1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.1 relative to sbcl-1.0:
3 * improvement: sb-sprof traces call stacks to an arbitrary depth on
4 x86/x86-64, rather than the previous fixed depth of 8
5 * bug fix: non-ascii command-line arguments are processed correctly
6 (thanks to Yaroslav Kavenchuk)
7 * bug fix: TRACE :ENCAPSULATE NIL (and function end breakpoints)
9 * bug fix: non-required arguments were not passed correctly when a method
10 defined using DEFMETHOD was called from a mop-generated method using
11 CALL-NEXT-METHOD (reported by Pascal Costanza)
12 * optimization: the FIND and POSITION family of sequence functions
13 are significantly faster on arrays whose element types have been
16 changes in sbcl-1.0 relative to sbcl-0.9.18:
17 * improvement: experimental support for threading on FreeBSD/x86.
18 (thanks to NIIMI Satoshi)
19 * improvement: runtime option --dynamic-space-size can be used
20 to set the size of the dynamic space reserved on startup.
21 * improvement: floating point modes in effect are now saved in
22 core, and restored on startup.
23 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
24 startup, not time since first call to GET-INTERNAL-REAL-TIME.
25 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
26 threads are not running after *SAVE-HOOKS* have run.
27 * improvement: writes to CLOS instance slots are type-checked in code
28 compiled with (SAFETY 3)
29 * improvement: floating-point exception handling on FreeBSD (thanks to
31 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
32 (thanks to Zach Beane)
33 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
35 * improvement: added support for the Shift-JIS external format.
36 (contributed by NIIMI Satoshi)
37 * improvement: callbacks are supported on Linux/PPC. (thanks to
39 * bug fix: compiler bug triggered by a (non-standard) VALUES
40 declaration in a LET* was fixed. (reported by Kaersten Poeck)
41 * bug fix: file compiler no longer confuses validated and already
42 dumped structurres. (reported by Kaersten Poeck)
43 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
44 fill-pointer signals a type-error as required. (thanks to
46 * bug fix: disassemly of funcallable instances works.
47 * bug fix: single stepping on PPC.
48 * bug fix: fix thread-safety problems in the type system (generally
49 manifesting as nonsensical errors like "STRING is a bad type specifier
50 for sequences" or "The value 1 is not of type FIXNUM")
51 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
52 (reported by Josip Gracin)
53 * bug fix: an error is signaled for attempts to displace arrays with
54 incompatible element types (thanks to Mario Mommer)
55 * bug fix: more correct handling of wide characters in debug info
56 (bug reported by Attila Lendvai and fixed by Juho Snellman)
57 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
58 and don't cause extra consing
59 * optimization: MAP and MAP-INTO are significantly faster on vectors
60 whose elements types have been declared.
61 * Improvements to SB-SPROF:
62 ** Support for allocation profiling
63 ** Reduced profiling overhead, especially for long profiling runs
64 * Improvements to the Windows port:
65 ** floating point exceptions are now reported correctly.
66 ** stack exhaustion detection works partially.
67 ** more accurate GET-INTERNAL-REAL-TIME.
68 ** better RUN-PROGRAM behaviour in case of errors in the child
70 ** PROBE-FILE now simplifies pathnames correctly.
71 ** DIRECTORY now works correctly with :WILD-INFERIORS.
72 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
73 1st 1970 (although time timezone data will be unreliable).
74 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
76 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
77 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
78 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
80 * bug fix: two potential GC deadlocks affecting threaded builds.
81 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
82 returns T (reported by Anton Kazennikov)
83 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
84 with non-variable places
85 * bug fix: remove a race condition in the setting of
86 funcallable-instance functions, this should make threaded CLOS
87 code more stable against memory faults.
88 * bug fix: corruption of specials when unbinding is interrupted by an
89 asynchronous unwind (reported by Hannu Koivisto)
90 * improvement: the debugger will now also display local variables that
91 are only used once, for code compiled with a DEBUG optimization quality
94 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
95 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
96 * incompatible change: External-format support for FFI calls. The
97 SB-ALIEN:C-STRING no longer implies an ASCII
98 external-format. Instead the string is subject to external-format
99 conversion. Additionally return values of type SB-ALIEN:C-STRING
100 are by default of type (SIMPLE-ARRAY CHARACTER), not
101 SIMPLE-BASE-STRING. For an alien type that behaves like the old
102 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
103 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
104 * incompatible change: SB-EXT package no longer contains the
105 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
106 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
108 * incompatible change: the single-stepper is no longer available
109 on Alpha, Mips and Sparc platforms.
110 * minor incompatible change: the direct superclasses of
111 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
112 not (STANDARD-OBJECT FUNCTION). This makes the
113 class-precedence-lists of GENERIC-FUNCTION and
114 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
116 * minor incompatible change: the default stream external format on
117 non-unicode SBCL is no longer detected from the locale, but is always
119 * new feature: Add a version of evaluator that uses an interpreter instead
120 of the compiler. EVAL still uses the compiler by default, to switch it
121 to use the interpreter, set the value of the variable
122 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
123 * minor incompatible change: the single-stepper REPL has been merged
124 with the normal debugger (see the "Stepping" heading of the debugger help
125 for more details). The debugger command STEP will no longer switch
126 to the single-stepper REPL.
127 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
128 for a type now works.
129 * bug fix: loading of default sysinit file works. (thanks to Leonid
131 * bug fix: better detection of circularities in the file-compiler.
132 (reported by Marco Monteiro)
133 * bug fix: the CL pathname functions now work with files that have
134 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
135 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
136 systems with Unix98 pty semantics.
137 * bug fix: ASDF-INSTALL will now work with bsd tar.
138 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
140 * bug fix: timers expiring in dead threads no longer cause a
141 type-error (reported by Paul "Nonny Mouse").
142 * bug fix: thanks to more lightweight single-stepper instrumentation,
143 code compiled with (DEBUG 3) will compile and execute significantly faster,
144 and will have more accurate type-inferencing than before
145 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
146 whose bindings are modified
147 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
148 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
149 ** replace dummy implementations of CL:MACHINE-INSTANCE and
150 CL:SOFTWARE-VERSION with proper version
152 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
153 * feature: implemented the READER-METHOD-CLASS and
154 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
155 as specified by AMOP.
156 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
158 * optimization: faster LOGCOUNT implementation on x86 and x86-64
159 (thanks to Lutz Euler)
160 * optimization: hashing of general arrays and vectors has been
161 improved. (reported by Any Fingerhut)
162 * enhancement: SB-INTROSPECT is now able to find definitions of
163 profiled functions. (thanks to Troels Henriksen)
164 * enhancement: compiler-macro expansion applies now to FUNCALL forms
165 as well, allowing compiler-macros for SETF-functions to expand.
166 * enhancement: step-instrumentation no longer wraps known
167 single-value functions with multiple-value context, allowing
168 better type inference.
169 * fixed bug #337: use of MAKE-METHOD in method combination now works
170 even in the presence of user-defined method classes. (reported by
171 Bruno Haible and Pascal Costanza)
172 * fixed bug #339(c): if there are applicable methods not part of any
173 long-form method-combination group, call INVALID-METHOD-ERROR.
174 (reported by Bruno Haible)
175 * fixed bug #361: the :FUNCTION initarg in the protocol for
176 initialization of methods can now be used to override
177 internally-produced optimized functions. (reported by Bruno
179 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
180 system-provided lambda expression no longer cause warnings about
181 unbound #:|pv-table| symbols.
182 * bug fix: improved the handling of type declarations and the
183 detection of violations for keyword arguments with non-constant
185 * bug fix: potentially erronous calls to PATHNAME and
186 MERGE-PATHNAMES were being flushed in some cases.
187 (reported by Richard Kreuter)
188 * bug fix: compiled calls to TYPEP were mishandling obsolete
189 instances. (reported by James Bielman and Attila Lendvai)
190 * bug fix: format strings with ~> without matching ~< no longer
191 trigger an AVER, but signal an understandable error instead.
192 (reported by Antonio Martinez)
193 * bug fix: specifying an output-file without a pathname-type for
194 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
195 of input-file instead of "fasl". (reported by Robert Dodier)
196 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
197 (reported by James Y Knight).
198 * bug fix: compiler-macros-function did not consider the environment
199 argument for shadowing by local functions.
200 * bug fix: compiler-macros expansion was inhibited by local INLINE
202 * bug fix: inline expansions of known functions were subject to
203 step-instrumentation in high debug policies, leading to problems
205 * bug fix: compiler failed to differentiate between different CONS
207 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
208 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
209 * bug fix: #368: incorrect use of expressed vs. upgraded array
211 * bug fix: #306a: more precise unions of array types.
212 * thread-safety improvements:
213 ** CONDITION-WAIT could return early on Linux, if the thread was
214 interrupted and subsequently continued with SIGCONT.
215 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
217 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
218 * added support for the ucs-2 external format. (contributed by Ivan
220 * minor incompatible change: pretty printing of objects of type
221 (cons symbol) is, in the default pprint-dispatch-table, now
222 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
224 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
225 called later in a class's lifetime, possibly as late as when the
226 first instance of the class is created. Previously,
227 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
228 class became finalizeable.
229 * fixed bug: FILE-POSITION sometimes returned inconsistent results
230 for multibyte external-format streams. (thanks to "vbzoli")
231 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
232 with :ALLOCATION :CLASS inherited from superclasses of the
234 * fixed bug: anonymous classes can now be created using the :NAME
235 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
236 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
237 * fixed bug: core-files saved with :EXECUTABLE T can again be
238 executed when SBCL_HOME isn't set. (reported by James Knight)
239 * fixed bug: toplevel LOCALLY forms with declarations could
240 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
241 * fixed bug: printing from several different threads using different
242 values of *print-case* could cause invalid output, due to
243 some internal special variables of the printer not being bound
244 thread-locally (reported by Max Mikhanosha)
245 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
246 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
247 (reported by Pascal Costanza)
248 * fixed bug: FINALIZE-INHERITANCE is called from
249 REINITIALIZE-INSTANCE on classes when the class has previously
250 been finalized, as required by AMOP.
251 * minor code generation optimizations:
252 ** better register allocation in CLOS dispatching functions
253 ** overflow detection when coercing signed bytes to fixnums on x86-64
254 ** is now implemented with one IMUL instruction instead of three shifts
255 ** more efficient bit-vector access on x86 and x86-64
256 ** more efficient access to raw structure slots on x86 and x86-64
257 * fixed some bugs revealed by Paul Dietz' test suite:
258 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
261 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
262 * feature: thread support on Solaris/x86, and experimental thread support
264 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
266 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
267 default method for SLOT-UNBOUND.
268 * minor incompatible change: prevent the user from specializing the
269 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
270 somewhat counter to the intent of the protocol, I (CSR) think, and
271 additionally it just doesn't work in SBCL as currently
272 implemented, thanks to optimizations (that are always valid for
273 the other three SLOT-VALUEish functions, but not for the setter).
274 * minor incompatible change: the :SB-LDB feature is now enabled by
275 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
276 the low-level debugger.
277 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
278 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
279 down to an SB-SYS:FD-STREAM, allowing direct communication in
280 more cases. Temporary files and pipes are still used as a fallback
282 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
283 on the non-reentrant C functions localtime(3) and gmtime(3).
284 * bug fix: global optimization policy was not visible in LOCALLY and
286 * bug fix: class objects can be used as specializers in methods.
287 (reported by Pascal Costanza)
288 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
289 with a type error. (reported by blitz_ on #lisp)
290 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
291 failed with a type error. (reported by Pascal Bourguignon)
292 * bug fix: merging pathnames against defaults with :DIRECTORY
293 starting with '(:RELATIVE :BACK) should preserve the :BACK.
294 (reported by James Y Knight)
295 * bug fix: saving large (>2GB) cores on x86-64 now works
296 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
297 constant index and a (simple-array (signed-byte 32)) array
298 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
299 error (patch by Robert J. Macomber)
300 * feature: TIME now displays the amount of run-time spent in GC
301 * bug fix: The debugger now does a better job of respecting
302 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
303 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
304 workaround for bug 403.)
305 * bug fix: floating point exception handling now works on FreeBSD
306 (thanks to NIIMI Satoshi)
307 * fixed some bugs revealed by Paul Dietz' test suite:
308 ** MISC.641: LET-conversion were not supposed to work in late
310 * improvements to the Win32/x86 port:
311 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
312 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
314 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
315 * new feature: source path information is generated for macro-expansion
316 errors for use in IDE's like Slime (thanks to Helmut Eller)
317 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
318 * bug fix: compiler does not loop forever on an invalid type in
320 * improvement: compilation of most CLOS applications is significantly
322 * optimization: added a limited bytecode compiler for simple toplevel
323 forms, speeding up compilation and FASL loading
324 * bug fix: the statistical profiler now properly distinguishes anonymous
327 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
328 * minor incompatible change: in sbcl-0.9.11 (but not earlier
329 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
330 into subclasses of the finalized class. Now user calls to
331 FINALIZE-INHERITANCE finalize just the one class, and calls by the
332 system return before any subclasses are finalized.
333 * minor incompatible change: The reader no longer ignores errors
334 regarding non-existent packages in #+ and #- feature tests.
335 * new feature: command line options --no-sysinit, --no-userinit to
336 inhibit loading the corresponding init files
337 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
338 for sending data through UDP sockets (thanks to François-René Rideau)
339 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
340 error is signalled instead (thanks to François-René Rideau)
341 * minor incompatible change: Error signalling behaviour of lexical
342 operations violating package locks has changed slightly. Refer to
343 documentation on package locks for details.
344 * enhancement: EVAL can process IF-expressions without resorting to the
346 * bug fix: Some operations involving SIN, COS, or TAN weren't being
347 constant-folded properly due to missing out-of-line functions.
348 (reported by Mika Pihlajamäki)
349 * bug fix: LISTEN sometimes returned T even in cases where no data was
350 immediately available from the stream
351 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
352 were reversed. (reported by Levente Mészáros)
353 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
354 when given a vector displaced to an adjustable array. (reported
356 * bug fix: derivation of float boundaries from numbers outside the
357 appropriate float range (reported by John Wiseman)
358 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
360 * bug fix: precomputing cache entries for generic functions with
361 some subclasses of specializers as yet invalid does not attempt to
362 fill a cache line with a negative offset. (reported by Levente
364 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
365 allow condition class objects as arguments to DOCUMENTATION and
366 (SETF DOCUMENTATION); only find and set documentation for
367 structure names for the STRUCTURE doc-type. (suggested by Gary
369 * improvements to the Win32/x86 port:
370 ** (user-homedir-pathname) and default initialization file
371 locations now know about the user's "Documents and Settings"
372 directory (thanks to Yaroslav Kavenchuk)
373 ** run-program is implemented (thanks to Mike Thomas)
374 ** sockets support (thanks to Timothy Ritchey)
375 ** better backtrace support (thanks to Alastair Bridgewater)
376 ** sb-grovel supported
377 ** asdf-install and sb-posix work somewhat
378 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
379 * improvements to the Solaris/x86 port:
380 ** works on Solaris 11/Solaris Express
381 ** floating-point exception handling support
382 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
383 * fixed some bugs revealed by Paul Dietz' test suite:
384 ** REMOVE-METHOD returns its generic function argument even when
385 the method is not one of the generic functions' methods.
386 ** objects declared as MEMBER types can be admissible arguments to
388 ** printing characters should simply be printed by the FORMAT ~:C
390 ** compiler failure when compiling functions with hairy constant
391 defaults for optional parameters.
392 ** compiler produces wrong code when MAYBE-INLINE-expanding a
393 function, which is already optimized.
395 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
396 * new platform: experimental support for SBCL x86/Darwin, including
397 MacOS X 10.4.5 on Intel.
398 * new feature: Unicode character names are now known to the system
399 (through CHAR-NAME and NAME-CHAR).
400 * new feature: the filesystem location of SBCL's core file is
401 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
402 * minor incompatible change: the contrib modules SB-POSIX and
403 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
404 this change is to make it easier to distribute
405 location-independent binaries.
406 * bug fix: as implied by AMOP, standardized classes no longer have
407 slots named by external symbols of public packages. (reported by
409 * optimization: calling functions via constant symbols -- as in
410 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
411 via the function object as in (FUNCALL #'FOO).
412 * optimization: on x86 and x86-64, the calling convention for the
413 case of dealing with an unknown number of values has been altered
414 to be friendlier to the prediction heuristics implemented,
415 particularly with reference to CALL and RET pairing. (thanks to
416 Alastair Bridgewater)
417 * optimization: on x86, the code for access to thread-local symbol
418 values and binding/unbinding of thread-local symbols is smaller.
419 (thanks to Alastair Bridgewater)
420 * enhancement: CONSTANTP is now able to determine constantness of
421 more complex forms, including calls to constant-foldable standardized
422 functions and some special forms beyond QUOTE.
423 * fixed bug: occasional GC crashes on Solaris/x86
424 * optimization: x86-64 supports stack allocation of results of simple
425 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
426 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
427 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
428 (thanks to James Knight)
429 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
430 by the standard (thanks to Stephen Compall)
432 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
433 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
434 be used for bundling the runtime and the core file into one
435 executable binary. This feature is not currently supported on all SBCL
436 platforms. (thanks to James Bielman and NIIMI Satoshi)
437 * new feature: a generational or ephemeral garbage collector is now
438 the default on the PowerPC platform (both Linux and Darwin). The
439 old Cheney (stop and copy) collector is a build-time option.
440 (thanks to Cyrus Harmon, after Raymond Toy)
441 * minor incompatible change: the method by which SBCL finds its
442 contributed modules has changed; it no longer relies on symbolic
443 links from an $SBCL_HOME/systems directory, but searches directly
444 in the subdirectories of $SBCL_HOME.
445 * enhancement: the dynamic heap size on the Linux/PPC platform is
446 markedly larger, even using the older Cheney garbage collector.
447 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
448 certain complicated string types. (reported by Gary King)
449 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
450 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
451 * fixed bug: DOCUMENTATION on structures no longer signals an
452 error if no structure type of the right name exists. (reported by
454 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
455 applicable non-standard primary or :AROUND method on
456 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
457 no longer cause unbound variable errors. (reported by Kevin Reid)
458 * fixed bug: metacircle resolution in cases where methods have slots
459 added before the slots from STANDARD-METHOD. (reported by Jean
461 * fixed bug: the Power PC assembly code for calling into lisp
462 sometimes computed its return address wrongly (depending on the
463 15th bit of the address, fixed at link time). This used to
464 manifest itself in a segmentation violation while building PCL.
465 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
466 many others over the years)
467 * fixed bug: ROOM no longer reports silly numbers for stack usage if
468 the stack pointer (treated as a signed integer) is negative.
469 (thanks to Peter van Eynde)
471 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
472 * new platform: experimental support for the Windows operating
473 system has been added. (thanks to Alastair Bridgewater)
474 * fixed several bugs in and robustified the PPC FFI (including
475 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
476 * bug fix: allow non-simple string symbol names (reported by Paul
478 * bug fix: interrupt handling on NetBSD (thanks to Richard M
480 * bug fix: saving a core corrupted callbacks on x86/x86-64
481 * bug fix: closed a loophole in metacircularity detection and
482 grounding in the PCL implementation of CLOS.
483 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
484 (reported by tomppa on #lisp)
485 * bug fix: clear the direction flag on Lisp -> C transitions, as
486 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
487 (reported by Andrej Grozin and Hendrik Maryns)
488 * optimization: major improvements to GC efficiency on GENCGC platforms
489 * optimization: faster implementation of EQUAL
490 * optimization: emit more efficient opcodes for some common
491 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
492 * optimization: several other minor code-generation improvements on x86-64
493 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
495 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
496 * minor incompatible change: (SETF CLASS-NAME) and (SETF
497 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
498 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
499 by ANSI as a generic function, and for consistency (SETF
500 GENERIC-FUNCTION-NAME) is treated likewise.
501 * fixed bug #233.b: make constraint propagation notice when a variable
502 value is changed after it is referenced but before it is used
503 * fixed bug #296: no more arbitrary behaviour from filenames with
504 odd characters as --load arguments. (reported by Adam Warner)
505 * fixed bug #390: :CHARACTER-SET pathname components now work as
506 expected. (reported by Tim Daly Jr)
507 * fixed bug #391: complicated :TYPE intersections in slot
508 definitions no longer cause an error in PCL internals.
509 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
510 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
511 returning the number of octets which would be written to the
512 file-stream. (thanks to Robert J. Macomber)
513 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
514 arguments to the internal stream functions. (thanks to David
516 * bug fix: stack exhaustion is now handled on the x86/SunOS
517 platform. (thanks to Daisuke Homma)
518 * bug fix: more accurate ROOM results on GENCGC platforms
519 * optimization: improved type inference for arithmetic-for
520 index variables in LOOP
521 * optimization: faster floating-point SQRT on x86-64
522 * fixed some bugs revealed by Paul Dietz' test suite:
523 ** DOCUMENTATION returns NIL instead of "" for method combinations
524 that don't have a docstring
526 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
527 * minor incompatible change: (SETF CLASS-NAME) and (SETF
528 GENERIC-FUNCTION-NAME) are no longer generic functions, and
529 therefore are not customizeable by user code (as seems to be at
530 least permitted and maybe required by AMOP). As a consolation,
531 however, the SBCL implementation of these functions now calls
532 REINITIALIZE-INSTANCE as specified by AMOP.
533 * bug fix: slot-definition documentation is propagated as per ANSI
534 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
535 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
536 argument, as do eslotd initialization methods. (from Pascal
537 Costanza's "Closer" project)
538 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
539 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
541 * bug fix: it is now possible to have more than one subclass of
542 STANDARD-GENERIC-FUNCTION without causing stack overflow.
543 (reported by Bruno Haible, Pascal Costanza and others)
544 * bug fix: the dependent update protocol now works for generic
545 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
547 * bug fix: condition-class instances corresponding to
548 DEFINE-CONDITION forms are now created eagerly. (reported by
549 Kalle Olavi Niemitalo on comp.lang.lisp)
550 * bug fix: floating point printing is more accurate in some
551 circumstances. (thanks to Simon Alexander)
552 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
553 merged with *DEFAULT-PATHNAME-DEFAULTS*.
554 * bug fix: callbacks on OS X now preserve stack-alignment.
555 (thanks to Cyrus Harmon)
556 * enhancement: the x86-64 disassembler is much better at
557 disassembling SSE instructions. (thanks to Lutz Euler)
558 * enhancement: CHECK-TYPE on a variable now allows type-inference
559 (thanks to Vincent Arkesteijn)
560 * optimization: improved performance of EUC-JP external format.
561 (thanks to NIIMI Satoshi)
562 * optimization: performance improvements to IO on file streams of
563 :ELEMENT-TYPE CHARACTER
564 * optimization: much faster memory allocation on x86-64
565 * optimization: faster garbage collections (latency and throughput)
567 * optimization: faster bignum subtraction and division on x86 and x86-64
568 (thanks to Lutz Euler)
569 * optimization: more accurate type-inference for loops that use a
570 floating point index variable or a negative step.
572 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
573 * bug fix: add a workaround to SBCL looping infinitely at startup on
574 Linux kernels with apparently buggy implementations of personality().
575 (thanks to Svein Ove Aas)
576 * bug fix: Unicode symbols are correctly printed in LDB backtraces
577 (thanks to David Lichteblau)
578 * bug fix: local bindings shadow global symbol macros for
579 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
580 * optimization: non-open coded uses of numeric comparison operators
581 (e.g. >) no longer cons when called with more than one parameter
582 on platforms supporting dynamic-extent allocation.
583 * enhancement: saving cores with foreign code loaded is now supported
584 on MIPS/Linux in addition to the previously supported platforms.
585 * bug fix: FIND-RESTART now tests for activity, not applicability when given
586 a restart object as identifier. (reported by Helmut Eller for CMUCL)
587 * bug fix: division by zero in sb-sprof when no samples were collected
588 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
589 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
590 (thanks to Alexander Kjeldaas)
591 * new feature: ignoring the return values of destructive functions
592 like NREVERSE signals a compile-time style-warning.
593 (thanks to Kevin Reid)
594 * new feature: the alignment of alien structure fields can be
595 explicitly specified. (thanks to Cyrus Harmon)
596 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
597 and dump core on SIGQUIT
599 ** incompatible change: threads do not inherit values of specials
600 from their parents (see manual)
601 ** bug fix: threads stacks belonging to dead threads are freed by the
602 next exiting thread, no need to gc to collect thread stacks anymore
603 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
604 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
605 ** bug fix: dynamic variable and thread start related gc lossage
606 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
608 * fixed some bugs revealed by Paul Dietz' test suite:
609 ** SUBTYPEP is slightly more accurate on heinously complicated
610 CONS types where some of the members have uncertain (in the
611 NIL, NIL sense) type relationships to each other.
613 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
614 no longer purified unless :PURIFY T is explicitly specified.
615 ** Non-purified cores are significantly smaller than before
617 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
618 * new feature: timers based on Zach Beane's excellent timer package
619 * added support for the following external formats: koi8-u,
620 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
621 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
622 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
623 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
624 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
625 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
626 * incompatible change: a threaded SBCL will no longer revert to
627 non-threaded mode on non-NPTL systems, but refuse to start entirely.
628 * bug fix: interrupts are disabled until startup is complete; no
629 more sigsegvs when receiving a signal to soon
630 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
632 * bug fix: add a workaround for the memory randomization features in
633 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
634 workaround will only be in effect on systems with the proc filesystem
636 * bug fix: printing objects of type HASH-TABLE signals a
637 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
639 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
640 have been running for over 50 days. (reported by Gilbert Baumann)
641 * bug fix: the logic for getting names of functions gets less
642 confused when confronded with alternate-metaclass
643 funcallable-instances. (reported by Cyrus Harmon)
644 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
645 consistent, even on internal alternate-metaclass objects.
646 * bug fix: flush closure information collected by physical
647 environment analysis prepass before the main pass. (bug reported
649 * bug fix: compiler pack phase does not modify a hashtable, which is
650 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
651 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
652 may be expanded. (reported by Alexander <malishev> on c.l.l)
653 * bug fix: no more style-warnings when compiling code using LAST
654 under high-SPEED low-SPACE settings. (reported by David Wallin
656 * fixed some incompatibilities between SBCL's MOP and the MOP
658 ** the METAOBJECT class is now implemented;
659 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
660 STANDARD-OBJECT, as required;
661 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
662 now compatible (as required by VALIDATE-SUPERCLASS); there
663 remains a consistency requirement over the presence or absence
664 of the FUNCTION class in the superclass list of finalized
665 classes; see the manual for more details;
666 ** the :AROUND method for COMPUTE-SLOTS on
667 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
668 requested slot ordering.
670 ** bug fix: parent thread now can be gc'ed even with a live
672 ** bug fix: binding a special with PROGV to no value is not
673 the same as that symbol not having been bound (thanks to
675 ** bug fix: binding specials is thread safe (thanks to
677 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
678 and SB-SPROF do not die with 'no handler for signal XX in
679 interrupt_handle_now(..)' anymore
680 ** bug fix: WITH-TIMEOUT works with multiple running threads
681 * fixed some bugs revealed by Paul Dietz' test suite:
682 ** ENSURE-GENERIC-FUNCTION should take a method class object for
683 the :method-class keyword argument.
685 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
686 * new port: the Solaris operating system on x86 processors is now
687 mostly supported, though some rough edges in the environment
688 remain. (thanks to Daisuke Homma)
689 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
690 space for its heap. (thanks to Thiemo Seufer)
691 * optimization: SBCL on MIPS platforms now supports dynamic-extent
692 closures. (thanks to Thiemo Seufer)
693 * minor incompatible change: eof selects abort in the debugger.
694 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
695 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
697 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
698 deprecated, and will go away in a future revision of SBCL.
699 * minor incompatible change: GC-ON and GC-OFF are no longer
700 implemented with a counter, it does not matter how many times gc
701 is switched on or off
702 * bug fix: discriminating functions for generic function classes
703 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
704 make invalid assumptions about method precedence order. (reported
706 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
707 platforms. (thanks to Thiemo Seufer)
708 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
709 * bug fix: correct pathname printing with printer escaping is on.
710 (thanks to Kevin Reid)
711 * bug fix: complex VOP definitions in "user-space" no longer trigger
712 package locks. (reported by Zach Beane)
713 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
714 for funcallable-instance functions, and is no different from
716 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
717 characters in the current readtable. (reported by Nicholas Neuss)
718 * bug fix: SUBTYPEP on various CONS types returns more a more
719 accurate acknowledgment of its certainty.
720 * optimizations: REMOVE-DUPLICATES now runs in linear time on
721 lists in some cases. This partially fixes bug 384.
722 * flush all standard streams before prompting in the REPL and the
724 * bug fix: signal handling and triggering gc do not conflict
725 directly anymore, in particular a high frequency sb-sprof does
726 not prevent gc from running
727 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
728 approximation for timezone and DST information between the
729 universal time epoch and the smallest negative 32-bit time_t.
730 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
731 year 1899 so long as the encoded time is non-negative.
732 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
733 the correct answer for non-constant rotate amounts when performing
734 an inline 32-bit rotation.
736 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
737 there is only one thread in the session
738 ** bug fix: memory leak for streams created in one thread and
739 written to in another
740 ** bug fix: lockup when compiled with gcc4
741 ** bug fix: race that allows the gc to be triggered when gc is
743 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
744 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
746 ** bug fix: finished threads are gc'ed properly
747 * fixed some bugs revealed by Paul Dietz' test suite:
748 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
749 has arguments with a different set of applicable methods from
750 the orignal arguments.
751 ** The type error thrown by MAP now has a correct expected-type
753 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
754 name a compiled function.
755 ** Three MISC tests where a large bignum was improperly coerced to
756 a float that couldn't represent that bignum during type
757 derivation were fixed.
758 ** SUBTYPEP can now handle types involving the negation of a
759 list-form FUNCTION type.
760 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
761 as (COMPLEX (AND RATIO (NOT FIXNUM))).
762 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
764 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
765 * New feature: Experimental support for bivalent streams: streams
766 opened with :element-type :default now allow character and binary
767 (unsigned-byte 8) I/O
768 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
769 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
771 * bug fix: no more highly sporadic "couldn't check whether ~S is
772 readable" when reading a stream and an interrupt hits in the middle
773 of a select system call
774 * compiler better recognizes complex arrays (reported by Eduardo
776 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
777 for variables on SBCL built with linkage-tables. (reported by Luis
779 * various error reporting improvements.
780 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
781 (Thanks to Thiemo Seufer)
782 * bug fix: nested reader invokations work correctly
783 * bug fix: it is possible to have simultaneous references to foreign
784 code and foreign data with the same name.
786 ** added x86-64 support
787 ** incompatible change: the threading api now works with thread
788 objects instead of thread ids
789 ** bug fix: threads are protected from signals and interruption when
790 starting up or going down
791 ** bug fix: a race where an exiting thread could lose its stack to gc
792 ** bug fix: don't halt on infinite error in threads if possible
793 ** fixed numerous gc deadlocks introduced in the pthread merge
794 ** bug fix: fixed thread safety issues in read and print
795 ** bug fix: debugger doesn't hang on session lock if interrupted at
796 an inappropriate moment
797 ** bug fix: run-program is now thread safe(r)
798 ** bug fix: inner with-recursive-lock no longer releases the mutex
799 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
800 * fixed some bugs revealed by Paul Dietz' test suite:
801 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
802 EXPECTED-TYPE slots filled.
803 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
805 ** MAKE-CONDITION accepts classes as type-designators.
806 ** COMPILE may never return NIL.
807 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
808 range before calling Unix time functions
810 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
811 * numerous signal handling fixes to increase stability
812 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
813 * minor incompatible change: we now correctly canonize default
814 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
815 as per the MOP, rather than the historical (INITARG INITFUNCTION
816 INITFORM). (reported by Bruno Haible)
817 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
818 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
819 for more information.
820 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
821 pathname is a directory pathname.
822 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
823 * dynamic space size on PPC has been increased to 768Mb. (thanks to
825 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
826 :METACLASS argument in addition to a class name. (reported by
827 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
828 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
829 (reported by Bruno Haible; patch from Gerd Moellmann)
830 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
832 * bug fix: more cleanups to the floating point exception handling on
833 x86-64 (thanks to James Knight)
834 * bug fix: the compiler does not try to constant fold calls of
835 COERCE and BIT-* functions when they return freshly constructed
836 objects. (reported by Nikodemus Siivola and Paul Dietz)
837 * optimization: improved the MIPS versions of generic (in the
838 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
839 * optimization: direct conversion of (unsigned-byte 32) to floats on
840 the PowerPC platform.
841 * optimization: structure instances with raw slots now use less
842 memory, and probably show better memory locality. (thanks to
844 * optimization: DYNAMIC-EXTENT declarations for lists and closures
845 are treated as requests for stack allocation on the x86-64,
846 Alpha-32, PPC and SPARC platforms.
847 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
848 the wrong file descriptor; implementation of SOCKET-OPEN-P.
849 (thanks to Tony Martinez)
851 ** gcing a dead thread can no longer lead to lockups
852 ** threads block signals until they are set up properly
853 ** errno is no longer shared by threads
854 ** interrupt-thread restores the eflags register on x86
855 ** fixed some lockups due to gc/thread interaction
856 * fixed some bugs revealed by Paul Dietz' test suite:
857 ** invalid dotted lists no longer raise a read error when
859 ** signal an error if a symbol that names a declaration is used
860 as the name of a type, or vice versa
861 ** allow using the (declare (typespec var*)) abbreviation for
862 (declare (type typespec var*)) with all type specifiers
863 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
864 ** always return NIL from PPRINT-POP when OBJECT is NIL
865 ** don't signal errors when pretty-printing malformed LABELS,
866 FLET or MACROLET forms
867 ** declarations in a DOLIST body are also in scope for the
869 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
871 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
874 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
875 * fixed cross-compiler leakages that prevented building a 32-bit
876 target with a 64-bit host compiler.
877 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
878 opened with :IF-EXISTS :OVERWRITE.
879 * fixed bug 281: error for an invalid qualifier in a short-form method
880 combination method is not signalled until the faulty method is called.
881 * bug fix: iteration variable type inferrer failed to deal with open
882 intervals. (reported by Alan Shields)
883 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
884 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
885 or more runtime options were provided to the sbcl binary.
886 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
887 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
888 * optimization: there's now a fast-path for fixnum arguments in the
889 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
890 * optimization: the code generated on x86-64 is more compact thanks
891 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
892 * bug fix: floating-point exception handling now partly works on
893 x86-64. (Thanks to James Knight)
894 * improvement to the MIPS backend from Thiemo Seufer: C-style
895 64-bit long long arguments and return values to alien functions
897 * fixed some bugs revealed by Paul Dietz' test suite:
898 ** the type-error signalled from WARN has a filled-in DATUM slot.
899 ** the type-error required when a stream is not associated with
900 a file has the stream as its datum.
901 ** type-errors on single-floats on x86-64 no longer have
902 :INVALID-OBJECT as the datum
903 ** the type-errors signalled for invalid function names now have
904 a correct expected type
905 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
906 for structures defined with a :TYPE. Documentation strings for
907 typed structures are no longer immediately discarded
908 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
909 broadcast streams and synonym streams. FILE-LENGTH now also works
910 on broadcast streams.
912 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
913 * incompatible change: the --noprogrammer option, deprecated since
914 version 0.7.5, has been removed. Please use the equivalent
915 --disable-debugger option instead.
916 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
918 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
919 inoperational for a while now) has been completely removed.
920 * null lexical environments are now printed as #<NULL-LEXENV>,
921 significantly reducing the amount of clutter in typical backtraces.
922 * documentation on weak pointers, finalization, and after GC hooks
923 has been added to the manual.
924 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
925 as other specialized array types, is much faster. SUBSEQ and
926 COPY-SEQ on such arrays have also been sped up.
927 * optimization: EQL is now more efficient when at least other argument
928 is known to be of type (OR FIXNUM (NOT NUMBER)).
929 * fixed bug: the runtime is now less vulnerable to changes in the
930 size of the SBCL object on OS X, and virtual memory is reserved for
931 all spaces that need to be at a fixed address.
932 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
933 * fixed bug: finalizers and after GC hooks that cause consing are now
935 * fixed bug: compiler error messages and summaries are now printed to
936 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
937 * fixed inference of the upper bound of an iteration variable.
938 (reported by Rajat Datta).
939 * fixed bug 211e: calling local functions with duplicated constant
940 keyword argument no longer causes a bogus style warning about an
942 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
943 ftype declaration to be disregarded. (reported by Dave Roberts)
944 * fixed bug 373: caused by erronous compilation of references to alien
945 variables in the runtime on ppc/darwin.
946 * fixed bug 376: CONJUGATE type deriver.
947 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
948 (reported by Baughn on #lisp)
949 * a cross-compiler bug on non-x86 platforms has been identified and
950 fixed. (thanks to Bruno Haible)
951 * improvements to the MIPS runtime code for increased stability.
952 (thanks to Thiemo Seufer)
953 * increased the maximimum compact environment size to allow
954 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
955 * partial workaround for bug 135: don't name the function we're
956 calling for hairy cases of EVAL, so as not to accumulate one environment
957 entry per call to EVAL. (reported by Kevin Reid)
958 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
959 * optimization: type testing for non-vector arrays should be faster.
960 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
962 * bug fix: setting 31st element of a bit vector to zero did not work
964 * bug fix: redefining a class definition which failed due to a
965 previous accessor / function clash now works (but see BUGS entry
966 #380 for more problems in this area). (thanks to Zach Beane)
967 * the long form of DEFINE-METHOD-COMBINATION disables method group
968 checking when given a single method group with pattern *. (thanks
970 * on x86 compiler supports stack allocation of results of simple
971 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
972 * support for building and running on Mac OS X 10.4 "Tiger" has been added
973 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
974 * fixed some bugs related to Unicode integration:
975 ** the restarts for recovering from input and output encoding
976 errors only appear when there is in fact such an error to
978 * fixed some bugs revealed by Paul Dietz' test suite:
979 ** MISC.549 and similar: late transformation of unsafe type
980 assertions into derived types caused unexpected code
982 ** SCALE-FLOAT type deriver is less wrong.
983 ** type derivers for EXP, LOG and similar functions compute result
984 types for complex arguments better.
985 ** (MISC.563) CONJUGATE type deriver works for very restricted
987 ** out-of-line type testers for character strings are available.
988 ** EQUAL compiler transform understands specialness of objects
990 ** accessing double-floats stored on the stack now works on x86-64.
991 ** debugger internals could sometimes create invalid lispobjs,
992 resulting in GC crashes.
993 ** MISC.548: type check weakening can convert required type into
995 ** initialization forms for bindings are not in scope of free special
998 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
999 * incompatible change: thread support for non-NPTL systems has
1000 been removed - locking is buggy and unreliable. A threaded
1001 SBCL build will now warn at startup and refuse to create
1002 new threads, unless futex support is detected in the kernel
1003 * incompatible change: the top level REPL now has only an ABORT
1004 restart associated with it, not TOPLEVEL and ABORT as it used to.
1005 TOP and TOPLEVEL are now available as debugger commands for
1006 returning to the top level.
1007 * incompatible change: forms evaluated in the REPL now use the
1008 global optimization policy.
1009 * incompatible change: user- and system-initialization files are
1010 no longer processed with LOAD, but by READ and EVAL; hence the
1011 global optimization policy, startup package, readtable, etc,
1013 * The .fasl file format number has been incremented because of
1014 various incompatible changes.
1015 * internal entry point details and argument counts no longer appear
1016 in backtraces unless explicitly requested by setting
1017 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1018 * built-in and standard functions no longer have names like "top
1019 level local call to FOO".
1020 * fixed bug 32: functions defined in non-null lexical environments
1021 now have more legible printed representation
1022 * fixed bug 33: functions defined in non-null lexical environemnts
1023 are now more amenable to inspection by INSPECT.
1024 * workaround for bug 354: XEPs no longer appear in backtraces unless
1025 explicitly requested.
1026 * fixed bug: receiving the signal which results from attempting to
1027 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1028 SIGBUS on other *BSDs) on architectures where the C stack is also
1029 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1030 to debugger confusion or wild instability. (reported by Cheuksan
1032 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1033 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1034 (reported by Lutz Euler)
1035 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1036 compiled code. (reported by Paul Dietz)
1037 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1038 specialized parameter to a method to something that is not TYPEP
1039 the specializer is now possible.
1040 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1041 face of package deletion.
1042 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1043 pathname. (reported by David Tolpin on c.l.l)
1044 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1045 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1046 STRING-STREAM and FILE-STREAM, respectively.
1047 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1048 than previously; in addition, its test suite is now run on build.
1049 (thanks to Luke Gorrie)
1050 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1051 * optimization: added a immediate representation for single-floats
1053 * fixed some bugs revealed by Paul Dietz' test suite:
1054 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1055 ** attempting to create a package with a colliding nickname causes
1056 correctable errors to be signalled.
1057 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1058 ** several x86-64 backend bugs related to sign-extension of immediate
1061 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1062 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1063 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1065 * fixed loading of multiply forward-referenced layouts.
1066 (thanks to Cheuksan Wang)
1067 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1068 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1069 (see COMPILE-FILE documentation for details of :PRINT option)
1070 * fixed bugs 19 and 317: fixed-format floating point printing is
1071 more accurate. This also fixes a bug reported by Adam Warner
1072 related to the ~@F format directive.
1073 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1075 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1076 dispatch table if the from-char is a dispatch macro character.
1077 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1078 by the wordsize no longer ignore the last word. (reported by Lutz
1080 * fixed bug in type checking of dynamic-extent variables. (reported
1082 * optimization: sequence traversal functions use their freedom to
1083 coerce function designators to functions.
1084 * optimization: code with many calls to CLOS methods specialized on
1085 CLOS classes has had redundant type checks removed.
1086 * optimization: added declarations to speed up operations that access
1087 the internal character database (for example STRING-UPCASE)
1088 * optimization: comparison operations between floats and sufficiently small
1089 fixnums no longer create extra rationals
1090 * fixed some bugs related to Unicode integration:
1091 ** portions of multibyte characters at the end of buffers for
1092 character-based file input are correctly transferred to the
1093 start of the buffer at the next read.
1094 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1095 passing it through to OPEN.
1096 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1097 argument given to internal calls to OPEN.
1098 ** the built-in comment readers (introduced by character sequences
1099 ";" and "#|") are more forgiving to encoding errors; they will
1100 STYLE-WARN and then attempt to resync the stream at a character
1101 boundary. (thanks to Teemu Kalvas)
1102 * fixed some bugs in the x86-64 port:
1103 ** Negative short int return values from c-calls are sign-extended
1105 ** The stack is aligned to 16-bytes for c-calls, as required by
1106 the ABI. (thanks to Cheuksan Wang)
1107 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1108 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1109 for enums. (thanks to Vincent Arkesteijn)
1110 ** Multiple small optimizations and bugfixes for floating point
1112 * fixed some bugs revealed by Paul Dietz' test suite:
1113 ** Space, Tab, Linefeed, Return and Page have the invalid
1114 secondary constituent character trait.
1115 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1117 ** WITH-INPUT-FROM-STRING should only update the index place on
1119 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1120 works more reliably.
1121 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1122 with one-dimensional arrays with fill pointers.
1123 ** TYPE-OF failed on a complex with an integer realpart and a
1125 ** compiler failure during type inference for the code of form
1126 (IF (EQL X (THE ...)) ...) (MISC.535).
1128 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1129 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1130 platform. The port supports 61-bit fixnums, large memory spaces
1131 and reloading shared object files.
1132 * enhancement: saving cores with foreign code loaded is now
1133 supported on ppc/Darwin in addition to the previously supported
1135 * enhancement: the statistical profiler now walks deeper into the
1136 call stack for more meaningful call-graphs and accrued time
1137 reports (x86/x86-64 only). It also now reports time spent in
1139 * enhancement: it is now possible to trace most individual methods
1140 of a generic function in addition to tracing the generic function
1142 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1144 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1145 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1146 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1147 produces an error. (thanks to Vincent Arkesteijn)
1148 * bug fix: NAMESTRING on pathnames with :WILD components in their
1149 directories works correctly. (thanks to Artem V. Andreev)
1150 * fixed bug 125: compiler preserves identity of closures. (reported
1152 * bug fixed: functions with &REST arguments sometimes failed with
1153 "Undefined function" when compiled with (DEBUG 3). (reported by
1155 * bug fix: overflow during compiling of setting element of a bit
1156 vector with constant index and value. (reported by Timmy Douglas)
1157 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1158 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1159 * on x86 compiler partially supports stack allocation of dynamic-extent
1161 * GO and RETURN-FROM do not check the extent of their exit points
1162 when compiled with SAFETY 0.
1163 * fixed some bugs related to Unicode integration:
1164 ** encoding and decoding errors are now much more robustly
1165 handled; it should now be possible to recover even from invalid
1166 input or output to the terminal. (thanks to Teemu Kalvas)
1167 ** provided a first cut at implementing STRING-TO-OCTETS and
1168 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1169 ** altered the SB-MD5 contributed module slightly, changing the
1170 interface just enough for it to be supportable for builds where
1171 lisp characters are not eight bits.
1172 * fixed some bugs revealed by Paul Dietz' test suite:
1173 ** the FORMATTER-generated functions for ~V[ conditionals require
1174 the correct number of arguments.
1175 ** READ-FROM-STRING returns the mandated second value when applied
1176 to displaced strings.
1177 ** the #\Rubout and #\Backspace characters are treated as invalid
1178 constituent characters by the tokenizer.
1180 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1181 * new feature: reloading changed shared object files with
1182 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1183 * new feature: references to foreign variables and functions
1184 can now be compiled and loaded before the corresponding shared
1185 object file is loaded, as long as the foreign definitions are
1186 available at runtime.
1187 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1188 supported. (thanks to Dan Debertin)
1189 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1190 just keywords. (thanks to Vincent Arkesteijn)
1191 * fixed bug #331: structure-class instances corresponding to
1192 DEFSTRUCT forms are now created eagerly.
1193 * fixed bug #345: backtraces from calls to undefined functions work
1194 on x86 as well. Related bug #61 is now also partially fixed on x86
1195 (backtraces from throws to unknown catch tags.)
1196 * bug fix: lambda-list parsing is now stricter vrt. order and number
1197 of lambda-list keywords.
1198 * bug fix: as specified by AMOP, an error is signalled if a
1199 class-option appears multiple times in a DEFCLASS form. (reported
1201 * bug fix: RANDOM can be compiled when the compiler derives the type
1202 of its numeric argument as a disjoint set of small integers.
1203 (reported by Paul Dietz)
1204 * bug fix: starting a core saved with shared objects loaded when
1205 those objects are not available no longer causes threaded SBCL to
1206 hang. (reported by Sean Ross)
1207 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1209 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1210 argument, which must always be NIL. (reported by Kalle Niemitalo)
1211 * bug fix: printing 1.0d+23 no longer results in an error.
1212 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1213 * bug fix: structure slot setters preserve evaluation order. (thanks
1215 * fixed some bugs related to Unicode integration:
1216 ** RUN-PROGRAM can allow its child to take input from a Lisp
1217 stream. (reported by Stefan Scholl)
1218 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1219 locales. (reported by Ken Causey)
1220 * fixed some bugs revealed by Paul Dietz' test suite:
1221 ** INCF, DECF and REMF evaluate their place form as specified in
1223 ** FORMATTER expands ~{ ~} iteration directives with V or #
1224 parameters correctly.
1225 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1226 consequent uses no arguments correctly.
1227 ** the system has a partial understanding of the (COMPLEX RATIO)
1229 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1232 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1233 * new feature: a build-time option (controlled by the :SB-UNICODE
1234 keyword feature, enabled by default) for building the system with
1235 support for the entire 21-bit character space defined by the
1237 * new feature: the system now has rudimentary external-format
1238 support; the primary user-visible change at this time is that
1239 characters with the high bit set (such as lower-case-e-acute) will
1240 print correctly to a terminal in a UTF-8 environment.
1241 * minor incompatible change: BASE-CHAR no longer names a class;
1242 however, CHARACTER continues to do so, as required by ANSI.
1243 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1244 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1245 has been moved to the SB-EXT package (temporarily re-exported from
1247 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1248 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1250 * fixed bug #351: better error-handlind and reporting for malformed
1252 * fixed bug #350: bignum-printing is now more memory-efficient,
1253 allowing printing of very large bignums, eg. (expt 2 10000000).
1254 (reported by Bruno Haible)
1255 * fixed bug #302: better primitive-type selection for intersection
1257 * fixed bug #308: non-graphic characters now all have names, as
1258 required. (reported by Bruno Haible)
1259 * bug fix: redefining a class with different superclasses now correctly
1260 removes it from the direct-subclasses of its previous superclasses.
1261 (reported by David Morse)
1262 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1263 new value now works. (reported by Bruno Haible)
1264 * bug fix: correct canonicalization of multiple non-standard slot
1265 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1266 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1267 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1268 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1269 now exists, an signals an error.
1270 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1271 during macroexpansion for non-iteration variables as well. (reported
1272 by Bruno Haible for CMUCL)
1273 * bug fix: Cyclic structures and unprintable objects in compiler
1274 messages no longer cause errors. (reported by Bruno Haible)
1275 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1276 and READ-FROM-STRING all now return a primary value of NIL if
1277 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1278 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1279 (reported by Bruno Haible for CMUCL)
1280 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1281 do not designate a condition. (reported by Bruno Haible for
1283 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1284 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1285 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1286 permitting resolution in favour of any of the conflicting symbols.
1287 (reported by Bruno Haible for CMUCL)
1288 * bug fix: EQUAL compiler optimizations is less aggressive on
1289 strings which can potentially compare true despite having distinct
1290 specialized array element types.
1291 * bug fix: unit enumerations can be defined without dividing by
1292 zero. (thanks to Vincent Arkesteijn)
1293 * FORMAT compile-time argument count checking has been enhanced.
1294 (report from Bruno Haible for CMUCL)
1295 * a partial workaround for the bug 262: the compiler does not try to
1296 inline-expand a local function doing RETURN-FROM from a deleted
1297 BLOCK. (thanks to Peter Denno for the bug report and to David
1298 Wragg for the simple test case)
1299 * fixed some bugs revealed by Paul Dietz' test suite:
1300 ** RENAME-PACKAGE allows all package designators as new package
1302 ** constraint propagation and lambda variable substitution are
1303 more cautious in dealing with partially deleted code.
1304 ** compiler sometimes lost reoptimization passes.
1305 ** CERROR, given a condition as condition designator, passes its
1306 remaining arguments to the continue format control without
1308 ** Case-altering FORMAT directives work correctly on non-ASCII
1310 ** The REAL type specifier handles bounds outside the floating
1311 point ranges without signalling FLOATING-POINT-OVERFLOW.
1312 ** Functions with IR1-transformations can create intercomponent
1313 references to global functions.
1314 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1316 ** FORMAT ~R treats a nil value for its first parameter correctly.
1318 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1319 * enhancement: saving cores with foreign code loaded is now
1320 supported on x86/NetBSD and sparc/Linux in addition to the previously
1321 supported platforms.
1322 * bug fix: on some platforms repeated installations caused multiple
1323 copies of HTML documentation to be installed -- should not happen
1324 any more. (reported by Stefan Scholl)
1325 * bug fix: parsing self-recursive alien record types multiple times
1326 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1327 original patch by Helmut Eller for CMUCL)
1328 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1329 (thanks to Richard Kreuter)
1330 * bug fix: defining classes whose accessors are methods on existing
1331 generic functions in other (locked) packages no longer signals
1332 bogus package lock violations. (reported by François-René Rideau)
1333 * bug fix: special variables as DEFMETHOD parameters no longer have
1334 associated bogus type declarations. (reported by David Wragg and
1336 * bug fix: read-write consistency on streams of element-type
1337 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1338 * bug fix: redefiniton of the only method of a generic function with
1339 no DEFGENERIC no longer emits a full WARNING. In addition,
1340 redefinition of generic functions with no DEFGENERIC to an
1341 incompatible lambda list now signals an error. (thanks to Zach
1343 * bug fix: DEFGENERIC now works even when there's a function of the
1344 same name in an enclosing lexical environment. (thanks to Zach
1346 * fixed compiler failure, caused by instrumenting code during
1347 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1348 * optimization: added loop analysis and improved register allocation
1349 * fixed some bugs revealed by Paul Dietz' test suite:
1350 ** POSITION on displaced vectors with non-zero displacement
1351 returns the right answer.
1352 ** (SIMPLE-STRING) is a valid type specifier for sequence
1354 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1356 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1357 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1359 ** The justification version of the FORMAT ~< directive treats
1360 non-zero minpad parameter correctly.
1362 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1363 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1364 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1365 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1366 the supported interface.
1367 * new feature: Single-stepping of code compiled with DEBUG 2 or
1368 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1369 * new feature: saving cores with foreign code loaded is now
1370 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1371 Timothy Moore's work for CMUCL)
1372 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1373 parameters to * instead of NIL if no initform is supplied.
1374 (reported by Johan Bockgaard)
1375 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1376 correctly when FUNCALL appears as the car of the form. Note:
1377 despite this FUNCALL forms are not currently subject to
1378 compiler-macro expansion. (port of Raymond Toy's fix for the
1379 same from CMUCL, reported by Johan Bockgaard)
1380 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1381 (thanks for Teemu Kalvas)
1382 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1383 case instead of terminating the iteration (thanks for Julian
1384 Squires, Sean Champ and Raymond Toy)
1385 * bug fix: incorrect expansion of defgeneric that caused a style
1386 warning. (thanks for Zach Beane)
1387 * bug fix: slot accessor effective method computation works properly
1388 for classes with multiple non-standard applicable methods on
1389 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1390 * on x86 compiler supports stack allocation of results of LIST and
1391 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1392 CMUCL implementation by Gerd Moellmann)
1393 * fixed some bugs revealed by Paul Dietz' test suite:
1394 ** FORMAT strings with both the ~<~:;~> form of the justification
1395 directive and pretty-printing directives cause an error.
1397 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1398 * incompatible change: the internal functions
1399 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1400 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1401 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1402 instead of the old functions.
1403 * new feature: on platforms where "dladdr" is available foreign
1404 function names now appear in backtraces. (based on Helmut Eller's
1406 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1407 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1409 * documentation: ASDF manual is now distributed with SBCL.
1410 * bug fix: SBCL can now load its contributed modules with REQUIRE
1411 even if the system-provided entries have been removed from
1412 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1414 * bug fix: backtraces involving undefined functions or assembly
1415 routines are more informative. (thanks to Brian Downing)
1416 * bug fix: mutually referent alien structures now work correctly.
1417 (reported by Rick Taube)
1418 * bug fix: structures defined by WITH-ALIEN can be referred to
1419 within other definitions in the same WITH-ALIEN.
1420 * bug fix: division operators (MOD, TRUNCATE and the like) with
1421 constant zero divisors and integer dividends no longer generate
1423 * bug fix: provide default methods for INPUT-STREAM-P and
1424 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1425 * bug fix: improve the detection and resolution of MOP metacycles
1426 (where effective method computation depends on the generic function
1427 having its effective method computed). (reported by Bruno Haible)
1428 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1429 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1430 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1431 * bug fix: pathnames with relative directory components can be
1432 represented relative to default pathnames.
1433 * optimization: in taking the GCD of bignums, reduce the two bignums
1434 to approximately the same size (using Euclid's algorithm) before
1435 applying the more sophisticated binary GCD. (thanks to Juho
1437 * optimization: COUNT on bitvectors now operates word-at-a-time.
1438 * optimization: ASH with a positive, but not necessarily constant,
1439 (leftwards) shift, when in a modular context, is compiled to a
1441 * fixed some bugs revealed by Paul Dietz' test suite:
1442 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1443 if the corresponding argument is NIL.
1444 ** FORMAT directives accepting a minpad parameter treat negative
1446 ** literal commas as character parameters to FORMAT directives are
1448 ** literal spaces directly after ~<Newline> directives within a
1449 format-logical-block (~:< ~@:>) do not induce :FILL-style
1450 conditional newlines.
1451 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1452 as specified: it no longer includes conditional newlines).
1453 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1455 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1456 :PER-LINE-PREFIX argument does not evaluate to a string.
1458 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1459 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1460 the manual for details; for now, package locks can be disabled by
1461 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1462 if no major problems are found then it is likely that they will be
1463 compiled in unconditionally.
1464 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1465 unsupported operators on all platforms. To load a shared library
1466 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1467 object file, link it into a shared library outside of SBCL and
1468 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1470 * bug workaround: ROOM T and the GC were not working together
1471 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1472 reporting facility aren't preserved. That reporting has been
1473 disabled, so now until and unless someone figures out how to make it
1474 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1475 * minor incompatible change: as threatened around sbcl-0.8.0, the
1476 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1477 The symbols are also exported from SB-PCL for backwards
1478 compatibility, but more so than before SB-PCL should be treated as
1479 an implementation-internal package.
1480 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1482 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1483 It is known as of this release not to work on the Alpha, however.
1484 * fixed bug #167: errors signalled due to illegal syntax in method
1485 bodies are now more legible.
1486 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1487 designators and can hence be used with TYPEP.
1488 * fixed bug #333: CHECK-TYPE now ensures that the type error
1489 signalled, if any, has the right object to be accessed by
1490 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1491 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1492 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1494 * fixed bug #334: programmatic addition of slots using specialized
1495 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1496 and :ALLOCATION :CLASS slots.
1497 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1498 float exponents. (rereported by Peter Seibel)
1499 * fixed a bug: #\Space (and other whitespace characters) are no
1500 longer considered to be macro characters in standard syntax by
1501 GET-MACRO-CHARACTER.
1502 * fixed bug: initialization of condition class metaobjects no longer
1503 causes an instance of the condition to be created. (reported by
1505 * fixed bug: it is now possible to have slots such that
1506 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1507 description is neither :INSTANCE nor :CLASS.
1508 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1509 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1510 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1511 non-standard SLOT-DEFINITION-ALLOCATION values.
1512 * fixed bugs: various race conditions handling exiting threads.
1513 CL-PPCRE's thread torture test now passes.
1514 * fixed arguably-a-bug: GC time proportional to number of threads in
1515 system even when most of them are idle
1516 * optimization: improved performance of BIT and SBIT on bit-vectors.
1517 * .fasl file incompatibility: The fasl file version number has
1518 been incremented because of changes associated with package locks.
1520 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1521 * minor incompatible change: the system no longer provides
1522 optimization hints (conditions which are TYPEP
1523 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1524 mode; these hints are still emitted when higher SPEED optimization
1526 * new contrib module: a sampling profiler (profiling by statistical
1527 sampling, rather than by instrumenting functions) is available as
1528 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1529 * the behaviour of the standard function ED is now customizeable by
1530 third parties through a hook variable: see ED's documentation
1531 string for information on the protocol.
1532 * the compiler no longer emits efficiency notes for (FUNCALL X)
1533 when the type of X is uncertain under default optimization
1535 * fixed bug 276: mutating a binding of a specialized parameter to a
1536 method to something that is not TYPEP the specializer is now
1538 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1539 exhibit the required behaviour on the x86 platform. (thanks to
1540 Peter van Eynde, Eric Marsden and Bruno Haible)
1541 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1542 even for difficult arguments. (reported by Peter Graves)
1543 * fixed bug 141a: the backquote printer now descends quoted
1545 * fixed another bug in backquote printing: no more destructive
1546 modification of the form's list structure. (reported by Brian
1548 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1549 move between its address being taken and the call to
1550 interrupt_thread, fixing a crashing race condition.
1551 * the SB-POSIX contrib implementation has been adjusted so that it
1552 no longer exhibits ridiculously poor performance when constructing
1553 instances corresponding to C structs.
1555 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1556 * minor incompatible change: the sb-grovel contrib now treats C
1557 structures as alien (in the sense of SB-ALIEN) objects rather than
1558 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1559 has implications for memory management of client code
1560 (sb-grovel-returned objects must now be manually managed) and for
1561 type safety (alien objects now have full types).
1562 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1563 used to control emission of compiler diagnostics, rather than the
1564 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1565 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1566 quality should be considered deprecated.
1567 * install.sh now installs the user manual as well
1568 * (not quite a new documentable feature, but worth considering in
1569 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1570 beginnings of a semantically meaningful condition hierarchy is
1571 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1573 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1574 designator as the defaults argument.
1575 * fixed bug: Displaced arrays whose displaced-to array has become
1576 too small now cause ARRAY-DIMENSION to signal an error, providing
1577 for safer bounds-checking. (reported by Bruno Haible)
1578 * fixed bug: DEFCLASS slot definitions with identical :READER and
1579 :WRITER names now signal a reasonable error. (reported by Thomas
1581 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1583 * fixed bug: Invalid slot specification errors now print correctly.
1584 (thanks to Zach Beane)
1585 * fixed bug 320: Shared to local slot value transfers in class
1586 redefinitions now happen corrently from superclasses as
1587 well. (reported by Bruno Haible)
1588 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1590 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1591 improper lists correctly. (reported by Bruno Haible)
1592 * fixed bug 313: source-transform for <fun-name> was erroneously
1593 applied to a call of a value of a variable with name <fun-name>.
1594 (reported by Antonio Menezes Leitao)
1595 * fixed bug 307: The obsolete instance protocol ensures that
1596 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1597 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1598 remove dead unknown-values globs from the middle of the stack.
1599 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1600 easier to resume long computations after using *BREAK-ON-SIGNALS*
1601 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1602 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1603 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1604 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1605 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1606 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1608 * fixed a (fairly theoretical) bug in string printing: if
1609 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1610 does not have array-element-type equal to the most general string
1612 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1613 function name. (thanks to Nikodemus Siivola)
1614 * fixed bug: compile-time format string checker failed on
1615 non-closed ~{. (reported by Thomas F Burdick)
1616 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1617 don't warn on legal loop code involving more than one aggregate
1618 boolean. (thanks to Nikodemus Siivola)
1619 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1620 in keyword position to MAKE-PACKAGE was not being reported as
1622 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1623 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1625 * fixed bugs 280 and 312: the checking for multiple definitions in a
1626 file is less likely to become confused by uses of inline
1628 * fixed bug: the #S reader macro performs the keyword coercion
1629 specified for slot names. (reported by Kalle Niemitalo)
1630 * fixed bug: lambda lists may contain symbols whose names start with
1631 & but are not lambda-list-keywords; their occurrence triggers a
1633 * fixed bug 321: define-method-combination argument lambda lists do
1634 not cause invalid code to be generated when &optional or &aux
1635 variables have default values. (reported by Bruno Haible)
1636 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1637 have CLOS classes; this appears also to have fixed reported
1638 instability in stack exhaustion detection.
1639 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1640 MUFFLE-WARNING when no associated restart is present is now
1642 * optimization: rearranged the expansion of various defining macros
1643 so that each expands into only one top-level form in a
1644 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1646 * optimization: used a previously little-used slot in symbols to
1647 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1649 * fixed some bugs revealed by Paul Dietz' test suite:
1650 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1652 ** Fixed an optimization bug related to inheritance of initforms
1653 from local to shared slots.
1654 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1655 ** CAST optimizer forgot to flush argument derived type.
1656 ** print/read consistency on floats is now orders of magnitude
1657 more likely. (thanks also to Bruno Haible for a similar report
1659 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1660 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1661 for objects of type REAL. Make it so.
1662 ** SXHASH is specified (infelicitously) to respect similarity,
1663 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1664 it so. (thanks to Markus Ziegler)
1665 ** on the Alpha, the compiler succeeds in compiling functions
1666 returning a known number of arguments greater than 63.
1667 ** fixed handling of invalid NIL arguments in keyword position
1669 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1671 ** provided a readably-printable representation for RANDOM-STATE
1673 ** ensured that pathnames, where they have a namestring, always
1674 print using #P"..." syntax.
1676 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1677 * Support for the forthcoming 2.0 version of the NetBSD kernel
1678 running on x86 hardware has been added. (thanks to Perry
1679 E. Metzger most immediately, and others for their past work)
1680 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1681 current release version; SBCL's OpenBSD support had been broken
1682 since about the time of OpenBSD's switch to ELF binary format.)
1683 * [placeholder for DX summary]
1684 ** user code with &REST lists declared dynamic-extent, under high
1685 speed or space and low safety and debug optimization policy.
1686 * The manual has been converted to Texinfo format and the debugger
1687 chapter from the cmucl manual has been added.
1688 * A facility has been added to extract documentation strings from
1689 sbcl and store them as Texinfo-formatted snippets for inclusion in
1690 the manual (via Texinfo's @include directive)
1691 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1692 the test case to Dave Roberts)
1693 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1694 pointers. (reported by Sean Ross)
1695 * bug fix: PROFILE output is printed nicely even for large numerical
1696 values. (thanks to Zach Beane)
1697 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1698 greater than 32 handle EOF correctly.
1699 * bug fix: on X86 an immediate argument of the IMUL instruction is
1700 correctly printed in disassembly. (thanks to Lutz Euler)
1701 * bug fix: class slots in redefined classes preserve their old
1702 values. (thanks to Bruno Haible and Nikodemus Siivola)
1703 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1704 succeeds. (reported by Marco Baringer)
1705 * fixed some bugs revealed by Paul Dietz' test suite:
1706 ** READ-SEQUENCE now works on ECHO-STREAMs.
1707 ** RATIONALIZE works more according to its specification. (thanks
1710 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1711 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1712 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1713 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1714 should matter to you only if you rebind the printer control
1715 variables and then find you want different bindings in the
1716 debugger than in the ordinary execution of your program.)
1717 * The runtime build system has been tweaked to support building
1718 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1719 assembler and linker. (thanks to Nikodemus Siivola)
1720 * Unbound, undefined, undeclared variables now trigger full
1721 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1722 more often programmer error than deliberate exploitation of undefined
1724 * optimization: the hash algorithm for strings has changed to one
1725 that is less vulnerable to spurious collisions. (thanks to Juho
1727 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1728 needless bounds checking. (thanks to Juho Snellman)
1729 * optimization: implemented multiplication as a modular
1730 (UNSIGNED-BYTE 32) operation on the PPC backend.
1731 * fixed some bugs revealed by Paul Dietz' test suite:
1732 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1734 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1735 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1736 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1738 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1740 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1742 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1743 their output stream on EOF from read.
1744 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1745 have been read to end-of-file.
1746 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1748 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1749 description of determination of which consecutive characters
1751 ** Printing the "Space" character with escaping on now yields "#\\ ",
1752 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1753 ** Reading floating-point numbers with *READ-BASE* set to a number
1754 less than 10 works correctly.
1755 ** Reading floating-point numbers with *READ-BASE* set to a number
1756 more than 10 works correctly.
1757 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1758 the readtable currently in effect.
1760 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1761 * minor incompatible change: parsing of namestrings on a physical
1762 (Unix) host has changed; numbers after the final #\. in a
1763 namestring are no longer interpreted as a version field. This is
1764 intented to be largely invisible to the user, except that the
1765 meaning of the namestring "*.*.*" has changed: it now refers to a
1766 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1767 should usually be replaced by
1768 (make-pathname :name :wild :type :wild :version :wild)
1769 with the added benefit that this is more likely to be portable.
1770 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1771 option to OPEN now signals an error if the file being opened
1772 exists; this may have an impact on existing code.
1773 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1774 signals received, on the PowerPC platforms (both Linux and
1775 Darwin). (thanks to Pierre Mai for pointing out the location of
1777 * several fixes on OS X: The system now builds and runs cleanly on
1778 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1779 (thanks to Brian Mastenbrook)
1780 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1781 second-resolution: integer multiples of 1/3600 between -24 and 24.
1782 (thanks to Vincent Arkesteijn)
1783 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1784 type when called with 1 argument; PEEK-CHAR checked type of
1785 PEEK-TYPE only after having read first character from a
1786 stream. (reported by Peter Graves)
1787 * bug fix: the garbage collector now has much better locality
1788 behaviour, and in particular no longer treats all memory as being
1789 exhausted when many small objects point to each other in a deeply
1791 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1792 recognized as being TYPEP their class.
1793 * bug fix: the PUSHNEW documentation string has been corrected.
1794 (thanks to Vincent Arkesteijn)
1795 * bug fix: defaulting of the value for the last of an atypically
1796 large number of multiple values being bound was not being
1797 performed correctly on the Alpha or PPC platforms
1798 * optimization: implemented multiplication as a modular
1799 (UNSIGNED-BYTE 32) operation on the x86 backend.
1800 * optimization: SEARCH on simple-base-strings can now be open-coded.
1801 (see also contrib/compiler-extras.lisp for inspiration for
1802 teaching the compiler about the Boyer-Moore algorithm).
1803 * value, returned by MAX (and MIN) called with several EQUALP, but
1804 not EQL, arguments now does not depend on compiler settings.
1805 * fixed some bugs revealed by Paul Dietz' test suite:
1806 ** in stack analysis liveness information is propagated from
1807 non-local entry points.
1808 ** pathwise CAST removing failed when the CAST node did not start
1810 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1811 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1813 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1814 host is already defined.
1815 ** RENAME-FILE works on streams instead of signalling an internal
1817 ** PEEK-CHAR uses the current readtable when determining whether
1818 or not a character is whitespace.
1819 ** MERGE-PATHNAMES handles the case when the pathname does not
1820 specify a name while the default-pathname specifies a version.
1821 ** Pathnames now stand a better chance of respecting print/read
1823 ** Attempting to use standardized file system operators with a
1824 pathname with invalid :DIRECTORY components signals a
1826 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1828 ** DIRECTORY on logical pathnames is more correct.
1829 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1830 signal a TYPE-ERROR if their argument is not a stream.
1831 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1832 designator argument does not designate a stream.
1833 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1834 examining the synonym.
1835 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1837 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1838 element-type larger than ([UN]SIGNED-BYTE 32).
1840 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1841 * When built with the :SB-FUTEX feature, threaded builds now take
1842 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1843 for faster/more reliable mutex and condition variable support.
1844 * Incompatible change (but one you probably shouldn't have been using
1845 anyway): the interface and code for arbitrating between multiple
1846 threads in the same user session has been redesigned.
1847 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1848 error when called without an explicit environment argument.
1849 (thanks to Nikodemus Siivola)
1850 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1851 confused about their position. (thanks to Adam Warner and Gerd
1853 * bug fix: Pretty printing backquoted forms with unquotations in the
1854 argument list position of various code constructs such as LAMBDA
1855 now works correctly. (reported by Paul Dietz)
1856 * bug fix: Pretty printing unquotations no longer loses all
1857 stream position information.
1858 * optimization: performance of string output streams is now less
1859 poor for multiple small sequence writes.
1860 * optimization: performance of CSUBTYPEP in the presence of complex
1861 expressions involving CONS and NOT many times has been improved.
1862 (reported by Paul Dietz)
1863 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1865 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1866 streams for the REPL, for the debugger, and so on. (thanks to
1868 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1869 * fixed some bugs revealed by Paul Dietz' test suite:
1870 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1871 is ignored if it is too small, rather than propagating through
1872 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1873 ** extremely complex negations of CONS types were not being
1874 sufficiently canonicalized, leading to inconsistencies in
1876 ** VALUES tranformer lost derived type.
1878 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1879 * fixed a bootstrapping bug: the build process no longer assumes
1880 that the various BOOLE-related constants have the same value in
1881 host and target lisps. (noted by Paul Dietz' test suite on an
1882 SBCL binary built from CLISP)
1883 * The system can now be dynamically linked on the MIPS platform,
1884 which enables dynamic loading of foreign code from Lisp. (thanks
1885 to Ralf Baechle for discussions on the MIPS ABI)
1886 * The system now records debugging information for its own source
1887 files in a filesystem-position-independent manner, relative to
1888 the "SYS" logical host.
1889 * fixed a compiler bug: MV-LET convertion did not check references
1890 to the "max args" entry point. (reported by Brian Downing)
1891 * tweaked disassembly notes to be less confident about proclaiming
1892 some instruction as an LRA. (thanks to Brian Downing)
1893 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1894 now each have their own history, command character, and other
1895 characteristics. (thanks to David Lichteblau)
1896 * fixed some bugs revealed by Paul Dietz' test suite:
1897 ** compiler failure in compiling LOGAND expressions including a
1899 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1900 shift greater than 32.
1901 ** FLUSH-DEST did not mark blocks for type check regeneration.
1902 ** HANDLER-CASE failed to accept declarations in handler clauses
1903 in some circumstances.
1905 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1906 * New code in contrib/sb-introspect (still probably not entirely
1907 stable yet) provides some support for smart Lisp development
1908 environments like SLIME.
1909 * The conditions signalled for errors occurring when loading .fasl
1910 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1911 in a way which should help ASDF recover gracefully.
1912 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1913 stuff has been cleaned up. If you code contrib/ stuff, this might
1914 affect you, and you can look at contrib/README, contrib/STANDARDS,
1915 and/or the 0.8.4.27 diff to check.
1916 * In full calls the compiler now does not generate checks for declared
1917 argument types for all arguments.
1918 * various threading fixes
1919 ** and some experimental patches which didn't make it into
1920 the main tree for this release, but which are shipped in
1921 contrib/experimental-thread.patch as a possible fix for some
1922 failures (deadlock, spinning...) in GC-intensive multithreaded
1924 * fixed PPC build problem (source code incompatibility of different
1925 library versions): added offsetof() hackery which attempts to divine
1926 where glibc maintainers put uc_mcontext today
1927 * fixed bug 282: compiler does not trust type assertions while passing
1928 arguments to a full call.
1929 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1930 &OPTIONAL VALUES type parameter.
1931 * fix bug 214: algorithm for noting rejected templates is now more
1932 similar to that of template seletion. (also reported by rydis on
1934 * fixed bug 141b: printing backquoted information readably and prettily
1935 inserts a space where necessary.
1936 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1937 (CHARACTER 10) are now reported as errors, rather than propagated
1938 as unknown types. (reported by piso on #lisp)
1939 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1940 with values NIL and :ERROR. (thanks to Milan Zamazal)
1941 * fixed bug 191c: CLOS now does proper keyword argument checking as
1942 described in CLHS 7.6.5 and 7.6.5.1.
1943 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1944 longer raise an error; further, using a list as a for-as-arithmetic
1945 counter now raises a meaningful error.
1946 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1947 now understood by sequence creation functions such as MAKE-SEQUENCE
1949 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1950 a STRING-INPUT-STREAM.
1951 * compiler enhancement: SIGNUM is now better able to derive the type
1953 * type declarations inside WITH-SLOTS are checked. (reported by
1955 * fixed some bugs revealed by Paul Dietz' test suite:
1956 ** incorrect optimization of TRUNCATE for a positive first
1957 argument and negative second.
1958 ** compiler failure in let-convertion during flushing dead code.
1959 ** compiler failure while deriving type of TRUNCATE on an
1960 interval, containing 0.
1961 ** ASH of a negative bignum by a negative bignum count now returns
1963 ** intersection of CONS types now canonicalizes properly, fixing
1964 inconsistencies in SUBTYPEP.
1966 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1967 * incompatible change: The --disable-debugger command line
1968 option now clobbers the debugger at a more fundamental
1969 level, by redefining #'INVOKE-DEBUGGER instead of by
1970 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1971 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1972 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1973 enter the debugger REPL and then suffer recursive errors
1974 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1975 while under the new system, BREAK in --disable-debugger mode
1976 terminates the system just as an unhandled error would.
1977 * fixed compiler performance when processing loops with a step >1;
1978 * bug fix: DOCUMENTATION now retrieves generic function
1979 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1980 support has been systematized, and now supports the methods
1981 specified by ANSI, along with a default method and a method for
1982 slot documentation. (reported by Nathan Froyd)
1983 * bug fix: effective methods associated with a generic function are
1984 no longer cached over a change of that generic function's method
1985 combination. (reported by Andreas Fuchs)
1986 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1987 names a non-existent directory. (thanks to Andreas Fuchs)
1988 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1989 the PPC platform, lead to stack corruption; this has been fixed.
1990 (reported by Rainer Joswig)
1991 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1992 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1993 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1994 arithmetic in general, is now much more reliable on the PPC
1996 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1997 platform now returns the right answer.
1998 * optimization: restored some effective method precomputation in
1999 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2000 precomputation is now tunable.
2001 * optimization: compiler-internal data structure use has been
2002 reviewed, and changes have been made that should improve the
2003 performance of the compiler by about 20%.
2004 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2005 simple-base-strings and simple-bit-vectors is improved.
2006 * optimization: the optimization of 32-bit logical and arithmetic
2007 functions introduced in version 0.8.3 on the x86 has been
2008 implemented on the mips, ppc and sparc platforms; an
2009 implementation of the same facility, but for 64-bit arithmetic,
2010 has been added for the alpha.
2011 * microoptimization: the compiler is better able to make use of the
2012 x86 LEA instruction for multiplication by constants.
2013 * bug fix: in some situations compiler did not report usage of
2014 generic arithmetic in (SPEED 3) policy.
2015 * bug 145b fix: compiler used wrong type specifier while converting
2016 MEMBER-types to numeric.
2017 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2019 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2020 index argument now works properly on the Alpha platform.
2021 * bug fix: floating point exception treatment on the Alpha platform
2023 * bug fix: FILE-POSITION works much better on string input and
2024 output streams. (thanks to Nikodemus Siivola)
2025 * bug fix: many threading/garbage collection symptoms sorted.
2026 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2027 might be pseudo-atomic.
2028 * internal change: Stopping for GC is now done with signals not
2029 ptrace. GC is now done in whichever thread wanted it, instead of
2031 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2033 * bug fix: result form in DO is not contained in the implicit
2035 * incompatible change: ICR structure is changed; the value part of
2036 CONTINUATION is now called LVAR; corresponding functions are
2037 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2038 * added type deriver for ISQRT (thanks to Robert E. Brown).
2039 * bug fix: better support for loading from the command line when an
2040 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2042 * fixed some bugs revealed by Paul Dietz' test suite:
2043 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2044 ** ROUND and FROUND now give the right answer when given very
2045 small float arguments.
2046 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2048 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2049 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2050 ** LCM should return a non-negative integer.
2051 ** PARSE-INTEGER returned the index of a terminator instead of the
2052 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2053 ** PARSE-INTEGER returned an incorrect index being applied to a
2055 ** LCM with two arguments of 0 returns 0 rather than signalling
2057 ** unsigned addition of a 32-bit constant with the high bit set no
2058 longer causes an internal compiler error on the x86.
2059 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2060 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2061 with negative last argument.
2062 ** byte specifiers with zero size and position no longer cause
2063 an error during type derivation.
2064 ** bignum multiplication on the Alpha platform now returns the
2066 * porting: The system now builds on SuSE AMD64, although it still
2067 generates a 32-bit binary.
2068 * .fasl file incompatibility: The fasl file version number has
2069 been incremented (because of the changes to internal compiler
2070 data structures referred to above).
2072 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2073 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2074 more accurately, on the Darwin kernel running on PowerPC hardware.
2075 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2076 * Compiler code deletion notes now signal a condition of type
2077 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2078 an associated MUFFLE-WARNING restart.
2079 * The compiler now performs limited argument count validation of
2080 constant format strings in FORMAT, and where appropriate in ERROR,
2081 CERROR and WARN. (thanks to Gerd Moellmann)
2082 * New ASDF-INSTALL contrib can be used for automatic download and
2083 installation of third-party Lisp code from CCLAN or other sites
2085 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2086 which forces another thread to execute a function supplied by the
2088 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2089 now accept and act upon their :ELEMENT-TYPE keyword argument.
2090 (reported by Martin Atzmueller, Edi Weitz)
2091 * bug fix: FILE-POSITION now accepts position designators up to
2092 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2093 is the greater. (thanks to Patrik Nordebo)
2094 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2095 Kalle Olavi Niemitalo)
2096 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2097 printable. (reported by Eric Marsden)
2098 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2099 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2100 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2101 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2102 (thanks to Stig Sandoe)
2103 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2104 UNREAD-CHAR now correctly includes the unread character in the
2105 target sequence. (thanks to Gerd Moellmann)
2106 * bug fix in threaded builds: the system can now be suspended and
2107 resumed by shell job control with minimal disruption.
2108 * bug fixes in times and timezones >2038AD
2109 * better handling of "where is GNU make?" problem in build scripts
2110 (thanks to Nikodemus Siivola)
2111 * new optimization: inside a named function any reference to a
2112 function with the same name is considered to be a self-reference;
2113 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2114 optimization quality.
2115 * new optimization on x86: logical functions and + now have
2116 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2117 used when the result is truncated to 32 bits.
2118 * VALUES declaration is partially enabled.
2119 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2120 * fixed some bugs revealed by Paul Dietz' test suite:
2121 ** The system now obeys the constraint imposed by
2122 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2123 types form a lattice under type intersection.
2124 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2125 ** ASSOC now ignores NIL elements in an alist.
2126 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2127 and (1+ MOST-POSITIVE-FIXNUM) answers.
2128 ** The addition of a method with invalid qualifiers to a generic
2129 function does not cause an error to be signalled immediately;
2130 a warning is signalled, and the error is generated only on
2131 calling the generic function.
2132 * changed .fasl file version number, in part to add type codes for
2133 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2134 obscure ANSI requirements
2136 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2137 * fixed bug 148: failure to inline-expand a local function left
2138 garbage, confusing the compiler.
2139 * fixed bugs 3cd: structure slot readers perform type check if the
2140 slot can have an invalid value (i.e. it is either not initialized
2141 or can be written with a less specific slot writer).
2142 * bug fix: the compiler now traps array references to elements off
2143 the end of an array; previously, the bounds checking in some
2144 circumstances could go off-by-one.
2145 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2147 * type declarations for array element types now obey the description
2148 on the CLHS page "Declaration TYPE", as per discussions on
2149 sbcl-help around 2003-05-08. This means that a declaration
2150 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2151 declaration, all references to BAR will be asserted or assumed
2152 (with THE, so dependent on compiler policy) to involve objects of
2153 type FOO. Note that no such declaration is implied in
2154 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2155 * declared types of functions from the "Conditions"
2156 chapter. (reported by Paul Dietz)
2157 * bug fix: CERROR accepts a function as its first argument.
2158 * bug fix: NTH an NTHCDR accept a bignum as index
2159 arguments. (reported by Adam Warner)
2160 * optimization: character compare routines now optimize comparing
2161 against a constant character. (reported by Gilbert Baumann)
2162 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2163 argument now works properly on the MIPS platform.
2164 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2165 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2166 declared to have a functional type, does not cause an error
2168 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2169 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2170 * fixed bug: if last continuation of a deleted block has a
2171 destination, this destination should be deleted too. (reported by
2173 * fixed a bug in the bootstrap process: the host compiler's values
2174 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2175 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2176 test case from Patrik Nordebo)
2177 * improved the ability of the disassembler on the PPC platform to
2178 provide helpful disassembly notes.
2179 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2180 the class in more cases than previously.
2181 * bug fix: FILE-POSITION now understands :START and :END for
2182 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2183 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2184 8) data. (thanks to David Lichteblau)
2185 * bug fix: it is possible to add a method to a generic function
2186 without lambda list.
2187 * bug fix: reader failed to signal END-OF-FILE inside an
2188 object representation. (reported by Nikodemus Siivola)
2189 * fixed some bugs revealed by Paul Dietz' test suite:
2190 ** LAST and [N]BUTLAST should accept a bignum.
2191 ** condition slot accessors are methods.
2192 ** (VECTOR NIL) is a subtype of STRING.
2194 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2195 * minor incompatible change: some nonsensical specialized lambda
2196 lists (used in DEFMETHOD) which were previously ignored now signal
2198 * minor incompatible change: the system is now aware of the types of
2199 variables in the COMMON-LISP package, and will signal errors for
2200 most violations of these type constraints (where previously they
2201 were silently accepted).
2202 * minor incompatible change: COMPILE-FILE now uses the freedom
2203 afforded (ANSI 3.2.2.3) to use derived function types for
2204 functions defined in the same file. This also permits the system
2205 to warn on static type mismatches and function
2206 redefinition. (Currently it does not work with high DEBUG level.)
2207 * minor incompatible change: VALUES declaration is disabled.
2208 * When issuing notes, the compiler now signals a condition of type
2209 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2210 restart for use in user handlers. It is expected that the
2211 COMPILER-NOTE condition will eventually become a condition
2212 supertype to a hierarchy of note types, which will then be
2213 handleable in a similar fashion. However, at the moment, no such
2214 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2215 but it's an implementation detail, not a classification for the
2217 * Changes in type checking closed the following bugs:
2218 ** type checking of unused values (192b, 194d, 203);
2219 ** template selection based on unsafe type assertions (192c, 236);
2220 ** type checking in branches (194bc).
2221 * A short form of VALUES type specifier has ANSI meaning (it has
2222 increased the number of situations when SBCL cannot perform type
2224 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2225 keywords or constants is permissible.
2226 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2227 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2228 classes. (thanks to Antonio Martinez)
2229 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2230 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2231 argument to a pretty-print stream. Instead, it leaves any such
2232 operations to DESCRIBE-OBJECT methods.
2233 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2235 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2236 lambda lists. (thanks to David Lichteblau)
2237 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2238 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2239 (thanks to Gerd Moellmann)
2240 * fixed bug 246: increased compilation speed of long
2241 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2243 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2245 * DEFCONSTANT now throws a condition of type
2246 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2247 constant to a non-EQL value; CONTINUE and ABORT restarts
2248 respectively change and preserve the value.
2249 * fixed bug 63: The code walker, part of the implementation of CLOS,
2250 is now better at handling symbol macros.
2251 * bug fix: There is no longer an internal implementation type named
2252 CL:LENGTH. (reported by Raymond Toy)
2253 * bug fix: In macro-like defining macros/special operators the
2254 implicit block does not enclose the lambda list.
2255 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2256 enclosed in parentheses) are not suitable as type specifiers, and
2257 their use properly signals an error now.
2258 * bug fix: An argument count mismatch for a type specifier in code
2259 being compiled no longer causes an unhandled error at compile
2260 time, but signals a compile-time warning.
2261 * fixed simple vector readable printing
2262 * bug fix: DESCRIBE takes more care over whether the class
2263 precedence list slot of a class is bound before accessing it.
2264 (reported by Markus Krummenacker)
2265 * bug fix: FORMATTER can successfully compile pretty-printer format
2266 strings which use variants of the ~* directive inside.
2267 * bug fix: SEARCH now applies its TEST predicate to the elements of
2268 the arguments in the correct order. (thanks to Wolfhard Buss)
2269 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2270 in type inference. (thanks to Robert E. Brown)
2271 * bug fix: Reading in symbols with an explicit package name of ""
2272 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2273 * fixed some bugs revealed by Paul Dietz' test suite:
2274 ** NIL is now allowed as a structure slot name.
2275 ** Arbitrary numbers, not just REALs, are allowed in certain
2276 circumstances in LOOP for-as-arithmetic clauses.
2277 ** Multiple class redefinitions before slot access no longer
2278 causes a type error.
2279 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2280 association between the name and a class.
2281 ** Generic functions with non-standard method-combination and over
2282 five methods all of which return constants no longer return NIL
2283 after the first few invocations. (thanks to Gerd Moellmann)
2284 ** CALL-NEXT-METHOD with no arguments now passes the original
2285 values of the arguments, even in the presence of assignment.
2286 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2288 ** DISASSEMBLE works with closures and funcallable instances.
2289 ** ADD-METHOD now returns the generic function, not the new method.
2290 ** FIND-METHOD signals an error if the lengths of the specializers
2291 is incompatible with the generic function, even if the ERRORP
2293 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2294 which its argument is a member.
2295 ** DEFCLASS only redefines the class named by its class-name
2296 argument if that name is the proper name of the class;
2297 otherwise, it creates a new class.
2298 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2299 of the UNBOUND-SLOT condition to the name of the slot.
2300 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2301 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2302 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2303 SLOT-MAKUNBOUND in the specified fashion.
2305 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2306 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2307 cross-compilation host. As a consequence, we can now bootstrap our
2308 way up to SBCL starting with a bare gcc toolchain and human-readable
2309 source code (first the source to CLISP, then the source to SBCL).
2310 * A contributed module containing a partial implementation of the
2311 simple-streams interface has been included. (thanks to Rudi
2313 * A contributed module implementing the RFC1321 Message Digest
2314 Algorithm, known as MD5, has been included.
2315 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2316 no longer has any effect, as the code controlled by this feature
2317 has been deleted. (As far as we know, no-one has ever built using
2318 this feature, and its semantics were confused in any case).
2319 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2320 work on conditions (as required by the ANSI specification),
2321 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2322 expected behaviour on conditions. Users should note, however,
2323 that such behaviour is not required by the ANSI specification,
2324 and so use of this behaviour may render their code unportable.
2325 * fixed some bugs revealed by Paul Dietz' test suite:
2326 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2328 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2329 ** SXHASH on condition objects no longer returns NIL.
2330 ** :ALLOCATION :CLASS slots are better treated; their values are
2331 updated on class redefinition, and initforms inherited from
2332 superclasses are applied.
2333 ** REMOVE-METHOD returns its generic function argument even when
2334 no method was removed.
2335 ** SHARED-INITIALIZE now initializes the values of the requested
2336 slots, including those with :ALLOCATION :CLASS.
2337 ** ALLOCATE-INSTANCE now works on structure classes defined via
2338 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2340 ** SLOT-EXISTS-P now works on conditions, as well as structures
2342 ** MAKE-LOAD-FORM now has the required methods on
2343 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2344 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2345 keyword, and now implements the SLOT-NAMES argument.
2346 ** methods with &OPTIONAL arguments no longer allow too many
2347 arguments to be passed in the call without error.
2348 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2349 option is consistent with the required arguments of the generic
2350 function lambda list.
2351 * bug fix: REQUIRE accepts a string designator. (Thanks to
2353 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2354 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2355 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2356 * bug fix: The compiler now checks for duplicated variables in macro
2358 * bug fix: SETQ on globals returns the correct value.
2359 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2360 not a COMPILER-ERROR (followed by some other strange error on
2361 choosing the CONTINUE restart).
2362 * bug fix: make.sh and friends are now more consistent in the way that
2363 they look for GNU "make".
2365 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2366 * experimental native threads support (on x86 Linux >=2.4 only).
2367 This is not compiled in by default: you need to add :SB-THREAD to
2368 the target features. See the "Beyond ANSI" chapter of the manual
2370 * fix for longstanding nonANSIism: The old distinction between
2371 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2372 The return value from CL:FIND-CLASS is now a CLOS class, and
2373 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2374 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2375 * An interface to the MetaObject Protocol, as described in Kiczales,
2376 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2377 MIT Press, 1991, is available from the SB-MOP package.
2378 * incompatible change: the SB-PCL package should now be considered
2379 a private implementation detail, and no longer a semi-private MOP
2381 * minor incompatible change: due to rearrangement for threads, the
2382 control stack and binding stack are now allocated at arbitrary
2383 addresses instead of being hardcoded per-port. Users affected by
2384 this probably have to be doing advanced things with shared
2385 libraries, and will know who they are.
2386 * minor incompatible change: Previously, all --eval forms used were
2387 processed with READ before any of them were processed with EVAL.
2388 Now each --eval form is processed with both READ and EVAL before
2389 the next --eval form is processed. (Thus package operations like
2390 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2391 work as the user might reasonably expect.)
2392 * minor incompatible change: *STANDARD-INPUT* is now only an
2393 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2395 * minor incompatible change: Y-OR-N-P is now character-oriented,
2396 not line oriented. Also, YES-OR-NO-P now works without errors.
2397 (thanks to Antonio Martinez)
2398 * sb-aclrepl module improvements: an integrated inspector, added
2399 repl features, and a bug fix to :trace command.
2400 * Known functions, which cannot be open coded by the backend, are
2401 considered to be able to check types of their arguments. (fixing
2402 a bug report by Nathan J. Froyd)
2403 * fixed a bug in computing method discriminating functions: It is
2404 now possible to define methods specialized on classes which have
2405 forward-referenced superclasses. (thanks to Gerd Moellmann)
2406 * fixed evaluation order in optional entries (reported by Gilbert
2408 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2409 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2410 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2411 required-by-AMOP NAME argument, as well as CLASS and
2412 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2413 * fixed bug 20: DEFMETHOD can define methods using names that are
2414 not the proper names of classes to designate class specializers.
2415 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2416 Unix file descriptors, instead of blowing up. (thanks to Antonio
2418 * Garbage collection refactoring: user-visible change is that a
2419 call to the GC function during WITHOUT-GCING will not do garbage
2420 collection until the end of the WITHOUT-GCING. If you were doing
2421 this you were probably losing anyway.
2422 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2423 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2424 TYPEP the latter but not the former.
2425 * The compiler issues a full WARNING for calls to undefined functions
2426 with names from the CL package.
2427 * MAP-INTO for a vector destination is open coded. (reported by
2428 Brian Downing on c.l.l)
2429 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2430 documentation string.
2431 * fixed some bugs revealed by Paul Dietz' test suite:
2432 ** COPY-ALIST now signals an error if its argument is a dotted
2434 ** Condition slots are now accessed more correctly in the presence
2435 of multiple initargs for a given slot.
2436 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2437 exclude restarts of the same name associated with a different
2439 ** DEFCLASS of forward-referenced classes with another
2440 forward-referenced class in the superclasses list no longer
2442 ** Condition slots are now initialized once each, not multiple
2443 times. (thanks to Gerd Moellmann)
2444 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2446 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2447 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2448 arguments contain duplicated elements.
2449 ** RESTART-CASE understands local macros.
2450 ** RESTART-CASE associates exactly its own restarts with a condition.
2451 ** ENDP in safe mode checks its argument to be of type LIST.
2452 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2453 in question is unbound.
2454 ** Optimized MAKE-INSTANCE functions no longer cause internal
2455 assertion failures in the presence of duplicate initargs.
2456 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2457 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2458 propagated even in the ctor optimized implementation of
2460 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2462 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2463 * a better implementation of SXHASH on (simple) bit vectors,
2464 measured both in execution speed and in distribution of results
2465 over the positive fixnums, has been installed. Likewise, a better
2466 implementation of EQUAL for simple bit vectors is now available.
2467 * fixed CEILING optimization for a divisor of form 2^k.
2468 * fixed bug 240 (emitting extra style warnings "using the lexical
2469 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2470 by Antonio Martinez)
2471 * fixed SXHASH, giving different results for NIL depending on type
2472 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2473 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2474 variable. (found by Rolf Wester)
2475 * SBCL does not ignore type declarations for special
2476 variables. (reported by rif on c.l.l 2003-03-05)
2477 * some bug fixes in contrib/sb-aclrepl/
2478 * fixed some bugs revealed by Paul Dietz' test suite:
2479 ** a bug in the CONS type specifier, whereby the CAR and CDR
2480 types got intertwined, has been fixed;
2481 ** the type system is now able to reason about the interaction
2482 between INTEGER and RATIO types more completely;
2483 ** APPEND, [N]REVERSE and NRECONC check that those their
2484 arguments, which must be proper lists, are really so;
2485 ** An array specialized to be unable to hold elements has been
2486 implemented, as required -- yes, really -- by ANSI;
2487 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2488 on malformed property lists;
2490 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2491 * incompatible packaging change: in line with Unix convention,
2492 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2493 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2494 * REQUIRE and PROVIDE are now optionally capable of doing something
2495 useful. See the documentation string for REQUIRE.
2496 * infrastructure for a managed SBCL contrib system: contributed
2497 modules in this release include:
2498 ** the ASDF system definition facility;
2499 ** an interface to the BSD Sockets API;
2500 ** an ACL-like convenience interface to the repl;
2501 (thanks to Kevin Rosenberg)
2502 ** an implementation of ROTATE-BYTE, with efficient implementation
2504 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2505 gives a closer approximation to the right answer than previously.
2506 (thanks to Raymond Toy)
2507 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2508 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2509 optional environment arguments, as required by ANSI.
2510 * fixed bugs in other functions taking environment objects, allowing
2511 calls with an explicit NIL environment argument to be compiled
2513 * fixed bug 228: primary return values from
2514 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2515 COMPILE or FUNCTION.
2516 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2517 :TYPE will no longer signal errors on innocuous objects.
2518 * fixed bug 231b: SETQ is better at respecting type declarations in
2519 the lexical environment.
2520 * fixed a bug in DEFCLASS: classes named by symbols with no or
2521 unprintable packages can now be defined.
2522 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2523 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2524 * fixed compiler failure related to checking types of functions.
2525 (reported by Robert E. Brown)
2526 * the compiler is now much more consistent in its error-checking
2527 treatment of bounding index arguments to sequence functions: in
2528 (SAFETY 3) code, errors will be signalled in almost all cases if
2529 invalid sequence bounding indices are passed to functions defined
2530 by ANSI to operate on sequences.
2531 * fixed a bug in the build procedure: documentation of SBCL-specific
2532 packages is now preserved and available in the final Lisp image.
2533 * lifted FDEFINITION lookup out of loops in the implementation of
2534 many list operations. (thanks to Robert E. Brown)
2535 * fixed a bug in the reader: the #n# reader macro now works for
2536 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2537 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2538 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2539 E. Brown and Tony Martinez respectively)
2540 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2541 SLOT-BOUNDP: in particular, it is now able to optimize them much
2542 better, and is now not vulnerable to having packages renamed.
2543 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2544 always called when a slot is not present in an instance. (thanks
2546 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2547 longer signal an error on unreading a character following EOF on
2548 the previous constituent stream. (thanks to Tony Martinez)
2549 * fixed some bugs revealed by Paul Dietz' test suite:
2550 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2551 not just nonnegative fixnums;
2552 ** the logical bit-array operators such as BIT-AND now accept an
2553 explicit NIL for their "opt-arg" argument (to indicate a
2554 freshly-consed result bit-array);
2555 ** ELT now signals an error on an invalid sequence index in safe
2557 ** the type system is now cleverer about negations of numeric
2558 types, and consequently understands the BIGNUM and RATIO types
2560 ** the type system is now cleverer about the interaction between
2561 INTEGER and RATIO types: while bugs still remain, many more
2562 cases are accurately computed;
2563 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2564 if it is in the last clause;
2565 ** CONSTANTLY now correctly returns a side-effect-free function in
2567 ** DECLARE is no longer treated as a special-operator; in
2568 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2569 * incremented fasl file version number due to the change in the
2570 DEFSTRUCT-SLOT-DESCRIPTION structure.
2572 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2573 * minor incompatible change: code processed by the "interpreter" or
2574 EVAL now has a compilation optimization policy of (DEBUG 2)
2575 (changed from (DEBUG 1)) to improve debuggability of interactive
2576 development, and to allow the use of the debug RETURN command in
2578 * an experimental implementation of the RETURN command for the
2579 debugger has been included. (thanks to Frederik Kuivinen)
2580 * fixed bug 62: constraints were not propagated into a loop.
2581 * fixed bug in embedded calls of SORT (reported and investigated by
2583 * fixed some bugs revealed by Paul Dietz' test suite:
2584 ** printing and reading of arrays with some dimensions having
2585 length 0 (thanks to Gerd Moellmann);
2586 ** BOA constructor with &AUX argument without a default value does
2587 not cause a type error;
2588 ** CONSTANTP now returns true for all self-evaluating objects.
2590 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2591 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2592 accessors that are related by inheritance, as specified in the
2593 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2594 Valtteri Vuorikoski)
2595 * The compiler is now able to inline functions that were defined in
2596 a complex lexical environment (e.g. inside a MACROLET).
2597 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2599 * Support for the upcoming FreeBSD-5.0 release has been included.
2600 (thanks to Dag-Erling Smorgrav)
2601 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2602 effect when it is not in a toplevel context.
2603 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2604 stand a better chance of being correct. (thanks to Gerd
2606 * fixed bug in COERCE, which now signals an error on coercing a
2607 rational to a bounded real type which excludes the expected
2609 * The compiler is now able to derive types more accurately from the
2610 COERCE and COMPILE functions.
2611 * fixed bug 223: functional binding is considered to be constant
2612 only for symbols in the CL package.
2613 * fixed bug 231: SETQ did not check the type of a variable being set
2614 (reported by Robert E. Brown)
2615 * A new optimization for MAKE-INSTANCE has been included, fixing
2616 various bugs (including relating to :ALLOCATION :CLASS slots and
2617 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2619 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2620 ** As required by ANSI, LOOP now disallows anonymous collection
2621 clauses such as COLLECT I in conjunction with aggregate boolean
2622 clauses such as THEREIS (= I 1);
2623 ** LOOP now signals an error when any variable is reused in the
2624 same loop (including the potentially useful construct analogous
2625 to WITH A = 1 WITH A = (1+ A);
2626 ** IT is only a special loop symbol within the first clause of a
2627 conditional loop clause;
2628 ** LOOP with a typed iteration variable over a hashtable now
2629 signals a type error iff it should.
2630 * fixed some other bugs revealed by Paul Dietz' test suite:
2631 ** FILE-STREAM now names the class previously known as FD-STREAM;
2632 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2633 argument) no longer signals an error;
2634 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2635 of the slot symbol, rather than using the current package
2636 ((:CONC-NAME "") continues to intern the slot's name in the
2638 * incremented fasl file version number, because of the incompatible
2639 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2640 of the new implementation of DEFINE-COMPILER-MACRO.
2642 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2643 * Support for building SBCL for MIPS platforms running in
2644 little-endian mode has now been checked in, and basic
2645 functionality on said platforms verified.
2646 * minor incompatible change: PCL now records the pathname of a file
2647 in which methods and the like are defined, rather than its
2649 * minor incompatible change: TRUENAME now considers the truename of
2650 a file naming a directory to be the pathname with :DIRECTORY
2651 component indicating that directory.
2652 * minor incompatible change: a NAMED clause in the extended form of
2653 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2654 reason for the previous behaviour is unclear.
2655 * more systematization and improvement of CLOS and MOP conformance
2656 in PCL (thanks to Gerd Moellman and Pierre Mai):
2657 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2659 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2660 primary methods with no specializers;
2661 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2663 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2664 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2665 FORWARD-REFERENCED-CLASSes; error reporting on
2666 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2668 ** SXHASH on CLOS instances now uses a slot internal to the
2669 instance to return different numbers on distinct instances,
2670 while preserving the same return value through invocations of
2672 ** DEFMETHOD signals errors when methods with longer incongruent
2673 lambda lists are added to generic functions;
2674 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2675 CLASS, as specified in AMOP;
2676 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2677 on the order returned by the primary method for classes of
2678 class STANDARD-CLASS;
2679 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2680 * fixed some bugs shown by Paul Dietz' test suite:
2681 ** DOLIST puts its body in TAGBODY;
2682 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2684 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2685 value producing form;
2686 ** if more variables are given to PROGV than values, extra
2687 variables are bound and made to have no value;
2688 ** NSUBSTITUTE on list arguments gets the right answer with
2690 ** ELT signals an error of type TYPE-ERROR when the index argument
2691 is not a valid sequence index;
2692 ** LOOP signals (at macroexpansion time) an error of type
2693 PROGRAM-ERROR when duplicate variable names are found;
2694 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2695 ** FUNCALL of special-operators now cause an error of type
2697 ** PSETQ now works as required in the presence of side-effecting
2698 symbol-macro places;
2699 ** NCONC accepts any object as its last argument;
2700 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2702 ** loop-for-as-package does not require a package to be explicitely
2704 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2706 * fixed bug 166: compiler preserves "there is a way to go"
2707 invariant when deleting code.
2708 * fixed bug 172: macro lambda lists with required arguments after
2709 &REST arguments now cause an error to be signalled. (thanks to
2711 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2712 bug. (thanks to Matthew Danish)
2713 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2715 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2716 arguments are lexically rebound. (thanks to Gerd Moellmann and
2718 * fixed bug 194: error messages are now more informative when there
2719 is no primary method applicable in a call to a generic
2720 function. (thanks to Gerd Moellmann)
2721 * fixed bug in command line argument checking (thanks to Julian
2723 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2724 (thanks to Matthew Danish)
2725 * incremented fasl file version number, because of the
2726 SXHASH-related changes in the layout of CLOS data structures
2728 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2729 * minor incompatible change: The runtime (the Unix executable named
2730 "sbcl") is now much pickier about the .core files it will load.
2731 Essentially it now requires .core files to descend from the same
2732 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2733 as the runtime does. (The intent is to prevent the crashes which
2734 can occur, and which can even be reported as mysterious failures,
2735 when people patch the sources or change the build parameters
2736 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2737 sbcl and .core files.)
2738 * fixed bug: VALUES-LIST is no longer optimized away.
2739 * fixed bug 142: The FFI conversion of C string values to Lisp
2740 string values no longer conses excessively. (thanks to Nathan
2741 Froyd porting Raymond Toy's fix to CMU CL)
2742 * began to systematize and improve MOP conformance in PCL (thanks to
2743 Nathan Froyd, Gerd Moellman and Pierre Mai):
2744 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2746 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2747 ** FINALIZE-INHERITANCE is now called on class finalization;
2748 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2749 argument precedence order.
2750 * fixed bug 202: The compiler no longer fails on functions whose
2751 derived types contradict their declared type.
2752 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2753 so it can be non-toplevel.
2754 * The fasl file version number has changed (because of the new
2755 implementation of DEFMACRO).
2756 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2757 safe code, that any length requirement by their type-specifier
2758 argument is valid. The exceptions to this are described in bug
2760 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2761 functions now check the types of their inputs as required by ANSI.
2762 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2763 introduced symbol is DECLAREd to be SPECIAL.
2764 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2765 * fixed bug: :COUNT argument to sequence functions may be negative
2766 * fixed bug: body of DO-SYMBOLS may contain declarations
2767 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2768 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2769 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2770 symbol macro only once
2771 * fixed printing of call frame when argument list is unavailable
2772 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2773 * compiler no longer signals WARNING on unknown keyword
2776 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2777 * A beta-quality port to the mips architecture running Linux,
2778 based on the old CMUCL backend, has been made. It has been tested
2779 on a big-endian kernel, and works sufficiently well to be able to
2780 rebuild itself; it has not been tested in little-endian mode.
2781 * fixed an inconsistency between gencgc.c and purify.c which made
2782 dumping/loading .core files unreliable
2783 * fixed bug 120a: The compiler now deals correctly with IFs where
2784 the consequent is the same as the alternative, instead of
2785 misderiving the return type. (thanks to Alexey Dejneka)
2786 * fixed bug 113: Logical pathnames are now dumpable (the logical
2787 host is resolved at load-time, throwing an error if it is not
2789 * fixed bug 174: FORMAT's error message is slightly clearer when a
2790 non-printing character is used in a format directive.
2791 * fixed several bugs in compiler checking of type declarations, i.e.
2792 violations of the Python "declarations are assertions" principle
2793 (thanks to Alexey Dejneka)
2794 * fixed several bugs in PCL's error checking (thanks to Gerd
2796 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2798 * fixed bug in compilation of functions as first class values
2799 (thanks to Antonio Martinez-Shotton)
2800 * The compiler's handling TYPE-ERRORs which it can prove will
2801 inevitably happen at runtime has been cleaned up and corrected
2802 in several ways. (thanks to Alexey Dejneka)
2803 * improved argument type checking for various basic arithmetic
2804 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2805 lavished on them in the past that they can be compiled in many
2806 ways in different special cases
2807 * fixed bug 181: compiler checks validity of user supplied type
2809 * cleaned up code flushing in optimization: Function calls which
2810 should signal errors for safety purposes (e.g. which ANSI says
2811 should signal errors when their arguments are of incorrect type)
2812 are no longer optimized away.
2813 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2814 * incremented fasl file version number, because changes in the
2815 implementation of sequence functions like COERCE caused
2816 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2817 in old inline expansions) to become undefined. (Actually these
2818 changes were later undone, so we might very well be binary
2819 compatible with 0.7.7 after all, but leaving the version number
2820 incremented seemed like the simplest and most conservative
2823 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2824 * An alpha-quality port to the parisc architecture running Linux,
2825 based on the old CMUCL backend, has been made. This, even more so
2826 than the other backends, should be considered still a work in
2827 progress; known problems include that the Linux kernel in 64-bit
2828 mode does not propagate the correct sigcontext structure to
2829 userspace, and consequently SBCL on a parisc64 kernel will not
2831 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2832 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2833 suppressed.) Also now that the compiler is looking at declarations
2834 in the environment, it checks optimization declarations as well,
2835 and suppresses inlining when (> DEBUG SPEED).
2836 * More fixes have been made to treatment of floating point exception
2837 treatment and other Unix signals. In particular, floating point
2838 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2839 * The detection and handling of control stack exhaustion (infinite
2840 or very deeply nested recursion) has changed. Stack exhaustion
2841 detection is now done by write-protecting pages at the OS level
2842 and applies at all optimization settings; when found, a
2843 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2844 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2845 be caught using IGNORE-ERRORS.
2846 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2847 correctly when transforming recursive LABELS functions to LETs.
2848 (thanks to Alexey Dejneka)
2849 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2850 that are names of constants or global variables.
2851 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2852 alien routines with docstrings.
2853 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2854 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2856 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2857 (thanks to Eric Marsden)
2858 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2859 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2860 * Bug fix: The SPARC backend can now compile functions involving
2861 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2862 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2863 to a functions expecting a PATHNAME-DESIGNATOR.
2864 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2865 lambda lists. (thanks to Alexey Dejneka)
2866 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2867 functions so that EXPORTing the name of the class doesn't cause
2868 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2869 to Antonio Martinez for reporting this)
2870 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2871 constant-folded without failing an assertion. (thanks to Einar
2872 Floystad Dorum for reporting this)
2873 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2874 types are now dealt with more correctly by the compiler.
2875 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2876 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2877 behaviour analogous to RENAME-FILE. This puts its behaviour more
2878 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2880 * The fasl file version number has changed again. (because of the
2881 bug fix involving the names of PCL MAKE-INSTANCE functions)
2883 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2884 * bug fix: Floating point exceptions are treated much more
2885 consistently on the x86/Linux and PPC/Linux platforms.
2886 * Array initialization with :INITIAL-ELEMENT is now much faster for
2887 cases when the compiler cannot open code the array creation, but
2888 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2889 array accesses have also seen a speed increase.
2890 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2891 specified. (This was at the root of some bad interactions between
2892 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2894 * bug fix: Internal error arguments for undefined functions are now
2895 computed correctly on the PPC/Linux platform.
2896 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2897 Raymond Toy's patch for CMU CL)
2898 * Support for the Solaris 9 operating environment has been included
2899 (thanks to Daniel Merritt)
2900 * A very ugly but hopefully complete draft of the missing FFI chapter
2901 of the manual has been created by reformatting the corresponding
2902 CMU CL manual chapter into (currently very ugly and incoherent)
2903 DocBook and bringing it up to date for SBCL behavior. Thus, the
2904 manual is now essentially complete, at least by my extreme
2905 once-and-only-once standards, whereby it's acceptable to refer to
2906 the doc strings of SB-EXT functions as the primary documentation.
2907 * The fasl file version number has changed again, due to cleanup of
2908 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2909 is no longer a static symbol.)
2911 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2912 * SBCL now builds with OpenMCL (version 0.12) as the
2913 cross-compilation host; also, more progress has been made toward
2914 bootstrapping under CLISP.
2915 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2917 * bug 158 fixed: The compiler can now deal with integer loop
2918 increments different from 1; fixing this turned out also to fix
2920 * bug 169 fixed: no more bogus warnings about using lexical bindings
2921 despite the presence of perfectly good SPECIAL declarations (thanks
2922 to David Lichteblau)
2923 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2924 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2925 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2927 * bug 180 fixed: Method combination specifications no longer ignore
2928 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2929 * bug fix: Structure type predicate functions now check their argument
2930 count as they should.
2931 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2932 correctly. (thanks to Pierre Mai)
2933 * minor incompatible change: The --noprogrammer option is deprecated
2934 in favor of the new --disable-debugger option, which is very similar.
2935 (The major difference is that it takes effect at a slightly different
2936 time at startup, causing handling of errors in --sysinit and
2937 --userinit files will be affected differently.) The
2938 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2939 been added to allow this functionality to be controlled from ordinary
2940 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2941 maintainers, who might want to run non-interactive scripts to
2942 build SBCL cores which will later be used interactively.)
2943 * minor incompatible change: The LOAD function no longer, when given
2944 a wild pathname to load, loads all files matching that pathname.
2945 Instead, an error of type FILE-ERROR is signalled.
2947 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2948 * bug 147 fixed: The compiler preserves its block link/count
2949 invariants more correctly now so that it doesn't crash. (thanks
2951 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2953 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2954 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2955 no-value-for-this-character as NIL (as specified by ANSI).
2956 * HOST-NAMESTRING on physical pathnames now returns a string that is
2957 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2958 (thanks to Christophe Rhodes)
2959 * The Alpha port handles icache flushing more correctly. (thanks to
2961 * More progress has been made toward bootstrapping under CLISP. (thanks
2962 to Christophe Rhodes)
2963 * The fasl file format has changed again, because dynamic loading
2964 on OpenBSD (which has non-ELF object files) motivated some cleanups
2965 in the way that foreign symbols are transformed and passed around.
2966 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2967 is no longer treated as whitespace by the reader, but instead as
2968 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2969 |A
\7fB|, instead of A as it used to.
2971 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2972 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2973 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2974 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2975 this as of 0.7.1.45, but was left out of the previous news section
2976 (thanks to Dan Barlow)
2977 * SBCL now runs on the Solaris operating system on SPARC architectures
2978 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2979 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2980 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2981 * SPARC backend cleanups, allowing builds of cores optimized for V8
2982 and V9 SPARCS, and also emission of code targeted to a particular
2983 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2985 * SBCL is closer to bootstrapping under CLISP, thanks to various
2986 fixes by Christophe Rhodes.
2987 * The fasl file format has changed again, to allow the compiler's
2988 INFO database to support symbol macros.
2989 * The user manual (in doc/) is formatted into HTML more nicely.
2990 (thanks to coreythomas)
2991 * The system is smarter about SUBTYPEP relationships, especially
2992 those involving NOT types (including types such as ATOM which are
2993 represented internally using NOT types). Thus SUBTYPEP is less
2994 likely to return (VALUES NIL NIL) in general, and in particular
2995 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2997 * The fasl file format has changed again, because the internal
2998 representation of types now includes a new slot to support the new
2999 SUBTYPEP-of-NOT-types logic.
3000 * (not a change in the main branch of SBCL, but a related prototype
3001 which can hopefully be merged into the main branch of SBCL in the
3002 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3003 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3004 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3005 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3006 dispatch macro character combinations. (thanks to Alexey Dejneka)
3007 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3008 that superficially look like logical namestrings correctly.
3009 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3011 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3012 default supplied-p) for &optional and &key arguments. (thanks to
3015 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3016 * incompatible change: The compiler is now less aggressive about
3017 tail call optimization, doing it only when (> SPACE DEBUG) or
3018 (> SPEED DEBUG). (This is an incompatible change because there are
3019 programs which relied on the old CMU-CL-style behavior to optimize
3020 away their unbounded recursion which will now die of stack overflow.)
3021 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3022 for non-GENCGC systems has been increased to 20M (since that
3023 seems much closer to the likely performance optimum for modern
3024 systems than the old 4M value was)
3025 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3026 and *DEBUG-PRINT-LEVEL*
3027 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3028 of CMU CL's support for SPARC, and various endianness and other
3029 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3030 * new syntactic sugar for the Unix command line: --load foo.bar is now
3031 an alternate notation for --eval '(load "foo.bar")'.
3033 ** The system now detects stack overflow and handles it gracefully,
3034 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3035 optimization settings. (This is a good thing in general, and
3036 its introduction in this version should be particularly timely
3037 for anyone whose code fails because of suppression of tail
3039 ** The system now hunts for the C variable "environ" in a more
3040 devious way, to avoid segfaults when the C library version differs
3041 between compile time and run time. (thanks to Christophe Rhodes)
3042 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3043 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3044 ** The compiler no longer issues bogus style warnings for undefined
3045 classes in the same source file as the DEFCLASSes which defined
3046 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3048 ** fixes in CONDITION class precedence list for undefined function
3049 errors (thanks to Alexei Dejneka)
3050 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3051 correctly. (thanks to Dan Barlow)
3052 ** portability fixes aiming at bootstrapping under CLISP (thanks
3053 to Dave McDonald and Christophe Rhodes)
3054 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3055 ** fixes in type translation and and type inference (thanks to
3057 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3058 ** various fixes in the new ports (thanks to Dan Barlow)
3059 * several changes related to debugging:
3060 ** suppression of tail recursion, as noted above
3061 ** stack overflow detection, as noted above
3062 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3063 is now the default. (For some time encapsulation has been more
3064 reliable than the breakpoint-based :ENCAPSULATE NIL
3065 implementation, at least on X86 systems; and I just noticed that
3066 encapsulation also seems closer to the spirit of the ANSI
3069 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3071 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3072 up properly again. (There was a packaging bug in 0.7.0 which
3073 left their definitions in SB-SYS::LOAD-FOREIGN and
3074 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3075 vital for most things which interface to C-level interfaces,
3076 like extensions working with sockets or databases or
3077 Perl-compatible regexes or whatever, and the need to fix
3078 this bug is the main reason that 0.7.1 was released so
3080 ** DEFGENERIC is now choosier about the methods it redefines, so that
3081 reLOADing a previously-LOADed file containing DEFGENERICs does
3082 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3083 cycle now works as it should. (thanks to Alexey Dejneka)
3084 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3085 fixed by Christophe Rhodes. (He actually submitted this patch
3086 months ago, and I delayed until after 0.7.0.)
3087 ** Bug 111 (internal compiler confusion about runtime checks on
3088 FUNCTION types) was fixed by Alexey Dejneka.
3089 * Some internal cleanups (getting rid of variables which aren't
3090 needed now that the byte interpreter is gone) caused the fasl
3091 file format number to change again.
3093 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3094 * major incompatible change: The default fasl file extension, i.e. the
3095 default extension for files produced by COMPILE-FILE, has changed
3096 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3098 ** There are many changes in the implementation of the compiler.
3099 SBCL is now essentially a compiler-only implementation of ANSI
3100 Common Lisp. EVAL still "interprets" a few special cases, but
3101 almost all the interesting cases are handled by creating
3102 a LAMBDA expression, calling COMPILE on it, then calling
3103 FUNCALL on the result.
3104 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3105 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3106 Since the code is newer, there might still be some new bugs
3107 (though not as many as before Martin Atzmueller's fixes:-). But
3108 the new code is substantially simpler and clearer, and hopefully
3109 any remaining bugs will be simpler, less fundamental, and more
3110 fixable then the bugs in the old code.
3111 ** The revised compiler is still a little unsteady on its feet.
3113 *** The debugging information it produces (particularly the names
3114 of FUNCTION objects) is sometimes much less useful than what
3115 the old compiler produced.
3116 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3117 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3118 within a MACROLET) has been temporarily weakened.
3119 ** There are new compiler optimizations for various functions:
3120 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3121 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3122 *** the math functions TRUNCATE, FLOOR, and CEILING
3123 *** the function-of-all-trades COERCE
3124 Mostly these should be transparent, but there's one
3125 potentially-annoying problem (bug 117): when the compiler
3126 inline-expands a function and does type analysis on the result,
3127 it can create control paths which have type mismatches, and
3128 when it can't prove that those control paths aren't taken,
3129 it will issue WARNINGs about the type mismatches. This is
3130 a particular problem in practice for the new sequence functions.
3131 It's not clear how this should be fixed, and for now, a
3132 workaround is given in the entry for 117 in the BUGS file.
3133 ** (Because of the interaction between the two previous items --
3134 occasional inlining problems and new inline expansions -- some
3135 of the new sequence function optimizations won't really kick in
3136 completely until debugging information, and then inlining, are
3137 straightened out in some future version.)
3138 * minor incompatible changes:
3139 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3140 DIRECTORY no longer implicitly promotes NIL slots of its
3141 pathname argument to :WILD. In particular, when you ask for the
3142 contents of a directory (which you used to be able to do without
3143 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3144 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3145 ** changes in behavior that ANSI explicitly defines to be
3146 implementation dependent:
3147 *** The new compiler-only implementation still conforms with ANSI,
3148 but acts a little different than before. Besides the obvious
3149 changes in performance tradeoffs (that the cost per form passed
3150 to EVAL has gone up, and the cost per form executed by EVAL
3151 has gone down), the behavior of the system changes a little
3152 because there are no longer any interpreted function objects.
3153 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3154 e.g. doing COMPILE on the output of interactive DEFUN is
3156 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3158 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3159 has changed from (:CL) to NIL.
3160 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3161 e.g. #\Newline and #\Tab, don't have names specified in the
3162 ANSI Common Lisp standard, is now based on their ASCII symbolic
3163 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3164 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3165 are no longer used for output.
3166 ** changes in internal implementation constants:
3167 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3168 4 million. (If your application spends a lot of time GCing and
3169 you have a lot of RAM, you might want to experiment with
3170 increasing it even more.)
3171 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3172 However, almost all old code should still continue to work without
3173 immediate update, as SB-C-CALL is now a (deprecated) nickname
3175 ** Old operator names in the style DEF-FOO are now deprecated in
3176 favor of new corresponding names DEFINE-FOO, for consistency with
3177 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3178 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3179 internal symbols, but a few supported extensions like
3180 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3181 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3182 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3183 etc. as you get deeper into recursive calls to the debugger
3184 command loop, instead of the old "5]", "5]]", "5]]]"
3185 sequence. (I was motivated to do this when squabbles between
3186 ILISP and SBCL left me very deeply nested in the debugger. In the
3187 short term, this change will probably provoke more ILISP/SBCL
3188 squabbles, but hopefully it will be an improvement in the long run.)
3189 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3190 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3191 function, since declaiming return types involving aliens is
3192 (1) annoyingly messy to do by hand and (2) vital to efficient
3193 compilation of code which calls such functions.
3194 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3195 longer reexported by the SB-EXT package. They're solely useful
3196 for alien code, so it seems more logical that you should get
3197 them from the SB-ALIEN package, not in SB-EXT.
3198 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3199 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3200 features. Instead, the code that they used to control is always
3201 built into the system.
3202 * many other bug fixes
3203 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3204 advantage of the new EVAL-WHEN stuff and to clean them up in
3205 general, and they are now more ANSI-compliant in a number of
3206 ways. Martin Atzmueller is responsible for a lot of this.
3207 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3209 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3210 *** correct ERROR type for various file operations
3211 *** some fixes for Lisp streams
3212 *** DEFMETHOD syntax checking
3213 *** changing old weird representation of debug information as
3214 strings (which, among their other deficiencies, don't transform
3215 correctly when you rename packages, and don't change their
3216 print representation when you change things like *PACKAGE*
3217 and *PRINT-LENGTH*) to symbols and lists of symbols
3218 He also made several improvements and fixed several bugs in DESCRIBE.
3219 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3221 *** misbehavior of WRITE-STRING/WRITE-LINE
3222 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3223 and several other LOOP problems as well
3224 *** DIRECTORY when similar filenames are present
3225 *** DEFGENERIC with :METHOD options
3226 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3227 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3228 *** argument ordering in FIND with :TEST option
3229 *** mishandled package designator argument in APROPOS-LIST
3230 *** various problems in the backquote readmacro
3231 *** a bug in APROPOS
3232 *** probably some others that I'm not describing very well here,
3233 since the CVS log documents them by reference to sbcl-devel
3234 messages, and the SourceForge archives aren't working well.:-(
3235 ** Dan Barlow improved the Alpha port (and is making progress on the
3236 PPC port, for those of you who think different).
3237 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3238 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3239 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3240 continues to work on the SPARC port (for those of you in a position
3241 to look down upon our little PC-compatible boxes from a great height).
3242 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3243 and uses that copy, rather than the current dynamic value, when
3244 it's trying to decide whether to truncate output. Thus e.g.
3245 (let ((*print-lines* 50))
3246 (pprint-logical-block (stream nil)
3248 (let ((*print-lines* 8))
3249 (print (aref possiblybigthings i) stream)))))
3250 should now truncate the logical block only at 50 lines, instead of
3251 often truncating it at 8 lines, as it did before.
3252 * The doc/cmucl/ directory, containing old CMU CL documentation
3253 from the time of the fork, is no longer part of the base system.
3254 SourceForge has shut down its anonymous FTP service, and with it
3255 my original plan for distributing the old CMU CL documentation
3256 there. For now, if you need these files you can download an old
3257 SBCL source release and extract them from it.
3258 * The fasl file version number changed again, for dozens of reasons,
3259 some of which are apparent above.
3261 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3262 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3263 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3264 patch, so that DISASSEMBLE output is much nicer.
3265 * The code in the SB-PROFILE package now seems reasonably stable.
3266 I still haven't decided what the final interface should look like
3267 (I'd like PROFILE to interact cleanly with TRACE, since both
3268 facilities use function encapsulation) but if you have a need
3269 for profiling now, you can probably use it successfully with
3270 the current CMU-CL-style interface.
3271 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3272 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3273 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3274 link now returns the dangling link itself, and for similar
3275 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3276 link itself. (In these cases the old code signalled an error and
3277 looped endlessly, respectively.) Thus, DIRECTORY now works even
3278 in the presence of dangling and cyclic symbolic links.
3279 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3280 is now a supported extension again, since the consensus on
3281 sbcl-devel was that it can be useful for ordinary development
3282 work, not just for debugging SBCL itself.
3283 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3284 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3285 that currently-defined functions might be redefined later with
3286 different return types.
3287 * Hash tables can be printed readably, as inspired by CMU CL code
3288 of Eric Marsden and SBCL code of Martin Atzmueller.
3289 * better error handling in CLOS method combination, thanks to
3290 Martin Atzmueller porting Pierre Mai's CMU CL patches
3291 * more overflow fixes for >16Mbyte I/O buffers
3292 * A bug in READ has been fixed, so that now a single Ctrl-D
3293 character suffices to cause end-of-file on character streams.
3294 In particular, now you only need one Ctrl-D at the command
3295 line (not two) to exit SBCL.
3296 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3298 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3299 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3300 does the right thing.
3301 * The implementation of some type tests, especially for CONDITION
3302 types, is now tidier and maybe faster, due to CMU CL code
3303 originally by Douglas Crosher, ported by Martin Atzmueller.
3304 * Some math functions have been fixed, and there are new
3305 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3306 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3307 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3308 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3309 likely to be built into sbcl-0.7.0.)
3310 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3311 CL, ported to SBCL by Martin Atzmueller)
3312 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3313 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3314 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3315 * There's a new slam.sh hack to shorten the edit/compile/debug
3316 cycle for low-level changes to SBCL itself, and a new
3317 :SB-AFTER-XC-CORE target feature to control the generation of
3318 the after-xc.core file needed by slam.sh.
3319 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3320 COMPILE-FILE is no longer supported, so that now every function
3321 gets an entry point, so that block compilation looks a little
3322 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3323 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3324 returns the number of bytes consed since the system started,
3325 rather than the number consed since the first time the function
3326 was called. (The new definition parallels ANSI functions like
3327 CL:GET-INTERNAL-RUN-TIME.)
3328 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3329 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3330 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3331 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3332 * Fasl file version numbers are now independent of the target CPU,
3333 since historically most system changes which required version
3334 number changes have affected all CPUs equally. Similarly,
3335 the byte fasl file version is now equal to the ordinary
3338 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3339 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3340 is no longer recognized. I apologize for this, because it was
3341 listed in SB-EXT as a supported extension, but I found that
3342 its existing behavior was poorly specified, as well as incorrectly
3343 specified, and it looked like too much of a mess to straighten it
3344 out. I have enough on my hands trying to get ANSI stuff to work..
3345 * many patches ported from CMU CL by Martin Atzmueller, with
3346 half a dozen bug fixes in pretty-printing and the debugger, and
3347 half a dozen others elsewhere
3348 * fixed bug 13: Floating point infinities are now supported again.
3349 They might still be a little bit flaky, but thanks to bug reports
3350 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3351 as flaky as they were.
3352 * The --noprogrammer command line option is now supported. (Its
3353 behavior is slightly different in detail from what the old man
3354 page claimed it would do, but it's still appropriate under the
3355 same circumstances that the man page talks about.)
3356 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3357 are now supported, and enabled by default. Thus, the compiler can
3358 handle many floating point and complex operations much less
3359 inefficiently. (Thus e.g. you can implement a complex FFT
3361 * The compiler now detects type mismatches between DECLAIM FTYPE
3362 and DEFUN better, and implements CHECK-TYPE more correctly, and
3363 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3364 thanks to patches from Martin Atzmueller.
3365 * various fixes to make the cross-compiler more portable to
3366 ANSI-conforming-but-different cross-compilation hosts (notably
3367 Lispworks for Windows, following bug reports from Arthur Lemmens)
3368 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3369 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3370 fixed thanks to Pierre Mai's CMU CL patches.
3371 * fixes to keep the system from overflowing internal counters when
3372 it tries to use i/o buffers larger than 16M bytes
3373 * fixed bug 45a: Various internal functions required to support
3374 complex special functions have been merged from CMU CL sources.
3375 (When I was first setting up SBCL, I misunderstood a compile-time
3376 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3377 * improved support for type intersection and union, fixing bug 12
3378 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3379 more obscure bugs as well
3380 * some steps toward byte-compiling non-performance-critical
3381 parts of the system, courtesy of patches from Martin Atzmueller
3382 * Christophe Rhodes has made some debian packages of sbcl at
3383 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3384 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3385 stable, but are nonetheless usable. When he's ready, I'd be happy
3386 to add them to the SourceForge "File Releases" section. (And if
3387 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3388 * new fasl file format version number (because of changes in
3389 internal representation of (OR ..) types to accommodate the new
3390 support for (AND ..) types, among other things)
3392 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3393 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3395 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3397 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3398 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3399 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3400 * DESCRIBE now works on structure objects again.
3401 * Most function call argument type mismatches are now handled as
3402 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3403 know whether the function will be redefined before the call is
3404 executed. (The compiler could flag local calls with full WARNINGs,
3405 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3406 it doesn't keep track of enough information to know whether calls
3407 are local in this sense.)
3408 * Compiler output is now more verbose, with messages truncated
3409 later than before. (There should be some supported way for users
3410 to override the default verbosity, but I haven't decided how to
3411 provide it yet, so this behavior is still controlled by the internal
3412 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3413 src/compiler/ir1util.lisp.)
3414 * Fasl file format version numbers have increased again, because
3415 support for the Gray streams extension changes the layout of the
3416 system's STREAM objects.
3417 * The Gray subclassable streams extension now works, thanks to a
3418 patch from Martin Atzmueller.
3419 * The full LOAD-FOREIGN extension (not just the primitive
3420 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3421 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3422 but like most other programs, it defaults to copying the Unix
3423 environment from the original process instead of starting the
3424 new process in an empty environment.
3425 * Extensions which manipulate the Unix environment now support
3426 an :ENVIRONMENT keyword option which doesn't smash case or
3427 do other bad things. The CMU-CL-style :ENV option is retained
3428 for porting convenience.
3429 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3430 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3432 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3434 * A patch from Martin Atzmueller seems to have solved the SIGINT
3435 problem, and as far as we know, signal-handling now works cleanly.
3436 (If you find any new bugs, please report them!)
3437 * The system no longer defaults Lisp source file names to types
3438 ".l", ".cl", or ".lsp", but only to ".lisp".
3439 * The compiler no longer uses special default file extensions for
3440 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3441 seems to expect a single default extension for all compiled code,
3442 and there's no compelling reason to try to stretch the standard
3443 to allow two different extensions.) Instead, byte-compiled files
3444 default to the same extension as native-compiled files.
3445 * Fasl file format version numbers have increased again, because
3446 a rearrangement of internal implementation packages made some
3447 dumped symbols in old fasl files unreadable in new cores.
3448 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3449 many fewer weird special cases.
3450 * Bug #17 (differing COMPILE-FILE behavior between logical and
3451 physical pathnames) has been fixed, and some related misbehavior too,
3452 thanks to a patch from Martin Atzmueller.
3453 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3454 by Tim Moore, ported to SBCL by Martin Atzmueller.
3455 * Martin Atzmueller fixed several filesystem-related problems,
3456 including bug #36, in part by porting CMU CL patches, which were
3457 written in part by Paul Werkowski.
3458 * More compiler warnings in src/runtime/ are gone, thanks to
3459 more patches from Martin Atzmueller.
3460 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3463 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3465 * DESCRIBE now works on CONDITION objects.
3466 * The debugger now handles errors which arise when trying to print
3467 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3469 * The build system now uses an additional file, customize-target-features.lisp,
3470 to allow local modifications to the target *FEATURES* list. (The point of
3471 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3472 debugging features enabled, without having to worry about propagating it
3473 into everyone's system when I do a "cvs update".) When no
3474 customize-target-features.lisp file exists, the target *FEATURES* list
3475 should be constructed the same way as before.
3476 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3477 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3478 warn about behavior which is undefined under the ANSI spec. Note: This
3479 is specified by ANSI, but it's not very popular with programmers.
3480 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3481 macro in the SBCL sources for an example of a workaround which you
3482 might use to make portable ANSI-standard code which does what you want.
3483 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3484 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3485 when it's used as a non-toplevel form. (This is required in order
3486 to implement the DEFCONSTANT-EQX macro.)
3487 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3488 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3489 funny things, due to bugs in the implementation of EVAL-WHEN
3490 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3491 non-toplevel DEFCONSTANTs are uncommon.)
3492 * The core file version number and fasl file version number have been
3493 incremented, because the old noncompliant DEFCONSTANT behavior involved
3494 calling functions which no longer exist, and because I also took the
3495 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3496 * fixed bug 1 (error handling before read-eval-print loop starts), and
3497 redid debugger restarts and related debugger commands somewhat while
3499 ** The QUIT debugger command is gone, since it did something
3500 rather different than the SB-EXT:QUIT command, and since it never
3501 worked properly outside the main toplevel read/eval/print loop.
3502 Invoking the new TOPLEVEL restart provides the same functionality.
3503 ** The GO debugger command is also gone, since you can just invoke
3504 the CONTINUE restart directly instead.
3505 ** The TOP debugger command is also gone, since it's redundant with the
3506 FRAME 0 command, and since it interfered with abbreviations for the
3508 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3510 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3511 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3512 some porting work by Martin Atzmueller)
3513 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3514 with a boatload of patches which he ported to SBCL. Now that those
3516 ** The system tries to make sure that its low-priority messages
3517 are prefixed by semicolons, to help people who like to use
3518 syntax highlighting in their ILISP buffer. (This patch
3519 was originally due to Raymond Toy.)
3520 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3521 patches originally written by Raymond Toy.
3522 ** The compiler understands coercion between single-value and
3523 multiple-VALUES type expressions better, getting rid of some very
3524 weird behavior, thanks to patches originally by Robert MacLachlan
3525 and Douglas Crosher.
3526 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3527 lambda lists, thanks to a patch originally by Pierre Mai.
3528 ** The system no longer bogusly warns about "abbreviated type
3530 ** The compiler gets less confused by inlining and RETURN-FROM,
3531 thanks to some patches originally by Tim Moore.
3532 ** The system no longer hangs when dumping circular lists to fasl
3533 files, thanks to a patch originally from Douglas Crosher.
3534 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3535 undefined function error.
3536 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3537 in structure slot initforms) for now, documented workaround instead:-|
3538 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3539 * fixed bug 5: added stubs for various Gray stream functions called
3540 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3541 installed, at least appropriate type errors are generated
3542 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3543 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3544 consistently in DEFMETHOD forms.
3545 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3546 it doesn't seem to affect SBCL after all
3547 * The C runtime system now builds with better optimization and many
3548 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3550 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3552 * The system is now under CVS at SourceForge (instead of the
3553 CVS repository on my home machine).
3554 * The new signal handling code has been tweaked to treat register
3555 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3556 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3557 downstream, e.g. in debug-int.lisp, has implicit dependencies
3558 on the unsignedness of integer representation of machine words,
3559 and that caused the system to bomb out with infinite regress
3560 when trying to recover from type errors involving signed values,
3561 e.g. (BUTLAST '(1 2 3) -1).)
3562 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3563 (This was one of the bugs Peter Van Eynde reported back in July.)
3564 * The system now uses code inspired by Colin Walters' O(N)
3565 implementation of MAP (from the cmucl-imp@cons.org mailing
3566 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3567 inline the MAP operation, and there is more than one
3568 sequence argument to the MAP call (so that it can't just
3569 do ETYPECASE once and for all based on the type of the
3570 single sequence argument). (The old non-inline implementation
3571 of the general M-argument sequence-of-length-N case required
3572 O(M*N*N) time when any of the sequence arguments were LISTs.)
3573 * The QUIT :UNIX-CODE keyword argument has been renamed to
3574 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3575 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3577 * Raymond Wiker's patches to port dynamic loading from Linux to
3578 FreeBSD have been added.
3579 * The BUGS file is now more nearly up to date, thanks in large part
3580 to Martin Atzmueller's review of it.
3581 * The debugger now flushes standard output streams before it begins
3582 its output ("debugger invoked" and so forth).
3583 * The core version number and fasl file version number have both
3584 been incremented, because of incompatible changes in the layout
3586 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3587 used to not be called for a saved Lisp image.)
3588 * Martin Atzmueller's version of a patch to fix a compiler crash,
3589 as posted on sbcl-devel 13 September 2000, has been installed.
3590 * Instead of installing Martin Atzmueller's patch for the
3591 compiler transform for SUBSEQ, I deleted the compiler transform,
3592 and transforms for some similar consing operations.
3593 * A bug in signal handling which kept TRACE from working on OpenBSD
3595 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3596 away by constant folding
3597 * The system now defines its address space constants in one place
3598 (in the Lisp sources), and propagates them automatically elsewhere
3599 (through GENESIS and the sbcl.h file). Therefore, patching the
3600 address map is less unnecessarily tedious and error-prone. The
3601 Lisp names of address space constants have also been systematized.
3602 * CVS tags like dollar-Header-dollar have been removed from
3603 the sources, because they have never saved me trouble and
3604 they've been source of trouble working with patches and other
3605 diff-related operations.
3606 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3607 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3609 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3611 * The system has been ported to OpenBSD.
3612 * The system now compiles with a simple "sh make.sh" on the systems
3613 that it's supported on. I.e., now you no longer need to tweak
3614 text in the target-features.lisp-expr and symlinks in src/runtime/
3615 by hand, the make.sh takes care of it for you.
3616 * The system is no longer so grossly inefficient when compiling code
3617 involving vectors implemented as general (not simple) vectors (VECTOR T),
3618 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3619 which dares to use the various sequence functions on non-simple
3620 vectors, takes less of a performance hit.
3621 * There is now a primitive type predicate VECTOR-T-P
3622 to test for the (VECTOR T) type, so that e.g.
3623 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3624 can now be compiled with some semblance of efficiency. (The old code
3625 turned the type declaration into a full call to %TYPEP at runtime!)
3626 * AREF on (VECTOR T) is still not fast, since it's still compiled
3627 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3628 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3629 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3630 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3631 without making full calls to SUBTYPEP at runtime.
3632 (Some analogous efficiency issues for non-simple vectors specialized to
3633 element types other than T, or for non-simple multidimensional arrays,
3634 have not been addressed. They could almost certainly be handled the
3635 same way if anyone is motivated to do so.)
3636 * The changes in array handling break binary compatibility, so
3637 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3638 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3639 instead of (VALUES T T).
3640 * By following the instructions that Dan Barlow posted to sbcl-devel
3641 on 2 July 2000, I was able to enable primitive dynamic object
3642 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3643 functionality is not implemented (since it calls ld to resolve
3644 library references automatically, requiring RUN-PROGRAM for its
3645 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3646 try to resolve library references) is now supported.
3647 * The system now flushes the standard output streams when it terminates,
3648 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3649 them at several other probably-convenient times, e.g. in each pass of
3650 the toplevel read-eval-print loop, and after evaluating a form given
3651 as an "--eval" command-line option. (These changes were motivated by a
3652 discussion of stream flushing issues on cmucl-imp in August 2000.)
3653 * The source transform for TYPEP of array types no longer assumes
3654 that an array whose element type is a not-yet-defined type
3655 is implemented as an array of T, but instead punts, so that the
3656 type will be interpreted at runtime.
3657 * There is now some support for cross-compiling in make.sh: each of
3658 the phases of make.sh has its own script. (This should be transparent
3659 to people doing ordinary, non-cross-compile builds.)
3660 * Since my laptop doesn't have hundreds of megabytes of memory like
3661 my desktop machine, I became more motivated to do some items on
3662 my to-do list in order to reduce the size of the system a little:
3663 ** Arrange for various needed-only-at-cold-init things to be
3664 uninterned after cold init. To support this, those things have
3665 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3666 symbols with such names are now uninterned after cold init).
3667 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3668 fasl files for cold load.
3669 ** Remove the old compiler structure pooling code (which used to
3670 be conditional on the target feature :SB-ALLOC) completely.
3671 ** Redo the representation of some data in cold init to be more compact.
3672 (I also looked into supporting byte compiled code at bootstrap time,
3673 which would probably reduce the size of the system a lot, but that
3674 looked too complicated, so I punted for now.)
3675 * The maximum signal nesting depth in the src/runtime/ support code has
3676 been reduced from 4096 to 256. (I don't know any reason for the very
3677 large old value. If the new smaller value turns out to break something,
3678 I'll probably just bump it back up.)
3679 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3681 * Many, many bugs reported by Peter Van Eynde have been added to
3682 the BUGS list; some have even been fixed.
3683 * While enabling dynamic object file loading, I tried to make the
3684 code easier to understand, renaming various functions and variables
3685 with less ambiguous names, and changing some function calling
3686 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3687 * While trying to figure out how to do the OpenBSD port, I tried to
3688 clean up some of the code in src/runtime/. In particular, I dropped
3689 support for non-POSIX signal handling, added various comments,
3690 tweaked the code to reduce the number of compilation warnings, and
3691 renamed some files to increase consistency.
3692 * To support the new automatic configuration functionality in make.sh,
3693 the source file target-features.lisp-expr has been replaced with the
3694 source file base-target-features.lisp-expr and the machine-generated
3695 file local-target-features.lisp-expr.
3696 * fixed a stupid quoting error in make.sh so that using CMU CL
3697 "lisp -batch" as cross-compilation host works again
3699 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3701 * DESCRIBE no longer tries to call itself recursively to describe
3702 bound/fbound values, so that it no longer fails on symbols which are
3703 bound to themselves (like keywords, T, and NIL).
3704 * DESCRIBE now works on generic functions.
3705 * The printer now prints less-screwed-up representations of closures
3706 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3707 * A private symbol is used instead of the :EMPTY keyword previously
3708 used to mark empty slots in hash tables. Thus
3709 (DEFVAR *HT* (MAKE-HASH-TABLE))
3710 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3711 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3712 now does what ANSI says that it should. (You can still get
3713 similar noncompliant behavior if bang on the hash table
3714 implementation with all the symbols you get back from
3715 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3716 This breaks binary compatibility, since tests for equality to
3717 :EMPTY are wired into things like the macroexpansion of
3718 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3720 * There's now a minimal placeholder implementation for CL:STEP,
3721 as required by ANSI.
3722 * An obscure bug in the interaction of the normal compiler, the byte
3723 compiler, inlining, and structure predicates has been patched
3724 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3725 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3727 * Missing ordinary arguments in a macro call are now detected even
3728 when the macro lambda list contains &KEY or &REST.
3729 * The debugger no longer complains about encountering the top of the
3730 stack when you type "FRAME 0" to explicitly instruct it to go to
3731 the top of the stack. And it now prints the frame you request even
3732 if it's the current frame (instead of saying "You are here.").
3733 * As specified by ANSI, the system now always prints keywords
3734 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3736 * The default initial SIZE of HASH-TABLEs is now smaller.
3737 * Type information from CLOS class dispatch is now propagated
3738 into DEFMETHOD bodies, so that e.g.
3739 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3741 is now basically equivalent to
3742 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3743 (DECLARE (TYPE SINGLE-FLOAT X))
3745 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3746 operation, without having to do run-time type dispatch.
3747 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3748 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3749 * Since I don't seem to be making big file reorganizations very often
3750 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3751 is overflowing my ability to conveniently back them up), I've finally
3752 checked the system into CVS. (The CVS repository is on my home system,
3753 not at SourceForge -- putting it on SourceForge might come later.)
3754 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3755 high-level GC-NOTIFY-FOO functions send their output. (There's
3756 still very little control of where low-level verbose GC functions
3757 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3758 less than it used to -- the GC-NOTIFY-FOO functions are now under
3759 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3760 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3761 in only one place in the source code, and propagates it automatically
3762 everywhere that it's needed. Thus e.g. when I bump the version from
3763 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3764 * The C source files now include boilerplate legalese and documentation
3765 at the head of each file (just as the Lisp source files already did).
3766 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3767 to his page will be replaced with a link to his new CLiki service.
3769 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3771 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3772 * The build process now looks for GNU make under the default name "gmake",
3773 instead of "make" as it used to. If GNU make is not available as "gmake"
3774 on your system, you can change this default behavior by setting the
3775 GNUMAKE environment variable.
3776 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3777 can build without error under CMU CL.
3779 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3781 * There is now a partial SBCL user manual (with some new text and some
3782 text cribbed from the CMU CL manual).
3783 * The beginnings of a profiler have been added (starting with the
3784 CMU CL profiler and simplifying and cleaning up). Eventually the
3785 main interface should be through the TRACE macro, but for now,
3786 it's still accessed through vaguely CMU-CL-style functions and macros
3787 exported from the package SB-PROFILE.
3788 * Some problems left over from porting CMU CL to the new
3789 cross-compilation bootstrap process have been cleaned up:
3790 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3791 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3792 part when getting rid of the compiler *BACKEND* stuff.)
3793 ** The value of *NULL-TYPE* was screwed up, because it was
3794 being initialized before the type system knew the final
3795 definition of the 'NULL type. This screwed up several key
3796 optimizations in the compiler, causing inefficiency in all sorts
3797 of places. (I found it because I wanted to understand why
3798 GET-INTERNAL-RUN-TIME was consing.)
3799 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3800 PROCLAIM FTYPE information. Unfortunately this broke binary
3801 compatibility again, since now the forms output by DEFGENERIC
3802 to refer to functions which didn't exist in 0.6.3.
3803 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3804 can use the new (as of 0.6.3) transform for SOME into MAP into
3806 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3807 values to (INTEGER 0 1000000), so that the time code will no longer
3808 occasionally get blown up by Linux returning 1000000 microseconds
3809 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3810 its output conform to the ANSI spec. (Alas, this makes its output
3811 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3812 * A full call to MAP NIL with a single sequence argument no longer conses.
3813 * fixes to problems pointed out by Martin Atzmueller:
3814 * The manual page no longer talks about multiprocessing as though
3815 it were currently supported.
3816 * The ILISP support patches have been removed from the distribution,
3817 because as of version 5.10.1, ILISP now supports SBCL without us
3818 having to maintain patches.
3819 * added a modified version of Raymond Toy's recent CMU CL patch for
3820 EQUALP comparison of HASH-TABLE
3822 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3824 * The system still can't cross-compile itself with
3825 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3826 entails), but at least it can compile more of itself that way
3827 than it used to be able to, and various buglets which were uncovered
3828 by trying to cross-compile itself that way have now been fixed.
3829 * This release breaks binary compatibility again. This time
3830 at least I've incremented the FASL file format version to 2, so that the
3831 problem can be detected reliably instead of just causing weird errors.
3832 * various new style warnings:
3833 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3834 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3835 ** using the lexical binding of a variable named in the *FOO* style
3836 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3837 as specified by ANSI.
3838 * *RANDOM-STATE* is no longer automatically initialized from
3839 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3840 default behavior of the system is to repeat its behavior every time
3841 it's run. If you'd like to change this behavior, you can always
3842 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3843 old convention there was no comparably easy way to get the system to
3844 repeat its behavior every time it was run.
3845 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3846 FTYPE declarations has been removed, in favor of their ANSI
3847 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3848 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3849 the types of their sequence arguments can be determined at compile time.
3850 This is done through a new open code expansion for MAP which eliminates
3851 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3852 argument types can be determined at compile time.
3853 * The optimizer now transforms COERCE into an identity operation when it
3854 can prove that the coerced object is already of the correct type. (This
3855 can be a win for machine generated code, including the output of other
3856 optimization transforms, such as the MAP transform above.)
3857 * Credit information has been moved from source file headers into CREDITS.
3858 * Source file headers have been made more standard.
3859 * The CASE macro now compiles without complaining even when it has
3862 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3864 * (Note that the way that the PCL macroexpansions were rewritten
3865 to accommodate the change in DEFGENERIC below breaks binary
3866 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3867 not run under sbcl-0.6.2. Once we get out of alpha releases,
3868 i.e. hit release 1.0.0, we'll probably try to maintain binary
3869 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3870 and sbcl-1.4.4. Until then, however, it might be fairly common
3871 for maintenance releases to break binary compatibility.)
3872 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3874 * The reporting of unhandled signals has been changed to print some
3875 explanatory text as well as the report form. (Previously only
3876 the report form was printed.)
3877 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3878 it defines, so that the compiler no longer issues undefined function
3879 warnings for compiled-but-not-yet-loaded generic functions.
3880 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3881 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3882 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3883 section of the ANSI spec are supported.
3884 * The "" nickname for the "KEYWORD" package has been removed.
3885 The reader still handles symbol tokens which begin with a package marker
3886 as keywords, but it doesn't expose its mechanism for doing so in the
3887 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3888 * The system now issues STYLE-WARNINGs for contradictory TYPE
3889 proclamations. (Warnings for contradictory FTYPE proclamations would
3890 be nice too, but those can't be done usefully unless the type system
3891 is made smarter about FUNCTION types.)
3892 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3893 systematized, so that "*target-*.lisp is supposed to exist only on the
3894 target and imply that there's a related file which exists on the
3895 host, and *host-*.lisp is supposed to exist only on the host and imply
3896 that there's a related file which exists on the target. This involves a
3897 lot of renaming. Hopefully the acute confusion caused by the renaming
3898 will be justified by the reduction in chronic confusion..
3899 ** runtime-type.lisp -> early-target-type.lisp
3900 ** target-type.lisp -> late-target-type.lisp
3901 ** early-host-format.lisp -> early-format.lisp
3902 ** late-host-format.lisp -> late-format.lisp
3903 ** host-error.lisp -> misc-error.lisp
3904 ** early-error.lisp -> early-target-error.lisp
3905 ** late-error.lisp -> late-target-error.lisp
3906 ** host-defboot.lisp -> early-defboot.lisp
3907 ** code/misc.lisp -> code/target-misc.lisp
3908 ** code/host-misc.lisp -> code/misc.lisp
3909 ** code/numbers.lisp -> code/target-numbers.lisp
3910 ** code/early-numbers.lisp -> numbers.lisp
3911 ** early-host-type.lisp -> early-type.lisp
3912 ** late-host-type.lisp -> late-type.lisp
3913 ** host-typep.lisp -> typep.lisp
3914 ** load.lisp -> target-load.lisp
3915 ** host-load.lisp -> load.lisp
3916 ** host-disassem.lisp -> disassem.lisp
3917 ** host-insts.lisp -> insts.lisp
3918 ** byte-comp.lisp -> target-byte-comp.lisp
3919 ** host-byte-comp.lisp -> byte-comp.lisp
3920 ** host-signal.lisp -> signal.lisp
3921 ** host-defstruct.lisp -> defstruct.lisp
3922 ** late-target-type.lisp -> deftypes-for-target.lisp
3923 Furthermore, several other previously target-only files foo.lisp (e.g.
3924 hash-table.lisp and random.lisp) have been split into a target-and-host
3925 foo.lisp file and a target-only target-foo.lisp file, with their key type
3926 definitions in the target-and-host part, so that the cross-compiler will
3927 know more about target types.
3928 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3929 gone away. In their place are various *BACKEND-FOO* variables
3930 corresponding to the slots of the old structure.
3931 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3932 into the target SBCL has been fixed.
3933 * The chill.lisp system for loading cold code into a running SBCL
3935 * Support for the CMU CL "scavenger hook" extension has been removed.
3936 (It was undocumented and unused in the CMU CL sources that SBCL was
3937 derived from, and stale in sbcl-0.6.1.)
3938 * Various errors in the cross-compiler type system were detected
3939 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3940 (enabling various consistency checks). Many of them were fixed,
3941 but some hard problems remain, so the compiler is back to
3942 running without *TYPE-SYSTEM-INITIALIZED* for now.
3943 * As part of the cross-compiler type system cleanup, I implemented
3944 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3945 * I have started adding UNCROSS calls throughout the type system
3946 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3947 on cross-compiler input files will be able to go away, and various
3949 * CONSTANTP now returns true for quoted forms (as explicitly required
3952 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3954 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3955 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3956 (MAKE-LIST -1). In the longer term, it ought to become true
3957 that declarations are assertions even at SAFETY 1. For now, it's not
3958 quite true even at SAFETY 3, but it's at least more nearly true..
3959 (Note that this change seems to increases the size of the system by
3960 O(5%) and to decrease the speed of the compiler by 20% or more.)
3961 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3962 for the problem of printing dozens of lines of distracting information
3963 about low-level system machinery as part of the top stack frame
3964 on entry to the debugger when an undefined function was called.
3965 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3966 is not reset to COMMON-LISP-USER.
3967 * Compilation of stuff related to dyncount.lisp has been made conditional
3968 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3969 smaller. The various dyncount-related symbols have been moved into
3970 a new "SB-DYNCOUNT" package.
3971 * tty-inspect.lisp has been renamed to inspect.lisp.
3972 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3973 feature has gone away. (When we eventually port to other flavors of
3974 libc and/or Unix, we'll try to make the differences between flavors
3975 invisible at the user level.)
3976 * Various other *FEATURES* tags, and/or their associated conditionals,
3977 have been removed if obsolescent, or given better documentation, or
3978 sometimes given more-mnemonic names.
3980 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3982 * tidied up "make.sh" script
3983 * tidied up system directory structure
3984 * better "clean.sh" behavior
3985 * added doc/FOR-CMUCL-DEVELOPERS
3986 * many many small tweaks to output format, e.g. removing possibly-confusing
3987 trailing #\. character in DESCRIBE-INSTANCE
3988 * (EQUALP #\A 'A) no longer signals an error.
3989 * new hashing code, including EQUALP hashing
3990 * tidied up Lisp initialization and toplevel
3991 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3992 * command line argument processing
3993 * added POSIX-GETENV function to deal with Unix-ish environment variables
3994 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3995 terminating SBCL on EOF
3996 * non-verbose GC by default
3997 * There is no more "sbcl" shell script; the sbcl file is now the C
3998 runtime executable (just like CMU CL).
3999 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4001 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4002 symbol and package information as Lisp native symbol and package objects
4003 instead of strings naming symbols and strings naming packages. This way,
4004 whenever packages are renamed (as in warm init), debug information is
4005 transformed along with everything else.
4006 * tweaked the optimization policy declarations which control the building
4007 of SBCL itself. Now, among other things, the system no longer saves
4008 source location debugging information. (This helps two problems at once
4009 by reducing SBCL size and by keeping SBCL from trying to look for its
4010 sources -- which may not exist -- when reporting errors.)
4011 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4012 debugging and testing purposes
4013 * cleaned up printing, making the printer call PRINT-OBJECT for
4014 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4015 methods, giving nearly-ANSI behavior
4016 * converted almost all special variables to use *FOO* naming convention
4017 * deleted PARSE-TIME functionality, since it can be done portably
4018 * moved some files out of cold init into warm init
4019 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4021 * regularized formatting of source files
4022 * added an install.sh script
4023 * fixed ridiculous memory usage of cross-compiler by making
4024 compiler/alloc.lisp not try to do pooling unless it can hook
4025 itself into the GC of the cross-compilation host. Now the system
4026 builds nicely on my old laptop.
4027 * added :SB-ALLOC in target-features.lisp-expr
4028 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4029 was not implemented)
4030 * re-did condition handling and note reporting in the compiler. Notes
4031 are no longer handled by signalling conditions. Style warnings
4032 and warnings are handled more correctly and reported in such a way
4033 that it's easy to find one or the other in your output (so that you
4034 can e.g. figure out which of many problems caused COMPILE-FILE to
4036 * changed the severity of several compiler warnings from full WARNING
4037 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4038 compiler note reporting so that it doesn't use the condition system
4039 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4040 COMPILE-FILE command)
4041 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4042 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4043 no longer supported, since their implementation was deeply intertwingled
4044 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4045 * removed bogus "support" for compiler macros named (SETF FOO), and
4046 removed the compiler macro for SETF INFO (but only after making a fool
4047 of myself on the cmucl-imp mailing list by posting a bogus patch for
4048 DEFINE-COMPILER-MACRO..)
4049 * Compiled files containing forms which have side effects on the Lisp
4050 reader (such as DEFPACKAGE forms) are now handled more correctly.
4051 (Compiler queuing of top level lambdas has been suppressed by setting
4052 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4053 * deleted various currently-unused source files, e.g. gengc.lisp. They
4054 may be added back at some point e.g. when porting to other architectures,
4055 but until they are it's distracting to distribute them and to try to
4057 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4058 there were so many of them they're just distractions, and UNCROSS is
4059 known to be able to handle the current sources
4060 * moved PROFILE functionality into TRACE, so that it will be clear
4061 how the wrapping and unwrapping of functions when you profile them
4062 interacts with the wrapping and unwrapping of functions when you
4063 trace them. (Actually, the functionality isn't there yet, but at least
4064 the interface specification is there. Hopefully, the functionality will
4065 arrive with some future maintenance release.)
4066 * removed host-oops.lisp
4067 * changed signature of QUIT function to allow UNIX-CODE argument
4068 * fixed READ-SEQUENCE bug
4069 * tweaked verbose GC output so that it looks more like the progress
4070 output that ANSI specifies for functions like LOAD
4071 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4072 * added <http://sbcl.sourceforge.com> to the banner information printed by