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)
13 changes in sbcl-1.0 relative to sbcl-0.9.18:
14 * improvement: experimental support for threading on FreeBSD/x86.
15 (thanks to NIIMI Satoshi)
16 * improvement: runtime option --dynamic-space-size can be used
17 to set the size of the dynamic space reserved on startup.
18 * improvement: floating point modes in effect are now saved in
19 core, and restored on startup.
20 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
21 startup, not time since first call to GET-INTERNAL-REAL-TIME.
22 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
23 threads are not running after *SAVE-HOOKS* have run.
24 * improvement: writes to CLOS instance slots are type-checked in code
25 compiled with (SAFETY 3)
26 * improvement: floating-point exception handling on FreeBSD (thanks to
28 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
29 (thanks to Zach Beane)
30 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
32 * improvement: added support for the Shift-JIS external format.
33 (contributed by NIIMI Satoshi)
34 * improvement: callbacks are supported on Linux/PPC. (thanks to
36 * bug fix: compiler bug triggered by a (non-standard) VALUES
37 declaration in a LET* was fixed. (reported by Kaersten Poeck)
38 * bug fix: file compiler no longer confuses validated and already
39 dumped structurres. (reported by Kaersten Poeck)
40 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
41 fill-pointer signals a type-error as required. (thanks to
43 * bug fix: disassemly of funcallable instances works.
44 * bug fix: single stepping on PPC.
45 * bug fix: fix thread-safety problems in the type system (generally
46 manifesting as nonsensical errors like "STRING is a bad type specifier
47 for sequences" or "The value 1 is not of type FIXNUM")
48 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
49 (reported by Josip Gracin)
50 * bug fix: an error is signaled for attempts to displace arrays with
51 incompatible element types (thanks to Mario Mommer)
52 * bug fix: more correct handling of wide characters in debug info
53 (bug reported by Attila Lendvai and fixed by Juho Snellman)
54 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
55 and don't cause extra consing
56 * optimization: MAP and MAP-INTO are significantly faster on vectors
57 whose elements types have been declared.
58 * Improvements to SB-SPROF:
59 ** Support for allocation profiling
60 ** Reduced profiling overhead, especially for long profiling runs
61 * Improvements to the Windows port:
62 ** floating point exceptions are now reported correctly.
63 ** stack exhaustion detection works partially.
64 ** more accurate GET-INTERNAL-REAL-TIME.
65 ** better RUN-PROGRAM behaviour in case of errors in the child
67 ** PROBE-FILE now simplifies pathnames correctly.
68 ** DIRECTORY now works correctly with :WILD-INFERIORS.
69 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
70 1st 1970 (although time timezone data will be unreliable).
71 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
73 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
74 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
75 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
77 * bug fix: two potential GC deadlocks affecting threaded builds.
78 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
79 returns T (reported by Anton Kazennikov)
80 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
81 with non-variable places
82 * bug fix: remove a race condition in the setting of
83 funcallable-instance functions, this should make threaded CLOS
84 code more stable against memory faults.
85 * bug fix: corruption of specials when unbinding is interrupted by an
86 asynchronous unwind (reported by Hannu Koivisto)
87 * improvement: the debugger will now also display local variables that
88 are only used once, for code compiled with a DEBUG optimization quality
91 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
92 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
93 * incompatible change: External-format support for FFI calls. The
94 SB-ALIEN:C-STRING no longer implies an ASCII
95 external-format. Instead the string is subject to external-format
96 conversion. Additionally return values of type SB-ALIEN:C-STRING
97 are by default of type (SIMPLE-ARRAY CHARACTER), not
98 SIMPLE-BASE-STRING. For an alien type that behaves like the old
99 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
100 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
101 * incompatible change: SB-EXT package no longer contains the
102 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
103 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
105 * incompatible change: the single-stepper is no longer available
106 on Alpha, Mips and Sparc platforms.
107 * minor incompatible change: the direct superclasses of
108 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
109 not (STANDARD-OBJECT FUNCTION). This makes the
110 class-precedence-lists of GENERIC-FUNCTION and
111 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
113 * minor incompatible change: the default stream external format on
114 non-unicode SBCL is no longer detected from the locale, but is always
116 * new feature: Add a version of evaluator that uses an interpreter instead
117 of the compiler. EVAL still uses the compiler by default, to switch it
118 to use the interpreter, set the value of the variable
119 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
120 * minor incompatible change: the single-stepper REPL has been merged
121 with the normal debugger (see the "Stepping" heading of the debugger help
122 for more details). The debugger command STEP will no longer switch
123 to the single-stepper REPL.
124 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
125 for a type now works.
126 * bug fix: loading of default sysinit file works. (thanks to Leonid
128 * bug fix: better detection of circularities in the file-compiler.
129 (reported by Marco Monteiro)
130 * bug fix: the CL pathname functions now work with files that have
131 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
132 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
133 systems with Unix98 pty semantics.
134 * bug fix: ASDF-INSTALL will now work with bsd tar.
135 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
137 * bug fix: timers expiring in dead threads no longer cause a
138 type-error (reported by Paul "Nonny Mouse").
139 * bug fix: thanks to more lightweight single-stepper instrumentation,
140 code compiled with (DEBUG 3) will compile and execute significantly faster,
141 and will have more accurate type-inferencing than before
142 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
143 whose bindings are modified
144 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
145 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
146 ** replace dummy implementations of CL:MACHINE-INSTANCE and
147 CL:SOFTWARE-VERSION with proper version
149 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
150 * feature: implemented the READER-METHOD-CLASS and
151 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
152 as specified by AMOP.
153 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
155 * optimization: faster LOGCOUNT implementation on x86 and x86-64
156 (thanks to Lutz Euler)
157 * optimization: hashing of general arrays and vectors has been
158 improved. (reported by Any Fingerhut)
159 * enhancement: SB-INTROSPECT is now able to find definitions of
160 profiled functions. (thanks to Troels Henriksen)
161 * enhancement: compiler-macro expansion applies now to FUNCALL forms
162 as well, allowing compiler-macros for SETF-functions to expand.
163 * enhancement: step-instrumentation no longer wraps known
164 single-value functions with multiple-value context, allowing
165 better type inference.
166 * fixed bug #337: use of MAKE-METHOD in method combination now works
167 even in the presence of user-defined method classes. (reported by
168 Bruno Haible and Pascal Costanza)
169 * fixed bug #339(c): if there are applicable methods not part of any
170 long-form method-combination group, call INVALID-METHOD-ERROR.
171 (reported by Bruno Haible)
172 * fixed bug #361: the :FUNCTION initarg in the protocol for
173 initialization of methods can now be used to override
174 internally-produced optimized functions. (reported by Bruno
176 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
177 system-provided lambda expression no longer cause warnings about
178 unbound #:|pv-table| symbols.
179 * bug fix: improved the handling of type declarations and the
180 detection of violations for keyword arguments with non-constant
182 * bug fix: potentially erronous calls to PATHNAME and
183 MERGE-PATHNAMES were being flushed in some cases.
184 (reported by Richard Kreuter)
185 * bug fix: compiled calls to TYPEP were mishandling obsolete
186 instances. (reported by James Bielman and Attila Lendvai)
187 * bug fix: format strings with ~> without matching ~< no longer
188 trigger an AVER, but signal an understandable error instead.
189 (reported by Antonio Martinez)
190 * bug fix: specifying an output-file without a pathname-type for
191 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
192 of input-file instead of "fasl". (reported by Robert Dodier)
193 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
194 (reported by James Y Knight).
195 * bug fix: compiler-macros-function did not consider the environment
196 argument for shadowing by local functions.
197 * bug fix: compiler-macros expansion was inhibited by local INLINE
199 * bug fix: inline expansions of known functions were subject to
200 step-instrumentation in high debug policies, leading to problems
202 * bug fix: compiler failed to differentiate between different CONS
204 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
205 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
206 * bug fix: #368: incorrect use of expressed vs. upgraded array
208 * bug fix: #306a: more precise unions of array types.
209 * thread-safety improvements:
210 ** CONDITION-WAIT could return early on Linux, if the thread was
211 interrupted and subsequently continued with SIGCONT.
212 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
214 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
215 * added support for the ucs-2 external format. (contributed by Ivan
217 * minor incompatible change: pretty printing of objects of type
218 (cons symbol) is, in the default pprint-dispatch-table, now
219 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
221 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
222 called later in a class's lifetime, possibly as late as when the
223 first instance of the class is created. Previously,
224 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
225 class became finalizeable.
226 * fixed bug: FILE-POSITION sometimes returned inconsistent results
227 for multibyte external-format streams. (thanks to "vbzoli")
228 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
229 with :ALLOCATION :CLASS inherited from superclasses of the
231 * fixed bug: anonymous classes can now be created using the :NAME
232 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
233 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
234 * fixed bug: core-files saved with :EXECUTABLE T can again be
235 executed when SBCL_HOME isn't set. (reported by James Knight)
236 * fixed bug: toplevel LOCALLY forms with declarations could
237 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
238 * fixed bug: printing from several different threads using different
239 values of *print-case* could cause invalid output, due to
240 some internal special variables of the printer not being bound
241 thread-locally (reported by Max Mikhanosha)
242 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
243 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
244 (reported by Pascal Costanza)
245 * fixed bug: FINALIZE-INHERITANCE is called from
246 REINITIALIZE-INSTANCE on classes when the class has previously
247 been finalized, as required by AMOP.
248 * minor code generation optimizations:
249 ** better register allocation in CLOS dispatching functions
250 ** overflow detection when coercing signed bytes to fixnums on x86-64
251 ** is now implemented with one IMUL instruction instead of three shifts
252 ** more efficient bit-vector access on x86 and x86-64
253 ** more efficient access to raw structure slots on x86 and x86-64
254 * fixed some bugs revealed by Paul Dietz' test suite:
255 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
258 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
259 * feature: thread support on Solaris/x86, and experimental thread support
261 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
263 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
264 default method for SLOT-UNBOUND.
265 * minor incompatible change: prevent the user from specializing the
266 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
267 somewhat counter to the intent of the protocol, I (CSR) think, and
268 additionally it just doesn't work in SBCL as currently
269 implemented, thanks to optimizations (that are always valid for
270 the other three SLOT-VALUEish functions, but not for the setter).
271 * minor incompatible change: the :SB-LDB feature is now enabled by
272 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
273 the low-level debugger.
274 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
275 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
276 down to an SB-SYS:FD-STREAM, allowing direct communication in
277 more cases. Temporary files and pipes are still used as a fallback
279 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
280 on the non-reentrant C functions localtime(3) and gmtime(3).
281 * bug fix: global optimization policy was not visible in LOCALLY and
283 * bug fix: class objects can be used as specializers in methods.
284 (reported by Pascal Costanza)
285 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
286 with a type error. (reported by blitz_ on #lisp)
287 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
288 failed with a type error. (reported by Pascal Bourguignon)
289 * bug fix: merging pathnames against defaults with :DIRECTORY
290 starting with '(:RELATIVE :BACK) should preserve the :BACK.
291 (reported by James Y Knight)
292 * bug fix: saving large (>2GB) cores on x86-64 now works
293 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
294 constant index and a (simple-array (signed-byte 32)) array
295 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
296 error (patch by Robert J. Macomber)
297 * feature: TIME now displays the amount of run-time spent in GC
298 * bug fix: The debugger now does a better job of respecting
299 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
300 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
301 workaround for bug 403.)
302 * bug fix: floating point exception handling now works on FreeBSD
303 (thanks to NIIMI Satoshi)
304 * fixed some bugs revealed by Paul Dietz' test suite:
305 ** MISC.641: LET-conversion were not supposed to work in late
307 * improvements to the Win32/x86 port:
308 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
309 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
311 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
312 * new feature: source path information is generated for macro-expansion
313 errors for use in IDE's like Slime (thanks to Helmut Eller)
314 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
315 * bug fix: compiler does not loop forever on an invalid type in
317 * improvement: compilation of most CLOS applications is significantly
319 * optimization: added a limited bytecode compiler for simple toplevel
320 forms, speeding up compilation and FASL loading
321 * bug fix: the statistical profiler now properly distinguishes anonymous
324 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
325 * minor incompatible change: in sbcl-0.9.11 (but not earlier
326 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
327 into subclasses of the finalized class. Now user calls to
328 FINALIZE-INHERITANCE finalize just the one class, and calls by the
329 system return before any subclasses are finalized.
330 * minor incompatible change: The reader no longer ignores errors
331 regarding non-existent packages in #+ and #- feature tests.
332 * new feature: command line options --no-sysinit, --no-userinit to
333 inhibit loading the corresponding init files
334 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
335 for sending data through UDP sockets (thanks to François-René Rideau)
336 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
337 error is signalled instead (thanks to François-René Rideau)
338 * minor incompatible change: Error signalling behaviour of lexical
339 operations violating package locks has changed slightly. Refer to
340 documentation on package locks for details.
341 * enhancement: EVAL can process IF-expressions without resorting to the
343 * bug fix: Some operations involving SIN, COS, or TAN weren't being
344 constant-folded properly due to missing out-of-line functions.
345 (reported by Mika Pihlajamäki)
346 * bug fix: LISTEN sometimes returned T even in cases where no data was
347 immediately available from the stream
348 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
349 were reversed. (reported by Levente Mészáros)
350 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
351 when given a vector displaced to an adjustable array. (reported
353 * bug fix: derivation of float boundaries from numbers outside the
354 appropriate float range (reported by John Wiseman)
355 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
357 * bug fix: precomputing cache entries for generic functions with
358 some subclasses of specializers as yet invalid does not attempt to
359 fill a cache line with a negative offset. (reported by Levente
361 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
362 allow condition class objects as arguments to DOCUMENTATION and
363 (SETF DOCUMENTATION); only find and set documentation for
364 structure names for the STRUCTURE doc-type. (suggested by Gary
366 * improvements to the Win32/x86 port:
367 ** (user-homedir-pathname) and default initialization file
368 locations now know about the user's "Documents and Settings"
369 directory (thanks to Yaroslav Kavenchuk)
370 ** run-program is implemented (thanks to Mike Thomas)
371 ** sockets support (thanks to Timothy Ritchey)
372 ** better backtrace support (thanks to Alastair Bridgewater)
373 ** sb-grovel supported
374 ** asdf-install and sb-posix work somewhat
375 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
376 * improvements to the Solaris/x86 port:
377 ** works on Solaris 11/Solaris Express
378 ** floating-point exception handling support
379 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
380 * fixed some bugs revealed by Paul Dietz' test suite:
381 ** REMOVE-METHOD returns its generic function argument even when
382 the method is not one of the generic functions' methods.
383 ** objects declared as MEMBER types can be admissible arguments to
385 ** printing characters should simply be printed by the FORMAT ~:C
387 ** compiler failure when compiling functions with hairy constant
388 defaults for optional parameters.
389 ** compiler produces wrong code when MAYBE-INLINE-expanding a
390 function, which is already optimized.
392 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
393 * new platform: experimental support for SBCL x86/Darwin, including
394 MacOS X 10.4.5 on Intel.
395 * new feature: Unicode character names are now known to the system
396 (through CHAR-NAME and NAME-CHAR).
397 * new feature: the filesystem location of SBCL's core file is
398 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
399 * minor incompatible change: the contrib modules SB-POSIX and
400 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
401 this change is to make it easier to distribute
402 location-independent binaries.
403 * bug fix: as implied by AMOP, standardized classes no longer have
404 slots named by external symbols of public packages. (reported by
406 * optimization: calling functions via constant symbols -- as in
407 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
408 via the function object as in (FUNCALL #'FOO).
409 * optimization: on x86 and x86-64, the calling convention for the
410 case of dealing with an unknown number of values has been altered
411 to be friendlier to the prediction heuristics implemented,
412 particularly with reference to CALL and RET pairing. (thanks to
413 Alastair Bridgewater)
414 * optimization: on x86, the code for access to thread-local symbol
415 values and binding/unbinding of thread-local symbols is smaller.
416 (thanks to Alastair Bridgewater)
417 * enhancement: CONSTANTP is now able to determine constantness of
418 more complex forms, including calls to constant-foldable standardized
419 functions and some special forms beyond QUOTE.
420 * fixed bug: occasional GC crashes on Solaris/x86
421 * optimization: x86-64 supports stack allocation of results of simple
422 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
423 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
424 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
425 (thanks to James Knight)
426 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
427 by the standard (thanks to Stephen Compall)
429 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
430 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
431 be used for bundling the runtime and the core file into one
432 executable binary. This feature is not currently supported on all SBCL
433 platforms. (thanks to James Bielman and NIIMI Satoshi)
434 * new feature: a generational or ephemeral garbage collector is now
435 the default on the PowerPC platform (both Linux and Darwin). The
436 old Cheney (stop and copy) collector is a build-time option.
437 (thanks to Cyrus Harmon, after Raymond Toy)
438 * minor incompatible change: the method by which SBCL finds its
439 contributed modules has changed; it no longer relies on symbolic
440 links from an $SBCL_HOME/systems directory, but searches directly
441 in the subdirectories of $SBCL_HOME.
442 * enhancement: the dynamic heap size on the Linux/PPC platform is
443 markedly larger, even using the older Cheney garbage collector.
444 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
445 certain complicated string types. (reported by Gary King)
446 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
447 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
448 * fixed bug: DOCUMENTATION on structures no longer signals an
449 error if no structure type of the right name exists. (reported by
451 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
452 applicable non-standard primary or :AROUND method on
453 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
454 no longer cause unbound variable errors. (reported by Kevin Reid)
455 * fixed bug: metacircle resolution in cases where methods have slots
456 added before the slots from STANDARD-METHOD. (reported by Jean
458 * fixed bug: the Power PC assembly code for calling into lisp
459 sometimes computed its return address wrongly (depending on the
460 15th bit of the address, fixed at link time). This used to
461 manifest itself in a segmentation violation while building PCL.
462 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
463 many others over the years)
464 * fixed bug: ROOM no longer reports silly numbers for stack usage if
465 the stack pointer (treated as a signed integer) is negative.
466 (thanks to Peter van Eynde)
468 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
469 * new platform: experimental support for the Windows operating
470 system has been added. (thanks to Alastair Bridgewater)
471 * fixed several bugs in and robustified the PPC FFI (including
472 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
473 * bug fix: allow non-simple string symbol names (reported by Paul
475 * bug fix: interrupt handling on NetBSD (thanks to Richard M
477 * bug fix: saving a core corrupted callbacks on x86/x86-64
478 * bug fix: closed a loophole in metacircularity detection and
479 grounding in the PCL implementation of CLOS.
480 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
481 (reported by tomppa on #lisp)
482 * bug fix: clear the direction flag on Lisp -> C transitions, as
483 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
484 (reported by Andrej Grozin and Hendrik Maryns)
485 * optimization: major improvements to GC efficiency on GENCGC platforms
486 * optimization: faster implementation of EQUAL
487 * optimization: emit more efficient opcodes for some common
488 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
489 * optimization: several other minor code-generation improvements on x86-64
490 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
492 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
493 * minor incompatible change: (SETF CLASS-NAME) and (SETF
494 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
495 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
496 by ANSI as a generic function, and for consistency (SETF
497 GENERIC-FUNCTION-NAME) is treated likewise.
498 * fixed bug #233.b: make constraint propagation notice when a variable
499 value is changed after it is referenced but before it is used
500 * fixed bug #296: no more arbitrary behaviour from filenames with
501 odd characters as --load arguments. (reported by Adam Warner)
502 * fixed bug #390: :CHARACTER-SET pathname components now work as
503 expected. (reported by Tim Daly Jr)
504 * fixed bug #391: complicated :TYPE intersections in slot
505 definitions no longer cause an error in PCL internals.
506 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
507 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
508 returning the number of octets which would be written to the
509 file-stream. (thanks to Robert J. Macomber)
510 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
511 arguments to the internal stream functions. (thanks to David
513 * bug fix: stack exhaustion is now handled on the x86/SunOS
514 platform. (thanks to Daisuke Homma)
515 * bug fix: more accurate ROOM results on GENCGC platforms
516 * optimization: improved type inference for arithmetic-for
517 index variables in LOOP
518 * optimization: faster floating-point SQRT on x86-64
519 * fixed some bugs revealed by Paul Dietz' test suite:
520 ** DOCUMENTATION returns NIL instead of "" for method combinations
521 that don't have a docstring
523 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
524 * minor incompatible change: (SETF CLASS-NAME) and (SETF
525 GENERIC-FUNCTION-NAME) are no longer generic functions, and
526 therefore are not customizeable by user code (as seems to be at
527 least permitted and maybe required by AMOP). As a consolation,
528 however, the SBCL implementation of these functions now calls
529 REINITIALIZE-INSTANCE as specified by AMOP.
530 * bug fix: slot-definition documentation is propagated as per ANSI
531 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
532 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
533 argument, as do eslotd initialization methods. (from Pascal
534 Costanza's "Closer" project)
535 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
536 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
538 * bug fix: it is now possible to have more than one subclass of
539 STANDARD-GENERIC-FUNCTION without causing stack overflow.
540 (reported by Bruno Haible, Pascal Costanza and others)
541 * bug fix: the dependent update protocol now works for generic
542 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
544 * bug fix: condition-class instances corresponding to
545 DEFINE-CONDITION forms are now created eagerly. (reported by
546 Kalle Olavi Niemitalo on comp.lang.lisp)
547 * bug fix: floating point printing is more accurate in some
548 circumstances. (thanks to Simon Alexander)
549 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
550 merged with *DEFAULT-PATHNAME-DEFAULTS*.
551 * bug fix: callbacks on OS X now preserve stack-alignment.
552 (thanks to Cyrus Harmon)
553 * enhancement: the x86-64 disassembler is much better at
554 disassembling SSE instructions. (thanks to Lutz Euler)
555 * enhancement: CHECK-TYPE on a variable now allows type-inference
556 (thanks to Vincent Arkesteijn)
557 * optimization: improved performance of EUC-JP external format.
558 (thanks to NIIMI Satoshi)
559 * optimization: performance improvements to IO on file streams of
560 :ELEMENT-TYPE CHARACTER
561 * optimization: much faster memory allocation on x86-64
562 * optimization: faster garbage collections (latency and throughput)
564 * optimization: faster bignum subtraction and division on x86 and x86-64
565 (thanks to Lutz Euler)
566 * optimization: more accurate type-inference for loops that use a
567 floating point index variable or a negative step.
569 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
570 * bug fix: add a workaround to SBCL looping infinitely at startup on
571 Linux kernels with apparently buggy implementations of personality().
572 (thanks to Svein Ove Aas)
573 * bug fix: Unicode symbols are correctly printed in LDB backtraces
574 (thanks to David Lichteblau)
575 * bug fix: local bindings shadow global symbol macros for
576 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
577 * optimization: non-open coded uses of numeric comparison operators
578 (e.g. >) no longer cons when called with more than one parameter
579 on platforms supporting dynamic-extent allocation.
580 * enhancement: saving cores with foreign code loaded is now supported
581 on MIPS/Linux in addition to the previously supported platforms.
582 * bug fix: FIND-RESTART now tests for activity, not applicability when given
583 a restart object as identifier. (reported by Helmut Eller for CMUCL)
584 * bug fix: division by zero in sb-sprof when no samples were collected
585 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
586 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
587 (thanks to Alexander Kjeldaas)
588 * new feature: ignoring the return values of destructive functions
589 like NREVERSE signals a compile-time style-warning.
590 (thanks to Kevin Reid)
591 * new feature: the alignment of alien structure fields can be
592 explicitly specified. (thanks to Cyrus Harmon)
593 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
594 and dump core on SIGQUIT
596 ** incompatible change: threads do not inherit values of specials
597 from their parents (see manual)
598 ** bug fix: threads stacks belonging to dead threads are freed by the
599 next exiting thread, no need to gc to collect thread stacks anymore
600 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
601 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
602 ** bug fix: dynamic variable and thread start related gc lossage
603 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
605 * fixed some bugs revealed by Paul Dietz' test suite:
606 ** SUBTYPEP is slightly more accurate on heinously complicated
607 CONS types where some of the members have uncertain (in the
608 NIL, NIL sense) type relationships to each other.
610 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
611 no longer purified unless :PURIFY T is explicitly specified.
612 ** Non-purified cores are significantly smaller than before
614 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
615 * new feature: timers based on Zach Beane's excellent timer package
616 * added support for the following external formats: koi8-u,
617 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
618 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
619 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
620 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
621 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
622 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
623 * incompatible change: a threaded SBCL will no longer revert to
624 non-threaded mode on non-NPTL systems, but refuse to start entirely.
625 * bug fix: interrupts are disabled until startup is complete; no
626 more sigsegvs when receiving a signal to soon
627 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
629 * bug fix: add a workaround for the memory randomization features in
630 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
631 workaround will only be in effect on systems with the proc filesystem
633 * bug fix: printing objects of type HASH-TABLE signals a
634 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
636 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
637 have been running for over 50 days. (reported by Gilbert Baumann)
638 * bug fix: the logic for getting names of functions gets less
639 confused when confronded with alternate-metaclass
640 funcallable-instances. (reported by Cyrus Harmon)
641 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
642 consistent, even on internal alternate-metaclass objects.
643 * bug fix: flush closure information collected by physical
644 environment analysis prepass before the main pass. (bug reported
646 * bug fix: compiler pack phase does not modify a hashtable, which is
647 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
648 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
649 may be expanded. (reported by Alexander <malishev> on c.l.l)
650 * bug fix: no more style-warnings when compiling code using LAST
651 under high-SPEED low-SPACE settings. (reported by David Wallin
653 * fixed some incompatibilities between SBCL's MOP and the MOP
655 ** the METAOBJECT class is now implemented;
656 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
657 STANDARD-OBJECT, as required;
658 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
659 now compatible (as required by VALIDATE-SUPERCLASS); there
660 remains a consistency requirement over the presence or absence
661 of the FUNCTION class in the superclass list of finalized
662 classes; see the manual for more details;
663 ** the :AROUND method for COMPUTE-SLOTS on
664 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
665 requested slot ordering.
667 ** bug fix: parent thread now can be gc'ed even with a live
669 ** bug fix: binding a special with PROGV to no value is not
670 the same as that symbol not having been bound (thanks to
672 ** bug fix: binding specials is thread safe (thanks to
674 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
675 and SB-SPROF do not die with 'no handler for signal XX in
676 interrupt_handle_now(..)' anymore
677 ** bug fix: WITH-TIMEOUT works with multiple running threads
678 * fixed some bugs revealed by Paul Dietz' test suite:
679 ** ENSURE-GENERIC-FUNCTION should take a method class object for
680 the :method-class keyword argument.
682 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
683 * new port: the Solaris operating system on x86 processors is now
684 mostly supported, though some rough edges in the environment
685 remain. (thanks to Daisuke Homma)
686 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
687 space for its heap. (thanks to Thiemo Seufer)
688 * optimization: SBCL on MIPS platforms now supports dynamic-extent
689 closures. (thanks to Thiemo Seufer)
690 * minor incompatible change: eof selects abort in the debugger.
691 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
692 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
694 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
695 deprecated, and will go away in a future revision of SBCL.
696 * minor incompatible change: GC-ON and GC-OFF are no longer
697 implemented with a counter, it does not matter how many times gc
698 is switched on or off
699 * bug fix: discriminating functions for generic function classes
700 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
701 make invalid assumptions about method precedence order. (reported
703 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
704 platforms. (thanks to Thiemo Seufer)
705 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
706 * bug fix: correct pathname printing with printer escaping is on.
707 (thanks to Kevin Reid)
708 * bug fix: complex VOP definitions in "user-space" no longer trigger
709 package locks. (reported by Zach Beane)
710 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
711 for funcallable-instance functions, and is no different from
713 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
714 characters in the current readtable. (reported by Nicholas Neuss)
715 * bug fix: SUBTYPEP on various CONS types returns more a more
716 accurate acknowledgment of its certainty.
717 * optimizations: REMOVE-DUPLICATES now runs in linear time on
718 lists in some cases. This partially fixes bug 384.
719 * flush all standard streams before prompting in the REPL and the
721 * bug fix: signal handling and triggering gc do not conflict
722 directly anymore, in particular a high frequency sb-sprof does
723 not prevent gc from running
724 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
725 approximation for timezone and DST information between the
726 universal time epoch and the smallest negative 32-bit time_t.
727 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
728 year 1899 so long as the encoded time is non-negative.
729 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
730 the correct answer for non-constant rotate amounts when performing
731 an inline 32-bit rotation.
733 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
734 there is only one thread in the session
735 ** bug fix: memory leak for streams created in one thread and
736 written to in another
737 ** bug fix: lockup when compiled with gcc4
738 ** bug fix: race that allows the gc to be triggered when gc is
740 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
741 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
743 ** bug fix: finished threads are gc'ed properly
744 * fixed some bugs revealed by Paul Dietz' test suite:
745 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
746 has arguments with a different set of applicable methods from
747 the orignal arguments.
748 ** The type error thrown by MAP now has a correct expected-type
750 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
751 name a compiled function.
752 ** Three MISC tests where a large bignum was improperly coerced to
753 a float that couldn't represent that bignum during type
754 derivation were fixed.
755 ** SUBTYPEP can now handle types involving the negation of a
756 list-form FUNCTION type.
757 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
758 as (COMPLEX (AND RATIO (NOT FIXNUM))).
759 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
761 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
762 * New feature: Experimental support for bivalent streams: streams
763 opened with :element-type :default now allow character and binary
764 (unsigned-byte 8) I/O
765 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
766 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
768 * bug fix: no more highly sporadic "couldn't check whether ~S is
769 readable" when reading a stream and an interrupt hits in the middle
770 of a select system call
771 * compiler better recognizes complex arrays (reported by Eduardo
773 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
774 for variables on SBCL built with linkage-tables. (reported by Luis
776 * various error reporting improvements.
777 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
778 (Thanks to Thiemo Seufer)
779 * bug fix: nested reader invokations work correctly
780 * bug fix: it is possible to have simultaneous references to foreign
781 code and foreign data with the same name.
783 ** added x86-64 support
784 ** incompatible change: the threading api now works with thread
785 objects instead of thread ids
786 ** bug fix: threads are protected from signals and interruption when
787 starting up or going down
788 ** bug fix: a race where an exiting thread could lose its stack to gc
789 ** bug fix: don't halt on infinite error in threads if possible
790 ** fixed numerous gc deadlocks introduced in the pthread merge
791 ** bug fix: fixed thread safety issues in read and print
792 ** bug fix: debugger doesn't hang on session lock if interrupted at
793 an inappropriate moment
794 ** bug fix: run-program is now thread safe(r)
795 ** bug fix: inner with-recursive-lock no longer releases the mutex
796 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
797 * fixed some bugs revealed by Paul Dietz' test suite:
798 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
799 EXPECTED-TYPE slots filled.
800 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
802 ** MAKE-CONDITION accepts classes as type-designators.
803 ** COMPILE may never return NIL.
804 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
805 range before calling Unix time functions
807 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
808 * numerous signal handling fixes to increase stability
809 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
810 * minor incompatible change: we now correctly canonize default
811 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
812 as per the MOP, rather than the historical (INITARG INITFUNCTION
813 INITFORM). (reported by Bruno Haible)
814 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
815 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
816 for more information.
817 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
818 pathname is a directory pathname.
819 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
820 * dynamic space size on PPC has been increased to 768Mb. (thanks to
822 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
823 :METACLASS argument in addition to a class name. (reported by
824 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
825 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
826 (reported by Bruno Haible; patch from Gerd Moellmann)
827 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
829 * bug fix: more cleanups to the floating point exception handling on
830 x86-64 (thanks to James Knight)
831 * bug fix: the compiler does not try to constant fold calls of
832 COERCE and BIT-* functions when they return freshly constructed
833 objects. (reported by Nikodemus Siivola and Paul Dietz)
834 * optimization: improved the MIPS versions of generic (in the
835 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
836 * optimization: direct conversion of (unsigned-byte 32) to floats on
837 the PowerPC platform.
838 * optimization: structure instances with raw slots now use less
839 memory, and probably show better memory locality. (thanks to
841 * optimization: DYNAMIC-EXTENT declarations for lists and closures
842 are treated as requests for stack allocation on the x86-64,
843 Alpha-32, PPC and SPARC platforms.
844 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
845 the wrong file descriptor; implementation of SOCKET-OPEN-P.
846 (thanks to Tony Martinez)
848 ** gcing a dead thread can no longer lead to lockups
849 ** threads block signals until they are set up properly
850 ** errno is no longer shared by threads
851 ** interrupt-thread restores the eflags register on x86
852 ** fixed some lockups due to gc/thread interaction
853 * fixed some bugs revealed by Paul Dietz' test suite:
854 ** invalid dotted lists no longer raise a read error when
856 ** signal an error if a symbol that names a declaration is used
857 as the name of a type, or vice versa
858 ** allow using the (declare (typespec var*)) abbreviation for
859 (declare (type typespec var*)) with all type specifiers
860 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
861 ** always return NIL from PPRINT-POP when OBJECT is NIL
862 ** don't signal errors when pretty-printing malformed LABELS,
863 FLET or MACROLET forms
864 ** declarations in a DOLIST body are also in scope for the
866 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
868 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
871 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
872 * fixed cross-compiler leakages that prevented building a 32-bit
873 target with a 64-bit host compiler.
874 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
875 opened with :IF-EXISTS :OVERWRITE.
876 * fixed bug 281: error for an invalid qualifier in a short-form method
877 combination method is not signalled until the faulty method is called.
878 * bug fix: iteration variable type inferrer failed to deal with open
879 intervals. (reported by Alan Shields)
880 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
881 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
882 or more runtime options were provided to the sbcl binary.
883 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
884 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
885 * optimization: there's now a fast-path for fixnum arguments in the
886 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
887 * optimization: the code generated on x86-64 is more compact thanks
888 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
889 * bug fix: floating-point exception handling now partly works on
890 x86-64. (Thanks to James Knight)
891 * improvement to the MIPS backend from Thiemo Seufer: C-style
892 64-bit long long arguments and return values to alien functions
894 * fixed some bugs revealed by Paul Dietz' test suite:
895 ** the type-error signalled from WARN has a filled-in DATUM slot.
896 ** the type-error required when a stream is not associated with
897 a file has the stream as its datum.
898 ** type-errors on single-floats on x86-64 no longer have
899 :INVALID-OBJECT as the datum
900 ** the type-errors signalled for invalid function names now have
901 a correct expected type
902 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
903 for structures defined with a :TYPE. Documentation strings for
904 typed structures are no longer immediately discarded
905 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
906 broadcast streams and synonym streams. FILE-LENGTH now also works
907 on broadcast streams.
909 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
910 * incompatible change: the --noprogrammer option, deprecated since
911 version 0.7.5, has been removed. Please use the equivalent
912 --disable-debugger option instead.
913 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
915 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
916 inoperational for a while now) has been completely removed.
917 * null lexical environments are now printed as #<NULL-LEXENV>,
918 significantly reducing the amount of clutter in typical backtraces.
919 * documentation on weak pointers, finalization, and after GC hooks
920 has been added to the manual.
921 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
922 as other specialized array types, is much faster. SUBSEQ and
923 COPY-SEQ on such arrays have also been sped up.
924 * optimization: EQL is now more efficient when at least other argument
925 is known to be of type (OR FIXNUM (NOT NUMBER)).
926 * fixed bug: the runtime is now less vulnerable to changes in the
927 size of the SBCL object on OS X, and virtual memory is reserved for
928 all spaces that need to be at a fixed address.
929 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
930 * fixed bug: finalizers and after GC hooks that cause consing are now
932 * fixed bug: compiler error messages and summaries are now printed to
933 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
934 * fixed inference of the upper bound of an iteration variable.
935 (reported by Rajat Datta).
936 * fixed bug 211e: calling local functions with duplicated constant
937 keyword argument no longer causes a bogus style warning about an
939 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
940 ftype declaration to be disregarded. (reported by Dave Roberts)
941 * fixed bug 373: caused by erronous compilation of references to alien
942 variables in the runtime on ppc/darwin.
943 * fixed bug 376: CONJUGATE type deriver.
944 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
945 (reported by Baughn on #lisp)
946 * a cross-compiler bug on non-x86 platforms has been identified and
947 fixed. (thanks to Bruno Haible)
948 * improvements to the MIPS runtime code for increased stability.
949 (thanks to Thiemo Seufer)
950 * increased the maximimum compact environment size to allow
951 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
952 * partial workaround for bug 135: don't name the function we're
953 calling for hairy cases of EVAL, so as not to accumulate one environment
954 entry per call to EVAL. (reported by Kevin Reid)
955 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
956 * optimization: type testing for non-vector arrays should be faster.
957 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
959 * bug fix: setting 31st element of a bit vector to zero did not work
961 * bug fix: redefining a class definition which failed due to a
962 previous accessor / function clash now works (but see BUGS entry
963 #380 for more problems in this area). (thanks to Zach Beane)
964 * the long form of DEFINE-METHOD-COMBINATION disables method group
965 checking when given a single method group with pattern *. (thanks
967 * on x86 compiler supports stack allocation of results of simple
968 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
969 * support for building and running on Mac OS X 10.4 "Tiger" has been added
970 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
971 * fixed some bugs related to Unicode integration:
972 ** the restarts for recovering from input and output encoding
973 errors only appear when there is in fact such an error to
975 * fixed some bugs revealed by Paul Dietz' test suite:
976 ** MISC.549 and similar: late transformation of unsafe type
977 assertions into derived types caused unexpected code
979 ** SCALE-FLOAT type deriver is less wrong.
980 ** type derivers for EXP, LOG and similar functions compute result
981 types for complex arguments better.
982 ** (MISC.563) CONJUGATE type deriver works for very restricted
984 ** out-of-line type testers for character strings are available.
985 ** EQUAL compiler transform understands specialness of objects
987 ** accessing double-floats stored on the stack now works on x86-64.
988 ** debugger internals could sometimes create invalid lispobjs,
989 resulting in GC crashes.
990 ** MISC.548: type check weakening can convert required type into
992 ** initialization forms for bindings are not in scope of free special
995 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
996 * incompatible change: thread support for non-NPTL systems has
997 been removed - locking is buggy and unreliable. A threaded
998 SBCL build will now warn at startup and refuse to create
999 new threads, unless futex support is detected in the kernel
1000 * incompatible change: the top level REPL now has only an ABORT
1001 restart associated with it, not TOPLEVEL and ABORT as it used to.
1002 TOP and TOPLEVEL are now available as debugger commands for
1003 returning to the top level.
1004 * incompatible change: forms evaluated in the REPL now use the
1005 global optimization policy.
1006 * incompatible change: user- and system-initialization files are
1007 no longer processed with LOAD, but by READ and EVAL; hence the
1008 global optimization policy, startup package, readtable, etc,
1010 * The .fasl file format number has been incremented because of
1011 various incompatible changes.
1012 * internal entry point details and argument counts no longer appear
1013 in backtraces unless explicitly requested by setting
1014 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1015 * built-in and standard functions no longer have names like "top
1016 level local call to FOO".
1017 * fixed bug 32: functions defined in non-null lexical environments
1018 now have more legible printed representation
1019 * fixed bug 33: functions defined in non-null lexical environemnts
1020 are now more amenable to inspection by INSPECT.
1021 * workaround for bug 354: XEPs no longer appear in backtraces unless
1022 explicitly requested.
1023 * fixed bug: receiving the signal which results from attempting to
1024 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1025 SIGBUS on other *BSDs) on architectures where the C stack is also
1026 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1027 to debugger confusion or wild instability. (reported by Cheuksan
1029 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1030 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1031 (reported by Lutz Euler)
1032 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1033 compiled code. (reported by Paul Dietz)
1034 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1035 specialized parameter to a method to something that is not TYPEP
1036 the specializer is now possible.
1037 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1038 face of package deletion.
1039 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1040 pathname. (reported by David Tolpin on c.l.l)
1041 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1042 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1043 STRING-STREAM and FILE-STREAM, respectively.
1044 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1045 than previously; in addition, its test suite is now run on build.
1046 (thanks to Luke Gorrie)
1047 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1048 * optimization: added a immediate representation for single-floats
1050 * fixed some bugs revealed by Paul Dietz' test suite:
1051 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1052 ** attempting to create a package with a colliding nickname causes
1053 correctable errors to be signalled.
1054 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1055 ** several x86-64 backend bugs related to sign-extension of immediate
1058 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1059 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1060 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1062 * fixed loading of multiply forward-referenced layouts.
1063 (thanks to Cheuksan Wang)
1064 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1065 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1066 (see COMPILE-FILE documentation for details of :PRINT option)
1067 * fixed bugs 19 and 317: fixed-format floating point printing is
1068 more accurate. This also fixes a bug reported by Adam Warner
1069 related to the ~@F format directive.
1070 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1072 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1073 dispatch table if the from-char is a dispatch macro character.
1074 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1075 by the wordsize no longer ignore the last word. (reported by Lutz
1077 * fixed bug in type checking of dynamic-extent variables. (reported
1079 * optimization: sequence traversal functions use their freedom to
1080 coerce function designators to functions.
1081 * optimization: code with many calls to CLOS methods specialized on
1082 CLOS classes has had redundant type checks removed.
1083 * optimization: added declarations to speed up operations that access
1084 the internal character database (for example STRING-UPCASE)
1085 * optimization: comparison operations between floats and sufficiently small
1086 fixnums no longer create extra rationals
1087 * fixed some bugs related to Unicode integration:
1088 ** portions of multibyte characters at the end of buffers for
1089 character-based file input are correctly transferred to the
1090 start of the buffer at the next read.
1091 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1092 passing it through to OPEN.
1093 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1094 argument given to internal calls to OPEN.
1095 ** the built-in comment readers (introduced by character sequences
1096 ";" and "#|") are more forgiving to encoding errors; they will
1097 STYLE-WARN and then attempt to resync the stream at a character
1098 boundary. (thanks to Teemu Kalvas)
1099 * fixed some bugs in the x86-64 port:
1100 ** Negative short int return values from c-calls are sign-extended
1102 ** The stack is aligned to 16-bytes for c-calls, as required by
1103 the ABI. (thanks to Cheuksan Wang)
1104 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1105 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1106 for enums. (thanks to Vincent Arkesteijn)
1107 ** Multiple small optimizations and bugfixes for floating point
1109 * fixed some bugs revealed by Paul Dietz' test suite:
1110 ** Space, Tab, Linefeed, Return and Page have the invalid
1111 secondary constituent character trait.
1112 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1114 ** WITH-INPUT-FROM-STRING should only update the index place on
1116 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1117 works more reliably.
1118 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1119 with one-dimensional arrays with fill pointers.
1120 ** TYPE-OF failed on a complex with an integer realpart and a
1122 ** compiler failure during type inference for the code of form
1123 (IF (EQL X (THE ...)) ...) (MISC.535).
1125 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1126 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1127 platform. The port supports 61-bit fixnums, large memory spaces
1128 and reloading shared object files.
1129 * enhancement: saving cores with foreign code loaded is now
1130 supported on ppc/Darwin in addition to the previously supported
1132 * enhancement: the statistical profiler now walks deeper into the
1133 call stack for more meaningful call-graphs and accrued time
1134 reports (x86/x86-64 only). It also now reports time spent in
1136 * enhancement: it is now possible to trace most individual methods
1137 of a generic function in addition to tracing the generic function
1139 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1141 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1142 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1143 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1144 produces an error. (thanks to Vincent Arkesteijn)
1145 * bug fix: NAMESTRING on pathnames with :WILD components in their
1146 directories works correctly. (thanks to Artem V. Andreev)
1147 * fixed bug 125: compiler preserves identity of closures. (reported
1149 * bug fixed: functions with &REST arguments sometimes failed with
1150 "Undefined function" when compiled with (DEBUG 3). (reported by
1152 * bug fix: overflow during compiling of setting element of a bit
1153 vector with constant index and value. (reported by Timmy Douglas)
1154 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1155 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1156 * on x86 compiler partially supports stack allocation of dynamic-extent
1158 * GO and RETURN-FROM do not check the extent of their exit points
1159 when compiled with SAFETY 0.
1160 * fixed some bugs related to Unicode integration:
1161 ** encoding and decoding errors are now much more robustly
1162 handled; it should now be possible to recover even from invalid
1163 input or output to the terminal. (thanks to Teemu Kalvas)
1164 ** provided a first cut at implementing STRING-TO-OCTETS and
1165 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1166 ** altered the SB-MD5 contributed module slightly, changing the
1167 interface just enough for it to be supportable for builds where
1168 lisp characters are not eight bits.
1169 * fixed some bugs revealed by Paul Dietz' test suite:
1170 ** the FORMATTER-generated functions for ~V[ conditionals require
1171 the correct number of arguments.
1172 ** READ-FROM-STRING returns the mandated second value when applied
1173 to displaced strings.
1174 ** the #\Rubout and #\Backspace characters are treated as invalid
1175 constituent characters by the tokenizer.
1177 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1178 * new feature: reloading changed shared object files with
1179 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1180 * new feature: references to foreign variables and functions
1181 can now be compiled and loaded before the corresponding shared
1182 object file is loaded, as long as the foreign definitions are
1183 available at runtime.
1184 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1185 supported. (thanks to Dan Debertin)
1186 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1187 just keywords. (thanks to Vincent Arkesteijn)
1188 * fixed bug #331: structure-class instances corresponding to
1189 DEFSTRUCT forms are now created eagerly.
1190 * fixed bug #345: backtraces from calls to undefined functions work
1191 on x86 as well. Related bug #61 is now also partially fixed on x86
1192 (backtraces from throws to unknown catch tags.)
1193 * bug fix: lambda-list parsing is now stricter vrt. order and number
1194 of lambda-list keywords.
1195 * bug fix: as specified by AMOP, an error is signalled if a
1196 class-option appears multiple times in a DEFCLASS form. (reported
1198 * bug fix: RANDOM can be compiled when the compiler derives the type
1199 of its numeric argument as a disjoint set of small integers.
1200 (reported by Paul Dietz)
1201 * bug fix: starting a core saved with shared objects loaded when
1202 those objects are not available no longer causes threaded SBCL to
1203 hang. (reported by Sean Ross)
1204 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1206 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1207 argument, which must always be NIL. (reported by Kalle Niemitalo)
1208 * bug fix: printing 1.0d+23 no longer results in an error.
1209 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1210 * bug fix: structure slot setters preserve evaluation order. (thanks
1212 * fixed some bugs related to Unicode integration:
1213 ** RUN-PROGRAM can allow its child to take input from a Lisp
1214 stream. (reported by Stefan Scholl)
1215 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1216 locales. (reported by Ken Causey)
1217 * fixed some bugs revealed by Paul Dietz' test suite:
1218 ** INCF, DECF and REMF evaluate their place form as specified in
1220 ** FORMATTER expands ~{ ~} iteration directives with V or #
1221 parameters correctly.
1222 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1223 consequent uses no arguments correctly.
1224 ** the system has a partial understanding of the (COMPLEX RATIO)
1226 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1229 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1230 * new feature: a build-time option (controlled by the :SB-UNICODE
1231 keyword feature, enabled by default) for building the system with
1232 support for the entire 21-bit character space defined by the
1234 * new feature: the system now has rudimentary external-format
1235 support; the primary user-visible change at this time is that
1236 characters with the high bit set (such as lower-case-e-acute) will
1237 print correctly to a terminal in a UTF-8 environment.
1238 * minor incompatible change: BASE-CHAR no longer names a class;
1239 however, CHARACTER continues to do so, as required by ANSI.
1240 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1241 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1242 has been moved to the SB-EXT package (temporarily re-exported from
1244 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1245 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1247 * fixed bug #351: better error-handlind and reporting for malformed
1249 * fixed bug #350: bignum-printing is now more memory-efficient,
1250 allowing printing of very large bignums, eg. (expt 2 10000000).
1251 (reported by Bruno Haible)
1252 * fixed bug #302: better primitive-type selection for intersection
1254 * fixed bug #308: non-graphic characters now all have names, as
1255 required. (reported by Bruno Haible)
1256 * bug fix: redefining a class with different superclasses now correctly
1257 removes it from the direct-subclasses of its previous superclasses.
1258 (reported by David Morse)
1259 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1260 new value now works. (reported by Bruno Haible)
1261 * bug fix: correct canonicalization of multiple non-standard slot
1262 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1263 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1264 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1265 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1266 now exists, an signals an error.
1267 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1268 during macroexpansion for non-iteration variables as well. (reported
1269 by Bruno Haible for CMUCL)
1270 * bug fix: Cyclic structures and unprintable objects in compiler
1271 messages no longer cause errors. (reported by Bruno Haible)
1272 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1273 and READ-FROM-STRING all now return a primary value of NIL if
1274 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1275 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1276 (reported by Bruno Haible for CMUCL)
1277 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1278 do not designate a condition. (reported by Bruno Haible for
1280 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1281 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1282 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1283 permitting resolution in favour of any of the conflicting symbols.
1284 (reported by Bruno Haible for CMUCL)
1285 * bug fix: EQUAL compiler optimizations is less aggressive on
1286 strings which can potentially compare true despite having distinct
1287 specialized array element types.
1288 * bug fix: unit enumerations can be defined without dividing by
1289 zero. (thanks to Vincent Arkesteijn)
1290 * FORMAT compile-time argument count checking has been enhanced.
1291 (report from Bruno Haible for CMUCL)
1292 * a partial workaround for the bug 262: the compiler does not try to
1293 inline-expand a local function doing RETURN-FROM from a deleted
1294 BLOCK. (thanks to Peter Denno for the bug report and to David
1295 Wragg for the simple test case)
1296 * fixed some bugs revealed by Paul Dietz' test suite:
1297 ** RENAME-PACKAGE allows all package designators as new package
1299 ** constraint propagation and lambda variable substitution are
1300 more cautious in dealing with partially deleted code.
1301 ** compiler sometimes lost reoptimization passes.
1302 ** CERROR, given a condition as condition designator, passes its
1303 remaining arguments to the continue format control without
1305 ** Case-altering FORMAT directives work correctly on non-ASCII
1307 ** The REAL type specifier handles bounds outside the floating
1308 point ranges without signalling FLOATING-POINT-OVERFLOW.
1309 ** Functions with IR1-transformations can create intercomponent
1310 references to global functions.
1311 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1313 ** FORMAT ~R treats a nil value for its first parameter correctly.
1315 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1316 * enhancement: saving cores with foreign code loaded is now
1317 supported on x86/NetBSD and sparc/Linux in addition to the previously
1318 supported platforms.
1319 * bug fix: on some platforms repeated installations caused multiple
1320 copies of HTML documentation to be installed -- should not happen
1321 any more. (reported by Stefan Scholl)
1322 * bug fix: parsing self-recursive alien record types multiple times
1323 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1324 original patch by Helmut Eller for CMUCL)
1325 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1326 (thanks to Richard Kreuter)
1327 * bug fix: defining classes whose accessors are methods on existing
1328 generic functions in other (locked) packages no longer signals
1329 bogus package lock violations. (reported by François-René Rideau)
1330 * bug fix: special variables as DEFMETHOD parameters no longer have
1331 associated bogus type declarations. (reported by David Wragg and
1333 * bug fix: read-write consistency on streams of element-type
1334 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1335 * bug fix: redefiniton of the only method of a generic function with
1336 no DEFGENERIC no longer emits a full WARNING. In addition,
1337 redefinition of generic functions with no DEFGENERIC to an
1338 incompatible lambda list now signals an error. (thanks to Zach
1340 * bug fix: DEFGENERIC now works even when there's a function of the
1341 same name in an enclosing lexical environment. (thanks to Zach
1343 * fixed compiler failure, caused by instrumenting code during
1344 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1345 * optimization: added loop analysis and improved register allocation
1346 * fixed some bugs revealed by Paul Dietz' test suite:
1347 ** POSITION on displaced vectors with non-zero displacement
1348 returns the right answer.
1349 ** (SIMPLE-STRING) is a valid type specifier for sequence
1351 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1353 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1354 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1356 ** The justification version of the FORMAT ~< directive treats
1357 non-zero minpad parameter correctly.
1359 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1360 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1361 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1362 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1363 the supported interface.
1364 * new feature: Single-stepping of code compiled with DEBUG 2 or
1365 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1366 * new feature: saving cores with foreign code loaded is now
1367 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1368 Timothy Moore's work for CMUCL)
1369 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1370 parameters to * instead of NIL if no initform is supplied.
1371 (reported by Johan Bockgaard)
1372 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1373 correctly when FUNCALL appears as the car of the form. Note:
1374 despite this FUNCALL forms are not currently subject to
1375 compiler-macro expansion. (port of Raymond Toy's fix for the
1376 same from CMUCL, reported by Johan Bockgaard)
1377 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1378 (thanks for Teemu Kalvas)
1379 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1380 case instead of terminating the iteration (thanks for Julian
1381 Squires, Sean Champ and Raymond Toy)
1382 * bug fix: incorrect expansion of defgeneric that caused a style
1383 warning. (thanks for Zach Beane)
1384 * bug fix: slot accessor effective method computation works properly
1385 for classes with multiple non-standard applicable methods on
1386 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1387 * on x86 compiler supports stack allocation of results of LIST and
1388 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1389 CMUCL implementation by Gerd Moellmann)
1390 * fixed some bugs revealed by Paul Dietz' test suite:
1391 ** FORMAT strings with both the ~<~:;~> form of the justification
1392 directive and pretty-printing directives cause an error.
1394 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1395 * incompatible change: the internal functions
1396 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1397 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1398 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1399 instead of the old functions.
1400 * new feature: on platforms where "dladdr" is available foreign
1401 function names now appear in backtraces. (based on Helmut Eller's
1403 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1404 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1406 * documentation: ASDF manual is now distributed with SBCL.
1407 * bug fix: SBCL can now load its contributed modules with REQUIRE
1408 even if the system-provided entries have been removed from
1409 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1411 * bug fix: backtraces involving undefined functions or assembly
1412 routines are more informative. (thanks to Brian Downing)
1413 * bug fix: mutually referent alien structures now work correctly.
1414 (reported by Rick Taube)
1415 * bug fix: structures defined by WITH-ALIEN can be referred to
1416 within other definitions in the same WITH-ALIEN.
1417 * bug fix: division operators (MOD, TRUNCATE and the like) with
1418 constant zero divisors and integer dividends no longer generate
1420 * bug fix: provide default methods for INPUT-STREAM-P and
1421 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1422 * bug fix: improve the detection and resolution of MOP metacycles
1423 (where effective method computation depends on the generic function
1424 having its effective method computed). (reported by Bruno Haible)
1425 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1426 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1427 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1428 * bug fix: pathnames with relative directory components can be
1429 represented relative to default pathnames.
1430 * optimization: in taking the GCD of bignums, reduce the two bignums
1431 to approximately the same size (using Euclid's algorithm) before
1432 applying the more sophisticated binary GCD. (thanks to Juho
1434 * optimization: COUNT on bitvectors now operates word-at-a-time.
1435 * optimization: ASH with a positive, but not necessarily constant,
1436 (leftwards) shift, when in a modular context, is compiled to a
1438 * fixed some bugs revealed by Paul Dietz' test suite:
1439 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1440 if the corresponding argument is NIL.
1441 ** FORMAT directives accepting a minpad parameter treat negative
1443 ** literal commas as character parameters to FORMAT directives are
1445 ** literal spaces directly after ~<Newline> directives within a
1446 format-logical-block (~:< ~@:>) do not induce :FILL-style
1447 conditional newlines.
1448 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1449 as specified: it no longer includes conditional newlines).
1450 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1452 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1453 :PER-LINE-PREFIX argument does not evaluate to a string.
1455 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1456 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1457 the manual for details; for now, package locks can be disabled by
1458 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1459 if no major problems are found then it is likely that they will be
1460 compiled in unconditionally.
1461 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1462 unsupported operators on all platforms. To load a shared library
1463 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1464 object file, link it into a shared library outside of SBCL and
1465 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1467 * bug workaround: ROOM T and the GC were not working together
1468 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1469 reporting facility aren't preserved. That reporting has been
1470 disabled, so now until and unless someone figures out how to make it
1471 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1472 * minor incompatible change: as threatened around sbcl-0.8.0, the
1473 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1474 The symbols are also exported from SB-PCL for backwards
1475 compatibility, but more so than before SB-PCL should be treated as
1476 an implementation-internal package.
1477 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1479 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1480 It is known as of this release not to work on the Alpha, however.
1481 * fixed bug #167: errors signalled due to illegal syntax in method
1482 bodies are now more legible.
1483 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1484 designators and can hence be used with TYPEP.
1485 * fixed bug #333: CHECK-TYPE now ensures that the type error
1486 signalled, if any, has the right object to be accessed by
1487 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1488 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1489 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1491 * fixed bug #334: programmatic addition of slots using specialized
1492 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1493 and :ALLOCATION :CLASS slots.
1494 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1495 float exponents. (rereported by Peter Seibel)
1496 * fixed a bug: #\Space (and other whitespace characters) are no
1497 longer considered to be macro characters in standard syntax by
1498 GET-MACRO-CHARACTER.
1499 * fixed bug: initialization of condition class metaobjects no longer
1500 causes an instance of the condition to be created. (reported by
1502 * fixed bug: it is now possible to have slots such that
1503 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1504 description is neither :INSTANCE nor :CLASS.
1505 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1506 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1507 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1508 non-standard SLOT-DEFINITION-ALLOCATION values.
1509 * fixed bugs: various race conditions handling exiting threads.
1510 CL-PPCRE's thread torture test now passes.
1511 * fixed arguably-a-bug: GC time proportional to number of threads in
1512 system even when most of them are idle
1513 * optimization: improved performance of BIT and SBIT on bit-vectors.
1514 * .fasl file incompatibility: The fasl file version number has
1515 been incremented because of changes associated with package locks.
1517 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1518 * minor incompatible change: the system no longer provides
1519 optimization hints (conditions which are TYPEP
1520 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1521 mode; these hints are still emitted when higher SPEED optimization
1523 * new contrib module: a sampling profiler (profiling by statistical
1524 sampling, rather than by instrumenting functions) is available as
1525 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1526 * the behaviour of the standard function ED is now customizeable by
1527 third parties through a hook variable: see ED's documentation
1528 string for information on the protocol.
1529 * the compiler no longer emits efficiency notes for (FUNCALL X)
1530 when the type of X is uncertain under default optimization
1532 * fixed bug 276: mutating a binding of a specialized parameter to a
1533 method to something that is not TYPEP the specializer is now
1535 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1536 exhibit the required behaviour on the x86 platform. (thanks to
1537 Peter van Eynde, Eric Marsden and Bruno Haible)
1538 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1539 even for difficult arguments. (reported by Peter Graves)
1540 * fixed bug 141a: the backquote printer now descends quoted
1542 * fixed another bug in backquote printing: no more destructive
1543 modification of the form's list structure. (reported by Brian
1545 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1546 move between its address being taken and the call to
1547 interrupt_thread, fixing a crashing race condition.
1548 * the SB-POSIX contrib implementation has been adjusted so that it
1549 no longer exhibits ridiculously poor performance when constructing
1550 instances corresponding to C structs.
1552 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1553 * minor incompatible change: the sb-grovel contrib now treats C
1554 structures as alien (in the sense of SB-ALIEN) objects rather than
1555 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1556 has implications for memory management of client code
1557 (sb-grovel-returned objects must now be manually managed) and for
1558 type safety (alien objects now have full types).
1559 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1560 used to control emission of compiler diagnostics, rather than the
1561 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1562 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1563 quality should be considered deprecated.
1564 * install.sh now installs the user manual as well
1565 * (not quite a new documentable feature, but worth considering in
1566 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1567 beginnings of a semantically meaningful condition hierarchy is
1568 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1570 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1571 designator as the defaults argument.
1572 * fixed bug: Displaced arrays whose displaced-to array has become
1573 too small now cause ARRAY-DIMENSION to signal an error, providing
1574 for safer bounds-checking. (reported by Bruno Haible)
1575 * fixed bug: DEFCLASS slot definitions with identical :READER and
1576 :WRITER names now signal a reasonable error. (reported by Thomas
1578 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1580 * fixed bug: Invalid slot specification errors now print correctly.
1581 (thanks to Zach Beane)
1582 * fixed bug 320: Shared to local slot value transfers in class
1583 redefinitions now happen corrently from superclasses as
1584 well. (reported by Bruno Haible)
1585 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1587 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1588 improper lists correctly. (reported by Bruno Haible)
1589 * fixed bug 313: source-transform for <fun-name> was erroneously
1590 applied to a call of a value of a variable with name <fun-name>.
1591 (reported by Antonio Menezes Leitao)
1592 * fixed bug 307: The obsolete instance protocol ensures that
1593 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1594 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1595 remove dead unknown-values globs from the middle of the stack.
1596 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1597 easier to resume long computations after using *BREAK-ON-SIGNALS*
1598 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1599 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1600 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1601 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1602 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1603 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1605 * fixed a (fairly theoretical) bug in string printing: if
1606 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1607 does not have array-element-type equal to the most general string
1609 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1610 function name. (thanks to Nikodemus Siivola)
1611 * fixed bug: compile-time format string checker failed on
1612 non-closed ~{. (reported by Thomas F Burdick)
1613 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1614 don't warn on legal loop code involving more than one aggregate
1615 boolean. (thanks to Nikodemus Siivola)
1616 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1617 in keyword position to MAKE-PACKAGE was not being reported as
1619 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1620 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1622 * fixed bugs 280 and 312: the checking for multiple definitions in a
1623 file is less likely to become confused by uses of inline
1625 * fixed bug: the #S reader macro performs the keyword coercion
1626 specified for slot names. (reported by Kalle Niemitalo)
1627 * fixed bug: lambda lists may contain symbols whose names start with
1628 & but are not lambda-list-keywords; their occurrence triggers a
1630 * fixed bug 321: define-method-combination argument lambda lists do
1631 not cause invalid code to be generated when &optional or &aux
1632 variables have default values. (reported by Bruno Haible)
1633 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1634 have CLOS classes; this appears also to have fixed reported
1635 instability in stack exhaustion detection.
1636 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1637 MUFFLE-WARNING when no associated restart is present is now
1639 * optimization: rearranged the expansion of various defining macros
1640 so that each expands into only one top-level form in a
1641 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1643 * optimization: used a previously little-used slot in symbols to
1644 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1646 * fixed some bugs revealed by Paul Dietz' test suite:
1647 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1649 ** Fixed an optimization bug related to inheritance of initforms
1650 from local to shared slots.
1651 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1652 ** CAST optimizer forgot to flush argument derived type.
1653 ** print/read consistency on floats is now orders of magnitude
1654 more likely. (thanks also to Bruno Haible for a similar report
1656 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1657 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1658 for objects of type REAL. Make it so.
1659 ** SXHASH is specified (infelicitously) to respect similarity,
1660 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1661 it so. (thanks to Markus Ziegler)
1662 ** on the Alpha, the compiler succeeds in compiling functions
1663 returning a known number of arguments greater than 63.
1664 ** fixed handling of invalid NIL arguments in keyword position
1666 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1668 ** provided a readably-printable representation for RANDOM-STATE
1670 ** ensured that pathnames, where they have a namestring, always
1671 print using #P"..." syntax.
1673 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1674 * Support for the forthcoming 2.0 version of the NetBSD kernel
1675 running on x86 hardware has been added. (thanks to Perry
1676 E. Metzger most immediately, and others for their past work)
1677 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1678 current release version; SBCL's OpenBSD support had been broken
1679 since about the time of OpenBSD's switch to ELF binary format.)
1680 * [placeholder for DX summary]
1681 ** user code with &REST lists declared dynamic-extent, under high
1682 speed or space and low safety and debug optimization policy.
1683 * The manual has been converted to Texinfo format and the debugger
1684 chapter from the cmucl manual has been added.
1685 * A facility has been added to extract documentation strings from
1686 sbcl and store them as Texinfo-formatted snippets for inclusion in
1687 the manual (via Texinfo's @include directive)
1688 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1689 the test case to Dave Roberts)
1690 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1691 pointers. (reported by Sean Ross)
1692 * bug fix: PROFILE output is printed nicely even for large numerical
1693 values. (thanks to Zach Beane)
1694 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1695 greater than 32 handle EOF correctly.
1696 * bug fix: on X86 an immediate argument of the IMUL instruction is
1697 correctly printed in disassembly. (thanks to Lutz Euler)
1698 * bug fix: class slots in redefined classes preserve their old
1699 values. (thanks to Bruno Haible and Nikodemus Siivola)
1700 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1701 succeeds. (reported by Marco Baringer)
1702 * fixed some bugs revealed by Paul Dietz' test suite:
1703 ** READ-SEQUENCE now works on ECHO-STREAMs.
1704 ** RATIONALIZE works more according to its specification. (thanks
1707 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1708 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1709 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1710 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1711 should matter to you only if you rebind the printer control
1712 variables and then find you want different bindings in the
1713 debugger than in the ordinary execution of your program.)
1714 * The runtime build system has been tweaked to support building
1715 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1716 assembler and linker. (thanks to Nikodemus Siivola)
1717 * Unbound, undefined, undeclared variables now trigger full
1718 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1719 more often programmer error than deliberate exploitation of undefined
1721 * optimization: the hash algorithm for strings has changed to one
1722 that is less vulnerable to spurious collisions. (thanks to Juho
1724 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1725 needless bounds checking. (thanks to Juho Snellman)
1726 * optimization: implemented multiplication as a modular
1727 (UNSIGNED-BYTE 32) operation on the PPC backend.
1728 * fixed some bugs revealed by Paul Dietz' test suite:
1729 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1731 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1732 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1733 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1735 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1737 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1739 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1740 their output stream on EOF from read.
1741 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1742 have been read to end-of-file.
1743 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1745 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1746 description of determination of which consecutive characters
1748 ** Printing the "Space" character with escaping on now yields "#\\ ",
1749 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1750 ** Reading floating-point numbers with *READ-BASE* set to a number
1751 less than 10 works correctly.
1752 ** Reading floating-point numbers with *READ-BASE* set to a number
1753 more than 10 works correctly.
1754 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1755 the readtable currently in effect.
1757 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1758 * minor incompatible change: parsing of namestrings on a physical
1759 (Unix) host has changed; numbers after the final #\. in a
1760 namestring are no longer interpreted as a version field. This is
1761 intented to be largely invisible to the user, except that the
1762 meaning of the namestring "*.*.*" has changed: it now refers to a
1763 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1764 should usually be replaced by
1765 (make-pathname :name :wild :type :wild :version :wild)
1766 with the added benefit that this is more likely to be portable.
1767 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1768 option to OPEN now signals an error if the file being opened
1769 exists; this may have an impact on existing code.
1770 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1771 signals received, on the PowerPC platforms (both Linux and
1772 Darwin). (thanks to Pierre Mai for pointing out the location of
1774 * several fixes on OS X: The system now builds and runs cleanly on
1775 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1776 (thanks to Brian Mastenbrook)
1777 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1778 second-resolution: integer multiples of 1/3600 between -24 and 24.
1779 (thanks to Vincent Arkesteijn)
1780 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1781 type when called with 1 argument; PEEK-CHAR checked type of
1782 PEEK-TYPE only after having read first character from a
1783 stream. (reported by Peter Graves)
1784 * bug fix: the garbage collector now has much better locality
1785 behaviour, and in particular no longer treats all memory as being
1786 exhausted when many small objects point to each other in a deeply
1788 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1789 recognized as being TYPEP their class.
1790 * bug fix: the PUSHNEW documentation string has been corrected.
1791 (thanks to Vincent Arkesteijn)
1792 * bug fix: defaulting of the value for the last of an atypically
1793 large number of multiple values being bound was not being
1794 performed correctly on the Alpha or PPC platforms
1795 * optimization: implemented multiplication as a modular
1796 (UNSIGNED-BYTE 32) operation on the x86 backend.
1797 * optimization: SEARCH on simple-base-strings can now be open-coded.
1798 (see also contrib/compiler-extras.lisp for inspiration for
1799 teaching the compiler about the Boyer-Moore algorithm).
1800 * value, returned by MAX (and MIN) called with several EQUALP, but
1801 not EQL, arguments now does not depend on compiler settings.
1802 * fixed some bugs revealed by Paul Dietz' test suite:
1803 ** in stack analysis liveness information is propagated from
1804 non-local entry points.
1805 ** pathwise CAST removing failed when the CAST node did not start
1807 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1808 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1810 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1811 host is already defined.
1812 ** RENAME-FILE works on streams instead of signalling an internal
1814 ** PEEK-CHAR uses the current readtable when determining whether
1815 or not a character is whitespace.
1816 ** MERGE-PATHNAMES handles the case when the pathname does not
1817 specify a name while the default-pathname specifies a version.
1818 ** Pathnames now stand a better chance of respecting print/read
1820 ** Attempting to use standardized file system operators with a
1821 pathname with invalid :DIRECTORY components signals a
1823 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1825 ** DIRECTORY on logical pathnames is more correct.
1826 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1827 signal a TYPE-ERROR if their argument is not a stream.
1828 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1829 designator argument does not designate a stream.
1830 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1831 examining the synonym.
1832 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1834 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1835 element-type larger than ([UN]SIGNED-BYTE 32).
1837 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1838 * When built with the :SB-FUTEX feature, threaded builds now take
1839 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1840 for faster/more reliable mutex and condition variable support.
1841 * Incompatible change (but one you probably shouldn't have been using
1842 anyway): the interface and code for arbitrating between multiple
1843 threads in the same user session has been redesigned.
1844 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1845 error when called without an explicit environment argument.
1846 (thanks to Nikodemus Siivola)
1847 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1848 confused about their position. (thanks to Adam Warner and Gerd
1850 * bug fix: Pretty printing backquoted forms with unquotations in the
1851 argument list position of various code constructs such as LAMBDA
1852 now works correctly. (reported by Paul Dietz)
1853 * bug fix: Pretty printing unquotations no longer loses all
1854 stream position information.
1855 * optimization: performance of string output streams is now less
1856 poor for multiple small sequence writes.
1857 * optimization: performance of CSUBTYPEP in the presence of complex
1858 expressions involving CONS and NOT many times has been improved.
1859 (reported by Paul Dietz)
1860 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1862 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1863 streams for the REPL, for the debugger, and so on. (thanks to
1865 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1866 * fixed some bugs revealed by Paul Dietz' test suite:
1867 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1868 is ignored if it is too small, rather than propagating through
1869 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1870 ** extremely complex negations of CONS types were not being
1871 sufficiently canonicalized, leading to inconsistencies in
1873 ** VALUES tranformer lost derived type.
1875 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1876 * fixed a bootstrapping bug: the build process no longer assumes
1877 that the various BOOLE-related constants have the same value in
1878 host and target lisps. (noted by Paul Dietz' test suite on an
1879 SBCL binary built from CLISP)
1880 * The system can now be dynamically linked on the MIPS platform,
1881 which enables dynamic loading of foreign code from Lisp. (thanks
1882 to Ralf Baechle for discussions on the MIPS ABI)
1883 * The system now records debugging information for its own source
1884 files in a filesystem-position-independent manner, relative to
1885 the "SYS" logical host.
1886 * fixed a compiler bug: MV-LET convertion did not check references
1887 to the "max args" entry point. (reported by Brian Downing)
1888 * tweaked disassembly notes to be less confident about proclaiming
1889 some instruction as an LRA. (thanks to Brian Downing)
1890 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1891 now each have their own history, command character, and other
1892 characteristics. (thanks to David Lichteblau)
1893 * fixed some bugs revealed by Paul Dietz' test suite:
1894 ** compiler failure in compiling LOGAND expressions including a
1896 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1897 shift greater than 32.
1898 ** FLUSH-DEST did not mark blocks for type check regeneration.
1899 ** HANDLER-CASE failed to accept declarations in handler clauses
1900 in some circumstances.
1902 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1903 * New code in contrib/sb-introspect (still probably not entirely
1904 stable yet) provides some support for smart Lisp development
1905 environments like SLIME.
1906 * The conditions signalled for errors occurring when loading .fasl
1907 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1908 in a way which should help ASDF recover gracefully.
1909 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1910 stuff has been cleaned up. If you code contrib/ stuff, this might
1911 affect you, and you can look at contrib/README, contrib/STANDARDS,
1912 and/or the 0.8.4.27 diff to check.
1913 * In full calls the compiler now does not generate checks for declared
1914 argument types for all arguments.
1915 * various threading fixes
1916 ** and some experimental patches which didn't make it into
1917 the main tree for this release, but which are shipped in
1918 contrib/experimental-thread.patch as a possible fix for some
1919 failures (deadlock, spinning...) in GC-intensive multithreaded
1921 * fixed PPC build problem (source code incompatibility of different
1922 library versions): added offsetof() hackery which attempts to divine
1923 where glibc maintainers put uc_mcontext today
1924 * fixed bug 282: compiler does not trust type assertions while passing
1925 arguments to a full call.
1926 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1927 &OPTIONAL VALUES type parameter.
1928 * fix bug 214: algorithm for noting rejected templates is now more
1929 similar to that of template seletion. (also reported by rydis on
1931 * fixed bug 141b: printing backquoted information readably and prettily
1932 inserts a space where necessary.
1933 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1934 (CHARACTER 10) are now reported as errors, rather than propagated
1935 as unknown types. (reported by piso on #lisp)
1936 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1937 with values NIL and :ERROR. (thanks to Milan Zamazal)
1938 * fixed bug 191c: CLOS now does proper keyword argument checking as
1939 described in CLHS 7.6.5 and 7.6.5.1.
1940 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1941 longer raise an error; further, using a list as a for-as-arithmetic
1942 counter now raises a meaningful error.
1943 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1944 now understood by sequence creation functions such as MAKE-SEQUENCE
1946 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1947 a STRING-INPUT-STREAM.
1948 * compiler enhancement: SIGNUM is now better able to derive the type
1950 * type declarations inside WITH-SLOTS are checked. (reported by
1952 * fixed some bugs revealed by Paul Dietz' test suite:
1953 ** incorrect optimization of TRUNCATE for a positive first
1954 argument and negative second.
1955 ** compiler failure in let-convertion during flushing dead code.
1956 ** compiler failure while deriving type of TRUNCATE on an
1957 interval, containing 0.
1958 ** ASH of a negative bignum by a negative bignum count now returns
1960 ** intersection of CONS types now canonicalizes properly, fixing
1961 inconsistencies in SUBTYPEP.
1963 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1964 * incompatible change: The --disable-debugger command line
1965 option now clobbers the debugger at a more fundamental
1966 level, by redefining #'INVOKE-DEBUGGER instead of by
1967 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1968 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1969 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1970 enter the debugger REPL and then suffer recursive errors
1971 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1972 while under the new system, BREAK in --disable-debugger mode
1973 terminates the system just as an unhandled error would.
1974 * fixed compiler performance when processing loops with a step >1;
1975 * bug fix: DOCUMENTATION now retrieves generic function
1976 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1977 support has been systematized, and now supports the methods
1978 specified by ANSI, along with a default method and a method for
1979 slot documentation. (reported by Nathan Froyd)
1980 * bug fix: effective methods associated with a generic function are
1981 no longer cached over a change of that generic function's method
1982 combination. (reported by Andreas Fuchs)
1983 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1984 names a non-existent directory. (thanks to Andreas Fuchs)
1985 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1986 the PPC platform, lead to stack corruption; this has been fixed.
1987 (reported by Rainer Joswig)
1988 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1989 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1990 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1991 arithmetic in general, is now much more reliable on the PPC
1993 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1994 platform now returns the right answer.
1995 * optimization: restored some effective method precomputation in
1996 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1997 precomputation is now tunable.
1998 * optimization: compiler-internal data structure use has been
1999 reviewed, and changes have been made that should improve the
2000 performance of the compiler by about 20%.
2001 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2002 simple-base-strings and simple-bit-vectors is improved.
2003 * optimization: the optimization of 32-bit logical and arithmetic
2004 functions introduced in version 0.8.3 on the x86 has been
2005 implemented on the mips, ppc and sparc platforms; an
2006 implementation of the same facility, but for 64-bit arithmetic,
2007 has been added for the alpha.
2008 * microoptimization: the compiler is better able to make use of the
2009 x86 LEA instruction for multiplication by constants.
2010 * bug fix: in some situations compiler did not report usage of
2011 generic arithmetic in (SPEED 3) policy.
2012 * bug 145b fix: compiler used wrong type specifier while converting
2013 MEMBER-types to numeric.
2014 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2016 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2017 index argument now works properly on the Alpha platform.
2018 * bug fix: floating point exception treatment on the Alpha platform
2020 * bug fix: FILE-POSITION works much better on string input and
2021 output streams. (thanks to Nikodemus Siivola)
2022 * bug fix: many threading/garbage collection symptoms sorted.
2023 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2024 might be pseudo-atomic.
2025 * internal change: Stopping for GC is now done with signals not
2026 ptrace. GC is now done in whichever thread wanted it, instead of
2028 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2030 * bug fix: result form in DO is not contained in the implicit
2032 * incompatible change: ICR structure is changed; the value part of
2033 CONTINUATION is now called LVAR; corresponding functions are
2034 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2035 * added type deriver for ISQRT (thanks to Robert E. Brown).
2036 * bug fix: better support for loading from the command line when an
2037 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2039 * fixed some bugs revealed by Paul Dietz' test suite:
2040 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2041 ** ROUND and FROUND now give the right answer when given very
2042 small float arguments.
2043 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2045 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2046 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2047 ** LCM should return a non-negative integer.
2048 ** PARSE-INTEGER returned the index of a terminator instead of the
2049 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2050 ** PARSE-INTEGER returned an incorrect index being applied to a
2052 ** LCM with two arguments of 0 returns 0 rather than signalling
2054 ** unsigned addition of a 32-bit constant with the high bit set no
2055 longer causes an internal compiler error on the x86.
2056 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2057 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2058 with negative last argument.
2059 ** byte specifiers with zero size and position no longer cause
2060 an error during type derivation.
2061 ** bignum multiplication on the Alpha platform now returns the
2063 * porting: The system now builds on SuSE AMD64, although it still
2064 generates a 32-bit binary.
2065 * .fasl file incompatibility: The fasl file version number has
2066 been incremented (because of the changes to internal compiler
2067 data structures referred to above).
2069 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2070 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2071 more accurately, on the Darwin kernel running on PowerPC hardware.
2072 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2073 * Compiler code deletion notes now signal a condition of type
2074 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2075 an associated MUFFLE-WARNING restart.
2076 * The compiler now performs limited argument count validation of
2077 constant format strings in FORMAT, and where appropriate in ERROR,
2078 CERROR and WARN. (thanks to Gerd Moellmann)
2079 * New ASDF-INSTALL contrib can be used for automatic download and
2080 installation of third-party Lisp code from CCLAN or other sites
2082 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2083 which forces another thread to execute a function supplied by the
2085 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2086 now accept and act upon their :ELEMENT-TYPE keyword argument.
2087 (reported by Martin Atzmueller, Edi Weitz)
2088 * bug fix: FILE-POSITION now accepts position designators up to
2089 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2090 is the greater. (thanks to Patrik Nordebo)
2091 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2092 Kalle Olavi Niemitalo)
2093 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2094 printable. (reported by Eric Marsden)
2095 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2096 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2097 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2098 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2099 (thanks to Stig Sandoe)
2100 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2101 UNREAD-CHAR now correctly includes the unread character in the
2102 target sequence. (thanks to Gerd Moellmann)
2103 * bug fix in threaded builds: the system can now be suspended and
2104 resumed by shell job control with minimal disruption.
2105 * bug fixes in times and timezones >2038AD
2106 * better handling of "where is GNU make?" problem in build scripts
2107 (thanks to Nikodemus Siivola)
2108 * new optimization: inside a named function any reference to a
2109 function with the same name is considered to be a self-reference;
2110 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2111 optimization quality.
2112 * new optimization on x86: logical functions and + now have
2113 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2114 used when the result is truncated to 32 bits.
2115 * VALUES declaration is partially enabled.
2116 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2117 * fixed some bugs revealed by Paul Dietz' test suite:
2118 ** The system now obeys the constraint imposed by
2119 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2120 types form a lattice under type intersection.
2121 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2122 ** ASSOC now ignores NIL elements in an alist.
2123 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2124 and (1+ MOST-POSITIVE-FIXNUM) answers.
2125 ** The addition of a method with invalid qualifiers to a generic
2126 function does not cause an error to be signalled immediately;
2127 a warning is signalled, and the error is generated only on
2128 calling the generic function.
2129 * changed .fasl file version number, in part to add type codes for
2130 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2131 obscure ANSI requirements
2133 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2134 * fixed bug 148: failure to inline-expand a local function left
2135 garbage, confusing the compiler.
2136 * fixed bugs 3cd: structure slot readers perform type check if the
2137 slot can have an invalid value (i.e. it is either not initialized
2138 or can be written with a less specific slot writer).
2139 * bug fix: the compiler now traps array references to elements off
2140 the end of an array; previously, the bounds checking in some
2141 circumstances could go off-by-one.
2142 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2144 * type declarations for array element types now obey the description
2145 on the CLHS page "Declaration TYPE", as per discussions on
2146 sbcl-help around 2003-05-08. This means that a declaration
2147 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2148 declaration, all references to BAR will be asserted or assumed
2149 (with THE, so dependent on compiler policy) to involve objects of
2150 type FOO. Note that no such declaration is implied in
2151 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2152 * declared types of functions from the "Conditions"
2153 chapter. (reported by Paul Dietz)
2154 * bug fix: CERROR accepts a function as its first argument.
2155 * bug fix: NTH an NTHCDR accept a bignum as index
2156 arguments. (reported by Adam Warner)
2157 * optimization: character compare routines now optimize comparing
2158 against a constant character. (reported by Gilbert Baumann)
2159 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2160 argument now works properly on the MIPS platform.
2161 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2162 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2163 declared to have a functional type, does not cause an error
2165 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2166 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2167 * fixed bug: if last continuation of a deleted block has a
2168 destination, this destination should be deleted too. (reported by
2170 * fixed a bug in the bootstrap process: the host compiler's values
2171 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2172 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2173 test case from Patrik Nordebo)
2174 * improved the ability of the disassembler on the PPC platform to
2175 provide helpful disassembly notes.
2176 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2177 the class in more cases than previously.
2178 * bug fix: FILE-POSITION now understands :START and :END for
2179 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2180 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2181 8) data. (thanks to David Lichteblau)
2182 * bug fix: it is possible to add a method to a generic function
2183 without lambda list.
2184 * bug fix: reader failed to signal END-OF-FILE inside an
2185 object representation. (reported by Nikodemus Siivola)
2186 * fixed some bugs revealed by Paul Dietz' test suite:
2187 ** LAST and [N]BUTLAST should accept a bignum.
2188 ** condition slot accessors are methods.
2189 ** (VECTOR NIL) is a subtype of STRING.
2191 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2192 * minor incompatible change: some nonsensical specialized lambda
2193 lists (used in DEFMETHOD) which were previously ignored now signal
2195 * minor incompatible change: the system is now aware of the types of
2196 variables in the COMMON-LISP package, and will signal errors for
2197 most violations of these type constraints (where previously they
2198 were silently accepted).
2199 * minor incompatible change: COMPILE-FILE now uses the freedom
2200 afforded (ANSI 3.2.2.3) to use derived function types for
2201 functions defined in the same file. This also permits the system
2202 to warn on static type mismatches and function
2203 redefinition. (Currently it does not work with high DEBUG level.)
2204 * minor incompatible change: VALUES declaration is disabled.
2205 * When issuing notes, the compiler now signals a condition of type
2206 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2207 restart for use in user handlers. It is expected that the
2208 COMPILER-NOTE condition will eventually become a condition
2209 supertype to a hierarchy of note types, which will then be
2210 handleable in a similar fashion. However, at the moment, no such
2211 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2212 but it's an implementation detail, not a classification for the
2214 * Changes in type checking closed the following bugs:
2215 ** type checking of unused values (192b, 194d, 203);
2216 ** template selection based on unsafe type assertions (192c, 236);
2217 ** type checking in branches (194bc).
2218 * A short form of VALUES type specifier has ANSI meaning (it has
2219 increased the number of situations when SBCL cannot perform type
2221 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2222 keywords or constants is permissible.
2223 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2224 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2225 classes. (thanks to Antonio Martinez)
2226 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2227 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2228 argument to a pretty-print stream. Instead, it leaves any such
2229 operations to DESCRIBE-OBJECT methods.
2230 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2232 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2233 lambda lists. (thanks to David Lichteblau)
2234 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2235 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2236 (thanks to Gerd Moellmann)
2237 * fixed bug 246: increased compilation speed of long
2238 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2240 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2242 * DEFCONSTANT now throws a condition of type
2243 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2244 constant to a non-EQL value; CONTINUE and ABORT restarts
2245 respectively change and preserve the value.
2246 * fixed bug 63: The code walker, part of the implementation of CLOS,
2247 is now better at handling symbol macros.
2248 * bug fix: There is no longer an internal implementation type named
2249 CL:LENGTH. (reported by Raymond Toy)
2250 * bug fix: In macro-like defining macros/special operators the
2251 implicit block does not enclose the lambda list.
2252 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2253 enclosed in parentheses) are not suitable as type specifiers, and
2254 their use properly signals an error now.
2255 * bug fix: An argument count mismatch for a type specifier in code
2256 being compiled no longer causes an unhandled error at compile
2257 time, but signals a compile-time warning.
2258 * fixed simple vector readable printing
2259 * bug fix: DESCRIBE takes more care over whether the class
2260 precedence list slot of a class is bound before accessing it.
2261 (reported by Markus Krummenacker)
2262 * bug fix: FORMATTER can successfully compile pretty-printer format
2263 strings which use variants of the ~* directive inside.
2264 * bug fix: SEARCH now applies its TEST predicate to the elements of
2265 the arguments in the correct order. (thanks to Wolfhard Buss)
2266 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2267 in type inference. (thanks to Robert E. Brown)
2268 * bug fix: Reading in symbols with an explicit package name of ""
2269 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2270 * fixed some bugs revealed by Paul Dietz' test suite:
2271 ** NIL is now allowed as a structure slot name.
2272 ** Arbitrary numbers, not just REALs, are allowed in certain
2273 circumstances in LOOP for-as-arithmetic clauses.
2274 ** Multiple class redefinitions before slot access no longer
2275 causes a type error.
2276 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2277 association between the name and a class.
2278 ** Generic functions with non-standard method-combination and over
2279 five methods all of which return constants no longer return NIL
2280 after the first few invocations. (thanks to Gerd Moellmann)
2281 ** CALL-NEXT-METHOD with no arguments now passes the original
2282 values of the arguments, even in the presence of assignment.
2283 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2285 ** DISASSEMBLE works with closures and funcallable instances.
2286 ** ADD-METHOD now returns the generic function, not the new method.
2287 ** FIND-METHOD signals an error if the lengths of the specializers
2288 is incompatible with the generic function, even if the ERRORP
2290 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2291 which its argument is a member.
2292 ** DEFCLASS only redefines the class named by its class-name
2293 argument if that name is the proper name of the class;
2294 otherwise, it creates a new class.
2295 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2296 of the UNBOUND-SLOT condition to the name of the slot.
2297 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2298 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2299 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2300 SLOT-MAKUNBOUND in the specified fashion.
2302 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2303 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2304 cross-compilation host. As a consequence, we can now bootstrap our
2305 way up to SBCL starting with a bare gcc toolchain and human-readable
2306 source code (first the source to CLISP, then the source to SBCL).
2307 * A contributed module containing a partial implementation of the
2308 simple-streams interface has been included. (thanks to Rudi
2310 * A contributed module implementing the RFC1321 Message Digest
2311 Algorithm, known as MD5, has been included.
2312 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2313 no longer has any effect, as the code controlled by this feature
2314 has been deleted. (As far as we know, no-one has ever built using
2315 this feature, and its semantics were confused in any case).
2316 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2317 work on conditions (as required by the ANSI specification),
2318 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2319 expected behaviour on conditions. Users should note, however,
2320 that such behaviour is not required by the ANSI specification,
2321 and so use of this behaviour may render their code unportable.
2322 * fixed some bugs revealed by Paul Dietz' test suite:
2323 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2325 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2326 ** SXHASH on condition objects no longer returns NIL.
2327 ** :ALLOCATION :CLASS slots are better treated; their values are
2328 updated on class redefinition, and initforms inherited from
2329 superclasses are applied.
2330 ** REMOVE-METHOD returns its generic function argument even when
2331 no method was removed.
2332 ** SHARED-INITIALIZE now initializes the values of the requested
2333 slots, including those with :ALLOCATION :CLASS.
2334 ** ALLOCATE-INSTANCE now works on structure classes defined via
2335 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2337 ** SLOT-EXISTS-P now works on conditions, as well as structures
2339 ** MAKE-LOAD-FORM now has the required methods on
2340 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2341 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2342 keyword, and now implements the SLOT-NAMES argument.
2343 ** methods with &OPTIONAL arguments no longer allow too many
2344 arguments to be passed in the call without error.
2345 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2346 option is consistent with the required arguments of the generic
2347 function lambda list.
2348 * bug fix: REQUIRE accepts a string designator. (Thanks to
2350 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2351 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2352 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2353 * bug fix: The compiler now checks for duplicated variables in macro
2355 * bug fix: SETQ on globals returns the correct value.
2356 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2357 not a COMPILER-ERROR (followed by some other strange error on
2358 choosing the CONTINUE restart).
2359 * bug fix: make.sh and friends are now more consistent in the way that
2360 they look for GNU "make".
2362 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2363 * experimental native threads support (on x86 Linux >=2.4 only).
2364 This is not compiled in by default: you need to add :SB-THREAD to
2365 the target features. See the "Beyond ANSI" chapter of the manual
2367 * fix for longstanding nonANSIism: The old distinction between
2368 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2369 The return value from CL:FIND-CLASS is now a CLOS class, and
2370 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2371 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2372 * An interface to the MetaObject Protocol, as described in Kiczales,
2373 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2374 MIT Press, 1991, is available from the SB-MOP package.
2375 * incompatible change: the SB-PCL package should now be considered
2376 a private implementation detail, and no longer a semi-private MOP
2378 * minor incompatible change: due to rearrangement for threads, the
2379 control stack and binding stack are now allocated at arbitrary
2380 addresses instead of being hardcoded per-port. Users affected by
2381 this probably have to be doing advanced things with shared
2382 libraries, and will know who they are.
2383 * minor incompatible change: Previously, all --eval forms used were
2384 processed with READ before any of them were processed with EVAL.
2385 Now each --eval form is processed with both READ and EVAL before
2386 the next --eval form is processed. (Thus package operations like
2387 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2388 work as the user might reasonably expect.)
2389 * minor incompatible change: *STANDARD-INPUT* is now only an
2390 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2392 * minor incompatible change: Y-OR-N-P is now character-oriented,
2393 not line oriented. Also, YES-OR-NO-P now works without errors.
2394 (thanks to Antonio Martinez)
2395 * sb-aclrepl module improvements: an integrated inspector, added
2396 repl features, and a bug fix to :trace command.
2397 * Known functions, which cannot be open coded by the backend, are
2398 considered to be able to check types of their arguments. (fixing
2399 a bug report by Nathan J. Froyd)
2400 * fixed a bug in computing method discriminating functions: It is
2401 now possible to define methods specialized on classes which have
2402 forward-referenced superclasses. (thanks to Gerd Moellmann)
2403 * fixed evaluation order in optional entries (reported by Gilbert
2405 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2406 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2407 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2408 required-by-AMOP NAME argument, as well as CLASS and
2409 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2410 * fixed bug 20: DEFMETHOD can define methods using names that are
2411 not the proper names of classes to designate class specializers.
2412 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2413 Unix file descriptors, instead of blowing up. (thanks to Antonio
2415 * Garbage collection refactoring: user-visible change is that a
2416 call to the GC function during WITHOUT-GCING will not do garbage
2417 collection until the end of the WITHOUT-GCING. If you were doing
2418 this you were probably losing anyway.
2419 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2420 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2421 TYPEP the latter but not the former.
2422 * The compiler issues a full WARNING for calls to undefined functions
2423 with names from the CL package.
2424 * MAP-INTO for a vector destination is open coded. (reported by
2425 Brian Downing on c.l.l)
2426 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2427 documentation string.
2428 * fixed some bugs revealed by Paul Dietz' test suite:
2429 ** COPY-ALIST now signals an error if its argument is a dotted
2431 ** Condition slots are now accessed more correctly in the presence
2432 of multiple initargs for a given slot.
2433 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2434 exclude restarts of the same name associated with a different
2436 ** DEFCLASS of forward-referenced classes with another
2437 forward-referenced class in the superclasses list no longer
2439 ** Condition slots are now initialized once each, not multiple
2440 times. (thanks to Gerd Moellmann)
2441 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2443 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2444 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2445 arguments contain duplicated elements.
2446 ** RESTART-CASE understands local macros.
2447 ** RESTART-CASE associates exactly its own restarts with a condition.
2448 ** ENDP in safe mode checks its argument to be of type LIST.
2449 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2450 in question is unbound.
2451 ** Optimized MAKE-INSTANCE functions no longer cause internal
2452 assertion failures in the presence of duplicate initargs.
2453 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2454 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2455 propagated even in the ctor optimized implementation of
2457 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2459 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2460 * a better implementation of SXHASH on (simple) bit vectors,
2461 measured both in execution speed and in distribution of results
2462 over the positive fixnums, has been installed. Likewise, a better
2463 implementation of EQUAL for simple bit vectors is now available.
2464 * fixed CEILING optimization for a divisor of form 2^k.
2465 * fixed bug 240 (emitting extra style warnings "using the lexical
2466 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2467 by Antonio Martinez)
2468 * fixed SXHASH, giving different results for NIL depending on type
2469 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2470 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2471 variable. (found by Rolf Wester)
2472 * SBCL does not ignore type declarations for special
2473 variables. (reported by rif on c.l.l 2003-03-05)
2474 * some bug fixes in contrib/sb-aclrepl/
2475 * fixed some bugs revealed by Paul Dietz' test suite:
2476 ** a bug in the CONS type specifier, whereby the CAR and CDR
2477 types got intertwined, has been fixed;
2478 ** the type system is now able to reason about the interaction
2479 between INTEGER and RATIO types more completely;
2480 ** APPEND, [N]REVERSE and NRECONC check that those their
2481 arguments, which must be proper lists, are really so;
2482 ** An array specialized to be unable to hold elements has been
2483 implemented, as required -- yes, really -- by ANSI;
2484 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2485 on malformed property lists;
2487 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2488 * incompatible packaging change: in line with Unix convention,
2489 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2490 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2491 * REQUIRE and PROVIDE are now optionally capable of doing something
2492 useful. See the documentation string for REQUIRE.
2493 * infrastructure for a managed SBCL contrib system: contributed
2494 modules in this release include:
2495 ** the ASDF system definition facility;
2496 ** an interface to the BSD Sockets API;
2497 ** an ACL-like convenience interface to the repl;
2498 (thanks to Kevin Rosenberg)
2499 ** an implementation of ROTATE-BYTE, with efficient implementation
2501 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2502 gives a closer approximation to the right answer than previously.
2503 (thanks to Raymond Toy)
2504 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2505 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2506 optional environment arguments, as required by ANSI.
2507 * fixed bugs in other functions taking environment objects, allowing
2508 calls with an explicit NIL environment argument to be compiled
2510 * fixed bug 228: primary return values from
2511 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2512 COMPILE or FUNCTION.
2513 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2514 :TYPE will no longer signal errors on innocuous objects.
2515 * fixed bug 231b: SETQ is better at respecting type declarations in
2516 the lexical environment.
2517 * fixed a bug in DEFCLASS: classes named by symbols with no or
2518 unprintable packages can now be defined.
2519 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2520 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2521 * fixed compiler failure related to checking types of functions.
2522 (reported by Robert E. Brown)
2523 * the compiler is now much more consistent in its error-checking
2524 treatment of bounding index arguments to sequence functions: in
2525 (SAFETY 3) code, errors will be signalled in almost all cases if
2526 invalid sequence bounding indices are passed to functions defined
2527 by ANSI to operate on sequences.
2528 * fixed a bug in the build procedure: documentation of SBCL-specific
2529 packages is now preserved and available in the final Lisp image.
2530 * lifted FDEFINITION lookup out of loops in the implementation of
2531 many list operations. (thanks to Robert E. Brown)
2532 * fixed a bug in the reader: the #n# reader macro now works for
2533 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2534 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2535 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2536 E. Brown and Tony Martinez respectively)
2537 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2538 SLOT-BOUNDP: in particular, it is now able to optimize them much
2539 better, and is now not vulnerable to having packages renamed.
2540 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2541 always called when a slot is not present in an instance. (thanks
2543 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2544 longer signal an error on unreading a character following EOF on
2545 the previous constituent stream. (thanks to Tony Martinez)
2546 * fixed some bugs revealed by Paul Dietz' test suite:
2547 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2548 not just nonnegative fixnums;
2549 ** the logical bit-array operators such as BIT-AND now accept an
2550 explicit NIL for their "opt-arg" argument (to indicate a
2551 freshly-consed result bit-array);
2552 ** ELT now signals an error on an invalid sequence index in safe
2554 ** the type system is now cleverer about negations of numeric
2555 types, and consequently understands the BIGNUM and RATIO types
2557 ** the type system is now cleverer about the interaction between
2558 INTEGER and RATIO types: while bugs still remain, many more
2559 cases are accurately computed;
2560 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2561 if it is in the last clause;
2562 ** CONSTANTLY now correctly returns a side-effect-free function in
2564 ** DECLARE is no longer treated as a special-operator; in
2565 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2566 * incremented fasl file version number due to the change in the
2567 DEFSTRUCT-SLOT-DESCRIPTION structure.
2569 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2570 * minor incompatible change: code processed by the "interpreter" or
2571 EVAL now has a compilation optimization policy of (DEBUG 2)
2572 (changed from (DEBUG 1)) to improve debuggability of interactive
2573 development, and to allow the use of the debug RETURN command in
2575 * an experimental implementation of the RETURN command for the
2576 debugger has been included. (thanks to Frederik Kuivinen)
2577 * fixed bug 62: constraints were not propagated into a loop.
2578 * fixed bug in embedded calls of SORT (reported and investigated by
2580 * fixed some bugs revealed by Paul Dietz' test suite:
2581 ** printing and reading of arrays with some dimensions having
2582 length 0 (thanks to Gerd Moellmann);
2583 ** BOA constructor with &AUX argument without a default value does
2584 not cause a type error;
2585 ** CONSTANTP now returns true for all self-evaluating objects.
2587 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2588 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2589 accessors that are related by inheritance, as specified in the
2590 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2591 Valtteri Vuorikoski)
2592 * The compiler is now able to inline functions that were defined in
2593 a complex lexical environment (e.g. inside a MACROLET).
2594 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2596 * Support for the upcoming FreeBSD-5.0 release has been included.
2597 (thanks to Dag-Erling Smorgrav)
2598 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2599 effect when it is not in a toplevel context.
2600 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2601 stand a better chance of being correct. (thanks to Gerd
2603 * fixed bug in COERCE, which now signals an error on coercing a
2604 rational to a bounded real type which excludes the expected
2606 * The compiler is now able to derive types more accurately from the
2607 COERCE and COMPILE functions.
2608 * fixed bug 223: functional binding is considered to be constant
2609 only for symbols in the CL package.
2610 * fixed bug 231: SETQ did not check the type of a variable being set
2611 (reported by Robert E. Brown)
2612 * A new optimization for MAKE-INSTANCE has been included, fixing
2613 various bugs (including relating to :ALLOCATION :CLASS slots and
2614 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2616 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2617 ** As required by ANSI, LOOP now disallows anonymous collection
2618 clauses such as COLLECT I in conjunction with aggregate boolean
2619 clauses such as THEREIS (= I 1);
2620 ** LOOP now signals an error when any variable is reused in the
2621 same loop (including the potentially useful construct analogous
2622 to WITH A = 1 WITH A = (1+ A);
2623 ** IT is only a special loop symbol within the first clause of a
2624 conditional loop clause;
2625 ** LOOP with a typed iteration variable over a hashtable now
2626 signals a type error iff it should.
2627 * fixed some other bugs revealed by Paul Dietz' test suite:
2628 ** FILE-STREAM now names the class previously known as FD-STREAM;
2629 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2630 argument) no longer signals an error;
2631 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2632 of the slot symbol, rather than using the current package
2633 ((:CONC-NAME "") continues to intern the slot's name in the
2635 * incremented fasl file version number, because of the incompatible
2636 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2637 of the new implementation of DEFINE-COMPILER-MACRO.
2639 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2640 * Support for building SBCL for MIPS platforms running in
2641 little-endian mode has now been checked in, and basic
2642 functionality on said platforms verified.
2643 * minor incompatible change: PCL now records the pathname of a file
2644 in which methods and the like are defined, rather than its
2646 * minor incompatible change: TRUENAME now considers the truename of
2647 a file naming a directory to be the pathname with :DIRECTORY
2648 component indicating that directory.
2649 * minor incompatible change: a NAMED clause in the extended form of
2650 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2651 reason for the previous behaviour is unclear.
2652 * more systematization and improvement of CLOS and MOP conformance
2653 in PCL (thanks to Gerd Moellman and Pierre Mai):
2654 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2656 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2657 primary methods with no specializers;
2658 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2660 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2661 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2662 FORWARD-REFERENCED-CLASSes; error reporting on
2663 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2665 ** SXHASH on CLOS instances now uses a slot internal to the
2666 instance to return different numbers on distinct instances,
2667 while preserving the same return value through invocations of
2669 ** DEFMETHOD signals errors when methods with longer incongruent
2670 lambda lists are added to generic functions;
2671 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2672 CLASS, as specified in AMOP;
2673 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2674 on the order returned by the primary method for classes of
2675 class STANDARD-CLASS;
2676 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2677 * fixed some bugs shown by Paul Dietz' test suite:
2678 ** DOLIST puts its body in TAGBODY;
2679 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2681 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2682 value producing form;
2683 ** if more variables are given to PROGV than values, extra
2684 variables are bound and made to have no value;
2685 ** NSUBSTITUTE on list arguments gets the right answer with
2687 ** ELT signals an error of type TYPE-ERROR when the index argument
2688 is not a valid sequence index;
2689 ** LOOP signals (at macroexpansion time) an error of type
2690 PROGRAM-ERROR when duplicate variable names are found;
2691 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2692 ** FUNCALL of special-operators now cause an error of type
2694 ** PSETQ now works as required in the presence of side-effecting
2695 symbol-macro places;
2696 ** NCONC accepts any object as its last argument;
2697 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2699 ** loop-for-as-package does not require a package to be explicitely
2701 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2703 * fixed bug 166: compiler preserves "there is a way to go"
2704 invariant when deleting code.
2705 * fixed bug 172: macro lambda lists with required arguments after
2706 &REST arguments now cause an error to be signalled. (thanks to
2708 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2709 bug. (thanks to Matthew Danish)
2710 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2712 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2713 arguments are lexically rebound. (thanks to Gerd Moellmann and
2715 * fixed bug 194: error messages are now more informative when there
2716 is no primary method applicable in a call to a generic
2717 function. (thanks to Gerd Moellmann)
2718 * fixed bug in command line argument checking (thanks to Julian
2720 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2721 (thanks to Matthew Danish)
2722 * incremented fasl file version number, because of the
2723 SXHASH-related changes in the layout of CLOS data structures
2725 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2726 * minor incompatible change: The runtime (the Unix executable named
2727 "sbcl") is now much pickier about the .core files it will load.
2728 Essentially it now requires .core files to descend from the same
2729 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2730 as the runtime does. (The intent is to prevent the crashes which
2731 can occur, and which can even be reported as mysterious failures,
2732 when people patch the sources or change the build parameters
2733 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2734 sbcl and .core files.)
2735 * fixed bug: VALUES-LIST is no longer optimized away.
2736 * fixed bug 142: The FFI conversion of C string values to Lisp
2737 string values no longer conses excessively. (thanks to Nathan
2738 Froyd porting Raymond Toy's fix to CMU CL)
2739 * began to systematize and improve MOP conformance in PCL (thanks to
2740 Nathan Froyd, Gerd Moellman and Pierre Mai):
2741 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2743 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2744 ** FINALIZE-INHERITANCE is now called on class finalization;
2745 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2746 argument precedence order.
2747 * fixed bug 202: The compiler no longer fails on functions whose
2748 derived types contradict their declared type.
2749 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2750 so it can be non-toplevel.
2751 * The fasl file version number has changed (because of the new
2752 implementation of DEFMACRO).
2753 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2754 safe code, that any length requirement by their type-specifier
2755 argument is valid. The exceptions to this are described in bug
2757 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2758 functions now check the types of their inputs as required by ANSI.
2759 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2760 introduced symbol is DECLAREd to be SPECIAL.
2761 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2762 * fixed bug: :COUNT argument to sequence functions may be negative
2763 * fixed bug: body of DO-SYMBOLS may contain declarations
2764 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2765 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2766 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2767 symbol macro only once
2768 * fixed printing of call frame when argument list is unavailable
2769 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2770 * compiler no longer signals WARNING on unknown keyword
2773 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2774 * A beta-quality port to the mips architecture running Linux,
2775 based on the old CMUCL backend, has been made. It has been tested
2776 on a big-endian kernel, and works sufficiently well to be able to
2777 rebuild itself; it has not been tested in little-endian mode.
2778 * fixed an inconsistency between gencgc.c and purify.c which made
2779 dumping/loading .core files unreliable
2780 * fixed bug 120a: The compiler now deals correctly with IFs where
2781 the consequent is the same as the alternative, instead of
2782 misderiving the return type. (thanks to Alexey Dejneka)
2783 * fixed bug 113: Logical pathnames are now dumpable (the logical
2784 host is resolved at load-time, throwing an error if it is not
2786 * fixed bug 174: FORMAT's error message is slightly clearer when a
2787 non-printing character is used in a format directive.
2788 * fixed several bugs in compiler checking of type declarations, i.e.
2789 violations of the Python "declarations are assertions" principle
2790 (thanks to Alexey Dejneka)
2791 * fixed several bugs in PCL's error checking (thanks to Gerd
2793 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2795 * fixed bug in compilation of functions as first class values
2796 (thanks to Antonio Martinez-Shotton)
2797 * The compiler's handling TYPE-ERRORs which it can prove will
2798 inevitably happen at runtime has been cleaned up and corrected
2799 in several ways. (thanks to Alexey Dejneka)
2800 * improved argument type checking for various basic arithmetic
2801 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2802 lavished on them in the past that they can be compiled in many
2803 ways in different special cases
2804 * fixed bug 181: compiler checks validity of user supplied type
2806 * cleaned up code flushing in optimization: Function calls which
2807 should signal errors for safety purposes (e.g. which ANSI says
2808 should signal errors when their arguments are of incorrect type)
2809 are no longer optimized away.
2810 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2811 * incremented fasl file version number, because changes in the
2812 implementation of sequence functions like COERCE caused
2813 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2814 in old inline expansions) to become undefined. (Actually these
2815 changes were later undone, so we might very well be binary
2816 compatible with 0.7.7 after all, but leaving the version number
2817 incremented seemed like the simplest and most conservative
2820 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2821 * An alpha-quality port to the parisc architecture running Linux,
2822 based on the old CMUCL backend, has been made. This, even more so
2823 than the other backends, should be considered still a work in
2824 progress; known problems include that the Linux kernel in 64-bit
2825 mode does not propagate the correct sigcontext structure to
2826 userspace, and consequently SBCL on a parisc64 kernel will not
2828 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2829 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2830 suppressed.) Also now that the compiler is looking at declarations
2831 in the environment, it checks optimization declarations as well,
2832 and suppresses inlining when (> DEBUG SPEED).
2833 * More fixes have been made to treatment of floating point exception
2834 treatment and other Unix signals. In particular, floating point
2835 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2836 * The detection and handling of control stack exhaustion (infinite
2837 or very deeply nested recursion) has changed. Stack exhaustion
2838 detection is now done by write-protecting pages at the OS level
2839 and applies at all optimization settings; when found, a
2840 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2841 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2842 be caught using IGNORE-ERRORS.
2843 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2844 correctly when transforming recursive LABELS functions to LETs.
2845 (thanks to Alexey Dejneka)
2846 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2847 that are names of constants or global variables.
2848 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2849 alien routines with docstrings.
2850 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2851 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2853 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2854 (thanks to Eric Marsden)
2855 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2856 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2857 * Bug fix: The SPARC backend can now compile functions involving
2858 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2859 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2860 to a functions expecting a PATHNAME-DESIGNATOR.
2861 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2862 lambda lists. (thanks to Alexey Dejneka)
2863 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2864 functions so that EXPORTing the name of the class doesn't cause
2865 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2866 to Antonio Martinez for reporting this)
2867 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2868 constant-folded without failing an assertion. (thanks to Einar
2869 Floystad Dorum for reporting this)
2870 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2871 types are now dealt with more correctly by the compiler.
2872 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2873 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2874 behaviour analogous to RENAME-FILE. This puts its behaviour more
2875 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2877 * The fasl file version number has changed again. (because of the
2878 bug fix involving the names of PCL MAKE-INSTANCE functions)
2880 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2881 * bug fix: Floating point exceptions are treated much more
2882 consistently on the x86/Linux and PPC/Linux platforms.
2883 * Array initialization with :INITIAL-ELEMENT is now much faster for
2884 cases when the compiler cannot open code the array creation, but
2885 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2886 array accesses have also seen a speed increase.
2887 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2888 specified. (This was at the root of some bad interactions between
2889 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2891 * bug fix: Internal error arguments for undefined functions are now
2892 computed correctly on the PPC/Linux platform.
2893 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2894 Raymond Toy's patch for CMU CL)
2895 * Support for the Solaris 9 operating environment has been included
2896 (thanks to Daniel Merritt)
2897 * A very ugly but hopefully complete draft of the missing FFI chapter
2898 of the manual has been created by reformatting the corresponding
2899 CMU CL manual chapter into (currently very ugly and incoherent)
2900 DocBook and bringing it up to date for SBCL behavior. Thus, the
2901 manual is now essentially complete, at least by my extreme
2902 once-and-only-once standards, whereby it's acceptable to refer to
2903 the doc strings of SB-EXT functions as the primary documentation.
2904 * The fasl file version number has changed again, due to cleanup of
2905 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2906 is no longer a static symbol.)
2908 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2909 * SBCL now builds with OpenMCL (version 0.12) as the
2910 cross-compilation host; also, more progress has been made toward
2911 bootstrapping under CLISP.
2912 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2914 * bug 158 fixed: The compiler can now deal with integer loop
2915 increments different from 1; fixing this turned out also to fix
2917 * bug 169 fixed: no more bogus warnings about using lexical bindings
2918 despite the presence of perfectly good SPECIAL declarations (thanks
2919 to David Lichteblau)
2920 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2921 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2922 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2924 * bug 180 fixed: Method combination specifications no longer ignore
2925 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2926 * bug fix: Structure type predicate functions now check their argument
2927 count as they should.
2928 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2929 correctly. (thanks to Pierre Mai)
2930 * minor incompatible change: The --noprogrammer option is deprecated
2931 in favor of the new --disable-debugger option, which is very similar.
2932 (The major difference is that it takes effect at a slightly different
2933 time at startup, causing handling of errors in --sysinit and
2934 --userinit files will be affected differently.) The
2935 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2936 been added to allow this functionality to be controlled from ordinary
2937 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2938 maintainers, who might want to run non-interactive scripts to
2939 build SBCL cores which will later be used interactively.)
2940 * minor incompatible change: The LOAD function no longer, when given
2941 a wild pathname to load, loads all files matching that pathname.
2942 Instead, an error of type FILE-ERROR is signalled.
2944 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2945 * bug 147 fixed: The compiler preserves its block link/count
2946 invariants more correctly now so that it doesn't crash. (thanks
2948 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2950 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2951 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2952 no-value-for-this-character as NIL (as specified by ANSI).
2953 * HOST-NAMESTRING on physical pathnames now returns a string that is
2954 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2955 (thanks to Christophe Rhodes)
2956 * The Alpha port handles icache flushing more correctly. (thanks to
2958 * More progress has been made toward bootstrapping under CLISP. (thanks
2959 to Christophe Rhodes)
2960 * The fasl file format has changed again, because dynamic loading
2961 on OpenBSD (which has non-ELF object files) motivated some cleanups
2962 in the way that foreign symbols are transformed and passed around.
2963 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2964 is no longer treated as whitespace by the reader, but instead as
2965 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2966 |A
\7fB|, instead of A as it used to.
2968 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2969 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2970 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2971 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2972 this as of 0.7.1.45, but was left out of the previous news section
2973 (thanks to Dan Barlow)
2974 * SBCL now runs on the Solaris operating system on SPARC architectures
2975 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2976 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2977 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2978 * SPARC backend cleanups, allowing builds of cores optimized for V8
2979 and V9 SPARCS, and also emission of code targeted to a particular
2980 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2982 * SBCL is closer to bootstrapping under CLISP, thanks to various
2983 fixes by Christophe Rhodes.
2984 * The fasl file format has changed again, to allow the compiler's
2985 INFO database to support symbol macros.
2986 * The user manual (in doc/) is formatted into HTML more nicely.
2987 (thanks to coreythomas)
2988 * The system is smarter about SUBTYPEP relationships, especially
2989 those involving NOT types (including types such as ATOM which are
2990 represented internally using NOT types). Thus SUBTYPEP is less
2991 likely to return (VALUES NIL NIL) in general, and in particular
2992 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2994 * The fasl file format has changed again, because the internal
2995 representation of types now includes a new slot to support the new
2996 SUBTYPEP-of-NOT-types logic.
2997 * (not a change in the main branch of SBCL, but a related prototype
2998 which can hopefully be merged into the main branch of SBCL in the
2999 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3000 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3001 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3002 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3003 dispatch macro character combinations. (thanks to Alexey Dejneka)
3004 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3005 that superficially look like logical namestrings correctly.
3006 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3008 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3009 default supplied-p) for &optional and &key arguments. (thanks to
3012 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3013 * incompatible change: The compiler is now less aggressive about
3014 tail call optimization, doing it only when (> SPACE DEBUG) or
3015 (> SPEED DEBUG). (This is an incompatible change because there are
3016 programs which relied on the old CMU-CL-style behavior to optimize
3017 away their unbounded recursion which will now die of stack overflow.)
3018 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3019 for non-GENCGC systems has been increased to 20M (since that
3020 seems much closer to the likely performance optimum for modern
3021 systems than the old 4M value was)
3022 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3023 and *DEBUG-PRINT-LEVEL*
3024 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3025 of CMU CL's support for SPARC, and various endianness and other
3026 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3027 * new syntactic sugar for the Unix command line: --load foo.bar is now
3028 an alternate notation for --eval '(load "foo.bar")'.
3030 ** The system now detects stack overflow and handles it gracefully,
3031 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3032 optimization settings. (This is a good thing in general, and
3033 its introduction in this version should be particularly timely
3034 for anyone whose code fails because of suppression of tail
3036 ** The system now hunts for the C variable "environ" in a more
3037 devious way, to avoid segfaults when the C library version differs
3038 between compile time and run time. (thanks to Christophe Rhodes)
3039 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3040 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3041 ** The compiler no longer issues bogus style warnings for undefined
3042 classes in the same source file as the DEFCLASSes which defined
3043 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3045 ** fixes in CONDITION class precedence list for undefined function
3046 errors (thanks to Alexei Dejneka)
3047 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3048 correctly. (thanks to Dan Barlow)
3049 ** portability fixes aiming at bootstrapping under CLISP (thanks
3050 to Dave McDonald and Christophe Rhodes)
3051 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3052 ** fixes in type translation and and type inference (thanks to
3054 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3055 ** various fixes in the new ports (thanks to Dan Barlow)
3056 * several changes related to debugging:
3057 ** suppression of tail recursion, as noted above
3058 ** stack overflow detection, as noted above
3059 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3060 is now the default. (For some time encapsulation has been more
3061 reliable than the breakpoint-based :ENCAPSULATE NIL
3062 implementation, at least on X86 systems; and I just noticed that
3063 encapsulation also seems closer to the spirit of the ANSI
3066 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3068 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3069 up properly again. (There was a packaging bug in 0.7.0 which
3070 left their definitions in SB-SYS::LOAD-FOREIGN and
3071 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3072 vital for most things which interface to C-level interfaces,
3073 like extensions working with sockets or databases or
3074 Perl-compatible regexes or whatever, and the need to fix
3075 this bug is the main reason that 0.7.1 was released so
3077 ** DEFGENERIC is now choosier about the methods it redefines, so that
3078 reLOADing a previously-LOADed file containing DEFGENERICs does
3079 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3080 cycle now works as it should. (thanks to Alexey Dejneka)
3081 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3082 fixed by Christophe Rhodes. (He actually submitted this patch
3083 months ago, and I delayed until after 0.7.0.)
3084 ** Bug 111 (internal compiler confusion about runtime checks on
3085 FUNCTION types) was fixed by Alexey Dejneka.
3086 * Some internal cleanups (getting rid of variables which aren't
3087 needed now that the byte interpreter is gone) caused the fasl
3088 file format number to change again.
3090 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3091 * major incompatible change: The default fasl file extension, i.e. the
3092 default extension for files produced by COMPILE-FILE, has changed
3093 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3095 ** There are many changes in the implementation of the compiler.
3096 SBCL is now essentially a compiler-only implementation of ANSI
3097 Common Lisp. EVAL still "interprets" a few special cases, but
3098 almost all the interesting cases are handled by creating
3099 a LAMBDA expression, calling COMPILE on it, then calling
3100 FUNCALL on the result.
3101 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3102 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3103 Since the code is newer, there might still be some new bugs
3104 (though not as many as before Martin Atzmueller's fixes:-). But
3105 the new code is substantially simpler and clearer, and hopefully
3106 any remaining bugs will be simpler, less fundamental, and more
3107 fixable then the bugs in the old code.
3108 ** The revised compiler is still a little unsteady on its feet.
3110 *** The debugging information it produces (particularly the names
3111 of FUNCTION objects) is sometimes much less useful than what
3112 the old compiler produced.
3113 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3114 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3115 within a MACROLET) has been temporarily weakened.
3116 ** There are new compiler optimizations for various functions:
3117 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3118 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3119 *** the math functions TRUNCATE, FLOOR, and CEILING
3120 *** the function-of-all-trades COERCE
3121 Mostly these should be transparent, but there's one
3122 potentially-annoying problem (bug 117): when the compiler
3123 inline-expands a function and does type analysis on the result,
3124 it can create control paths which have type mismatches, and
3125 when it can't prove that those control paths aren't taken,
3126 it will issue WARNINGs about the type mismatches. This is
3127 a particular problem in practice for the new sequence functions.
3128 It's not clear how this should be fixed, and for now, a
3129 workaround is given in the entry for 117 in the BUGS file.
3130 ** (Because of the interaction between the two previous items --
3131 occasional inlining problems and new inline expansions -- some
3132 of the new sequence function optimizations won't really kick in
3133 completely until debugging information, and then inlining, are
3134 straightened out in some future version.)
3135 * minor incompatible changes:
3136 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3137 DIRECTORY no longer implicitly promotes NIL slots of its
3138 pathname argument to :WILD. In particular, when you ask for the
3139 contents of a directory (which you used to be able to do without
3140 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3141 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3142 ** changes in behavior that ANSI explicitly defines to be
3143 implementation dependent:
3144 *** The new compiler-only implementation still conforms with ANSI,
3145 but acts a little different than before. Besides the obvious
3146 changes in performance tradeoffs (that the cost per form passed
3147 to EVAL has gone up, and the cost per form executed by EVAL
3148 has gone down), the behavior of the system changes a little
3149 because there are no longer any interpreted function objects.
3150 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3151 e.g. doing COMPILE on the output of interactive DEFUN is
3153 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3155 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3156 has changed from (:CL) to NIL.
3157 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3158 e.g. #\Newline and #\Tab, don't have names specified in the
3159 ANSI Common Lisp standard, is now based on their ASCII symbolic
3160 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3161 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3162 are no longer used for output.
3163 ** changes in internal implementation constants:
3164 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3165 4 million. (If your application spends a lot of time GCing and
3166 you have a lot of RAM, you might want to experiment with
3167 increasing it even more.)
3168 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3169 However, almost all old code should still continue to work without
3170 immediate update, as SB-C-CALL is now a (deprecated) nickname
3172 ** Old operator names in the style DEF-FOO are now deprecated in
3173 favor of new corresponding names DEFINE-FOO, for consistency with
3174 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3175 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3176 internal symbols, but a few supported extensions like
3177 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3178 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3179 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3180 etc. as you get deeper into recursive calls to the debugger
3181 command loop, instead of the old "5]", "5]]", "5]]]"
3182 sequence. (I was motivated to do this when squabbles between
3183 ILISP and SBCL left me very deeply nested in the debugger. In the
3184 short term, this change will probably provoke more ILISP/SBCL
3185 squabbles, but hopefully it will be an improvement in the long run.)
3186 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3187 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3188 function, since declaiming return types involving aliens is
3189 (1) annoyingly messy to do by hand and (2) vital to efficient
3190 compilation of code which calls such functions.
3191 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3192 longer reexported by the SB-EXT package. They're solely useful
3193 for alien code, so it seems more logical that you should get
3194 them from the SB-ALIEN package, not in SB-EXT.
3195 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3196 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3197 features. Instead, the code that they used to control is always
3198 built into the system.
3199 * many other bug fixes
3200 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3201 advantage of the new EVAL-WHEN stuff and to clean them up in
3202 general, and they are now more ANSI-compliant in a number of
3203 ways. Martin Atzmueller is responsible for a lot of this.
3204 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3206 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3207 *** correct ERROR type for various file operations
3208 *** some fixes for Lisp streams
3209 *** DEFMETHOD syntax checking
3210 *** changing old weird representation of debug information as
3211 strings (which, among their other deficiencies, don't transform
3212 correctly when you rename packages, and don't change their
3213 print representation when you change things like *PACKAGE*
3214 and *PRINT-LENGTH*) to symbols and lists of symbols
3215 He also made several improvements and fixed several bugs in DESCRIBE.
3216 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3218 *** misbehavior of WRITE-STRING/WRITE-LINE
3219 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3220 and several other LOOP problems as well
3221 *** DIRECTORY when similar filenames are present
3222 *** DEFGENERIC with :METHOD options
3223 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3224 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3225 *** argument ordering in FIND with :TEST option
3226 *** mishandled package designator argument in APROPOS-LIST
3227 *** various problems in the backquote readmacro
3228 *** a bug in APROPOS
3229 *** probably some others that I'm not describing very well here,
3230 since the CVS log documents them by reference to sbcl-devel
3231 messages, and the SourceForge archives aren't working well.:-(
3232 ** Dan Barlow improved the Alpha port (and is making progress on the
3233 PPC port, for those of you who think different).
3234 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3235 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3236 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3237 continues to work on the SPARC port (for those of you in a position
3238 to look down upon our little PC-compatible boxes from a great height).
3239 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3240 and uses that copy, rather than the current dynamic value, when
3241 it's trying to decide whether to truncate output. Thus e.g.
3242 (let ((*print-lines* 50))
3243 (pprint-logical-block (stream nil)
3245 (let ((*print-lines* 8))
3246 (print (aref possiblybigthings i) stream)))))
3247 should now truncate the logical block only at 50 lines, instead of
3248 often truncating it at 8 lines, as it did before.
3249 * The doc/cmucl/ directory, containing old CMU CL documentation
3250 from the time of the fork, is no longer part of the base system.
3251 SourceForge has shut down its anonymous FTP service, and with it
3252 my original plan for distributing the old CMU CL documentation
3253 there. For now, if you need these files you can download an old
3254 SBCL source release and extract them from it.
3255 * The fasl file version number changed again, for dozens of reasons,
3256 some of which are apparent above.
3258 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3259 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3260 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3261 patch, so that DISASSEMBLE output is much nicer.
3262 * The code in the SB-PROFILE package now seems reasonably stable.
3263 I still haven't decided what the final interface should look like
3264 (I'd like PROFILE to interact cleanly with TRACE, since both
3265 facilities use function encapsulation) but if you have a need
3266 for profiling now, you can probably use it successfully with
3267 the current CMU-CL-style interface.
3268 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3269 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3270 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3271 link now returns the dangling link itself, and for similar
3272 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3273 link itself. (In these cases the old code signalled an error and
3274 looped endlessly, respectively.) Thus, DIRECTORY now works even
3275 in the presence of dangling and cyclic symbolic links.
3276 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3277 is now a supported extension again, since the consensus on
3278 sbcl-devel was that it can be useful for ordinary development
3279 work, not just for debugging SBCL itself.
3280 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3281 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3282 that currently-defined functions might be redefined later with
3283 different return types.
3284 * Hash tables can be printed readably, as inspired by CMU CL code
3285 of Eric Marsden and SBCL code of Martin Atzmueller.
3286 * better error handling in CLOS method combination, thanks to
3287 Martin Atzmueller porting Pierre Mai's CMU CL patches
3288 * more overflow fixes for >16Mbyte I/O buffers
3289 * A bug in READ has been fixed, so that now a single Ctrl-D
3290 character suffices to cause end-of-file on character streams.
3291 In particular, now you only need one Ctrl-D at the command
3292 line (not two) to exit SBCL.
3293 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3295 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3296 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3297 does the right thing.
3298 * The implementation of some type tests, especially for CONDITION
3299 types, is now tidier and maybe faster, due to CMU CL code
3300 originally by Douglas Crosher, ported by Martin Atzmueller.
3301 * Some math functions have been fixed, and there are new
3302 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3303 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3304 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3305 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3306 likely to be built into sbcl-0.7.0.)
3307 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3308 CL, ported to SBCL by Martin Atzmueller)
3309 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3310 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3311 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3312 * There's a new slam.sh hack to shorten the edit/compile/debug
3313 cycle for low-level changes to SBCL itself, and a new
3314 :SB-AFTER-XC-CORE target feature to control the generation of
3315 the after-xc.core file needed by slam.sh.
3316 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3317 COMPILE-FILE is no longer supported, so that now every function
3318 gets an entry point, so that block compilation looks a little
3319 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3320 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3321 returns the number of bytes consed since the system started,
3322 rather than the number consed since the first time the function
3323 was called. (The new definition parallels ANSI functions like
3324 CL:GET-INTERNAL-RUN-TIME.)
3325 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3326 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3327 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3328 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3329 * Fasl file version numbers are now independent of the target CPU,
3330 since historically most system changes which required version
3331 number changes have affected all CPUs equally. Similarly,
3332 the byte fasl file version is now equal to the ordinary
3335 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3336 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3337 is no longer recognized. I apologize for this, because it was
3338 listed in SB-EXT as a supported extension, but I found that
3339 its existing behavior was poorly specified, as well as incorrectly
3340 specified, and it looked like too much of a mess to straighten it
3341 out. I have enough on my hands trying to get ANSI stuff to work..
3342 * many patches ported from CMU CL by Martin Atzmueller, with
3343 half a dozen bug fixes in pretty-printing and the debugger, and
3344 half a dozen others elsewhere
3345 * fixed bug 13: Floating point infinities are now supported again.
3346 They might still be a little bit flaky, but thanks to bug reports
3347 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3348 as flaky as they were.
3349 * The --noprogrammer command line option is now supported. (Its
3350 behavior is slightly different in detail from what the old man
3351 page claimed it would do, but it's still appropriate under the
3352 same circumstances that the man page talks about.)
3353 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3354 are now supported, and enabled by default. Thus, the compiler can
3355 handle many floating point and complex operations much less
3356 inefficiently. (Thus e.g. you can implement a complex FFT
3358 * The compiler now detects type mismatches between DECLAIM FTYPE
3359 and DEFUN better, and implements CHECK-TYPE more correctly, and
3360 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3361 thanks to patches from Martin Atzmueller.
3362 * various fixes to make the cross-compiler more portable to
3363 ANSI-conforming-but-different cross-compilation hosts (notably
3364 Lispworks for Windows, following bug reports from Arthur Lemmens)
3365 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3366 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3367 fixed thanks to Pierre Mai's CMU CL patches.
3368 * fixes to keep the system from overflowing internal counters when
3369 it tries to use i/o buffers larger than 16M bytes
3370 * fixed bug 45a: Various internal functions required to support
3371 complex special functions have been merged from CMU CL sources.
3372 (When I was first setting up SBCL, I misunderstood a compile-time
3373 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3374 * improved support for type intersection and union, fixing bug 12
3375 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3376 more obscure bugs as well
3377 * some steps toward byte-compiling non-performance-critical
3378 parts of the system, courtesy of patches from Martin Atzmueller
3379 * Christophe Rhodes has made some debian packages of sbcl at
3380 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3381 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3382 stable, but are nonetheless usable. When he's ready, I'd be happy
3383 to add them to the SourceForge "File Releases" section. (And if
3384 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3385 * new fasl file format version number (because of changes in
3386 internal representation of (OR ..) types to accommodate the new
3387 support for (AND ..) types, among other things)
3389 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3390 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3392 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3394 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3395 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3396 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3397 * DESCRIBE now works on structure objects again.
3398 * Most function call argument type mismatches are now handled as
3399 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3400 know whether the function will be redefined before the call is
3401 executed. (The compiler could flag local calls with full WARNINGs,
3402 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3403 it doesn't keep track of enough information to know whether calls
3404 are local in this sense.)
3405 * Compiler output is now more verbose, with messages truncated
3406 later than before. (There should be some supported way for users
3407 to override the default verbosity, but I haven't decided how to
3408 provide it yet, so this behavior is still controlled by the internal
3409 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3410 src/compiler/ir1util.lisp.)
3411 * Fasl file format version numbers have increased again, because
3412 support for the Gray streams extension changes the layout of the
3413 system's STREAM objects.
3414 * The Gray subclassable streams extension now works, thanks to a
3415 patch from Martin Atzmueller.
3416 * The full LOAD-FOREIGN extension (not just the primitive
3417 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3418 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3419 but like most other programs, it defaults to copying the Unix
3420 environment from the original process instead of starting the
3421 new process in an empty environment.
3422 * Extensions which manipulate the Unix environment now support
3423 an :ENVIRONMENT keyword option which doesn't smash case or
3424 do other bad things. The CMU-CL-style :ENV option is retained
3425 for porting convenience.
3426 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3427 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3429 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3431 * A patch from Martin Atzmueller seems to have solved the SIGINT
3432 problem, and as far as we know, signal-handling now works cleanly.
3433 (If you find any new bugs, please report them!)
3434 * The system no longer defaults Lisp source file names to types
3435 ".l", ".cl", or ".lsp", but only to ".lisp".
3436 * The compiler no longer uses special default file extensions for
3437 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3438 seems to expect a single default extension for all compiled code,
3439 and there's no compelling reason to try to stretch the standard
3440 to allow two different extensions.) Instead, byte-compiled files
3441 default to the same extension as native-compiled files.
3442 * Fasl file format version numbers have increased again, because
3443 a rearrangement of internal implementation packages made some
3444 dumped symbols in old fasl files unreadable in new cores.
3445 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3446 many fewer weird special cases.
3447 * Bug #17 (differing COMPILE-FILE behavior between logical and
3448 physical pathnames) has been fixed, and some related misbehavior too,
3449 thanks to a patch from Martin Atzmueller.
3450 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3451 by Tim Moore, ported to SBCL by Martin Atzmueller.
3452 * Martin Atzmueller fixed several filesystem-related problems,
3453 including bug #36, in part by porting CMU CL patches, which were
3454 written in part by Paul Werkowski.
3455 * More compiler warnings in src/runtime/ are gone, thanks to
3456 more patches from Martin Atzmueller.
3457 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3460 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3462 * DESCRIBE now works on CONDITION objects.
3463 * The debugger now handles errors which arise when trying to print
3464 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3466 * The build system now uses an additional file, customize-target-features.lisp,
3467 to allow local modifications to the target *FEATURES* list. (The point of
3468 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3469 debugging features enabled, without having to worry about propagating it
3470 into everyone's system when I do a "cvs update".) When no
3471 customize-target-features.lisp file exists, the target *FEATURES* list
3472 should be constructed the same way as before.
3473 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3474 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3475 warn about behavior which is undefined under the ANSI spec. Note: This
3476 is specified by ANSI, but it's not very popular with programmers.
3477 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3478 macro in the SBCL sources for an example of a workaround which you
3479 might use to make portable ANSI-standard code which does what you want.
3480 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3481 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3482 when it's used as a non-toplevel form. (This is required in order
3483 to implement the DEFCONSTANT-EQX macro.)
3484 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3485 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3486 funny things, due to bugs in the implementation of EVAL-WHEN
3487 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3488 non-toplevel DEFCONSTANTs are uncommon.)
3489 * The core file version number and fasl file version number have been
3490 incremented, because the old noncompliant DEFCONSTANT behavior involved
3491 calling functions which no longer exist, and because I also took the
3492 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3493 * fixed bug 1 (error handling before read-eval-print loop starts), and
3494 redid debugger restarts and related debugger commands somewhat while
3496 ** The QUIT debugger command is gone, since it did something
3497 rather different than the SB-EXT:QUIT command, and since it never
3498 worked properly outside the main toplevel read/eval/print loop.
3499 Invoking the new TOPLEVEL restart provides the same functionality.
3500 ** The GO debugger command is also gone, since you can just invoke
3501 the CONTINUE restart directly instead.
3502 ** The TOP debugger command is also gone, since it's redundant with the
3503 FRAME 0 command, and since it interfered with abbreviations for the
3505 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3507 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3508 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3509 some porting work by Martin Atzmueller)
3510 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3511 with a boatload of patches which he ported to SBCL. Now that those
3513 ** The system tries to make sure that its low-priority messages
3514 are prefixed by semicolons, to help people who like to use
3515 syntax highlighting in their ILISP buffer. (This patch
3516 was originally due to Raymond Toy.)
3517 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3518 patches originally written by Raymond Toy.
3519 ** The compiler understands coercion between single-value and
3520 multiple-VALUES type expressions better, getting rid of some very
3521 weird behavior, thanks to patches originally by Robert MacLachlan
3522 and Douglas Crosher.
3523 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3524 lambda lists, thanks to a patch originally by Pierre Mai.
3525 ** The system no longer bogusly warns about "abbreviated type
3527 ** The compiler gets less confused by inlining and RETURN-FROM,
3528 thanks to some patches originally by Tim Moore.
3529 ** The system no longer hangs when dumping circular lists to fasl
3530 files, thanks to a patch originally from Douglas Crosher.
3531 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3532 undefined function error.
3533 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3534 in structure slot initforms) for now, documented workaround instead:-|
3535 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3536 * fixed bug 5: added stubs for various Gray stream functions called
3537 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3538 installed, at least appropriate type errors are generated
3539 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3540 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3541 consistently in DEFMETHOD forms.
3542 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3543 it doesn't seem to affect SBCL after all
3544 * The C runtime system now builds with better optimization and many
3545 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3547 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3549 * The system is now under CVS at SourceForge (instead of the
3550 CVS repository on my home machine).
3551 * The new signal handling code has been tweaked to treat register
3552 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3553 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3554 downstream, e.g. in debug-int.lisp, has implicit dependencies
3555 on the unsignedness of integer representation of machine words,
3556 and that caused the system to bomb out with infinite regress
3557 when trying to recover from type errors involving signed values,
3558 e.g. (BUTLAST '(1 2 3) -1).)
3559 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3560 (This was one of the bugs Peter Van Eynde reported back in July.)
3561 * The system now uses code inspired by Colin Walters' O(N)
3562 implementation of MAP (from the cmucl-imp@cons.org mailing
3563 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3564 inline the MAP operation, and there is more than one
3565 sequence argument to the MAP call (so that it can't just
3566 do ETYPECASE once and for all based on the type of the
3567 single sequence argument). (The old non-inline implementation
3568 of the general M-argument sequence-of-length-N case required
3569 O(M*N*N) time when any of the sequence arguments were LISTs.)
3570 * The QUIT :UNIX-CODE keyword argument has been renamed to
3571 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3572 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3574 * Raymond Wiker's patches to port dynamic loading from Linux to
3575 FreeBSD have been added.
3576 * The BUGS file is now more nearly up to date, thanks in large part
3577 to Martin Atzmueller's review of it.
3578 * The debugger now flushes standard output streams before it begins
3579 its output ("debugger invoked" and so forth).
3580 * The core version number and fasl file version number have both
3581 been incremented, because of incompatible changes in the layout
3583 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3584 used to not be called for a saved Lisp image.)
3585 * Martin Atzmueller's version of a patch to fix a compiler crash,
3586 as posted on sbcl-devel 13 September 2000, has been installed.
3587 * Instead of installing Martin Atzmueller's patch for the
3588 compiler transform for SUBSEQ, I deleted the compiler transform,
3589 and transforms for some similar consing operations.
3590 * A bug in signal handling which kept TRACE from working on OpenBSD
3592 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3593 away by constant folding
3594 * The system now defines its address space constants in one place
3595 (in the Lisp sources), and propagates them automatically elsewhere
3596 (through GENESIS and the sbcl.h file). Therefore, patching the
3597 address map is less unnecessarily tedious and error-prone. The
3598 Lisp names of address space constants have also been systematized.
3599 * CVS tags like dollar-Header-dollar have been removed from
3600 the sources, because they have never saved me trouble and
3601 they've been source of trouble working with patches and other
3602 diff-related operations.
3603 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3604 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3606 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3608 * The system has been ported to OpenBSD.
3609 * The system now compiles with a simple "sh make.sh" on the systems
3610 that it's supported on. I.e., now you no longer need to tweak
3611 text in the target-features.lisp-expr and symlinks in src/runtime/
3612 by hand, the make.sh takes care of it for you.
3613 * The system is no longer so grossly inefficient when compiling code
3614 involving vectors implemented as general (not simple) vectors (VECTOR T),
3615 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3616 which dares to use the various sequence functions on non-simple
3617 vectors, takes less of a performance hit.
3618 * There is now a primitive type predicate VECTOR-T-P
3619 to test for the (VECTOR T) type, so that e.g.
3620 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3621 can now be compiled with some semblance of efficiency. (The old code
3622 turned the type declaration into a full call to %TYPEP at runtime!)
3623 * AREF on (VECTOR T) is still not fast, since it's still compiled
3624 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3625 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3626 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3627 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3628 without making full calls to SUBTYPEP at runtime.
3629 (Some analogous efficiency issues for non-simple vectors specialized to
3630 element types other than T, or for non-simple multidimensional arrays,
3631 have not been addressed. They could almost certainly be handled the
3632 same way if anyone is motivated to do so.)
3633 * The changes in array handling break binary compatibility, so
3634 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3635 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3636 instead of (VALUES T T).
3637 * By following the instructions that Dan Barlow posted to sbcl-devel
3638 on 2 July 2000, I was able to enable primitive dynamic object
3639 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3640 functionality is not implemented (since it calls ld to resolve
3641 library references automatically, requiring RUN-PROGRAM for its
3642 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3643 try to resolve library references) is now supported.
3644 * The system now flushes the standard output streams when it terminates,
3645 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3646 them at several other probably-convenient times, e.g. in each pass of
3647 the toplevel read-eval-print loop, and after evaluating a form given
3648 as an "--eval" command-line option. (These changes were motivated by a
3649 discussion of stream flushing issues on cmucl-imp in August 2000.)
3650 * The source transform for TYPEP of array types no longer assumes
3651 that an array whose element type is a not-yet-defined type
3652 is implemented as an array of T, but instead punts, so that the
3653 type will be interpreted at runtime.
3654 * There is now some support for cross-compiling in make.sh: each of
3655 the phases of make.sh has its own script. (This should be transparent
3656 to people doing ordinary, non-cross-compile builds.)
3657 * Since my laptop doesn't have hundreds of megabytes of memory like
3658 my desktop machine, I became more motivated to do some items on
3659 my to-do list in order to reduce the size of the system a little:
3660 ** Arrange for various needed-only-at-cold-init things to be
3661 uninterned after cold init. To support this, those things have
3662 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3663 symbols with such names are now uninterned after cold init).
3664 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3665 fasl files for cold load.
3666 ** Remove the old compiler structure pooling code (which used to
3667 be conditional on the target feature :SB-ALLOC) completely.
3668 ** Redo the representation of some data in cold init to be more compact.
3669 (I also looked into supporting byte compiled code at bootstrap time,
3670 which would probably reduce the size of the system a lot, but that
3671 looked too complicated, so I punted for now.)
3672 * The maximum signal nesting depth in the src/runtime/ support code has
3673 been reduced from 4096 to 256. (I don't know any reason for the very
3674 large old value. If the new smaller value turns out to break something,
3675 I'll probably just bump it back up.)
3676 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3678 * Many, many bugs reported by Peter Van Eynde have been added to
3679 the BUGS list; some have even been fixed.
3680 * While enabling dynamic object file loading, I tried to make the
3681 code easier to understand, renaming various functions and variables
3682 with less ambiguous names, and changing some function calling
3683 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3684 * While trying to figure out how to do the OpenBSD port, I tried to
3685 clean up some of the code in src/runtime/. In particular, I dropped
3686 support for non-POSIX signal handling, added various comments,
3687 tweaked the code to reduce the number of compilation warnings, and
3688 renamed some files to increase consistency.
3689 * To support the new automatic configuration functionality in make.sh,
3690 the source file target-features.lisp-expr has been replaced with the
3691 source file base-target-features.lisp-expr and the machine-generated
3692 file local-target-features.lisp-expr.
3693 * fixed a stupid quoting error in make.sh so that using CMU CL
3694 "lisp -batch" as cross-compilation host works again
3696 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3698 * DESCRIBE no longer tries to call itself recursively to describe
3699 bound/fbound values, so that it no longer fails on symbols which are
3700 bound to themselves (like keywords, T, and NIL).
3701 * DESCRIBE now works on generic functions.
3702 * The printer now prints less-screwed-up representations of closures
3703 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3704 * A private symbol is used instead of the :EMPTY keyword previously
3705 used to mark empty slots in hash tables. Thus
3706 (DEFVAR *HT* (MAKE-HASH-TABLE))
3707 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3708 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3709 now does what ANSI says that it should. (You can still get
3710 similar noncompliant behavior if bang on the hash table
3711 implementation with all the symbols you get back from
3712 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3713 This breaks binary compatibility, since tests for equality to
3714 :EMPTY are wired into things like the macroexpansion of
3715 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3717 * There's now a minimal placeholder implementation for CL:STEP,
3718 as required by ANSI.
3719 * An obscure bug in the interaction of the normal compiler, the byte
3720 compiler, inlining, and structure predicates has been patched
3721 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3722 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3724 * Missing ordinary arguments in a macro call are now detected even
3725 when the macro lambda list contains &KEY or &REST.
3726 * The debugger no longer complains about encountering the top of the
3727 stack when you type "FRAME 0" to explicitly instruct it to go to
3728 the top of the stack. And it now prints the frame you request even
3729 if it's the current frame (instead of saying "You are here.").
3730 * As specified by ANSI, the system now always prints keywords
3731 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3733 * The default initial SIZE of HASH-TABLEs is now smaller.
3734 * Type information from CLOS class dispatch is now propagated
3735 into DEFMETHOD bodies, so that e.g.
3736 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3738 is now basically equivalent to
3739 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3740 (DECLARE (TYPE SINGLE-FLOAT X))
3742 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3743 operation, without having to do run-time type dispatch.
3744 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3745 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3746 * Since I don't seem to be making big file reorganizations very often
3747 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3748 is overflowing my ability to conveniently back them up), I've finally
3749 checked the system into CVS. (The CVS repository is on my home system,
3750 not at SourceForge -- putting it on SourceForge might come later.)
3751 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3752 high-level GC-NOTIFY-FOO functions send their output. (There's
3753 still very little control of where low-level verbose GC functions
3754 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3755 less than it used to -- the GC-NOTIFY-FOO functions are now under
3756 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3757 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3758 in only one place in the source code, and propagates it automatically
3759 everywhere that it's needed. Thus e.g. when I bump the version from
3760 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3761 * The C source files now include boilerplate legalese and documentation
3762 at the head of each file (just as the Lisp source files already did).
3763 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3764 to his page will be replaced with a link to his new CLiki service.
3766 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3768 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3769 * The build process now looks for GNU make under the default name "gmake",
3770 instead of "make" as it used to. If GNU make is not available as "gmake"
3771 on your system, you can change this default behavior by setting the
3772 GNUMAKE environment variable.
3773 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3774 can build without error under CMU CL.
3776 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3778 * There is now a partial SBCL user manual (with some new text and some
3779 text cribbed from the CMU CL manual).
3780 * The beginnings of a profiler have been added (starting with the
3781 CMU CL profiler and simplifying and cleaning up). Eventually the
3782 main interface should be through the TRACE macro, but for now,
3783 it's still accessed through vaguely CMU-CL-style functions and macros
3784 exported from the package SB-PROFILE.
3785 * Some problems left over from porting CMU CL to the new
3786 cross-compilation bootstrap process have been cleaned up:
3787 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3788 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3789 part when getting rid of the compiler *BACKEND* stuff.)
3790 ** The value of *NULL-TYPE* was screwed up, because it was
3791 being initialized before the type system knew the final
3792 definition of the 'NULL type. This screwed up several key
3793 optimizations in the compiler, causing inefficiency in all sorts
3794 of places. (I found it because I wanted to understand why
3795 GET-INTERNAL-RUN-TIME was consing.)
3796 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3797 PROCLAIM FTYPE information. Unfortunately this broke binary
3798 compatibility again, since now the forms output by DEFGENERIC
3799 to refer to functions which didn't exist in 0.6.3.
3800 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3801 can use the new (as of 0.6.3) transform for SOME into MAP into
3803 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3804 values to (INTEGER 0 1000000), so that the time code will no longer
3805 occasionally get blown up by Linux returning 1000000 microseconds
3806 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3807 its output conform to the ANSI spec. (Alas, this makes its output
3808 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3809 * A full call to MAP NIL with a single sequence argument no longer conses.
3810 * fixes to problems pointed out by Martin Atzmueller:
3811 * The manual page no longer talks about multiprocessing as though
3812 it were currently supported.
3813 * The ILISP support patches have been removed from the distribution,
3814 because as of version 5.10.1, ILISP now supports SBCL without us
3815 having to maintain patches.
3816 * added a modified version of Raymond Toy's recent CMU CL patch for
3817 EQUALP comparison of HASH-TABLE
3819 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3821 * The system still can't cross-compile itself with
3822 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3823 entails), but at least it can compile more of itself that way
3824 than it used to be able to, and various buglets which were uncovered
3825 by trying to cross-compile itself that way have now been fixed.
3826 * This release breaks binary compatibility again. This time
3827 at least I've incremented the FASL file format version to 2, so that the
3828 problem can be detected reliably instead of just causing weird errors.
3829 * various new style warnings:
3830 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3831 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3832 ** using the lexical binding of a variable named in the *FOO* style
3833 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3834 as specified by ANSI.
3835 * *RANDOM-STATE* is no longer automatically initialized from
3836 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3837 default behavior of the system is to repeat its behavior every time
3838 it's run. If you'd like to change this behavior, you can always
3839 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3840 old convention there was no comparably easy way to get the system to
3841 repeat its behavior every time it was run.
3842 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3843 FTYPE declarations has been removed, in favor of their ANSI
3844 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3845 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3846 the types of their sequence arguments can be determined at compile time.
3847 This is done through a new open code expansion for MAP which eliminates
3848 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3849 argument types can be determined at compile time.
3850 * The optimizer now transforms COERCE into an identity operation when it
3851 can prove that the coerced object is already of the correct type. (This
3852 can be a win for machine generated code, including the output of other
3853 optimization transforms, such as the MAP transform above.)
3854 * Credit information has been moved from source file headers into CREDITS.
3855 * Source file headers have been made more standard.
3856 * The CASE macro now compiles without complaining even when it has
3859 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3861 * (Note that the way that the PCL macroexpansions were rewritten
3862 to accommodate the change in DEFGENERIC below breaks binary
3863 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3864 not run under sbcl-0.6.2. Once we get out of alpha releases,
3865 i.e. hit release 1.0.0, we'll probably try to maintain binary
3866 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3867 and sbcl-1.4.4. Until then, however, it might be fairly common
3868 for maintenance releases to break binary compatibility.)
3869 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3871 * The reporting of unhandled signals has been changed to print some
3872 explanatory text as well as the report form. (Previously only
3873 the report form was printed.)
3874 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3875 it defines, so that the compiler no longer issues undefined function
3876 warnings for compiled-but-not-yet-loaded generic functions.
3877 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3878 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3879 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3880 section of the ANSI spec are supported.
3881 * The "" nickname for the "KEYWORD" package has been removed.
3882 The reader still handles symbol tokens which begin with a package marker
3883 as keywords, but it doesn't expose its mechanism for doing so in the
3884 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3885 * The system now issues STYLE-WARNINGs for contradictory TYPE
3886 proclamations. (Warnings for contradictory FTYPE proclamations would
3887 be nice too, but those can't be done usefully unless the type system
3888 is made smarter about FUNCTION types.)
3889 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3890 systematized, so that "*target-*.lisp is supposed to exist only on the
3891 target and imply that there's a related file which exists on the
3892 host, and *host-*.lisp is supposed to exist only on the host and imply
3893 that there's a related file which exists on the target. This involves a
3894 lot of renaming. Hopefully the acute confusion caused by the renaming
3895 will be justified by the reduction in chronic confusion..
3896 ** runtime-type.lisp -> early-target-type.lisp
3897 ** target-type.lisp -> late-target-type.lisp
3898 ** early-host-format.lisp -> early-format.lisp
3899 ** late-host-format.lisp -> late-format.lisp
3900 ** host-error.lisp -> misc-error.lisp
3901 ** early-error.lisp -> early-target-error.lisp
3902 ** late-error.lisp -> late-target-error.lisp
3903 ** host-defboot.lisp -> early-defboot.lisp
3904 ** code/misc.lisp -> code/target-misc.lisp
3905 ** code/host-misc.lisp -> code/misc.lisp
3906 ** code/numbers.lisp -> code/target-numbers.lisp
3907 ** code/early-numbers.lisp -> numbers.lisp
3908 ** early-host-type.lisp -> early-type.lisp
3909 ** late-host-type.lisp -> late-type.lisp
3910 ** host-typep.lisp -> typep.lisp
3911 ** load.lisp -> target-load.lisp
3912 ** host-load.lisp -> load.lisp
3913 ** host-disassem.lisp -> disassem.lisp
3914 ** host-insts.lisp -> insts.lisp
3915 ** byte-comp.lisp -> target-byte-comp.lisp
3916 ** host-byte-comp.lisp -> byte-comp.lisp
3917 ** host-signal.lisp -> signal.lisp
3918 ** host-defstruct.lisp -> defstruct.lisp
3919 ** late-target-type.lisp -> deftypes-for-target.lisp
3920 Furthermore, several other previously target-only files foo.lisp (e.g.
3921 hash-table.lisp and random.lisp) have been split into a target-and-host
3922 foo.lisp file and a target-only target-foo.lisp file, with their key type
3923 definitions in the target-and-host part, so that the cross-compiler will
3924 know more about target types.
3925 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3926 gone away. In their place are various *BACKEND-FOO* variables
3927 corresponding to the slots of the old structure.
3928 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3929 into the target SBCL has been fixed.
3930 * The chill.lisp system for loading cold code into a running SBCL
3932 * Support for the CMU CL "scavenger hook" extension has been removed.
3933 (It was undocumented and unused in the CMU CL sources that SBCL was
3934 derived from, and stale in sbcl-0.6.1.)
3935 * Various errors in the cross-compiler type system were detected
3936 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3937 (enabling various consistency checks). Many of them were fixed,
3938 but some hard problems remain, so the compiler is back to
3939 running without *TYPE-SYSTEM-INITIALIZED* for now.
3940 * As part of the cross-compiler type system cleanup, I implemented
3941 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3942 * I have started adding UNCROSS calls throughout the type system
3943 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3944 on cross-compiler input files will be able to go away, and various
3946 * CONSTANTP now returns true for quoted forms (as explicitly required
3949 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3951 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3952 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3953 (MAKE-LIST -1). In the longer term, it ought to become true
3954 that declarations are assertions even at SAFETY 1. For now, it's not
3955 quite true even at SAFETY 3, but it's at least more nearly true..
3956 (Note that this change seems to increases the size of the system by
3957 O(5%) and to decrease the speed of the compiler by 20% or more.)
3958 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3959 for the problem of printing dozens of lines of distracting information
3960 about low-level system machinery as part of the top stack frame
3961 on entry to the debugger when an undefined function was called.
3962 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3963 is not reset to COMMON-LISP-USER.
3964 * Compilation of stuff related to dyncount.lisp has been made conditional
3965 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3966 smaller. The various dyncount-related symbols have been moved into
3967 a new "SB-DYNCOUNT" package.
3968 * tty-inspect.lisp has been renamed to inspect.lisp.
3969 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3970 feature has gone away. (When we eventually port to other flavors of
3971 libc and/or Unix, we'll try to make the differences between flavors
3972 invisible at the user level.)
3973 * Various other *FEATURES* tags, and/or their associated conditionals,
3974 have been removed if obsolescent, or given better documentation, or
3975 sometimes given more-mnemonic names.
3977 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3979 * tidied up "make.sh" script
3980 * tidied up system directory structure
3981 * better "clean.sh" behavior
3982 * added doc/FOR-CMUCL-DEVELOPERS
3983 * many many small tweaks to output format, e.g. removing possibly-confusing
3984 trailing #\. character in DESCRIBE-INSTANCE
3985 * (EQUALP #\A 'A) no longer signals an error.
3986 * new hashing code, including EQUALP hashing
3987 * tidied up Lisp initialization and toplevel
3988 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3989 * command line argument processing
3990 * added POSIX-GETENV function to deal with Unix-ish environment variables
3991 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3992 terminating SBCL on EOF
3993 * non-verbose GC by default
3994 * There is no more "sbcl" shell script; the sbcl file is now the C
3995 runtime executable (just like CMU CL).
3996 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3998 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3999 symbol and package information as Lisp native symbol and package objects
4000 instead of strings naming symbols and strings naming packages. This way,
4001 whenever packages are renamed (as in warm init), debug information is
4002 transformed along with everything else.
4003 * tweaked the optimization policy declarations which control the building
4004 of SBCL itself. Now, among other things, the system no longer saves
4005 source location debugging information. (This helps two problems at once
4006 by reducing SBCL size and by keeping SBCL from trying to look for its
4007 sources -- which may not exist -- when reporting errors.)
4008 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4009 debugging and testing purposes
4010 * cleaned up printing, making the printer call PRINT-OBJECT for
4011 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4012 methods, giving nearly-ANSI behavior
4013 * converted almost all special variables to use *FOO* naming convention
4014 * deleted PARSE-TIME functionality, since it can be done portably
4015 * moved some files out of cold init into warm init
4016 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4018 * regularized formatting of source files
4019 * added an install.sh script
4020 * fixed ridiculous memory usage of cross-compiler by making
4021 compiler/alloc.lisp not try to do pooling unless it can hook
4022 itself into the GC of the cross-compilation host. Now the system
4023 builds nicely on my old laptop.
4024 * added :SB-ALLOC in target-features.lisp-expr
4025 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4026 was not implemented)
4027 * re-did condition handling and note reporting in the compiler. Notes
4028 are no longer handled by signalling conditions. Style warnings
4029 and warnings are handled more correctly and reported in such a way
4030 that it's easy to find one or the other in your output (so that you
4031 can e.g. figure out which of many problems caused COMPILE-FILE to
4033 * changed the severity of several compiler warnings from full WARNING
4034 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4035 compiler note reporting so that it doesn't use the condition system
4036 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4037 COMPILE-FILE command)
4038 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4039 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4040 no longer supported, since their implementation was deeply intertwingled
4041 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4042 * removed bogus "support" for compiler macros named (SETF FOO), and
4043 removed the compiler macro for SETF INFO (but only after making a fool
4044 of myself on the cmucl-imp mailing list by posting a bogus patch for
4045 DEFINE-COMPILER-MACRO..)
4046 * Compiled files containing forms which have side effects on the Lisp
4047 reader (such as DEFPACKAGE forms) are now handled more correctly.
4048 (Compiler queuing of top level lambdas has been suppressed by setting
4049 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4050 * deleted various currently-unused source files, e.g. gengc.lisp. They
4051 may be added back at some point e.g. when porting to other architectures,
4052 but until they are it's distracting to distribute them and to try to
4054 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4055 there were so many of them they're just distractions, and UNCROSS is
4056 known to be able to handle the current sources
4057 * moved PROFILE functionality into TRACE, so that it will be clear
4058 how the wrapping and unwrapping of functions when you profile them
4059 interacts with the wrapping and unwrapping of functions when you
4060 trace them. (Actually, the functionality isn't there yet, but at least
4061 the interface specification is there. Hopefully, the functionality will
4062 arrive with some future maintenance release.)
4063 * removed host-oops.lisp
4064 * changed signature of QUIT function to allow UNIX-CODE argument
4065 * fixed READ-SEQUENCE bug
4066 * tweaked verbose GC output so that it looks more like the progress
4067 output that ANSI specifies for functions like LOAD
4068 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4069 * added <http://sbcl.sourceforge.com> to the banner information printed by