp".
3178 * The compiler no longer uses special default file extensions for
3179 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3180 seems to expect a single default extension for all compiled code,
3181 and there's no compelling reason to try to stretch the standard
3182 to allow two different extensions.) Instead, byte-compiled files
3183 default to the same extension as native-compiled files.
3184 * Fasl file format version numbers have increased again, because
3185 a rearrangement of internal implementation packages made some
3186 dumped symbols in old fasl files unreadable in new cores.
3187 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3188 many fewer weird special cases.
3189 * Bug #17 (differing COMPILE-FILE behavior between logical and
3190 physical pathnames) has been fixed, and some related misbehavior too,
3191 thanks to a patch from Martin Atzmueller.
3192 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3193 by Tim Moore, ported to SBCL by Martin Atzmueller.
3194 * Martin Atzmueller fixed several filesystem-related problems,
3195 including bug #36, in part by porting CMU CL patches, which were
3196 written in part by Paul Werkowski.
3197 * More compiler warnings in src/runtime/ are gone, thanks to
3198 more patches from Martin Atzmueller.
3199 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3202 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3204 * DESCRIBE now works on CONDITION objects.
3205 * The debugger now handles errors which arise when trying to print
3206 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3208 * The build system now uses an additional file, customize-target-features.lisp,
3209 to allow local modifications to the target *FEATURES* list. (The point of
3210 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3211 debugging features enabled, without having to worry about propagating it
3212 into everyone's system when I do a "cvs update".) When no
3213 customize-target-features.lisp file exists, the target *FEATURES* list
3214 should be constructed the same way as before.
3215 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3216 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3217 warn about behavior which is undefined under the ANSI spec. Note: This
3218 is specified by ANSI, but it's not very popular with programmers.
3219 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3220 macro in the SBCL sources for an example of a workaround which you
3221 might use to make portable ANSI-standard code which does what you want.
3222 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3223 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3224 when it's used as a non-toplevel form. (This is required in order
3225 to implement the DEFCONSTANT-EQX macro.)
3226 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3227 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3228 funny things, due to bugs in the implementation of EVAL-WHEN
3229 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3230 non-toplevel DEFCONSTANTs are uncommon.)
3231 * The core file version number and fasl file version number have been
3232 incremented, because the old noncompliant DEFCONSTANT behavior involved
3233 calling functions which no longer exist, and because I also took the
3234 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3235 * fixed bug 1 (error handling before read-eval-print loop starts), and
3236 redid debugger restarts and related debugger commands somewhat while
3238 ** The QUIT debugger command is gone, since it did something
3239 rather different than the SB-EXT:QUIT command, and since it never
3240 worked properly outside the main toplevel read/eval/print loop.
3241 Invoking the new TOPLEVEL restart provides the same functionality.
3242 ** The GO debugger command is also gone, since you can just invoke
3243 the CONTINUE restart directly instead.
3244 ** The TOP debugger command is also gone, since it's redundant with the
3245 FRAME 0 command, and since it interfered with abbreviations for the
3247 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3249 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3250 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3251 some porting work by Martin Atzmueller)
3252 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3253 with a boatload of patches which he ported to SBCL. Now that those
3255 ** The system tries to make sure that its low-priority messages
3256 are prefixed by semicolons, to help people who like to use
3257 syntax highlighting in their ILISP buffer. (This patch
3258 was originally due to Raymond Toy.)
3259 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3260 patches originally written by Raymond Toy.
3261 ** The compiler understands coercion between single-value and
3262 multiple-VALUES type expressions better, getting rid of some very
3263 weird behavior, thanks to patches originally by Robert MacLachlan
3264 and Douglas Crosher.
3265 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3266 lambda lists, thanks to a patch originally by Pierre Mai.
3267 ** The system no longer bogusly warns about "abbreviated type
3269 ** The compiler gets less confused by inlining and RETURN-FROM,
3270 thanks to some patches originally by Tim Moore.
3271 ** The system no longer hangs when dumping circular lists to fasl
3272 files, thanks to a patch originally from Douglas Crosher.
3273 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3274 undefined function error.
3275 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3276 in structure slot initforms) for now, documented workaround instead:-|
3277 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3278 * fixed bug 5: added stubs for various Gray stream functions called
3279 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3280 installed, at least appropriate type errors are generated
3281 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3282 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3283 consistently in DEFMETHOD forms.
3284 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3285 it doesn't seem to affect SBCL after all
3286 * The C runtime system now builds with better optimization and many
3287 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3289 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3291 * The system is now under CVS at SourceForge (instead of the
3292 CVS repository on my home machine).
3293 * The new signal handling code has been tweaked to treat register
3294 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3295 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3296 downstream, e.g. in debug-int.lisp, has implicit dependencies
3297 on the unsignedness of integer representation of machine words,
3298 and that caused the system to bomb out with infinite regress
3299 when trying to recover from type errors involving signed values,
3300 e.g. (BUTLAST '(1 2 3) -1).)
3301 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3302 (This was one of the bugs Peter Van Eynde reported back in July.)
3303 * The system now uses code inspired by Colin Walters' O(N)
3304 implementation of MAP (from the cmucl-imp@cons.org mailing
3305 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3306 inline the MAP operation, and there is more than one
3307 sequence argument to the MAP call (so that it can't just
3308 do ETYPECASE once and for all based on the type of the
3309 single sequence argument). (The old non-inline implementation
3310 of the general M-argument sequence-of-length-N case required
3311 O(M*N*N) time when any of the sequence arguments were LISTs.)
3312 * The QUIT :UNIX-CODE keyword argument has been renamed to
3313 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3314 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3316 * Raymond Wiker's patches to port dynamic loading from Linux to
3317 FreeBSD have been added.
3318 * The BUGS file is now more nearly up to date, thanks in large part
3319 to Martin Atzmueller's review of it.
3320 * The debugger now flushes standard output streams before it begins
3321 its output ("debugger invoked" and so forth).
3322 * The core version number and fasl file version number have both
3323 been incremented, because of incompatible changes in the layout
3325 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3326 used to not be called for a saved Lisp image.)
3327 * Martin Atzmueller's version of a patch to fix a compiler crash,
3328 as posted on sbcl-devel 13 September 2000, has been installed.
3329 * Instead of installing Martin Atzmueller's patch for the
3330 compiler transform for SUBSEQ, I deleted the compiler transform,
3331 and transforms for some similar consing operations.
3332 * A bug in signal handling which kept TRACE from working on OpenBSD
3334 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3335 away by constant folding
3336 * The system now defines its address space constants in one place
3337 (in the Lisp sources), and propagates them automatically elsewhere
3338 (through GENESIS and the sbcl.h file). Therefore, patching the
3339 address map is less unnecessarily tedious and error-prone. The
3340 Lisp names of address space constants have also been systematized.
3341 * CVS tags like dollar-Header-dollar have been removed from
3342 the sources, because they have never saved me trouble and
3343 they've been source of trouble working with patches and other
3344 diff-related operations.
3345 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3346 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3348 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3350 * The system has been ported to OpenBSD.
3351 * The system now compiles with a simple "sh make.sh" on the systems
3352 that it's supported on. I.e., now you no longer need to tweak
3353 text in the target-features.lisp-expr and symlinks in src/runtime/
3354 by hand, the make.sh takes care of it for you.
3355 * The system is no longer so grossly inefficient when compiling code
3356 involving vectors implemented as general (not simple) vectors (VECTOR T),
3357 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3358 which dares to use the various sequence functions on non-simple
3359 vectors, takes less of a performance hit.
3360 * There is now a primitive type predicate VECTOR-T-P
3361 to test for the (VECTOR T) type, so that e.g.
3362 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3363 can now be compiled with some semblance of efficiency. (The old code
3364 turned the type declaration into a full call to %TYPEP at runtime!)
3365 * AREF on (VECTOR T) is still not fast, since it's still compiled
3366 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3367 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3368 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3369 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3370 without making full calls to SUBTYPEP at runtime.
3371 (Some analogous efficiency issues for non-simple vectors specialized to
3372 element types other than T, or for non-simple multidimensional arrays,
3373 have not been addressed. They could almost certainly be handled the
3374 same way if anyone is motivated to do so.)
3375 * The changes in array handling break binary compatibility, so
3376 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3377 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3378 instead of (VALUES T T).
3379 * By following the instructions that Dan Barlow posted to sbcl-devel
3380 on 2 July 2000, I was able to enable primitive dynamic object
3381 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3382 functionality is not implemented (since it calls ld to resolve
3383 library references automatically, requiring RUN-PROGRAM for its
3384 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3385 try to resolve library references) is now supported.
3386 * The system now flushes the standard output streams when it terminates,
3387 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3388 them at several other probably-convenient times, e.g. in each pass of
3389 the toplevel read-eval-print loop, and after evaluating a form given
3390 as an "--eval" command-line option. (These changes were motivated by a
3391 discussion of stream flushing issues on cmucl-imp in August 2000.)
3392 * The source transform for TYPEP of array types no longer assumes
3393 that an array whose element type is a not-yet-defined type
3394 is implemented as an array of T, but instead punts, so that the
3395 type will be interpreted at runtime.
3396 * There is now some support for cross-compiling in make.sh: each of
3397 the phases of make.sh has its own script. (This should be transparent
3398 to people doing ordinary, non-cross-compile builds.)
3399 * Since my laptop doesn't have hundreds of megabytes of memory like
3400 my desktop machine, I became more motivated to do some items on
3401 my to-do list in order to reduce the size of the system a little:
3402 ** Arrange for various needed-only-at-cold-init things to be
3403 uninterned after cold init. To support this, those things have
3404 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3405 symbols with such names are now uninterned after cold init).
3406 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3407 fasl files for cold load.
3408 ** Remove the old compiler structure pooling code (which used to
3409 be conditional on the target feature :SB-ALLOC) completely.
3410 ** Redo the representation of some data in cold init to be more compact.
3411 (I also looked into supporting byte compiled code at bootstrap time,
3412 which would probably reduce the size of the system a lot, but that
3413 looked too complicated, so I punted for now.)
3414 * The maximum signal nesting depth in the src/runtime/ support code has
3415 been reduced from 4096 to 256. (I don't know any reason for the very
3416 large old value. If the new smaller value turns out to break something,
3417 I'll probably just bump it back up.)
3418 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3420 * Many, many bugs reported by Peter Van Eynde have been added to
3421 the BUGS list; some have even been fixed.
3422 * While enabling dynamic object file loading, I tried to make the
3423 code easier to understand, renaming various functions and variables
3424 with less ambiguous names, and changing some function calling
3425 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3426 * While trying to figure out how to do the OpenBSD port, I tried to
3427 clean up some of the code in src/runtime/. In particular, I dropped
3428 support for non-POSIX signal handling, added various comments,
3429 tweaked the code to reduce the number of compilation warnings, and
3430 renamed some files to increase consistency.
3431 * To support the new automatic configuration functionality in make.sh,
3432 the source file target-features.lisp-expr has been replaced with the
3433 source file base-target-features.lisp-expr and the machine-generated
3434 file local-target-features.lisp-expr.
3435 * fixed a stupid quoting error in make.sh so that using CMU CL
3436 "lisp -batch" as cross-compilation host works again
3438 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3440 * DESCRIBE no longer tries to call itself recursively to describe
3441 bound/fbound values, so that it no longer fails on symbols which are
3442 bound to themselves (like keywords, T, and NIL).
3443 * DESCRIBE now works on generic functions.
3444 * The printer now prints less-screwed-up representations of closures
3445 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3446 * A private symbol is used instead of the :EMPTY keyword previously
3447 used to mark empty slots in hash tables. Thus
3448 (DEFVAR *HT* (MAKE-HASH-TABLE))
3449 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3450 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3451 now does what ANSI says that it should. (You can still get
3452 similar noncompliant behavior if bang on the hash table
3453 implementation with all the symbols you get back from
3454 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3455 This breaks binary compatibility, since tests for equality to
3456 :EMPTY are wired into things like the macroexpansion of
3457 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3459 * There's now a minimal placeholder implementation for CL:STEP,
3460 as required by ANSI.
3461 * An obscure bug in the interaction of the normal compiler, the byte
3462 compiler, inlining, and structure predicates has been patched
3463 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3464 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3466 * Missing ordinary arguments in a macro call are now detected even
3467 when the macro lambda list contains &KEY or &REST.
3468 * The debugger no longer complains about encountering the top of the
3469 stack when you type "FRAME 0" to explicitly instruct it to go to
3470 the top of the stack. And it now prints the frame you request even
3471 if it's the current frame (instead of saying "You are here.").
3472 * As specified by ANSI, the system now always prints keywords
3473 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3475 * The default initial SIZE of HASH-TABLEs is now smaller.
3476 * Type information from CLOS class dispatch is now propagated
3477 into DEFMETHOD bodies, so that e.g.
3478 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3480 is now basically equivalent to
3481 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3482 (DECLARE (TYPE SINGLE-FLOAT X))
3484 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3485 operation, without having to do run-time type dispatch.
3486 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3487 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3488 * Since I don't seem to be making big file reorganizations very often
3489 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3490 is overflowing my ability to conveniently back them up), I've finally
3491 checked the system into CVS. (The CVS repository is on my home system,
3492 not at SourceForge -- putting it on SourceForge might come later.)
3493 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3494 high-level GC-NOTIFY-FOO functions send their output. (There's
3495 still very little control of where low-level verbose GC functions
3496 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3497 less than it used to -- the GC-NOTIFY-FOO functions are now under
3498 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3499 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3500 in only one place in the source code, and propagates it automatically
3501 everywhere that it's needed. Thus e.g. when I bump the version from
3502 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3503 * The C source files now include boilerplate legalese and documentation
3504 at the head of each file (just as the Lisp source files already did).
3505 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3506 to his page will be replaced with a link to his new CLiki service.
3508 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3510 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3511 * The build process now looks for GNU make under the default name "gmake",
3512 instead of "make" as it used to. If GNU make is not available as "gmake"
3513 on your system, you can change this default behavior by setting the
3514 GNUMAKE environment variable.
3515 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3516 can build without error under CMU CL.
3518 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3520 * There is now a partial SBCL user manual (with some new text and some
3521 text cribbed from the CMU CL manual).
3522 * The beginnings of a profiler have been added (starting with the
3523 CMU CL profiler and simplifying and cleaning up). Eventually the
3524 main interface should be through the TRACE macro, but for now,
3525 it's still accessed through vaguely CMU-CL-style functions and macros
3526 exported from the package SB-PROFILE.
3527 * Some problems left over from porting CMU CL to the new
3528 cross-compilation bootstrap process have been cleaned up:
3529 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3530 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3531 part when getting rid of the compiler *BACKEND* stuff.)
3532 ** The value of *NULL-TYPE* was screwed up, because it was
3533 being initialized before the type system knew the final
3534 definition of the 'NULL type. This screwed up several key
3535 optimizations in the compiler, causing inefficiency in all sorts
3536 of places. (I found it because I wanted to understand why
3537 GET-INTERNAL-RUN-TIME was consing.)
3538 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3539 PROCLAIM FTYPE information. Unfortunately this broke binary
3540 compatibility again, since now the forms output by DEFGENERIC
3541 to refer to functions which didn't exist in 0.6.3.
3542 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3543 can use the new (as of 0.6.3) transform for SOME into MAP into
3545 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3546 values to (INTEGER 0 1000000), so that the time code will no longer
3547 occasionally get blown up by Linux returning 1000000 microseconds
3548 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3549 its output conform to the ANSI spec. (Alas, this makes its output
3550 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3551 * A full call to MAP NIL with a single sequence argument no longer conses.
3552 * fixes to problems pointed out by Martin Atzmueller:
3553 * The manual page no longer talks about multiprocessing as though
3554 it were currently supported.
3555 * The ILISP support patches have been removed from the distribution,
3556 because as of version 5.10.1, ILISP now supports SBCL without us
3557 having to maintain patches.
3558 * added a modified version of Raymond Toy's recent CMU CL patch for
3559 EQUALP comparison of HASH-TABLE
3561 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3563 * The system still can't cross-compile itself with
3564 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3565 entails), but at least it can compile more of itself that way
3566 than it used to be able to, and various buglets which were uncovered
3567 by trying to cross-compile itself that way have now been fixed.
3568 * This release breaks binary compatibility again. This time
3569 at least I've incremented the FASL file format version to 2, so that the
3570 problem can be detected reliably instead of just causing weird errors.
3571 * various new style warnings:
3572 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3573 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3574 ** using the lexical binding of a variable named in the *FOO* style
3575 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3576 as specified by ANSI.
3577 * *RANDOM-STATE* is no longer automatically initialized from
3578 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3579 default behavior of the system is to repeat its behavior every time
3580 it's run. If you'd like to change this behavior, you can always
3581 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3582 old convention there was no comparably easy way to get the system to
3583 repeat its behavior every time it was run.
3584 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3585 FTYPE declarations has been removed, in favor of their ANSI
3586 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3587 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3588 the types of their sequence arguments can be determined at compile time.
3589 This is done through a new open code expansion for MAP which eliminates
3590 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3591 argument types can be determined at compile time.
3592 * The optimizer now transforms COERCE into an identity operation when it
3593 can prove that the coerced object is already of the correct type. (This
3594 can be a win for machine generated code, including the output of other
3595 optimization transforms, such as the MAP transform above.)
3596 * Credit information has been moved from source file headers into CREDITS.
3597 * Source file headers have been made more standard.
3598 * The CASE macro now compiles without complaining even when it has
3601 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3603 * (Note that the way that the PCL macroexpansions were rewritten
3604 to accommodate the change in DEFGENERIC below breaks binary
3605 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3606 not run under sbcl-0.6.2. Once we get out of alpha releases,
3607 i.e. hit release 1.0.0, we'll probably try to maintain binary
3608 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3609 and sbcl-1.4.4. Until then, however, it might be fairly common
3610 for maintenance releases to break binary compatibility.)
3611 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3613 * The reporting of unhandled signals has been changed to print some
3614 explanatory text as well as the report form. (Previously only
3615 the report form was printed.)
3616 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3617 it defines, so that the compiler no longer issues undefined function
3618 warnings for compiled-but-not-yet-loaded generic functions.
3619 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3620 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3621 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3622 section of the ANSI spec are supported.
3623 * The "" nickname for the "KEYWORD" package has been removed.
3624 The reader still handles symbol tokens which begin with a package marker
3625 as keywords, but it doesn't expose its mechanism for doing so in the
3626 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3627 * The system now issues STYLE-WARNINGs for contradictory TYPE
3628 proclamations. (Warnings for contradictory FTYPE proclamations would
3629 be nice too, but those can't be done usefully unless the type system
3630 is made smarter about FUNCTION types.)
3631 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3632 systematized, so that "*target-*.lisp is supposed to exist only on the
3633 target and imply that there's a related file which exists on the
3634 host, and *host-*.lisp is supposed to exist only on the host and imply
3635 that there's a related file which exists on the target. This involves a
3636 lot of renaming. Hopefully the acute confusion caused by the renaming
3637 will be justified by the reduction in chronic confusion..
3638 ** runtime-type.lisp -> early-target-type.lisp
3639 ** target-type.lisp -> late-target-type.lisp
3640 ** early-host-format.lisp -> early-format.lisp
3641 ** late-host-format.lisp -> late-format.lisp
3642 ** host-error.lisp -> misc-error.lisp
3643 ** early-error.lisp -> early-target-error.lisp
3644 ** late-error.lisp -> late-target-error.lisp
3645 ** host-defboot.lisp -> early-defboot.lisp
3646 ** code/misc.lisp -> code/target-misc.lisp
3647 ** code/host-misc.lisp -> code/misc.lisp
3648 ** code/numbers.lisp -> code/target-numbers.lisp
3649 ** code/early-numbers.lisp -> numbers.lisp
3650 ** early-host-type.lisp -> early-type.lisp
3651 ** late-host-type.lisp -> late-type.lisp
3652 ** host-typep.lisp -> typep.lisp
3653 ** load.lisp -> target-load.lisp
3654 ** host-load.lisp -> load.lisp
3655 ** host-disassem.lisp -> disassem.lisp
3656 ** host-insts.lisp -> insts.lisp
3657 ** byte-comp.lisp -> target-byte-comp.lisp
3658 ** host-byte-comp.lisp -> byte-comp.lisp
3659 ** host-signal.lisp -> signal.lisp
3660 ** host-defstruct.lisp -> defstruct.lisp
3661 ** late-target-type.lisp -> deftypes-for-target.lisp
3662 Furthermore, several other previously target-only files foo.lisp (e.g.
3663 hash-table.lisp and random.lisp) have been split into a target-and-host
3664 foo.lisp file and a target-only target-foo.lisp file, with their key type
3665 definitions in the target-and-host part, so that the cross-compiler will
3666 know more about target types.
3667 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3668 gone away. In their place are various *BACKEND-FOO* variables
3669 corresponding to the slots of the old structure.
3670 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3671 into the target SBCL has been fixed.
3672 * The chill.lisp system for loading cold code into a running SBCL
3674 * Support for the CMU CL "scavenger hook" extension has been removed.
3675 (It was undocumented and unused in the CMU CL sources that SBCL was
3676 derived from, and stale in sbcl-0.6.1.)
3677 * Various errors in the cross-compiler type system were detected
3678 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3679 (enabling various consistency checks). Many of them were fixed,
3680 but some hard problems remain, so the compiler is back to
3681 running without *TYPE-SYSTEM-INITIALIZED* for now.
3682 * As part of the cross-compiler type system cleanup, I implemented
3683 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3684 * I have started adding UNCROSS calls throughout the type system
3685 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3686 on cross-compiler input files will be able to go away, and various
3688 * CONSTANTP now returns true for quoted forms (as explicitly required
3691 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3693 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3694 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3695 (MAKE-LIST -1). In the longer term, it ought to become true
3696 that declarations are assertions even at SAFETY 1. For now, it's not
3697 quite true even at SAFETY 3, but it's at least more nearly true..
3698 (Note that this change seems to increases the size of the system by
3699 O(5%) and to decrease the speed of the compiler by 20% or more.)
3700 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3701 for the problem of printing dozens of lines of distracting information
3702 about low-level system machinery as part of the top stack frame
3703 on entry to the debugger when an undefined function was called.
3704 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3705 is not reset to COMMON-LISP-USER.
3706 * Compilation of stuff related to dyncount.lisp has been made conditional
3707 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3708 smaller. The various dyncount-related symbols have been moved into
3709 a new "SB-DYNCOUNT" package.
3710 * tty-inspect.lisp has been renamed to inspect.lisp.
3711 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3712 feature has gone away. (When we eventually port to other flavors of
3713 libc and/or Unix, we'll try to make the differences between flavors
3714 invisible at the user level.)
3715 * Various other *FEATURES* tags, and/or their associated conditionals,
3716 have been removed if obsolescent, or given better documentation, or
3717 sometimes given more-mnemonic names.
3719 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3721 * tidied up "make.sh" script
3722 * tidied up system directory structure
3723 * better "clean.sh" behavior
3724 * added doc/FOR-CMUCL-DEVELOPERS
3725 * many many small tweaks to output format, e.g. removing possibly-confusing
3726 trailing #\. character in DESCRIBE-INSTANCE
3727 * (EQUALP #\A 'A) no longer signals an error.
3728 * new hashing code, including EQUALP hashing
3729 * tidied up Lisp initialization and toplevel
3730 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3731 * command line argument processing
3732 * added POSIX-GETENV function to deal with Unix-ish environment variables
3733 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3734 terminating SBCL on EOF
3735 * non-verbose GC by default
3736 * There is no more "sbcl" shell script; the sbcl file is now the C
3737 runtime executable (just like CMU CL).
3738 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3740 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3741 symbol and package information as Lisp native symbol and package objects
3742 instead of strings naming symbols and strings naming packages. This way,
3743 whenever packages are renamed (as in warm init), debug information is
3744 transformed along with everything else.
3745 * tweaked the optimization policy declarations which control the building
3746 of SBCL itself. Now, among other things, the system no longer saves
3747 source location debugging information. (This helps two problems at once
3748 by reducing SBCL size and by keeping SBCL from trying to look for its
3749 sources -- which may not exist -- when reporting errors.)
3750 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3751 debugging and testing purposes
3752 * cleaned up printing, making the printer call PRINT-OBJECT for
3753 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3754 methods, giving nearly-ANSI behavior
3755 * converted almost all special variables to use *FOO* naming convention
3756 * deleted PARSE-TIME functionality, since it can be done portably
3757 * moved some files out of cold init into warm init
3758 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3760 * regularized formatting of source files
3761 * added an install.sh script
3762 * fixed ridiculous memory usage of cross-compiler by making
3763 compiler/alloc.lisp not try to do pooling unless it can hook
3764 itself into the GC of the cross-compilation host. Now the system
3765 builds nicely on my old laptop.
3766 * added :SB-ALLOC in target-features.lisp-expr
3767 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3768 was not implemented)
3769 * re-did condition handling and note reporting in the compiler. Notes
3770 are no longer handled by signalling conditions. Style warnings
3771 and warnings are handled more correctly and reported in such a way
3772 that it's easy to find one or the other in your output (so that you
3773 can e.g. figure out which of many problems caused COMPILE-FILE to
3775 * changed the severity of several compiler warnings from full WARNING
3776 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3777 compiler note reporting so that it doesn't use the condition system
3778 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3779 COMPILE-FILE command)
3780 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3781 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3782 no longer supported, since their implementation was deeply intertwingled
3783 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3784 * removed bogus "support" for compiler macros named (SETF FOO), and
3785 removed the compiler macro for SETF INFO (but only after making a fool
3786 of myself on the cmucl-imp mailing list by posting a bogus patch for
3787 DEFINE-COMPILER-MACRO..)
3788 * Compiled files containing forms which have side effects on the Lisp
3789 reader (such as DEFPACKAGE forms) are now handled more correctly.
3790 (Compiler queuing of top level lambdas has been suppressed by setting
3791 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3792 * deleted various currently-unused source files, e.g. gengc.lisp. They
3793 may be added back at some point e.g. when porting to other architectures,
3794 but until they are it's distracting to distribute them and to try to
3796 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3797 there were so many of them they're just distractions, and UNCROSS is
3798 known to be able to handle the current sources
3799 * moved PROFILE functionality into TRACE, so that it will be clear
3800 how the wrapping and unwrapping of functions when you profile them
3801 interacts with the wrapping and unwrapping of functions when you
3802 trace them. (Actually, the functionality isn't there yet, but at least
3803 the interface specification is there. Hopefully, the functionality will
3804 arrive with some future maintenance release.)
3805 * removed host-oops.lisp
3806 * changed signature of QUIT function to allow UNIX-CODE argument
3807 * fixed READ-SEQUENCE bug
3808 * tweaked verbose GC output so that it looks more like the progress
3809 output that ANSI specifies for functions like LOAD
3810 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3811 * added <http://sbcl.sourceforge.com> to the banner information printed by