1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.1 relative to sbcl-1.0:
3 * improvement: sb-sprof traces call stacks to an arbitrary depth on
4 x86/x86-64, rather than the previous fixed depth of 8
5 * bug fix: non-ascii command-line arguments are processed correctly
6 (thanks to Yaroslav Kavenchuk)
7 * bug fix: TRACE :ENCAPSULATE NIL (and function end breakpoints)
9 * bug fix: non-required arguments were not passed correctly when a method
10 defined using DEFMETHOD was called from a mop-generated method using
11 CALL-NEXT-METHOD (reported by Pascal Costanza)
12 * bug fix: recursion is now permitted in accessors through
13 SLOT-UNBOUND. (reported by Pascal Costanza)
14 * bug fix: an error was signaled at startup if the HOME environment
15 variable was defined, but had an empty value (reported by Peter Van Eynde)
16 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
17 signal a PROGRAM-ERROR, not a BUG.
18 * bug fix: SB-POSIX:READDIR works when built with large file support.
19 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
21 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
22 for better automatic indentation support. (thanks to Matt Pillsbury)
23 * optimization: loading generic functions no longer takes O(n^2) time,
24 proportional to the amount of methods in the generic function
25 (reported by Todd Sabin and Jeremy Brown)
26 * optimization: the FIND and POSITION family of sequence functions
27 are significantly faster on arrays whose element types have been
30 changes in sbcl-1.0 relative to sbcl-0.9.18:
31 * improvement: experimental support for threading on FreeBSD/x86.
32 (thanks to NIIMI Satoshi)
33 * improvement: runtime option --dynamic-space-size can be used
34 to set the size of the dynamic space reserved on startup.
35 * improvement: floating point modes in effect are now saved in
36 core, and restored on startup.
37 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
38 startup, not time since first call to GET-INTERNAL-REAL-TIME.
39 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
40 threads are not running after *SAVE-HOOKS* have run.
41 * improvement: writes to CLOS instance slots are type-checked in code
42 compiled with (SAFETY 3)
43 * improvement: floating-point exception handling on FreeBSD (thanks to
45 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
46 (thanks to Zach Beane)
47 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
49 * improvement: added support for the Shift-JIS external format.
50 (contributed by NIIMI Satoshi)
51 * improvement: callbacks are supported on Linux/PPC. (thanks to
53 * bug fix: compiler bug triggered by a (non-standard) VALUES
54 declaration in a LET* was fixed. (reported by Kaersten Poeck)
55 * bug fix: file compiler no longer confuses validated and already
56 dumped structurres. (reported by Kaersten Poeck)
57 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
58 fill-pointer signals a type-error as required. (thanks to
60 * bug fix: disassemly of funcallable instances works.
61 * bug fix: single stepping on PPC.
62 * bug fix: fix thread-safety problems in the type system (generally
63 manifesting as nonsensical errors like "STRING is a bad type specifier
64 for sequences" or "The value 1 is not of type FIXNUM")
65 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
66 (reported by Josip Gracin)
67 * bug fix: an error is signaled for attempts to displace arrays with
68 incompatible element types (thanks to Mario Mommer)
69 * bug fix: more correct handling of wide characters in debug info
70 (bug reported by Attila Lendvai and fixed by Juho Snellman)
71 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
72 and don't cause extra consing
73 * optimization: MAP and MAP-INTO are significantly faster on vectors
74 whose elements types have been declared.
75 * Improvements to SB-SPROF:
76 ** Support for allocation profiling
77 ** Reduced profiling overhead, especially for long profiling runs
78 * Improvements to the Windows port:
79 ** floating point exceptions are now reported correctly.
80 ** stack exhaustion detection works partially.
81 ** more accurate GET-INTERNAL-REAL-TIME.
82 ** better RUN-PROGRAM behaviour in case of errors in the child
84 ** PROBE-FILE now simplifies pathnames correctly.
85 ** DIRECTORY now works correctly with :WILD-INFERIORS.
86 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
87 1st 1970 (although time timezone data will be unreliable).
88 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
90 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
91 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
92 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
94 * bug fix: two potential GC deadlocks affecting threaded builds.
95 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
96 returns T (reported by Anton Kazennikov)
97 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
98 with non-variable places
99 * bug fix: remove a race condition in the setting of
100 funcallable-instance functions, this should make threaded CLOS
101 code more stable against memory faults.
102 * bug fix: corruption of specials when unbinding is interrupted by an
103 asynchronous unwind (reported by Hannu Koivisto)
104 * improvement: the debugger will now also display local variables that
105 are only used once, for code compiled with a DEBUG optimization quality
108 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
109 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
110 * incompatible change: External-format support for FFI calls. The
111 SB-ALIEN:C-STRING no longer implies an ASCII
112 external-format. Instead the string is subject to external-format
113 conversion. Additionally return values of type SB-ALIEN:C-STRING
114 are by default of type (SIMPLE-ARRAY CHARACTER), not
115 SIMPLE-BASE-STRING. For an alien type that behaves like the old
116 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
117 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
118 * incompatible change: SB-EXT package no longer contains the
119 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
120 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
122 * incompatible change: the single-stepper is no longer available
123 on Alpha, Mips and Sparc platforms.
124 * minor incompatible change: the direct superclasses of
125 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
126 not (STANDARD-OBJECT FUNCTION). This makes the
127 class-precedence-lists of GENERIC-FUNCTION and
128 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
130 * minor incompatible change: the default stream external format on
131 non-unicode SBCL is no longer detected from the locale, but is always
133 * new feature: Add a version of evaluator that uses an interpreter instead
134 of the compiler. EVAL still uses the compiler by default, to switch it
135 to use the interpreter, set the value of the variable
136 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
137 * minor incompatible change: the single-stepper REPL has been merged
138 with the normal debugger (see the "Stepping" heading of the debugger help
139 for more details). The debugger command STEP will no longer switch
140 to the single-stepper REPL.
141 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
142 for a type now works.
143 * bug fix: loading of default sysinit file works. (thanks to Leonid
145 * bug fix: better detection of circularities in the file-compiler.
146 (reported by Marco Monteiro)
147 * bug fix: the CL pathname functions now work with files that have
148 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
149 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
150 systems with Unix98 pty semantics.
151 * bug fix: ASDF-INSTALL will now work with bsd tar.
152 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
154 * bug fix: timers expiring in dead threads no longer cause a
155 type-error (reported by Paul "Nonny Mouse").
156 * bug fix: thanks to more lightweight single-stepper instrumentation,
157 code compiled with (DEBUG 3) will compile and execute significantly faster,
158 and will have more accurate type-inferencing than before
159 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
160 whose bindings are modified
161 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
162 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
163 ** replace dummy implementations of CL:MACHINE-INSTANCE and
164 CL:SOFTWARE-VERSION with proper version
166 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
167 * feature: implemented the READER-METHOD-CLASS and
168 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
169 as specified by AMOP.
170 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
172 * optimization: faster LOGCOUNT implementation on x86 and x86-64
173 (thanks to Lutz Euler)
174 * optimization: hashing of general arrays and vectors has been
175 improved. (reported by Any Fingerhut)
176 * enhancement: SB-INTROSPECT is now able to find definitions of
177 profiled functions. (thanks to Troels Henriksen)
178 * enhancement: compiler-macro expansion applies now to FUNCALL forms
179 as well, allowing compiler-macros for SETF-functions to expand.
180 * enhancement: step-instrumentation no longer wraps known
181 single-value functions with multiple-value context, allowing
182 better type inference.
183 * fixed bug #337: use of MAKE-METHOD in method combination now works
184 even in the presence of user-defined method classes. (reported by
185 Bruno Haible and Pascal Costanza)
186 * fixed bug #339(c): if there are applicable methods not part of any
187 long-form method-combination group, call INVALID-METHOD-ERROR.
188 (reported by Bruno Haible)
189 * fixed bug #361: the :FUNCTION initarg in the protocol for
190 initialization of methods can now be used to override
191 internally-produced optimized functions. (reported by Bruno
193 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
194 system-provided lambda expression no longer cause warnings about
195 unbound #:|pv-table| symbols.
196 * bug fix: improved the handling of type declarations and the
197 detection of violations for keyword arguments with non-constant
199 * bug fix: potentially erronous calls to PATHNAME and
200 MERGE-PATHNAMES were being flushed in some cases.
201 (reported by Richard Kreuter)
202 * bug fix: compiled calls to TYPEP were mishandling obsolete
203 instances. (reported by James Bielman and Attila Lendvai)
204 * bug fix: format strings with ~> without matching ~< no longer
205 trigger an AVER, but signal an understandable error instead.
206 (reported by Antonio Martinez)
207 * bug fix: specifying an output-file without a pathname-type for
208 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
209 of input-file instead of "fasl". (reported by Robert Dodier)
210 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
211 (reported by James Y Knight).
212 * bug fix: compiler-macros-function did not consider the environment
213 argument for shadowing by local functions.
214 * bug fix: compiler-macros expansion was inhibited by local INLINE
216 * bug fix: inline expansions of known functions were subject to
217 step-instrumentation in high debug policies, leading to problems
219 * bug fix: compiler failed to differentiate between different CONS
221 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
222 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
223 * bug fix: #368: incorrect use of expressed vs. upgraded array
225 * bug fix: #306a: more precise unions of array types.
226 * thread-safety improvements:
227 ** CONDITION-WAIT could return early on Linux, if the thread was
228 interrupted and subsequently continued with SIGCONT.
229 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
231 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
232 * added support for the ucs-2 external format. (contributed by Ivan
234 * minor incompatible change: pretty printing of objects of type
235 (cons symbol) is, in the default pprint-dispatch-table, now
236 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
238 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
239 called later in a class's lifetime, possibly as late as when the
240 first instance of the class is created. Previously,
241 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
242 class became finalizeable.
243 * fixed bug: FILE-POSITION sometimes returned inconsistent results
244 for multibyte external-format streams. (thanks to "vbzoli")
245 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
246 with :ALLOCATION :CLASS inherited from superclasses of the
248 * fixed bug: anonymous classes can now be created using the :NAME
249 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
250 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
251 * fixed bug: core-files saved with :EXECUTABLE T can again be
252 executed when SBCL_HOME isn't set. (reported by James Knight)
253 * fixed bug: toplevel LOCALLY forms with declarations could
254 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
255 * fixed bug: printing from several different threads using different
256 values of *print-case* could cause invalid output, due to
257 some internal special variables of the printer not being bound
258 thread-locally (reported by Max Mikhanosha)
259 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
260 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
261 (reported by Pascal Costanza)
262 * fixed bug: FINALIZE-INHERITANCE is called from
263 REINITIALIZE-INSTANCE on classes when the class has previously
264 been finalized, as required by AMOP.
265 * minor code generation optimizations:
266 ** better register allocation in CLOS dispatching functions
267 ** overflow detection when coercing signed bytes to fixnums on x86-64
268 ** is now implemented with one IMUL instruction instead of three shifts
269 ** more efficient bit-vector access on x86 and x86-64
270 ** more efficient access to raw structure slots on x86 and x86-64
271 * fixed some bugs revealed by Paul Dietz' test suite:
272 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
275 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
276 * feature: thread support on Solaris/x86, and experimental thread support
278 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
280 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
281 default method for SLOT-UNBOUND.
282 * minor incompatible change: prevent the user from specializing the
283 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
284 somewhat counter to the intent of the protocol, I (CSR) think, and
285 additionally it just doesn't work in SBCL as currently
286 implemented, thanks to optimizations (that are always valid for
287 the other three SLOT-VALUEish functions, but not for the setter).
288 * minor incompatible change: the :SB-LDB feature is now enabled by
289 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
290 the low-level debugger.
291 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
292 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
293 down to an SB-SYS:FD-STREAM, allowing direct communication in
294 more cases. Temporary files and pipes are still used as a fallback
296 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
297 on the non-reentrant C functions localtime(3) and gmtime(3).
298 * bug fix: global optimization policy was not visible in LOCALLY and
300 * bug fix: class objects can be used as specializers in methods.
301 (reported by Pascal Costanza)
302 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
303 with a type error. (reported by blitz_ on #lisp)
304 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
305 failed with a type error. (reported by Pascal Bourguignon)
306 * bug fix: merging pathnames against defaults with :DIRECTORY
307 starting with '(:RELATIVE :BACK) should preserve the :BACK.
308 (reported by James Y Knight)
309 * bug fix: saving large (>2GB) cores on x86-64 now works
310 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
311 constant index and a (simple-array (signed-byte 32)) array
312 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
313 error (patch by Robert J. Macomber)
314 * feature: TIME now displays the amount of run-time spent in GC
315 * bug fix: The debugger now does a better job of respecting
316 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
317 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
318 workaround for bug 403.)
319 * bug fix: floating point exception handling now works on FreeBSD
320 (thanks to NIIMI Satoshi)
321 * fixed some bugs revealed by Paul Dietz' test suite:
322 ** MISC.641: LET-conversion were not supposed to work in late
324 * improvements to the Win32/x86 port:
325 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
326 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
328 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
329 * new feature: source path information is generated for macro-expansion
330 errors for use in IDE's like Slime (thanks to Helmut Eller)
331 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
332 * bug fix: compiler does not loop forever on an invalid type in
334 * improvement: compilation of most CLOS applications is significantly
336 * optimization: added a limited bytecode compiler for simple toplevel
337 forms, speeding up compilation and FASL loading
338 * bug fix: the statistical profiler now properly distinguishes anonymous
341 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
342 * minor incompatible change: in sbcl-0.9.11 (but not earlier
343 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
344 into subclasses of the finalized class. Now user calls to
345 FINALIZE-INHERITANCE finalize just the one class, and calls by the
346 system return before any subclasses are finalized.
347 * minor incompatible change: The reader no longer ignores errors
348 regarding non-existent packages in #+ and #- feature tests.
349 * new feature: command line options --no-sysinit, --no-userinit to
350 inhibit loading the corresponding init files
351 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
352 for sending data through UDP sockets (thanks to François-René Rideau)
353 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
354 error is signalled instead (thanks to François-René Rideau)
355 * minor incompatible change: Error signalling behaviour of lexical
356 operations violating package locks has changed slightly. Refer to
357 documentation on package locks for details.
358 * enhancement: EVAL can process IF-expressions without resorting to the
360 * bug fix: Some operations involving SIN, COS, or TAN weren't being
361 constant-folded properly due to missing out-of-line functions.
362 (reported by Mika Pihlajamäki)
363 * bug fix: LISTEN sometimes returned T even in cases where no data was
364 immediately available from the stream
365 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
366 were reversed. (reported by Levente Mészáros)
367 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
368 when given a vector displaced to an adjustable array. (reported
370 * bug fix: derivation of float boundaries from numbers outside the
371 appropriate float range (reported by John Wiseman)
372 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
374 * bug fix: precomputing cache entries for generic functions with
375 some subclasses of specializers as yet invalid does not attempt to
376 fill a cache line with a negative offset. (reported by Levente
378 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
379 allow condition class objects as arguments to DOCUMENTATION and
380 (SETF DOCUMENTATION); only find and set documentation for
381 structure names for the STRUCTURE doc-type. (suggested by Gary
383 * improvements to the Win32/x86 port:
384 ** (user-homedir-pathname) and default initialization file
385 locations now know about the user's "Documents and Settings"
386 directory (thanks to Yaroslav Kavenchuk)
387 ** run-program is implemented (thanks to Mike Thomas)
388 ** sockets support (thanks to Timothy Ritchey)
389 ** better backtrace support (thanks to Alastair Bridgewater)
390 ** sb-grovel supported
391 ** asdf-install and sb-posix work somewhat
392 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
393 * improvements to the Solaris/x86 port:
394 ** works on Solaris 11/Solaris Express
395 ** floating-point exception handling support
396 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
397 * fixed some bugs revealed by Paul Dietz' test suite:
398 ** REMOVE-METHOD returns its generic function argument even when
399 the method is not one of the generic functions' methods.
400 ** objects declared as MEMBER types can be admissible arguments to
402 ** printing characters should simply be printed by the FORMAT ~:C
404 ** compiler failure when compiling functions with hairy constant
405 defaults for optional parameters.
406 ** compiler produces wrong code when MAYBE-INLINE-expanding a
407 function, which is already optimized.
409 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
410 * new platform: experimental support for SBCL x86/Darwin, including
411 MacOS X 10.4.5 on Intel.
412 * new feature: Unicode character names are now known to the system
413 (through CHAR-NAME and NAME-CHAR).
414 * new feature: the filesystem location of SBCL's core file is
415 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
416 * minor incompatible change: the contrib modules SB-POSIX and
417 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
418 this change is to make it easier to distribute
419 location-independent binaries.
420 * bug fix: as implied by AMOP, standardized classes no longer have
421 slots named by external symbols of public packages. (reported by
423 * optimization: calling functions via constant symbols -- as in
424 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
425 via the function object as in (FUNCALL #'FOO).
426 * optimization: on x86 and x86-64, the calling convention for the
427 case of dealing with an unknown number of values has been altered
428 to be friendlier to the prediction heuristics implemented,
429 particularly with reference to CALL and RET pairing. (thanks to
430 Alastair Bridgewater)
431 * optimization: on x86, the code for access to thread-local symbol
432 values and binding/unbinding of thread-local symbols is smaller.
433 (thanks to Alastair Bridgewater)
434 * enhancement: CONSTANTP is now able to determine constantness of
435 more complex forms, including calls to constant-foldable standardized
436 functions and some special forms beyond QUOTE.
437 * fixed bug: occasional GC crashes on Solaris/x86
438 * optimization: x86-64 supports stack allocation of results of simple
439 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
440 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
441 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
442 (thanks to James Knight)
443 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
444 by the standard (thanks to Stephen Compall)
446 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
447 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
448 be used for bundling the runtime and the core file into one
449 executable binary. This feature is not currently supported on all SBCL
450 platforms. (thanks to James Bielman and NIIMI Satoshi)
451 * new feature: a generational or ephemeral garbage collector is now
452 the default on the PowerPC platform (both Linux and Darwin). The
453 old Cheney (stop and copy) collector is a build-time option.
454 (thanks to Cyrus Harmon, after Raymond Toy)
455 * minor incompatible change: the method by which SBCL finds its
456 contributed modules has changed; it no longer relies on symbolic
457 links from an $SBCL_HOME/systems directory, but searches directly
458 in the subdirectories of $SBCL_HOME.
459 * enhancement: the dynamic heap size on the Linux/PPC platform is
460 markedly larger, even using the older Cheney garbage collector.
461 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
462 certain complicated string types. (reported by Gary King)
463 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
464 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
465 * fixed bug: DOCUMENTATION on structures no longer signals an
466 error if no structure type of the right name exists. (reported by
468 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
469 applicable non-standard primary or :AROUND method on
470 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
471 no longer cause unbound variable errors. (reported by Kevin Reid)
472 * fixed bug: metacircle resolution in cases where methods have slots
473 added before the slots from STANDARD-METHOD. (reported by Jean
475 * fixed bug: the Power PC assembly code for calling into lisp
476 sometimes computed its return address wrongly (depending on the
477 15th bit of the address, fixed at link time). This used to
478 manifest itself in a segmentation violation while building PCL.
479 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
480 many others over the years)
481 * fixed bug: ROOM no longer reports silly numbers for stack usage if
482 the stack pointer (treated as a signed integer) is negative.
483 (thanks to Peter van Eynde)
485 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
486 * new platform: experimental support for the Windows operating
487 system has been added. (thanks to Alastair Bridgewater)
488 * fixed several bugs in and robustified the PPC FFI (including
489 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
490 * bug fix: allow non-simple string symbol names (reported by Paul
492 * bug fix: interrupt handling on NetBSD (thanks to Richard M
494 * bug fix: saving a core corrupted callbacks on x86/x86-64
495 * bug fix: closed a loophole in metacircularity detection and
496 grounding in the PCL implementation of CLOS.
497 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
498 (reported by tomppa on #lisp)
499 * bug fix: clear the direction flag on Lisp -> C transitions, as
500 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
501 (reported by Andrej Grozin and Hendrik Maryns)
502 * optimization: major improvements to GC efficiency on GENCGC platforms
503 * optimization: faster implementation of EQUAL
504 * optimization: emit more efficient opcodes for some common
505 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
506 * optimization: several other minor code-generation improvements on x86-64
507 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
509 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
510 * minor incompatible change: (SETF CLASS-NAME) and (SETF
511 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
512 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
513 by ANSI as a generic function, and for consistency (SETF
514 GENERIC-FUNCTION-NAME) is treated likewise.
515 * fixed bug #233.b: make constraint propagation notice when a variable
516 value is changed after it is referenced but before it is used
517 * fixed bug #296: no more arbitrary behaviour from filenames with
518 odd characters as --load arguments. (reported by Adam Warner)
519 * fixed bug #390: :CHARACTER-SET pathname components now work as
520 expected. (reported by Tim Daly Jr)
521 * fixed bug #391: complicated :TYPE intersections in slot
522 definitions no longer cause an error in PCL internals.
523 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
524 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
525 returning the number of octets which would be written to the
526 file-stream. (thanks to Robert J. Macomber)
527 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
528 arguments to the internal stream functions. (thanks to David
530 * bug fix: stack exhaustion is now handled on the x86/SunOS
531 platform. (thanks to Daisuke Homma)
532 * bug fix: more accurate ROOM results on GENCGC platforms
533 * optimization: improved type inference for arithmetic-for
534 index variables in LOOP
535 * optimization: faster floating-point SQRT on x86-64
536 * fixed some bugs revealed by Paul Dietz' test suite:
537 ** DOCUMENTATION returns NIL instead of "" for method combinations
538 that don't have a docstring
540 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
541 * minor incompatible change: (SETF CLASS-NAME) and (SETF
542 GENERIC-FUNCTION-NAME) are no longer generic functions, and
543 therefore are not customizeable by user code (as seems to be at
544 least permitted and maybe required by AMOP). As a consolation,
545 however, the SBCL implementation of these functions now calls
546 REINITIALIZE-INSTANCE as specified by AMOP.
547 * bug fix: slot-definition documentation is propagated as per ANSI
548 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
549 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
550 argument, as do eslotd initialization methods. (from Pascal
551 Costanza's "Closer" project)
552 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
553 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
555 * bug fix: it is now possible to have more than one subclass of
556 STANDARD-GENERIC-FUNCTION without causing stack overflow.
557 (reported by Bruno Haible, Pascal Costanza and others)
558 * bug fix: the dependent update protocol now works for generic
559 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
561 * bug fix: condition-class instances corresponding to
562 DEFINE-CONDITION forms are now created eagerly. (reported by
563 Kalle Olavi Niemitalo on comp.lang.lisp)
564 * bug fix: floating point printing is more accurate in some
565 circumstances. (thanks to Simon Alexander)
566 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
567 merged with *DEFAULT-PATHNAME-DEFAULTS*.
568 * bug fix: callbacks on OS X now preserve stack-alignment.
569 (thanks to Cyrus Harmon)
570 * enhancement: the x86-64 disassembler is much better at
571 disassembling SSE instructions. (thanks to Lutz Euler)
572 * enhancement: CHECK-TYPE on a variable now allows type-inference
573 (thanks to Vincent Arkesteijn)
574 * optimization: improved performance of EUC-JP external format.
575 (thanks to NIIMI Satoshi)
576 * optimization: performance improvements to IO on file streams of
577 :ELEMENT-TYPE CHARACTER
578 * optimization: much faster memory allocation on x86-64
579 * optimization: faster garbage collections (latency and throughput)
581 * optimization: faster bignum subtraction and division on x86 and x86-64
582 (thanks to Lutz Euler)
583 * optimization: more accurate type-inference for loops that use a
584 floating point index variable or a negative step.
586 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
587 * bug fix: add a workaround to SBCL looping infinitely at startup on
588 Linux kernels with apparently buggy implementations of personality().
589 (thanks to Svein Ove Aas)
590 * bug fix: Unicode symbols are correctly printed in LDB backtraces
591 (thanks to David Lichteblau)
592 * bug fix: local bindings shadow global symbol macros for
593 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
594 * optimization: non-open coded uses of numeric comparison operators
595 (e.g. >) no longer cons when called with more than one parameter
596 on platforms supporting dynamic-extent allocation.
597 * enhancement: saving cores with foreign code loaded is now supported
598 on MIPS/Linux in addition to the previously supported platforms.
599 * bug fix: FIND-RESTART now tests for activity, not applicability when given
600 a restart object as identifier. (reported by Helmut Eller for CMUCL)
601 * bug fix: division by zero in sb-sprof when no samples were collected
602 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
603 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
604 (thanks to Alexander Kjeldaas)
605 * new feature: ignoring the return values of destructive functions
606 like NREVERSE signals a compile-time style-warning.
607 (thanks to Kevin Reid)
608 * new feature: the alignment of alien structure fields can be
609 explicitly specified. (thanks to Cyrus Harmon)
610 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
611 and dump core on SIGQUIT
613 ** incompatible change: threads do not inherit values of specials
614 from their parents (see manual)
615 ** bug fix: threads stacks belonging to dead threads are freed by the
616 next exiting thread, no need to gc to collect thread stacks anymore
617 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
618 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
619 ** bug fix: dynamic variable and thread start related gc lossage
620 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
622 * fixed some bugs revealed by Paul Dietz' test suite:
623 ** SUBTYPEP is slightly more accurate on heinously complicated
624 CONS types where some of the members have uncertain (in the
625 NIL, NIL sense) type relationships to each other.
627 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
628 no longer purified unless :PURIFY T is explicitly specified.
629 ** Non-purified cores are significantly smaller than before
631 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
632 * new feature: timers based on Zach Beane's excellent timer package
633 * added support for the following external formats: koi8-u,
634 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
635 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
636 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
637 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
638 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
639 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
640 * incompatible change: a threaded SBCL will no longer revert to
641 non-threaded mode on non-NPTL systems, but refuse to start entirely.
642 * bug fix: interrupts are disabled until startup is complete; no
643 more sigsegvs when receiving a signal to soon
644 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
646 * bug fix: add a workaround for the memory randomization features in
647 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
648 workaround will only be in effect on systems with the proc filesystem
650 * bug fix: printing objects of type HASH-TABLE signals a
651 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
653 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
654 have been running for over 50 days. (reported by Gilbert Baumann)
655 * bug fix: the logic for getting names of functions gets less
656 confused when confronded with alternate-metaclass
657 funcallable-instances. (reported by Cyrus Harmon)
658 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
659 consistent, even on internal alternate-metaclass objects.
660 * bug fix: flush closure information collected by physical
661 environment analysis prepass before the main pass. (bug reported
663 * bug fix: compiler pack phase does not modify a hashtable, which is
664 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
665 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
666 may be expanded. (reported by Alexander <malishev> on c.l.l)
667 * bug fix: no more style-warnings when compiling code using LAST
668 under high-SPEED low-SPACE settings. (reported by David Wallin
670 * fixed some incompatibilities between SBCL's MOP and the MOP
672 ** the METAOBJECT class is now implemented;
673 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
674 STANDARD-OBJECT, as required;
675 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
676 now compatible (as required by VALIDATE-SUPERCLASS); there
677 remains a consistency requirement over the presence or absence
678 of the FUNCTION class in the superclass list of finalized
679 classes; see the manual for more details;
680 ** the :AROUND method for COMPUTE-SLOTS on
681 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
682 requested slot ordering.
684 ** bug fix: parent thread now can be gc'ed even with a live
686 ** bug fix: binding a special with PROGV to no value is not
687 the same as that symbol not having been bound (thanks to
689 ** bug fix: binding specials is thread safe (thanks to
691 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
692 and SB-SPROF do not die with 'no handler for signal XX in
693 interrupt_handle_now(..)' anymore
694 ** bug fix: WITH-TIMEOUT works with multiple running threads
695 * fixed some bugs revealed by Paul Dietz' test suite:
696 ** ENSURE-GENERIC-FUNCTION should take a method class object for
697 the :method-class keyword argument.
699 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
700 * new port: the Solaris operating system on x86 processors is now
701 mostly supported, though some rough edges in the environment
702 remain. (thanks to Daisuke Homma)
703 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
704 space for its heap. (thanks to Thiemo Seufer)
705 * optimization: SBCL on MIPS platforms now supports dynamic-extent
706 closures. (thanks to Thiemo Seufer)
707 * minor incompatible change: eof selects abort in the debugger.
708 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
709 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
711 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
712 deprecated, and will go away in a future revision of SBCL.
713 * minor incompatible change: GC-ON and GC-OFF are no longer
714 implemented with a counter, it does not matter how many times gc
715 is switched on or off
716 * bug fix: discriminating functions for generic function classes
717 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
718 make invalid assumptions about method precedence order. (reported
720 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
721 platforms. (thanks to Thiemo Seufer)
722 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
723 * bug fix: correct pathname printing with printer escaping is on.
724 (thanks to Kevin Reid)
725 * bug fix: complex VOP definitions in "user-space" no longer trigger
726 package locks. (reported by Zach Beane)
727 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
728 for funcallable-instance functions, and is no different from
730 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
731 characters in the current readtable. (reported by Nicholas Neuss)
732 * bug fix: SUBTYPEP on various CONS types returns more a more
733 accurate acknowledgment of its certainty.
734 * optimizations: REMOVE-DUPLICATES now runs in linear time on
735 lists in some cases. This partially fixes bug 384.
736 * flush all standard streams before prompting in the REPL and the
738 * bug fix: signal handling and triggering gc do not conflict
739 directly anymore, in particular a high frequency sb-sprof does
740 not prevent gc from running
741 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
742 approximation for timezone and DST information between the
743 universal time epoch and the smallest negative 32-bit time_t.
744 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
745 year 1899 so long as the encoded time is non-negative.
746 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
747 the correct answer for non-constant rotate amounts when performing
748 an inline 32-bit rotation.
750 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
751 there is only one thread in the session
752 ** bug fix: memory leak for streams created in one thread and
753 written to in another
754 ** bug fix: lockup when compiled with gcc4
755 ** bug fix: race that allows the gc to be triggered when gc is
757 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
758 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
760 ** bug fix: finished threads are gc'ed properly
761 * fixed some bugs revealed by Paul Dietz' test suite:
762 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
763 has arguments with a different set of applicable methods from
764 the orignal arguments.
765 ** The type error thrown by MAP now has a correct expected-type
767 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
768 name a compiled function.
769 ** Three MISC tests where a large bignum was improperly coerced to
770 a float that couldn't represent that bignum during type
771 derivation were fixed.
772 ** SUBTYPEP can now handle types involving the negation of a
773 list-form FUNCTION type.
774 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
775 as (COMPLEX (AND RATIO (NOT FIXNUM))).
776 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
778 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
779 * New feature: Experimental support for bivalent streams: streams
780 opened with :element-type :default now allow character and binary
781 (unsigned-byte 8) I/O
782 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
783 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
785 * bug fix: no more highly sporadic "couldn't check whether ~S is
786 readable" when reading a stream and an interrupt hits in the middle
787 of a select system call
788 * compiler better recognizes complex arrays (reported by Eduardo
790 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
791 for variables on SBCL built with linkage-tables. (reported by Luis
793 * various error reporting improvements.
794 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
795 (Thanks to Thiemo Seufer)
796 * bug fix: nested reader invokations work correctly
797 * bug fix: it is possible to have simultaneous references to foreign
798 code and foreign data with the same name.
800 ** added x86-64 support
801 ** incompatible change: the threading api now works with thread
802 objects instead of thread ids
803 ** bug fix: threads are protected from signals and interruption when
804 starting up or going down
805 ** bug fix: a race where an exiting thread could lose its stack to gc
806 ** bug fix: don't halt on infinite error in threads if possible
807 ** fixed numerous gc deadlocks introduced in the pthread merge
808 ** bug fix: fixed thread safety issues in read and print
809 ** bug fix: debugger doesn't hang on session lock if interrupted at
810 an inappropriate moment
811 ** bug fix: run-program is now thread safe(r)
812 ** bug fix: inner with-recursive-lock no longer releases the mutex
813 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
814 * fixed some bugs revealed by Paul Dietz' test suite:
815 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
816 EXPECTED-TYPE slots filled.
817 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
819 ** MAKE-CONDITION accepts classes as type-designators.
820 ** COMPILE may never return NIL.
821 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
822 range before calling Unix time functions
824 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
825 * numerous signal handling fixes to increase stability
826 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
827 * minor incompatible change: we now correctly canonize default
828 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
829 as per the MOP, rather than the historical (INITARG INITFUNCTION
830 INITFORM). (reported by Bruno Haible)
831 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
832 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
833 for more information.
834 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
835 pathname is a directory pathname.
836 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
837 * dynamic space size on PPC has been increased to 768Mb. (thanks to
839 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
840 :METACLASS argument in addition to a class name. (reported by
841 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
842 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
843 (reported by Bruno Haible; patch from Gerd Moellmann)
844 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
846 * bug fix: more cleanups to the floating point exception handling on
847 x86-64 (thanks to James Knight)
848 * bug fix: the compiler does not try to constant fold calls of
849 COERCE and BIT-* functions when they return freshly constructed
850 objects. (reported by Nikodemus Siivola and Paul Dietz)
851 * optimization: improved the MIPS versions of generic (in the
852 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
853 * optimization: direct conversion of (unsigned-byte 32) to floats on
854 the PowerPC platform.
855 * optimization: structure instances with raw slots now use less
856 memory, and probably show better memory locality. (thanks to
858 * optimization: DYNAMIC-EXTENT declarations for lists and closures
859 are treated as requests for stack allocation on the x86-64,
860 Alpha-32, PPC and SPARC platforms.
861 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
862 the wrong file descriptor; implementation of SOCKET-OPEN-P.
863 (thanks to Tony Martinez)
865 ** gcing a dead thread can no longer lead to lockups
866 ** threads block signals until they are set up properly
867 ** errno is no longer shared by threads
868 ** interrupt-thread restores the eflags register on x86
869 ** fixed some lockups due to gc/thread interaction
870 * fixed some bugs revealed by Paul Dietz' test suite:
871 ** invalid dotted lists no longer raise a read error when
873 ** signal an error if a symbol that names a declaration is used
874 as the name of a type, or vice versa
875 ** allow using the (declare (typespec var*)) abbreviation for
876 (declare (type typespec var*)) with all type specifiers
877 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
878 ** always return NIL from PPRINT-POP when OBJECT is NIL
879 ** don't signal errors when pretty-printing malformed LABELS,
880 FLET or MACROLET forms
881 ** declarations in a DOLIST body are also in scope for the
883 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
885 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
888 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
889 * fixed cross-compiler leakages that prevented building a 32-bit
890 target with a 64-bit host compiler.
891 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
892 opened with :IF-EXISTS :OVERWRITE.
893 * fixed bug 281: error for an invalid qualifier in a short-form method
894 combination method is not signalled until the faulty method is called.
895 * bug fix: iteration variable type inferrer failed to deal with open
896 intervals. (reported by Alan Shields)
897 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
898 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
899 or more runtime options were provided to the sbcl binary.
900 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
901 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
902 * optimization: there's now a fast-path for fixnum arguments in the
903 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
904 * optimization: the code generated on x86-64 is more compact thanks
905 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
906 * bug fix: floating-point exception handling now partly works on
907 x86-64. (Thanks to James Knight)
908 * improvement to the MIPS backend from Thiemo Seufer: C-style
909 64-bit long long arguments and return values to alien functions
911 * fixed some bugs revealed by Paul Dietz' test suite:
912 ** the type-error signalled from WARN has a filled-in DATUM slot.
913 ** the type-error required when a stream is not associated with
914 a file has the stream as its datum.
915 ** type-errors on single-floats on x86-64 no longer have
916 :INVALID-OBJECT as the datum
917 ** the type-errors signalled for invalid function names now have
918 a correct expected type
919 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
920 for structures defined with a :TYPE. Documentation strings for
921 typed structures are no longer immediately discarded
922 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
923 broadcast streams and synonym streams. FILE-LENGTH now also works
924 on broadcast streams.
926 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
927 * incompatible change: the --noprogrammer option, deprecated since
928 version 0.7.5, has been removed. Please use the equivalent
929 --disable-debugger option instead.
930 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
932 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
933 inoperational for a while now) has been completely removed.
934 * null lexical environments are now printed as #<NULL-LEXENV>,
935 significantly reducing the amount of clutter in typical backtraces.
936 * documentation on weak pointers, finalization, and after GC hooks
937 has been added to the manual.
938 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
939 as other specialized array types, is much faster. SUBSEQ and
940 COPY-SEQ on such arrays have also been sped up.
941 * optimization: EQL is now more efficient when at least other argument
942 is known to be of type (OR FIXNUM (NOT NUMBER)).
943 * fixed bug: the runtime is now less vulnerable to changes in the
944 size of the SBCL object on OS X, and virtual memory is reserved for
945 all spaces that need to be at a fixed address.
946 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
947 * fixed bug: finalizers and after GC hooks that cause consing are now
949 * fixed bug: compiler error messages and summaries are now printed to
950 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
951 * fixed inference of the upper bound of an iteration variable.
952 (reported by Rajat Datta).
953 * fixed bug 211e: calling local functions with duplicated constant
954 keyword argument no longer causes a bogus style warning about an
956 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
957 ftype declaration to be disregarded. (reported by Dave Roberts)
958 * fixed bug 373: caused by erronous compilation of references to alien
959 variables in the runtime on ppc/darwin.
960 * fixed bug 376: CONJUGATE type deriver.
961 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
962 (reported by Baughn on #lisp)
963 * a cross-compiler bug on non-x86 platforms has been identified and
964 fixed. (thanks to Bruno Haible)
965 * improvements to the MIPS runtime code for increased stability.
966 (thanks to Thiemo Seufer)
967 * increased the maximimum compact environment size to allow
968 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
969 * partial workaround for bug 135: don't name the function we're
970 calling for hairy cases of EVAL, so as not to accumulate one environment
971 entry per call to EVAL. (reported by Kevin Reid)
972 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
973 * optimization: type testing for non-vector arrays should be faster.
974 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
976 * bug fix: setting 31st element of a bit vector to zero did not work
978 * bug fix: redefining a class definition which failed due to a
979 previous accessor / function clash now works (but see BUGS entry
980 #380 for more problems in this area). (thanks to Zach Beane)
981 * the long form of DEFINE-METHOD-COMBINATION disables method group
982 checking when given a single method group with pattern *. (thanks
984 * on x86 compiler supports stack allocation of results of simple
985 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
986 * support for building and running on Mac OS X 10.4 "Tiger" has been added
987 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
988 * fixed some bugs related to Unicode integration:
989 ** the restarts for recovering from input and output encoding
990 errors only appear when there is in fact such an error to
992 * fixed some bugs revealed by Paul Dietz' test suite:
993 ** MISC.549 and similar: late transformation of unsafe type
994 assertions into derived types caused unexpected code
996 ** SCALE-FLOAT type deriver is less wrong.
997 ** type derivers for EXP, LOG and similar functions compute result
998 types for complex arguments better.
999 ** (MISC.563) CONJUGATE type deriver works for very restricted
1001 ** out-of-line type testers for character strings are available.
1002 ** EQUAL compiler transform understands specialness of objects
1004 ** accessing double-floats stored on the stack now works on x86-64.
1005 ** debugger internals could sometimes create invalid lispobjs,
1006 resulting in GC crashes.
1007 ** MISC.548: type check weakening can convert required type into
1009 ** initialization forms for bindings are not in scope of free special
1012 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1013 * incompatible change: thread support for non-NPTL systems has
1014 been removed - locking is buggy and unreliable. A threaded
1015 SBCL build will now warn at startup and refuse to create
1016 new threads, unless futex support is detected in the kernel
1017 * incompatible change: the top level REPL now has only an ABORT
1018 restart associated with it, not TOPLEVEL and ABORT as it used to.
1019 TOP and TOPLEVEL are now available as debugger commands for
1020 returning to the top level.
1021 * incompatible change: forms evaluated in the REPL now use the
1022 global optimization policy.
1023 * incompatible change: user- and system-initialization files are
1024 no longer processed with LOAD, but by READ and EVAL; hence the
1025 global optimization policy, startup package, readtable, etc,
1027 * The .fasl file format number has been incremented because of
1028 various incompatible changes.
1029 * internal entry point details and argument counts no longer appear
1030 in backtraces unless explicitly requested by setting
1031 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1032 * built-in and standard functions no longer have names like "top
1033 level local call to FOO".
1034 * fixed bug 32: functions defined in non-null lexical environments
1035 now have more legible printed representation
1036 * fixed bug 33: functions defined in non-null lexical environemnts
1037 are now more amenable to inspection by INSPECT.
1038 * workaround for bug 354: XEPs no longer appear in backtraces unless
1039 explicitly requested.
1040 * fixed bug: receiving the signal which results from attempting to
1041 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1042 SIGBUS on other *BSDs) on architectures where the C stack is also
1043 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1044 to debugger confusion or wild instability. (reported by Cheuksan
1046 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1047 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1048 (reported by Lutz Euler)
1049 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1050 compiled code. (reported by Paul Dietz)
1051 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1052 specialized parameter to a method to something that is not TYPEP
1053 the specializer is now possible.
1054 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1055 face of package deletion.
1056 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1057 pathname. (reported by David Tolpin on c.l.l)
1058 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1059 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1060 STRING-STREAM and FILE-STREAM, respectively.
1061 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1062 than previously; in addition, its test suite is now run on build.
1063 (thanks to Luke Gorrie)
1064 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1065 * optimization: added a immediate representation for single-floats
1067 * fixed some bugs revealed by Paul Dietz' test suite:
1068 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1069 ** attempting to create a package with a colliding nickname causes
1070 correctable errors to be signalled.
1071 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1072 ** several x86-64 backend bugs related to sign-extension of immediate
1075 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1076 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1077 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1079 * fixed loading of multiply forward-referenced layouts.
1080 (thanks to Cheuksan Wang)
1081 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1082 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1083 (see COMPILE-FILE documentation for details of :PRINT option)
1084 * fixed bugs 19 and 317: fixed-format floating point printing is
1085 more accurate. This also fixes a bug reported by Adam Warner
1086 related to the ~@F format directive.
1087 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1089 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1090 dispatch table if the from-char is a dispatch macro character.
1091 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1092 by the wordsize no longer ignore the last word. (reported by Lutz
1094 * fixed bug in type checking of dynamic-extent variables. (reported
1096 * optimization: sequence traversal functions use their freedom to
1097 coerce function designators to functions.
1098 * optimization: code with many calls to CLOS methods specialized on
1099 CLOS classes has had redundant type checks removed.
1100 * optimization: added declarations to speed up operations that access
1101 the internal character database (for example STRING-UPCASE)
1102 * optimization: comparison operations between floats and sufficiently small
1103 fixnums no longer create extra rationals
1104 * fixed some bugs related to Unicode integration:
1105 ** portions of multibyte characters at the end of buffers for
1106 character-based file input are correctly transferred to the
1107 start of the buffer at the next read.
1108 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1109 passing it through to OPEN.
1110 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1111 argument given to internal calls to OPEN.
1112 ** the built-in comment readers (introduced by character sequences
1113 ";" and "#|") are more forgiving to encoding errors; they will
1114 STYLE-WARN and then attempt to resync the stream at a character
1115 boundary. (thanks to Teemu Kalvas)
1116 * fixed some bugs in the x86-64 port:
1117 ** Negative short int return values from c-calls are sign-extended
1119 ** The stack is aligned to 16-bytes for c-calls, as required by
1120 the ABI. (thanks to Cheuksan Wang)
1121 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1122 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1123 for enums. (thanks to Vincent Arkesteijn)
1124 ** Multiple small optimizations and bugfixes for floating point
1126 * fixed some bugs revealed by Paul Dietz' test suite:
1127 ** Space, Tab, Linefeed, Return and Page have the invalid
1128 secondary constituent character trait.
1129 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1131 ** WITH-INPUT-FROM-STRING should only update the index place on
1133 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1134 works more reliably.
1135 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1136 with one-dimensional arrays with fill pointers.
1137 ** TYPE-OF failed on a complex with an integer realpart and a
1139 ** compiler failure during type inference for the code of form
1140 (IF (EQL X (THE ...)) ...) (MISC.535).
1142 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1143 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1144 platform. The port supports 61-bit fixnums, large memory spaces
1145 and reloading shared object files.
1146 * enhancement: saving cores with foreign code loaded is now
1147 supported on ppc/Darwin in addition to the previously supported
1149 * enhancement: the statistical profiler now walks deeper into the
1150 call stack for more meaningful call-graphs and accrued time
1151 reports (x86/x86-64 only). It also now reports time spent in
1153 * enhancement: it is now possible to trace most individual methods
1154 of a generic function in addition to tracing the generic function
1156 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1158 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1159 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1160 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1161 produces an error. (thanks to Vincent Arkesteijn)
1162 * bug fix: NAMESTRING on pathnames with :WILD components in their
1163 directories works correctly. (thanks to Artem V. Andreev)
1164 * fixed bug 125: compiler preserves identity of closures. (reported
1166 * bug fixed: functions with &REST arguments sometimes failed with
1167 "Undefined function" when compiled with (DEBUG 3). (reported by
1169 * bug fix: overflow during compiling of setting element of a bit
1170 vector with constant index and value. (reported by Timmy Douglas)
1171 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1172 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1173 * on x86 compiler partially supports stack allocation of dynamic-extent
1175 * GO and RETURN-FROM do not check the extent of their exit points
1176 when compiled with SAFETY 0.
1177 * fixed some bugs related to Unicode integration:
1178 ** encoding and decoding errors are now much more robustly
1179 handled; it should now be possible to recover even from invalid
1180 input or output to the terminal. (thanks to Teemu Kalvas)
1181 ** provided a first cut at implementing STRING-TO-OCTETS and
1182 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1183 ** altered the SB-MD5 contributed module slightly, changing the
1184 interface just enough for it to be supportable for builds where
1185 lisp characters are not eight bits.
1186 * fixed some bugs revealed by Paul Dietz' test suite:
1187 ** the FORMATTER-generated functions for ~V[ conditionals require
1188 the correct number of arguments.
1189 ** READ-FROM-STRING returns the mandated second value when applied
1190 to displaced strings.
1191 ** the #\Rubout and #\Backspace characters are treated as invalid
1192 constituent characters by the tokenizer.
1194 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1195 * new feature: reloading changed shared object files with
1196 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1197 * new feature: references to foreign variables and functions
1198 can now be compiled and loaded before the corresponding shared
1199 object file is loaded, as long as the foreign definitions are
1200 available at runtime.
1201 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1202 supported. (thanks to Dan Debertin)
1203 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1204 just keywords. (thanks to Vincent Arkesteijn)
1205 * fixed bug #331: structure-class instances corresponding to
1206 DEFSTRUCT forms are now created eagerly.
1207 * fixed bug #345: backtraces from calls to undefined functions work
1208 on x86 as well. Related bug #61 is now also partially fixed on x86
1209 (backtraces from throws to unknown catch tags.)
1210 * bug fix: lambda-list parsing is now stricter vrt. order and number
1211 of lambda-list keywords.
1212 * bug fix: as specified by AMOP, an error is signalled if a
1213 class-option appears multiple times in a DEFCLASS form. (reported
1215 * bug fix: RANDOM can be compiled when the compiler derives the type
1216 of its numeric argument as a disjoint set of small integers.
1217 (reported by Paul Dietz)
1218 * bug fix: starting a core saved with shared objects loaded when
1219 those objects are not available no longer causes threaded SBCL to
1220 hang. (reported by Sean Ross)
1221 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1223 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1224 argument, which must always be NIL. (reported by Kalle Niemitalo)
1225 * bug fix: printing 1.0d+23 no longer results in an error.
1226 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1227 * bug fix: structure slot setters preserve evaluation order. (thanks
1229 * fixed some bugs related to Unicode integration:
1230 ** RUN-PROGRAM can allow its child to take input from a Lisp
1231 stream. (reported by Stefan Scholl)
1232 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1233 locales. (reported by Ken Causey)
1234 * fixed some bugs revealed by Paul Dietz' test suite:
1235 ** INCF, DECF and REMF evaluate their place form as specified in
1237 ** FORMATTER expands ~{ ~} iteration directives with V or #
1238 parameters correctly.
1239 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1240 consequent uses no arguments correctly.
1241 ** the system has a partial understanding of the (COMPLEX RATIO)
1243 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1246 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1247 * new feature: a build-time option (controlled by the :SB-UNICODE
1248 keyword feature, enabled by default) for building the system with
1249 support for the entire 21-bit character space defined by the
1251 * new feature: the system now has rudimentary external-format
1252 support; the primary user-visible change at this time is that
1253 characters with the high bit set (such as lower-case-e-acute) will
1254 print correctly to a terminal in a UTF-8 environment.
1255 * minor incompatible change: BASE-CHAR no longer names a class;
1256 however, CHARACTER continues to do so, as required by ANSI.
1257 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1258 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1259 has been moved to the SB-EXT package (temporarily re-exported from
1261 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1262 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1264 * fixed bug #351: better error-handlind and reporting for malformed
1266 * fixed bug #350: bignum-printing is now more memory-efficient,
1267 allowing printing of very large bignums, eg. (expt 2 10000000).
1268 (reported by Bruno Haible)
1269 * fixed bug #302: better primitive-type selection for intersection
1271 * fixed bug #308: non-graphic characters now all have names, as
1272 required. (reported by Bruno Haible)
1273 * bug fix: redefining a class with different superclasses now correctly
1274 removes it from the direct-subclasses of its previous superclasses.
1275 (reported by David Morse)
1276 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1277 new value now works. (reported by Bruno Haible)
1278 * bug fix: correct canonicalization of multiple non-standard slot
1279 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1280 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1281 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1282 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1283 now exists, an signals an error.
1284 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1285 during macroexpansion for non-iteration variables as well. (reported
1286 by Bruno Haible for CMUCL)
1287 * bug fix: Cyclic structures and unprintable objects in compiler
1288 messages no longer cause errors. (reported by Bruno Haible)
1289 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1290 and READ-FROM-STRING all now return a primary value of NIL if
1291 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1292 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1293 (reported by Bruno Haible for CMUCL)
1294 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1295 do not designate a condition. (reported by Bruno Haible for
1297 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1298 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1299 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1300 permitting resolution in favour of any of the conflicting symbols.
1301 (reported by Bruno Haible for CMUCL)
1302 * bug fix: EQUAL compiler optimizations is less aggressive on
1303 strings which can potentially compare true despite having distinct
1304 specialized array element types.
1305 * bug fix: unit enumerations can be defined without dividing by
1306 zero. (thanks to Vincent Arkesteijn)
1307 * FORMAT compile-time argument count checking has been enhanced.
1308 (report from Bruno Haible for CMUCL)
1309 * a partial workaround for the bug 262: the compiler does not try to
1310 inline-expand a local function doing RETURN-FROM from a deleted
1311 BLOCK. (thanks to Peter Denno for the bug report and to David
1312 Wragg for the simple test case)
1313 * fixed some bugs revealed by Paul Dietz' test suite:
1314 ** RENAME-PACKAGE allows all package designators as new package
1316 ** constraint propagation and lambda variable substitution are
1317 more cautious in dealing with partially deleted code.
1318 ** compiler sometimes lost reoptimization passes.
1319 ** CERROR, given a condition as condition designator, passes its
1320 remaining arguments to the continue format control without
1322 ** Case-altering FORMAT directives work correctly on non-ASCII
1324 ** The REAL type specifier handles bounds outside the floating
1325 point ranges without signalling FLOATING-POINT-OVERFLOW.
1326 ** Functions with IR1-transformations can create intercomponent
1327 references to global functions.
1328 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1330 ** FORMAT ~R treats a nil value for its first parameter correctly.
1332 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1333 * enhancement: saving cores with foreign code loaded is now
1334 supported on x86/NetBSD and sparc/Linux in addition to the previously
1335 supported platforms.
1336 * bug fix: on some platforms repeated installations caused multiple
1337 copies of HTML documentation to be installed -- should not happen
1338 any more. (reported by Stefan Scholl)
1339 * bug fix: parsing self-recursive alien record types multiple times
1340 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1341 original patch by Helmut Eller for CMUCL)
1342 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1343 (thanks to Richard Kreuter)
1344 * bug fix: defining classes whose accessors are methods on existing
1345 generic functions in other (locked) packages no longer signals
1346 bogus package lock violations. (reported by François-René Rideau)
1347 * bug fix: special variables as DEFMETHOD parameters no longer have
1348 associated bogus type declarations. (reported by David Wragg and
1350 * bug fix: read-write consistency on streams of element-type
1351 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1352 * bug fix: redefiniton of the only method of a generic function with
1353 no DEFGENERIC no longer emits a full WARNING. In addition,
1354 redefinition of generic functions with no DEFGENERIC to an
1355 incompatible lambda list now signals an error. (thanks to Zach
1357 * bug fix: DEFGENERIC now works even when there's a function of the
1358 same name in an enclosing lexical environment. (thanks to Zach
1360 * fixed compiler failure, caused by instrumenting code during
1361 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1362 * optimization: added loop analysis and improved register allocation
1363 * fixed some bugs revealed by Paul Dietz' test suite:
1364 ** POSITION on displaced vectors with non-zero displacement
1365 returns the right answer.
1366 ** (SIMPLE-STRING) is a valid type specifier for sequence
1368 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1370 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1371 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1373 ** The justification version of the FORMAT ~< directive treats
1374 non-zero minpad parameter correctly.
1376 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1377 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1378 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1379 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1380 the supported interface.
1381 * new feature: Single-stepping of code compiled with DEBUG 2 or
1382 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1383 * new feature: saving cores with foreign code loaded is now
1384 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1385 Timothy Moore's work for CMUCL)
1386 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1387 parameters to * instead of NIL if no initform is supplied.
1388 (reported by Johan Bockgaard)
1389 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1390 correctly when FUNCALL appears as the car of the form. Note:
1391 despite this FUNCALL forms are not currently subject to
1392 compiler-macro expansion. (port of Raymond Toy's fix for the
1393 same from CMUCL, reported by Johan Bockgaard)
1394 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1395 (thanks for Teemu Kalvas)
1396 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1397 case instead of terminating the iteration (thanks for Julian
1398 Squires, Sean Champ and Raymond Toy)
1399 * bug fix: incorrect expansion of defgeneric that caused a style
1400 warning. (thanks for Zach Beane)
1401 * bug fix: slot accessor effective method computation works properly
1402 for classes with multiple non-standard applicable methods on
1403 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1404 * on x86 compiler supports stack allocation of results of LIST and
1405 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1406 CMUCL implementation by Gerd Moellmann)
1407 * fixed some bugs revealed by Paul Dietz' test suite:
1408 ** FORMAT strings with both the ~<~:;~> form of the justification
1409 directive and pretty-printing directives cause an error.
1411 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1412 * incompatible change: the internal functions
1413 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1414 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1415 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1416 instead of the old functions.
1417 * new feature: on platforms where "dladdr" is available foreign
1418 function names now appear in backtraces. (based on Helmut Eller's
1420 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1421 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1423 * documentation: ASDF manual is now distributed with SBCL.
1424 * bug fix: SBCL can now load its contributed modules with REQUIRE
1425 even if the system-provided entries have been removed from
1426 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1428 * bug fix: backtraces involving undefined functions or assembly
1429 routines are more informative. (thanks to Brian Downing)
1430 * bug fix: mutually referent alien structures now work correctly.
1431 (reported by Rick Taube)
1432 * bug fix: structures defined by WITH-ALIEN can be referred to
1433 within other definitions in the same WITH-ALIEN.
1434 * bug fix: division operators (MOD, TRUNCATE and the like) with
1435 constant zero divisors and integer dividends no longer generate
1437 * bug fix: provide default methods for INPUT-STREAM-P and
1438 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1439 * bug fix: improve the detection and resolution of MOP metacycles
1440 (where effective method computation depends on the generic function
1441 having its effective method computed). (reported by Bruno Haible)
1442 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1443 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1444 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1445 * bug fix: pathnames with relative directory components can be
1446 represented relative to default pathnames.
1447 * optimization: in taking the GCD of bignums, reduce the two bignums
1448 to approximately the same size (using Euclid's algorithm) before
1449 applying the more sophisticated binary GCD. (thanks to Juho
1451 * optimization: COUNT on bitvectors now operates word-at-a-time.
1452 * optimization: ASH with a positive, but not necessarily constant,
1453 (leftwards) shift, when in a modular context, is compiled to a
1455 * fixed some bugs revealed by Paul Dietz' test suite:
1456 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1457 if the corresponding argument is NIL.
1458 ** FORMAT directives accepting a minpad parameter treat negative
1460 ** literal commas as character parameters to FORMAT directives are
1462 ** literal spaces directly after ~<Newline> directives within a
1463 format-logical-block (~:< ~@:>) do not induce :FILL-style
1464 conditional newlines.
1465 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1466 as specified: it no longer includes conditional newlines).
1467 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1469 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1470 :PER-LINE-PREFIX argument does not evaluate to a string.
1472 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1473 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1474 the manual for details; for now, package locks can be disabled by
1475 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1476 if no major problems are found then it is likely that they will be
1477 compiled in unconditionally.
1478 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1479 unsupported operators on all platforms. To load a shared library
1480 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1481 object file, link it into a shared library outside of SBCL and
1482 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1484 * bug workaround: ROOM T and the GC were not working together
1485 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1486 reporting facility aren't preserved. That reporting has been
1487 disabled, so now until and unless someone figures out how to make it
1488 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1489 * minor incompatible change: as threatened around sbcl-0.8.0, the
1490 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1491 The symbols are also exported from SB-PCL for backwards
1492 compatibility, but more so than before SB-PCL should be treated as
1493 an implementation-internal package.
1494 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1496 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1497 It is known as of this release not to work on the Alpha, however.
1498 * fixed bug #167: errors signalled due to illegal syntax in method
1499 bodies are now more legible.
1500 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1501 designators and can hence be used with TYPEP.
1502 * fixed bug #333: CHECK-TYPE now ensures that the type error
1503 signalled, if any, has the right object to be accessed by
1504 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1505 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1506 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1508 * fixed bug #334: programmatic addition of slots using specialized
1509 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1510 and :ALLOCATION :CLASS slots.
1511 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1512 float exponents. (rereported by Peter Seibel)
1513 * fixed a bug: #\Space (and other whitespace characters) are no
1514 longer considered to be macro characters in standard syntax by
1515 GET-MACRO-CHARACTER.
1516 * fixed bug: initialization of condition class metaobjects no longer
1517 causes an instance of the condition to be created. (reported by
1519 * fixed bug: it is now possible to have slots such that
1520 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1521 description is neither :INSTANCE nor :CLASS.
1522 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1523 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1524 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1525 non-standard SLOT-DEFINITION-ALLOCATION values.
1526 * fixed bugs: various race conditions handling exiting threads.
1527 CL-PPCRE's thread torture test now passes.
1528 * fixed arguably-a-bug: GC time proportional to number of threads in
1529 system even when most of them are idle
1530 * optimization: improved performance of BIT and SBIT on bit-vectors.
1531 * .fasl file incompatibility: The fasl file version number has
1532 been incremented because of changes associated with package locks.
1534 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1535 * minor incompatible change: the system no longer provides
1536 optimization hints (conditions which are TYPEP
1537 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1538 mode; these hints are still emitted when higher SPEED optimization
1540 * new contrib module: a sampling profiler (profiling by statistical
1541 sampling, rather than by instrumenting functions) is available as
1542 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1543 * the behaviour of the standard function ED is now customizeable by
1544 third parties through a hook variable: see ED's documentation
1545 string for information on the protocol.
1546 * the compiler no longer emits efficiency notes for (FUNCALL X)
1547 when the type of X is uncertain under default optimization
1549 * fixed bug 276: mutating a binding of a specialized parameter to a
1550 method to something that is not TYPEP the specializer is now
1552 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1553 exhibit the required behaviour on the x86 platform. (thanks to
1554 Peter van Eynde, Eric Marsden and Bruno Haible)
1555 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1556 even for difficult arguments. (reported by Peter Graves)
1557 * fixed bug 141a: the backquote printer now descends quoted
1559 * fixed another bug in backquote printing: no more destructive
1560 modification of the form's list structure. (reported by Brian
1562 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1563 move between its address being taken and the call to
1564 interrupt_thread, fixing a crashing race condition.
1565 * the SB-POSIX contrib implementation has been adjusted so that it
1566 no longer exhibits ridiculously poor performance when constructing
1567 instances corresponding to C structs.
1569 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1570 * minor incompatible change: the sb-grovel contrib now treats C
1571 structures as alien (in the sense of SB-ALIEN) objects rather than
1572 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1573 has implications for memory management of client code
1574 (sb-grovel-returned objects must now be manually managed) and for
1575 type safety (alien objects now have full types).
1576 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1577 used to control emission of compiler diagnostics, rather than the
1578 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1579 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1580 quality should be considered deprecated.
1581 * install.sh now installs the user manual as well
1582 * (not quite a new documentable feature, but worth considering in
1583 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1584 beginnings of a semantically meaningful condition hierarchy is
1585 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1587 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1588 designator as the defaults argument.
1589 * fixed bug: Displaced arrays whose displaced-to array has become
1590 too small now cause ARRAY-DIMENSION to signal an error, providing
1591 for safer bounds-checking. (reported by Bruno Haible)
1592 * fixed bug: DEFCLASS slot definitions with identical :READER and
1593 :WRITER names now signal a reasonable error. (reported by Thomas
1595 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1597 * fixed bug: Invalid slot specification errors now print correctly.
1598 (thanks to Zach Beane)
1599 * fixed bug 320: Shared to local slot value transfers in class
1600 redefinitions now happen corrently from superclasses as
1601 well. (reported by Bruno Haible)
1602 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1604 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1605 improper lists correctly. (reported by Bruno Haible)
1606 * fixed bug 313: source-transform for <fun-name> was erroneously
1607 applied to a call of a value of a variable with name <fun-name>.
1608 (reported by Antonio Menezes Leitao)
1609 * fixed bug 307: The obsolete instance protocol ensures that
1610 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1611 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1612 remove dead unknown-values globs from the middle of the stack.
1613 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1614 easier to resume long computations after using *BREAK-ON-SIGNALS*
1615 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1616 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1617 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1618 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1619 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1620 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1622 * fixed a (fairly theoretical) bug in string printing: if
1623 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1624 does not have array-element-type equal to the most general string
1626 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1627 function name. (thanks to Nikodemus Siivola)
1628 * fixed bug: compile-time format string checker failed on
1629 non-closed ~{. (reported by Thomas F Burdick)
1630 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1631 don't warn on legal loop code involving more than one aggregate
1632 boolean. (thanks to Nikodemus Siivola)
1633 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1634 in keyword position to MAKE-PACKAGE was not being reported as
1636 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1637 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1639 * fixed bugs 280 and 312: the checking for multiple definitions in a
1640 file is less likely to become confused by uses of inline
1642 * fixed bug: the #S reader macro performs the keyword coercion
1643 specified for slot names. (reported by Kalle Niemitalo)
1644 * fixed bug: lambda lists may contain symbols whose names start with
1645 & but are not lambda-list-keywords; their occurrence triggers a
1647 * fixed bug 321: define-method-combination argument lambda lists do
1648 not cause invalid code to be generated when &optional or &aux
1649 variables have default values. (reported by Bruno Haible)
1650 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1651 have CLOS classes; this appears also to have fixed reported
1652 instability in stack exhaustion detection.
1653 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1654 MUFFLE-WARNING when no associated restart is present is now
1656 * optimization: rearranged the expansion of various defining macros
1657 so that each expands into only one top-level form in a
1658 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1660 * optimization: used a previously little-used slot in symbols to
1661 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1663 * fixed some bugs revealed by Paul Dietz' test suite:
1664 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1666 ** Fixed an optimization bug related to inheritance of initforms
1667 from local to shared slots.
1668 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1669 ** CAST optimizer forgot to flush argument derived type.
1670 ** print/read consistency on floats is now orders of magnitude
1671 more likely. (thanks also to Bruno Haible for a similar report
1673 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1674 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1675 for objects of type REAL. Make it so.
1676 ** SXHASH is specified (infelicitously) to respect similarity,
1677 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1678 it so. (thanks to Markus Ziegler)
1679 ** on the Alpha, the compiler succeeds in compiling functions
1680 returning a known number of arguments greater than 63.
1681 ** fixed handling of invalid NIL arguments in keyword position
1683 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1685 ** provided a readably-printable representation for RANDOM-STATE
1687 ** ensured that pathnames, where they have a namestring, always
1688 print using #P"..." syntax.
1690 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1691 * Support for the forthcoming 2.0 version of the NetBSD kernel
1692 running on x86 hardware has been added. (thanks to Perry
1693 E. Metzger most immediately, and others for their past work)
1694 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1695 current release version; SBCL's OpenBSD support had been broken
1696 since about the time of OpenBSD's switch to ELF binary format.)
1697 * [placeholder for DX summary]
1698 ** user code with &REST lists declared dynamic-extent, under high
1699 speed or space and low safety and debug optimization policy.
1700 * The manual has been converted to Texinfo format and the debugger
1701 chapter from the cmucl manual has been added.
1702 * A facility has been added to extract documentation strings from
1703 sbcl and store them as Texinfo-formatted snippets for inclusion in
1704 the manual (via Texinfo's @include directive)
1705 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1706 the test case to Dave Roberts)
1707 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1708 pointers. (reported by Sean Ross)
1709 * bug fix: PROFILE output is printed nicely even for large numerical
1710 values. (thanks to Zach Beane)
1711 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1712 greater than 32 handle EOF correctly.
1713 * bug fix: on X86 an immediate argument of the IMUL instruction is
1714 correctly printed in disassembly. (thanks to Lutz Euler)
1715 * bug fix: class slots in redefined classes preserve their old
1716 values. (thanks to Bruno Haible and Nikodemus Siivola)
1717 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1718 succeeds. (reported by Marco Baringer)
1719 * fixed some bugs revealed by Paul Dietz' test suite:
1720 ** READ-SEQUENCE now works on ECHO-STREAMs.
1721 ** RATIONALIZE works more according to its specification. (thanks
1724 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1725 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1726 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1727 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1728 should matter to you only if you rebind the printer control
1729 variables and then find you want different bindings in the
1730 debugger than in the ordinary execution of your program.)
1731 * The runtime build system has been tweaked to support building
1732 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1733 assembler and linker. (thanks to Nikodemus Siivola)
1734 * Unbound, undefined, undeclared variables now trigger full
1735 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1736 more often programmer error than deliberate exploitation of undefined
1738 * optimization: the hash algorithm for strings has changed to one
1739 that is less vulnerable to spurious collisions. (thanks to Juho
1741 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1742 needless bounds checking. (thanks to Juho Snellman)
1743 * optimization: implemented multiplication as a modular
1744 (UNSIGNED-BYTE 32) operation on the PPC backend.
1745 * fixed some bugs revealed by Paul Dietz' test suite:
1746 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1748 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1749 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1750 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1752 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1754 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1756 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1757 their output stream on EOF from read.
1758 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1759 have been read to end-of-file.
1760 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1762 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1763 description of determination of which consecutive characters
1765 ** Printing the "Space" character with escaping on now yields "#\\ ",
1766 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1767 ** Reading floating-point numbers with *READ-BASE* set to a number
1768 less than 10 works correctly.
1769 ** Reading floating-point numbers with *READ-BASE* set to a number
1770 more than 10 works correctly.
1771 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1772 the readtable currently in effect.
1774 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1775 * minor incompatible change: parsing of namestrings on a physical
1776 (Unix) host has changed; numbers after the final #\. in a
1777 namestring are no longer interpreted as a version field. This is
1778 intented to be largely invisible to the user, except that the
1779 meaning of the namestring "*.*.*" has changed: it now refers to a
1780 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1781 should usually be replaced by
1782 (make-pathname :name :wild :type :wild :version :wild)
1783 with the added benefit that this is more likely to be portable.
1784 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1785 option to OPEN now signals an error if the file being opened
1786 exists; this may have an impact on existing code.
1787 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1788 signals received, on the PowerPC platforms (both Linux and
1789 Darwin). (thanks to Pierre Mai for pointing out the location of
1791 * several fixes on OS X: The system now builds and runs cleanly on
1792 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1793 (thanks to Brian Mastenbrook)
1794 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1795 second-resolution: integer multiples of 1/3600 between -24 and 24.
1796 (thanks to Vincent Arkesteijn)
1797 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1798 type when called with 1 argument; PEEK-CHAR checked type of
1799 PEEK-TYPE only after having read first character from a
1800 stream. (reported by Peter Graves)
1801 * bug fix: the garbage collector now has much better locality
1802 behaviour, and in particular no longer treats all memory as being
1803 exhausted when many small objects point to each other in a deeply
1805 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1806 recognized as being TYPEP their class.
1807 * bug fix: the PUSHNEW documentation string has been corrected.
1808 (thanks to Vincent Arkesteijn)
1809 * bug fix: defaulting of the value for the last of an atypically
1810 large number of multiple values being bound was not being
1811 performed correctly on the Alpha or PPC platforms
1812 * optimization: implemented multiplication as a modular
1813 (UNSIGNED-BYTE 32) operation on the x86 backend.
1814 * optimization: SEARCH on simple-base-strings can now be open-coded.
1815 (see also contrib/compiler-extras.lisp for inspiration for
1816 teaching the compiler about the Boyer-Moore algorithm).
1817 * value, returned by MAX (and MIN) called with several EQUALP, but
1818 not EQL, arguments now does not depend on compiler settings.
1819 * fixed some bugs revealed by Paul Dietz' test suite:
1820 ** in stack analysis liveness information is propagated from
1821 non-local entry points.
1822 ** pathwise CAST removing failed when the CAST node did not start
1824 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1825 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1827 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1828 host is already defined.
1829 ** RENAME-FILE works on streams instead of signalling an internal
1831 ** PEEK-CHAR uses the current readtable when determining whether
1832 or not a character is whitespace.
1833 ** MERGE-PATHNAMES handles the case when the pathname does not
1834 specify a name while the default-pathname specifies a version.
1835 ** Pathnames now stand a better chance of respecting print/read
1837 ** Attempting to use standardized file system operators with a
1838 pathname with invalid :DIRECTORY components signals a
1840 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1842 ** DIRECTORY on logical pathnames is more correct.
1843 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1844 signal a TYPE-ERROR if their argument is not a stream.
1845 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1846 designator argument does not designate a stream.
1847 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1848 examining the synonym.
1849 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1851 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1852 element-type larger than ([UN]SIGNED-BYTE 32).
1854 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1855 * When built with the :SB-FUTEX feature, threaded builds now take
1856 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1857 for faster/more reliable mutex and condition variable support.
1858 * Incompatible change (but one you probably shouldn't have been using
1859 anyway): the interface and code for arbitrating between multiple
1860 threads in the same user session has been redesigned.
1861 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1862 error when called without an explicit environment argument.
1863 (thanks to Nikodemus Siivola)
1864 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1865 confused about their position. (thanks to Adam Warner and Gerd
1867 * bug fix: Pretty printing backquoted forms with unquotations in the
1868 argument list position of various code constructs such as LAMBDA
1869 now works correctly. (reported by Paul Dietz)
1870 * bug fix: Pretty printing unquotations no longer loses all
1871 stream position information.
1872 * optimization: performance of string output streams is now less
1873 poor for multiple small sequence writes.
1874 * optimization: performance of CSUBTYPEP in the presence of complex
1875 expressions involving CONS and NOT many times has been improved.
1876 (reported by Paul Dietz)
1877 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1879 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1880 streams for the REPL, for the debugger, and so on. (thanks to
1882 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1883 * fixed some bugs revealed by Paul Dietz' test suite:
1884 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1885 is ignored if it is too small, rather than propagating through
1886 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1887 ** extremely complex negations of CONS types were not being
1888 sufficiently canonicalized, leading to inconsistencies in
1890 ** VALUES tranformer lost derived type.
1892 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1893 * fixed a bootstrapping bug: the build process no longer assumes
1894 that the various BOOLE-related constants have the same value in
1895 host and target lisps. (noted by Paul Dietz' test suite on an
1896 SBCL binary built from CLISP)
1897 * The system can now be dynamically linked on the MIPS platform,
1898 which enables dynamic loading of foreign code from Lisp. (thanks
1899 to Ralf Baechle for discussions on the MIPS ABI)
1900 * The system now records debugging information for its own source
1901 files in a filesystem-position-independent manner, relative to
1902 the "SYS" logical host.
1903 * fixed a compiler bug: MV-LET convertion did not check references
1904 to the "max args" entry point. (reported by Brian Downing)
1905 * tweaked disassembly notes to be less confident about proclaiming
1906 some instruction as an LRA. (thanks to Brian Downing)
1907 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1908 now each have their own history, command character, and other
1909 characteristics. (thanks to David Lichteblau)
1910 * fixed some bugs revealed by Paul Dietz' test suite:
1911 ** compiler failure in compiling LOGAND expressions including a
1913 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1914 shift greater than 32.
1915 ** FLUSH-DEST did not mark blocks for type check regeneration.
1916 ** HANDLER-CASE failed to accept declarations in handler clauses
1917 in some circumstances.
1919 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1920 * New code in contrib/sb-introspect (still probably not entirely
1921 stable yet) provides some support for smart Lisp development
1922 environments like SLIME.
1923 * The conditions signalled for errors occurring when loading .fasl
1924 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1925 in a way which should help ASDF recover gracefully.
1926 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1927 stuff has been cleaned up. If you code contrib/ stuff, this might
1928 affect you, and you can look at contrib/README, contrib/STANDARDS,
1929 and/or the 0.8.4.27 diff to check.
1930 * In full calls the compiler now does not generate checks for declared
1931 argument types for all arguments.
1932 * various threading fixes
1933 ** and some experimental patches which didn't make it into
1934 the main tree for this release, but which are shipped in
1935 contrib/experimental-thread.patch as a possible fix for some
1936 failures (deadlock, spinning...) in GC-intensive multithreaded
1938 * fixed PPC build problem (source code incompatibility of different
1939 library versions): added offsetof() hackery which attempts to divine
1940 where glibc maintainers put uc_mcontext today
1941 * fixed bug 282: compiler does not trust type assertions while passing
1942 arguments to a full call.
1943 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1944 &OPTIONAL VALUES type parameter.
1945 * fix bug 214: algorithm for noting rejected templates is now more
1946 similar to that of template seletion. (also reported by rydis on
1948 * fixed bug 141b: printing backquoted information readably and prettily
1949 inserts a space where necessary.
1950 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1951 (CHARACTER 10) are now reported as errors, rather than propagated
1952 as unknown types. (reported by piso on #lisp)
1953 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1954 with values NIL and :ERROR. (thanks to Milan Zamazal)
1955 * fixed bug 191c: CLOS now does proper keyword argument checking as
1956 described in CLHS 7.6.5 and 7.6.5.1.
1957 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1958 longer raise an error; further, using a list as a for-as-arithmetic
1959 counter now raises a meaningful error.
1960 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1961 now understood by sequence creation functions such as MAKE-SEQUENCE
1963 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1964 a STRING-INPUT-STREAM.
1965 * compiler enhancement: SIGNUM is now better able to derive the type
1967 * type declarations inside WITH-SLOTS are checked. (reported by
1969 * fixed some bugs revealed by Paul Dietz' test suite:
1970 ** incorrect optimization of TRUNCATE for a positive first
1971 argument and negative second.
1972 ** compiler failure in let-convertion during flushing dead code.
1973 ** compiler failure while deriving type of TRUNCATE on an
1974 interval, containing 0.
1975 ** ASH of a negative bignum by a negative bignum count now returns
1977 ** intersection of CONS types now canonicalizes properly, fixing
1978 inconsistencies in SUBTYPEP.
1980 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1981 * incompatible change: The --disable-debugger command line
1982 option now clobbers the debugger at a more fundamental
1983 level, by redefining #'INVOKE-DEBUGGER instead of by
1984 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1985 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1986 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1987 enter the debugger REPL and then suffer recursive errors
1988 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1989 while under the new system, BREAK in --disable-debugger mode
1990 terminates the system just as an unhandled error would.
1991 * fixed compiler performance when processing loops with a step >1;
1992 * bug fix: DOCUMENTATION now retrieves generic function
1993 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1994 support has been systematized, and now supports the methods
1995 specified by ANSI, along with a default method and a method for
1996 slot documentation. (reported by Nathan Froyd)
1997 * bug fix: effective methods associated with a generic function are
1998 no longer cached over a change of that generic function's method
1999 combination. (reported by Andreas Fuchs)
2000 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2001 names a non-existent directory. (thanks to Andreas Fuchs)
2002 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2003 the PPC platform, lead to stack corruption; this has been fixed.
2004 (reported by Rainer Joswig)
2005 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2006 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2007 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2008 arithmetic in general, is now much more reliable on the PPC
2010 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2011 platform now returns the right answer.
2012 * optimization: restored some effective method precomputation in
2013 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2014 precomputation is now tunable.
2015 * optimization: compiler-internal data structure use has been
2016 reviewed, and changes have been made that should improve the
2017 performance of the compiler by about 20%.
2018 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2019 simple-base-strings and simple-bit-vectors is improved.
2020 * optimization: the optimization of 32-bit logical and arithmetic
2021 functions introduced in version 0.8.3 on the x86 has been
2022 implemented on the mips, ppc and sparc platforms; an
2023 implementation of the same facility, but for 64-bit arithmetic,
2024 has been added for the alpha.
2025 * microoptimization: the compiler is better able to make use of the
2026 x86 LEA instruction for multiplication by constants.
2027 * bug fix: in some situations compiler did not report usage of
2028 generic arithmetic in (SPEED 3) policy.
2029 * bug 145b fix: compiler used wrong type specifier while converting
2030 MEMBER-types to numeric.
2031 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2033 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2034 index argument now works properly on the Alpha platform.
2035 * bug fix: floating point exception treatment on the Alpha platform
2037 * bug fix: FILE-POSITION works much better on string input and
2038 output streams. (thanks to Nikodemus Siivola)
2039 * bug fix: many threading/garbage collection symptoms sorted.
2040 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2041 might be pseudo-atomic.
2042 * internal change: Stopping for GC is now done with signals not
2043 ptrace. GC is now done in whichever thread wanted it, instead of
2045 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2047 * bug fix: result form in DO is not contained in the implicit
2049 * incompatible change: ICR structure is changed; the value part of
2050 CONTINUATION is now called LVAR; corresponding functions are
2051 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2052 * added type deriver for ISQRT (thanks to Robert E. Brown).
2053 * bug fix: better support for loading from the command line when an
2054 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2056 * fixed some bugs revealed by Paul Dietz' test suite:
2057 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2058 ** ROUND and FROUND now give the right answer when given very
2059 small float arguments.
2060 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2062 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2063 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2064 ** LCM should return a non-negative integer.
2065 ** PARSE-INTEGER returned the index of a terminator instead of the
2066 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2067 ** PARSE-INTEGER returned an incorrect index being applied to a
2069 ** LCM with two arguments of 0 returns 0 rather than signalling
2071 ** unsigned addition of a 32-bit constant with the high bit set no
2072 longer causes an internal compiler error on the x86.
2073 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2074 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2075 with negative last argument.
2076 ** byte specifiers with zero size and position no longer cause
2077 an error during type derivation.
2078 ** bignum multiplication on the Alpha platform now returns the
2080 * porting: The system now builds on SuSE AMD64, although it still
2081 generates a 32-bit binary.
2082 * .fasl file incompatibility: The fasl file version number has
2083 been incremented (because of the changes to internal compiler
2084 data structures referred to above).
2086 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2087 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2088 more accurately, on the Darwin kernel running on PowerPC hardware.
2089 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2090 * Compiler code deletion notes now signal a condition of type
2091 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2092 an associated MUFFLE-WARNING restart.
2093 * The compiler now performs limited argument count validation of
2094 constant format strings in FORMAT, and where appropriate in ERROR,
2095 CERROR and WARN. (thanks to Gerd Moellmann)
2096 * New ASDF-INSTALL contrib can be used for automatic download and
2097 installation of third-party Lisp code from CCLAN or other sites
2099 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2100 which forces another thread to execute a function supplied by the
2102 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2103 now accept and act upon their :ELEMENT-TYPE keyword argument.
2104 (reported by Martin Atzmueller, Edi Weitz)
2105 * bug fix: FILE-POSITION now accepts position designators up to
2106 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2107 is the greater. (thanks to Patrik Nordebo)
2108 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2109 Kalle Olavi Niemitalo)
2110 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2111 printable. (reported by Eric Marsden)
2112 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2113 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2114 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2115 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2116 (thanks to Stig Sandoe)
2117 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2118 UNREAD-CHAR now correctly includes the unread character in the
2119 target sequence. (thanks to Gerd Moellmann)
2120 * bug fix in threaded builds: the system can now be suspended and
2121 resumed by shell job control with minimal disruption.
2122 * bug fixes in times and timezones >2038AD
2123 * better handling of "where is GNU make?" problem in build scripts
2124 (thanks to Nikodemus Siivola)
2125 * new optimization: inside a named function any reference to a
2126 function with the same name is considered to be a self-reference;
2127 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2128 optimization quality.
2129 * new optimization on x86: logical functions and + now have
2130 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2131 used when the result is truncated to 32 bits.
2132 * VALUES declaration is partially enabled.
2133 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2134 * fixed some bugs revealed by Paul Dietz' test suite:
2135 ** The system now obeys the constraint imposed by
2136 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2137 types form a lattice under type intersection.
2138 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2139 ** ASSOC now ignores NIL elements in an alist.
2140 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2141 and (1+ MOST-POSITIVE-FIXNUM) answers.
2142 ** The addition of a method with invalid qualifiers to a generic
2143 function does not cause an error to be signalled immediately;
2144 a warning is signalled, and the error is generated only on
2145 calling the generic function.
2146 * changed .fasl file version number, in part to add type codes for
2147 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2148 obscure ANSI requirements
2150 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2151 * fixed bug 148: failure to inline-expand a local function left
2152 garbage, confusing the compiler.
2153 * fixed bugs 3cd: structure slot readers perform type check if the
2154 slot can have an invalid value (i.e. it is either not initialized
2155 or can be written with a less specific slot writer).
2156 * bug fix: the compiler now traps array references to elements off
2157 the end of an array; previously, the bounds checking in some
2158 circumstances could go off-by-one.
2159 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2161 * type declarations for array element types now obey the description
2162 on the CLHS page "Declaration TYPE", as per discussions on
2163 sbcl-help around 2003-05-08. This means that a declaration
2164 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2165 declaration, all references to BAR will be asserted or assumed
2166 (with THE, so dependent on compiler policy) to involve objects of
2167 type FOO. Note that no such declaration is implied in
2168 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2169 * declared types of functions from the "Conditions"
2170 chapter. (reported by Paul Dietz)
2171 * bug fix: CERROR accepts a function as its first argument.
2172 * bug fix: NTH an NTHCDR accept a bignum as index
2173 arguments. (reported by Adam Warner)
2174 * optimization: character compare routines now optimize comparing
2175 against a constant character. (reported by Gilbert Baumann)
2176 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2177 argument now works properly on the MIPS platform.
2178 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2179 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2180 declared to have a functional type, does not cause an error
2182 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2183 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2184 * fixed bug: if last continuation of a deleted block has a
2185 destination, this destination should be deleted too. (reported by
2187 * fixed a bug in the bootstrap process: the host compiler's values
2188 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2189 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2190 test case from Patrik Nordebo)
2191 * improved the ability of the disassembler on the PPC platform to
2192 provide helpful disassembly notes.
2193 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2194 the class in more cases than previously.
2195 * bug fix: FILE-POSITION now understands :START and :END for
2196 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2197 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2198 8) data. (thanks to David Lichteblau)
2199 * bug fix: it is possible to add a method to a generic function
2200 without lambda list.
2201 * bug fix: reader failed to signal END-OF-FILE inside an
2202 object representation. (reported by Nikodemus Siivola)
2203 * fixed some bugs revealed by Paul Dietz' test suite:
2204 ** LAST and [N]BUTLAST should accept a bignum.
2205 ** condition slot accessors are methods.
2206 ** (VECTOR NIL) is a subtype of STRING.
2208 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2209 * minor incompatible change: some nonsensical specialized lambda
2210 lists (used in DEFMETHOD) which were previously ignored now signal
2212 * minor incompatible change: the system is now aware of the types of
2213 variables in the COMMON-LISP package, and will signal errors for
2214 most violations of these type constraints (where previously they
2215 were silently accepted).
2216 * minor incompatible change: COMPILE-FILE now uses the freedom
2217 afforded (ANSI 3.2.2.3) to use derived function types for
2218 functions defined in the same file. This also permits the system
2219 to warn on static type mismatches and function
2220 redefinition. (Currently it does not work with high DEBUG level.)
2221 * minor incompatible change: VALUES declaration is disabled.
2222 * When issuing notes, the compiler now signals a condition of type
2223 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2224 restart for use in user handlers. It is expected that the
2225 COMPILER-NOTE condition will eventually become a condition
2226 supertype to a hierarchy of note types, which will then be
2227 handleable in a similar fashion. However, at the moment, no such
2228 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2229 but it's an implementation detail, not a classification for the
2231 * Changes in type checking closed the following bugs:
2232 ** type checking of unused values (192b, 194d, 203);
2233 ** template selection based on unsafe type assertions (192c, 236);
2234 ** type checking in branches (194bc).
2235 * A short form of VALUES type specifier has ANSI meaning (it has
2236 increased the number of situations when SBCL cannot perform type
2238 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2239 keywords or constants is permissible.
2240 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2241 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2242 classes. (thanks to Antonio Martinez)
2243 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2244 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2245 argument to a pretty-print stream. Instead, it leaves any such
2246 operations to DESCRIBE-OBJECT methods.
2247 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2249 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2250 lambda lists. (thanks to David Lichteblau)
2251 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2252 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2253 (thanks to Gerd Moellmann)
2254 * fixed bug 246: increased compilation speed of long
2255 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2257 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2259 * DEFCONSTANT now throws a condition of type
2260 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2261 constant to a non-EQL value; CONTINUE and ABORT restarts
2262 respectively change and preserve the value.
2263 * fixed bug 63: The code walker, part of the implementation of CLOS,
2264 is now better at handling symbol macros.
2265 * bug fix: There is no longer an internal implementation type named
2266 CL:LENGTH. (reported by Raymond Toy)
2267 * bug fix: In macro-like defining macros/special operators the
2268 implicit block does not enclose the lambda list.
2269 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2270 enclosed in parentheses) are not suitable as type specifiers, and
2271 their use properly signals an error now.
2272 * bug fix: An argument count mismatch for a type specifier in code
2273 being compiled no longer causes an unhandled error at compile
2274 time, but signals a compile-time warning.
2275 * fixed simple vector readable printing
2276 * bug fix: DESCRIBE takes more care over whether the class
2277 precedence list slot of a class is bound before accessing it.
2278 (reported by Markus Krummenacker)
2279 * bug fix: FORMATTER can successfully compile pretty-printer format
2280 strings which use variants of the ~* directive inside.
2281 * bug fix: SEARCH now applies its TEST predicate to the elements of
2282 the arguments in the correct order. (thanks to Wolfhard Buss)
2283 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2284 in type inference. (thanks to Robert E. Brown)
2285 * bug fix: Reading in symbols with an explicit package name of ""
2286 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2287 * fixed some bugs revealed by Paul Dietz' test suite:
2288 ** NIL is now allowed as a structure slot name.
2289 ** Arbitrary numbers, not just REALs, are allowed in certain
2290 circumstances in LOOP for-as-arithmetic clauses.
2291 ** Multiple class redefinitions before slot access no longer
2292 causes a type error.
2293 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2294 association between the name and a class.
2295 ** Generic functions with non-standard method-combination and over
2296 five methods all of which return constants no longer return NIL
2297 after the first few invocations. (thanks to Gerd Moellmann)
2298 ** CALL-NEXT-METHOD with no arguments now passes the original
2299 values of the arguments, even in the presence of assignment.
2300 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2302 ** DISASSEMBLE works with closures and funcallable instances.
2303 ** ADD-METHOD now returns the generic function, not the new method.
2304 ** FIND-METHOD signals an error if the lengths of the specializers
2305 is incompatible with the generic function, even if the ERRORP
2307 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2308 which its argument is a member.
2309 ** DEFCLASS only redefines the class named by its class-name
2310 argument if that name is the proper name of the class;
2311 otherwise, it creates a new class.
2312 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2313 of the UNBOUND-SLOT condition to the name of the slot.
2314 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2315 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2316 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2317 SLOT-MAKUNBOUND in the specified fashion.
2319 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2320 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2321 cross-compilation host. As a consequence, we can now bootstrap our
2322 way up to SBCL starting with a bare gcc toolchain and human-readable
2323 source code (first the source to CLISP, then the source to SBCL).
2324 * A contributed module containing a partial implementation of the
2325 simple-streams interface has been included. (thanks to Rudi
2327 * A contributed module implementing the RFC1321 Message Digest
2328 Algorithm, known as MD5, has been included.
2329 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2330 no longer has any effect, as the code controlled by this feature
2331 has been deleted. (As far as we know, no-one has ever built using
2332 this feature, and its semantics were confused in any case).
2333 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2334 work on conditions (as required by the ANSI specification),
2335 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2336 expected behaviour on conditions. Users should note, however,
2337 that such behaviour is not required by the ANSI specification,
2338 and so use of this behaviour may render their code unportable.
2339 * fixed some bugs revealed by Paul Dietz' test suite:
2340 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2342 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2343 ** SXHASH on condition objects no longer returns NIL.
2344 ** :ALLOCATION :CLASS slots are better treated; their values are
2345 updated on class redefinition, and initforms inherited from
2346 superclasses are applied.
2347 ** REMOVE-METHOD returns its generic function argument even when
2348 no method was removed.
2349 ** SHARED-INITIALIZE now initializes the values of the requested
2350 slots, including those with :ALLOCATION :CLASS.
2351 ** ALLOCATE-INSTANCE now works on structure classes defined via
2352 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2354 ** SLOT-EXISTS-P now works on conditions, as well as structures
2356 ** MAKE-LOAD-FORM now has the required methods on
2357 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2358 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2359 keyword, and now implements the SLOT-NAMES argument.
2360 ** methods with &OPTIONAL arguments no longer allow too many
2361 arguments to be passed in the call without error.
2362 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2363 option is consistent with the required arguments of the generic
2364 function lambda list.
2365 * bug fix: REQUIRE accepts a string designator. (Thanks to
2367 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2368 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2369 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2370 * bug fix: The compiler now checks for duplicated variables in macro
2372 * bug fix: SETQ on globals returns the correct value.
2373 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2374 not a COMPILER-ERROR (followed by some other strange error on
2375 choosing the CONTINUE restart).
2376 * bug fix: make.sh and friends are now more consistent in the way that
2377 they look for GNU "make".
2379 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2380 * experimental native threads support (on x86 Linux >=2.4 only).
2381 This is not compiled in by default: you need to add :SB-THREAD to
2382 the target features. See the "Beyond ANSI" chapter of the manual
2384 * fix for longstanding nonANSIism: The old distinction between
2385 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2386 The return value from CL:FIND-CLASS is now a CLOS class, and
2387 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2388 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2389 * An interface to the MetaObject Protocol, as described in Kiczales,
2390 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2391 MIT Press, 1991, is available from the SB-MOP package.
2392 * incompatible change: the SB-PCL package should now be considered
2393 a private implementation detail, and no longer a semi-private MOP
2395 * minor incompatible change: due to rearrangement for threads, the
2396 control stack and binding stack are now allocated at arbitrary
2397 addresses instead of being hardcoded per-port. Users affected by
2398 this probably have to be doing advanced things with shared
2399 libraries, and will know who they are.
2400 * minor incompatible change: Previously, all --eval forms used were
2401 processed with READ before any of them were processed with EVAL.
2402 Now each --eval form is processed with both READ and EVAL before
2403 the next --eval form is processed. (Thus package operations like
2404 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2405 work as the user might reasonably expect.)
2406 * minor incompatible change: *STANDARD-INPUT* is now only an
2407 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2409 * minor incompatible change: Y-OR-N-P is now character-oriented,
2410 not line oriented. Also, YES-OR-NO-P now works without errors.
2411 (thanks to Antonio Martinez)
2412 * sb-aclrepl module improvements: an integrated inspector, added
2413 repl features, and a bug fix to :trace command.
2414 * Known functions, which cannot be open coded by the backend, are
2415 considered to be able to check types of their arguments. (fixing
2416 a bug report by Nathan J. Froyd)
2417 * fixed a bug in computing method discriminating functions: It is
2418 now possible to define methods specialized on classes which have
2419 forward-referenced superclasses. (thanks to Gerd Moellmann)
2420 * fixed evaluation order in optional entries (reported by Gilbert
2422 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2423 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2424 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2425 required-by-AMOP NAME argument, as well as CLASS and
2426 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2427 * fixed bug 20: DEFMETHOD can define methods using names that are
2428 not the proper names of classes to designate class specializers.
2429 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2430 Unix file descriptors, instead of blowing up. (thanks to Antonio
2432 * Garbage collection refactoring: user-visible change is that a
2433 call to the GC function during WITHOUT-GCING will not do garbage
2434 collection until the end of the WITHOUT-GCING. If you were doing
2435 this you were probably losing anyway.
2436 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2437 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2438 TYPEP the latter but not the former.
2439 * The compiler issues a full WARNING for calls to undefined functions
2440 with names from the CL package.
2441 * MAP-INTO for a vector destination is open coded. (reported by
2442 Brian Downing on c.l.l)
2443 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2444 documentation string.
2445 * fixed some bugs revealed by Paul Dietz' test suite:
2446 ** COPY-ALIST now signals an error if its argument is a dotted
2448 ** Condition slots are now accessed more correctly in the presence
2449 of multiple initargs for a given slot.
2450 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2451 exclude restarts of the same name associated with a different
2453 ** DEFCLASS of forward-referenced classes with another
2454 forward-referenced class in the superclasses list no longer
2456 ** Condition slots are now initialized once each, not multiple
2457 times. (thanks to Gerd Moellmann)
2458 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2460 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2461 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2462 arguments contain duplicated elements.
2463 ** RESTART-CASE understands local macros.
2464 ** RESTART-CASE associates exactly its own restarts with a condition.
2465 ** ENDP in safe mode checks its argument to be of type LIST.
2466 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2467 in question is unbound.
2468 ** Optimized MAKE-INSTANCE functions no longer cause internal
2469 assertion failures in the presence of duplicate initargs.
2470 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2471 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2472 propagated even in the ctor optimized implementation of
2474 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2476 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2477 * a better implementation of SXHASH on (simple) bit vectors,
2478 measured both in execution speed and in distribution of results
2479 over the positive fixnums, has been installed. Likewise, a better
2480 implementation of EQUAL for simple bit vectors is now available.
2481 * fixed CEILING optimization for a divisor of form 2^k.
2482 * fixed bug 240 (emitting extra style warnings "using the lexical
2483 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2484 by Antonio Martinez)
2485 * fixed SXHASH, giving different results for NIL depending on type
2486 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2487 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2488 variable. (found by Rolf Wester)
2489 * SBCL does not ignore type declarations for special
2490 variables. (reported by rif on c.l.l 2003-03-05)
2491 * some bug fixes in contrib/sb-aclrepl/
2492 * fixed some bugs revealed by Paul Dietz' test suite:
2493 ** a bug in the CONS type specifier, whereby the CAR and CDR
2494 types got intertwined, has been fixed;
2495 ** the type system is now able to reason about the interaction
2496 between INTEGER and RATIO types more completely;
2497 ** APPEND, [N]REVERSE and NRECONC check that those their
2498 arguments, which must be proper lists, are really so;
2499 ** An array specialized to be unable to hold elements has been
2500 implemented, as required -- yes, really -- by ANSI;
2501 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2502 on malformed property lists;
2504 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2505 * incompatible packaging change: in line with Unix convention,
2506 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2507 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2508 * REQUIRE and PROVIDE are now optionally capable of doing something
2509 useful. See the documentation string for REQUIRE.
2510 * infrastructure for a managed SBCL contrib system: contributed
2511 modules in this release include:
2512 ** the ASDF system definition facility;
2513 ** an interface to the BSD Sockets API;
2514 ** an ACL-like convenience interface to the repl;
2515 (thanks to Kevin Rosenberg)
2516 ** an implementation of ROTATE-BYTE, with efficient implementation
2518 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2519 gives a closer approximation to the right answer than previously.
2520 (thanks to Raymond Toy)
2521 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2522 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2523 optional environment arguments, as required by ANSI.
2524 * fixed bugs in other functions taking environment objects, allowing
2525 calls with an explicit NIL environment argument to be compiled
2527 * fixed bug 228: primary return values from
2528 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2529 COMPILE or FUNCTION.
2530 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2531 :TYPE will no longer signal errors on innocuous objects.
2532 * fixed bug 231b: SETQ is better at respecting type declarations in
2533 the lexical environment.
2534 * fixed a bug in DEFCLASS: classes named by symbols with no or
2535 unprintable packages can now be defined.
2536 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2537 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2538 * fixed compiler failure related to checking types of functions.
2539 (reported by Robert E. Brown)
2540 * the compiler is now much more consistent in its error-checking
2541 treatment of bounding index arguments to sequence functions: in
2542 (SAFETY 3) code, errors will be signalled in almost all cases if
2543 invalid sequence bounding indices are passed to functions defined
2544 by ANSI to operate on sequences.
2545 * fixed a bug in the build procedure: documentation of SBCL-specific
2546 packages is now preserved and available in the final Lisp image.
2547 * lifted FDEFINITION lookup out of loops in the implementation of
2548 many list operations. (thanks to Robert E. Brown)
2549 * fixed a bug in the reader: the #n# reader macro now works for
2550 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2551 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2552 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2553 E. Brown and Tony Martinez respectively)
2554 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2555 SLOT-BOUNDP: in particular, it is now able to optimize them much
2556 better, and is now not vulnerable to having packages renamed.
2557 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2558 always called when a slot is not present in an instance. (thanks
2560 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2561 longer signal an error on unreading a character following EOF on
2562 the previous constituent stream. (thanks to Tony Martinez)
2563 * fixed some bugs revealed by Paul Dietz' test suite:
2564 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2565 not just nonnegative fixnums;
2566 ** the logical bit-array operators such as BIT-AND now accept an
2567 explicit NIL for their "opt-arg" argument (to indicate a
2568 freshly-consed result bit-array);
2569 ** ELT now signals an error on an invalid sequence index in safe
2571 ** the type system is now cleverer about negations of numeric
2572 types, and consequently understands the BIGNUM and RATIO types
2574 ** the type system is now cleverer about the interaction between
2575 INTEGER and RATIO types: while bugs still remain, many more
2576 cases are accurately computed;
2577 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2578 if it is in the last clause;
2579 ** CONSTANTLY now correctly returns a side-effect-free function in
2581 ** DECLARE is no longer treated as a special-operator; in
2582 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2583 * incremented fasl file version number due to the change in the
2584 DEFSTRUCT-SLOT-DESCRIPTION structure.
2586 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2587 * minor incompatible change: code processed by the "interpreter" or
2588 EVAL now has a compilation optimization policy of (DEBUG 2)
2589 (changed from (DEBUG 1)) to improve debuggability of interactive
2590 development, and to allow the use of the debug RETURN command in
2592 * an experimental implementation of the RETURN command for the
2593 debugger has been included. (thanks to Frederik Kuivinen)
2594 * fixed bug 62: constraints were not propagated into a loop.
2595 * fixed bug in embedded calls of SORT (reported and investigated by
2597 * fixed some bugs revealed by Paul Dietz' test suite:
2598 ** printing and reading of arrays with some dimensions having
2599 length 0 (thanks to Gerd Moellmann);
2600 ** BOA constructor with &AUX argument without a default value does
2601 not cause a type error;
2602 ** CONSTANTP now returns true for all self-evaluating objects.
2604 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2605 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2606 accessors that are related by inheritance, as specified in the
2607 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2608 Valtteri Vuorikoski)
2609 * The compiler is now able to inline functions that were defined in
2610 a complex lexical environment (e.g. inside a MACROLET).
2611 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2613 * Support for the upcoming FreeBSD-5.0 release has been included.
2614 (thanks to Dag-Erling Smorgrav)
2615 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2616 effect when it is not in a toplevel context.
2617 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2618 stand a better chance of being correct. (thanks to Gerd
2620 * fixed bug in COERCE, which now signals an error on coercing a
2621 rational to a bounded real type which excludes the expected
2623 * The compiler is now able to derive types more accurately from the
2624 COERCE and COMPILE functions.
2625 * fixed bug 223: functional binding is considered to be constant
2626 only for symbols in the CL package.
2627 * fixed bug 231: SETQ did not check the type of a variable being set
2628 (reported by Robert E. Brown)
2629 * A new optimization for MAKE-INSTANCE has been included, fixing
2630 various bugs (including relating to :ALLOCATION :CLASS slots and
2631 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2633 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2634 ** As required by ANSI, LOOP now disallows anonymous collection
2635 clauses such as COLLECT I in conjunction with aggregate boolean
2636 clauses such as THEREIS (= I 1);
2637 ** LOOP now signals an error when any variable is reused in the
2638 same loop (including the potentially useful construct analogous
2639 to WITH A = 1 WITH A = (1+ A);
2640 ** IT is only a special loop symbol within the first clause of a
2641 conditional loop clause;
2642 ** LOOP with a typed iteration variable over a hashtable now
2643 signals a type error iff it should.
2644 * fixed some other bugs revealed by Paul Dietz' test suite:
2645 ** FILE-STREAM now names the class previously known as FD-STREAM;
2646 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2647 argument) no longer signals an error;
2648 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2649 of the slot symbol, rather than using the current package
2650 ((:CONC-NAME "") continues to intern the slot's name in the
2652 * incremented fasl file version number, because of the incompatible
2653 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2654 of the new implementation of DEFINE-COMPILER-MACRO.
2656 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2657 * Support for building SBCL for MIPS platforms running in
2658 little-endian mode has now been checked in, and basic
2659 functionality on said platforms verified.
2660 * minor incompatible change: PCL now records the pathname of a file
2661 in which methods and the like are defined, rather than its
2663 * minor incompatible change: TRUENAME now considers the truename of
2664 a file naming a directory to be the pathname with :DIRECTORY
2665 component indicating that directory.
2666 * minor incompatible change: a NAMED clause in the extended form of
2667 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2668 reason for the previous behaviour is unclear.
2669 * more systematization and improvement of CLOS and MOP conformance
2670 in PCL (thanks to Gerd Moellman and Pierre Mai):
2671 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2673 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2674 primary methods with no specializers;
2675 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2677 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2678 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2679 FORWARD-REFERENCED-CLASSes; error reporting on
2680 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2682 ** SXHASH on CLOS instances now uses a slot internal to the
2683 instance to return different numbers on distinct instances,
2684 while preserving the same return value through invocations of
2686 ** DEFMETHOD signals errors when methods with longer incongruent
2687 lambda lists are added to generic functions;
2688 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2689 CLASS, as specified in AMOP;
2690 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2691 on the order returned by the primary method for classes of
2692 class STANDARD-CLASS;
2693 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2694 * fixed some bugs shown by Paul Dietz' test suite:
2695 ** DOLIST puts its body in TAGBODY;
2696 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2698 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2699 value producing form;
2700 ** if more variables are given to PROGV than values, extra
2701 variables are bound and made to have no value;
2702 ** NSUBSTITUTE on list arguments gets the right answer with
2704 ** ELT signals an error of type TYPE-ERROR when the index argument
2705 is not a valid sequence index;
2706 ** LOOP signals (at macroexpansion time) an error of type
2707 PROGRAM-ERROR when duplicate variable names are found;
2708 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2709 ** FUNCALL of special-operators now cause an error of type
2711 ** PSETQ now works as required in the presence of side-effecting
2712 symbol-macro places;
2713 ** NCONC accepts any object as its last argument;
2714 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2716 ** loop-for-as-package does not require a package to be explicitely
2718 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2720 * fixed bug 166: compiler preserves "there is a way to go"
2721 invariant when deleting code.
2722 * fixed bug 172: macro lambda lists with required arguments after
2723 &REST arguments now cause an error to be signalled. (thanks to
2725 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2726 bug. (thanks to Matthew Danish)
2727 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2729 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2730 arguments are lexically rebound. (thanks to Gerd Moellmann and
2732 * fixed bug 194: error messages are now more informative when there
2733 is no primary method applicable in a call to a generic
2734 function. (thanks to Gerd Moellmann)
2735 * fixed bug in command line argument checking (thanks to Julian
2737 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2738 (thanks to Matthew Danish)
2739 * incremented fasl file version number, because of the
2740 SXHASH-related changes in the layout of CLOS data structures
2742 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2743 * minor incompatible change: The runtime (the Unix executable named
2744 "sbcl") is now much pickier about the .core files it will load.
2745 Essentially it now requires .core files to descend from the same
2746 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2747 as the runtime does. (The intent is to prevent the crashes which
2748 can occur, and which can even be reported as mysterious failures,
2749 when people patch the sources or change the build parameters
2750 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2751 sbcl and .core files.)
2752 * fixed bug: VALUES-LIST is no longer optimized away.
2753 * fixed bug 142: The FFI conversion of C string values to Lisp
2754 string values no longer conses excessively. (thanks to Nathan
2755 Froyd porting Raymond Toy's fix to CMU CL)
2756 * began to systematize and improve MOP conformance in PCL (thanks to
2757 Nathan Froyd, Gerd Moellman and Pierre Mai):
2758 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2760 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2761 ** FINALIZE-INHERITANCE is now called on class finalization;
2762 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2763 argument precedence order.
2764 * fixed bug 202: The compiler no longer fails on functions whose
2765 derived types contradict their declared type.
2766 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2767 so it can be non-toplevel.
2768 * The fasl file version number has changed (because of the new
2769 implementation of DEFMACRO).
2770 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2771 safe code, that any length requirement by their type-specifier
2772 argument is valid. The exceptions to this are described in bug
2774 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2775 functions now check the types of their inputs as required by ANSI.
2776 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2777 introduced symbol is DECLAREd to be SPECIAL.
2778 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2779 * fixed bug: :COUNT argument to sequence functions may be negative
2780 * fixed bug: body of DO-SYMBOLS may contain declarations
2781 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2782 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2783 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2784 symbol macro only once
2785 * fixed printing of call frame when argument list is unavailable
2786 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2787 * compiler no longer signals WARNING on unknown keyword
2790 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2791 * A beta-quality port to the mips architecture running Linux,
2792 based on the old CMUCL backend, has been made. It has been tested
2793 on a big-endian kernel, and works sufficiently well to be able to
2794 rebuild itself; it has not been tested in little-endian mode.
2795 * fixed an inconsistency between gencgc.c and purify.c which made
2796 dumping/loading .core files unreliable
2797 * fixed bug 120a: The compiler now deals correctly with IFs where
2798 the consequent is the same as the alternative, instead of
2799 misderiving the return type. (thanks to Alexey Dejneka)
2800 * fixed bug 113: Logical pathnames are now dumpable (the logical
2801 host is resolved at load-time, throwing an error if it is not
2803 * fixed bug 174: FORMAT's error message is slightly clearer when a
2804 non-printing character is used in a format directive.
2805 * fixed several bugs in compiler checking of type declarations, i.e.
2806 violations of the Python "declarations are assertions" principle
2807 (thanks to Alexey Dejneka)
2808 * fixed several bugs in PCL's error checking (thanks to Gerd
2810 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2812 * fixed bug in compilation of functions as first class values
2813 (thanks to Antonio Martinez-Shotton)
2814 * The compiler's handling TYPE-ERRORs which it can prove will
2815 inevitably happen at runtime has been cleaned up and corrected
2816 in several ways. (thanks to Alexey Dejneka)
2817 * improved argument type checking for various basic arithmetic
2818 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2819 lavished on them in the past that they can be compiled in many
2820 ways in different special cases
2821 * fixed bug 181: compiler checks validity of user supplied type
2823 * cleaned up code flushing in optimization: Function calls which
2824 should signal errors for safety purposes (e.g. which ANSI says
2825 should signal errors when their arguments are of incorrect type)
2826 are no longer optimized away.
2827 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2828 * incremented fasl file version number, because changes in the
2829 implementation of sequence functions like COERCE caused
2830 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2831 in old inline expansions) to become undefined. (Actually these
2832 changes were later undone, so we might very well be binary
2833 compatible with 0.7.7 after all, but leaving the version number
2834 incremented seemed like the simplest and most conservative
2837 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2838 * An alpha-quality port to the parisc architecture running Linux,
2839 based on the old CMUCL backend, has been made. This, even more so
2840 than the other backends, should be considered still a work in
2841 progress; known problems include that the Linux kernel in 64-bit
2842 mode does not propagate the correct sigcontext structure to
2843 userspace, and consequently SBCL on a parisc64 kernel will not
2845 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2846 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2847 suppressed.) Also now that the compiler is looking at declarations
2848 in the environment, it checks optimization declarations as well,
2849 and suppresses inlining when (> DEBUG SPEED).
2850 * More fixes have been made to treatment of floating point exception
2851 treatment and other Unix signals. In particular, floating point
2852 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2853 * The detection and handling of control stack exhaustion (infinite
2854 or very deeply nested recursion) has changed. Stack exhaustion
2855 detection is now done by write-protecting pages at the OS level
2856 and applies at all optimization settings; when found, a
2857 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2858 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2859 be caught using IGNORE-ERRORS.
2860 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2861 correctly when transforming recursive LABELS functions to LETs.
2862 (thanks to Alexey Dejneka)
2863 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2864 that are names of constants or global variables.
2865 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2866 alien routines with docstrings.
2867 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2868 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2870 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2871 (thanks to Eric Marsden)
2872 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2873 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2874 * Bug fix: The SPARC backend can now compile functions involving
2875 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2876 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2877 to a functions expecting a PATHNAME-DESIGNATOR.
2878 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2879 lambda lists. (thanks to Alexey Dejneka)
2880 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2881 functions so that EXPORTing the name of the class doesn't cause
2882 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2883 to Antonio Martinez for reporting this)
2884 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2885 constant-folded without failing an assertion. (thanks to Einar
2886 Floystad Dorum for reporting this)
2887 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2888 types are now dealt with more correctly by the compiler.
2889 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2890 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2891 behaviour analogous to RENAME-FILE. This puts its behaviour more
2892 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2894 * The fasl file version number has changed again. (because of the
2895 bug fix involving the names of PCL MAKE-INSTANCE functions)
2897 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2898 * bug fix: Floating point exceptions are treated much more
2899 consistently on the x86/Linux and PPC/Linux platforms.
2900 * Array initialization with :INITIAL-ELEMENT is now much faster for
2901 cases when the compiler cannot open code the array creation, but
2902 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2903 array accesses have also seen a speed increase.
2904 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2905 specified. (This was at the root of some bad interactions between
2906 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2908 * bug fix: Internal error arguments for undefined functions are now
2909 computed correctly on the PPC/Linux platform.
2910 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2911 Raymond Toy's patch for CMU CL)
2912 * Support for the Solaris 9 operating environment has been included
2913 (thanks to Daniel Merritt)
2914 * A very ugly but hopefully complete draft of the missing FFI chapter
2915 of the manual has been created by reformatting the corresponding
2916 CMU CL manual chapter into (currently very ugly and incoherent)
2917 DocBook and bringing it up to date for SBCL behavior. Thus, the
2918 manual is now essentially complete, at least by my extreme
2919 once-and-only-once standards, whereby it's acceptable to refer to
2920 the doc strings of SB-EXT functions as the primary documentation.
2921 * The fasl file version number has changed again, due to cleanup of
2922 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2923 is no longer a static symbol.)
2925 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2926 * SBCL now builds with OpenMCL (version 0.12) as the
2927 cross-compilation host; also, more progress has been made toward
2928 bootstrapping under CLISP.
2929 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2931 * bug 158 fixed: The compiler can now deal with integer loop
2932 increments different from 1; fixing this turned out also to fix
2934 * bug 169 fixed: no more bogus warnings about using lexical bindings
2935 despite the presence of perfectly good SPECIAL declarations (thanks
2936 to David Lichteblau)
2937 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2938 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2939 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2941 * bug 180 fixed: Method combination specifications no longer ignore
2942 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2943 * bug fix: Structure type predicate functions now check their argument
2944 count as they should.
2945 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2946 correctly. (thanks to Pierre Mai)
2947 * minor incompatible change: The --noprogrammer option is deprecated
2948 in favor of the new --disable-debugger option, which is very similar.
2949 (The major difference is that it takes effect at a slightly different
2950 time at startup, causing handling of errors in --sysinit and
2951 --userinit files will be affected differently.) The
2952 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2953 been added to allow this functionality to be controlled from ordinary
2954 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2955 maintainers, who might want to run non-interactive scripts to
2956 build SBCL cores which will later be used interactively.)
2957 * minor incompatible change: The LOAD function no longer, when given
2958 a wild pathname to load, loads all files matching that pathname.
2959 Instead, an error of type FILE-ERROR is signalled.
2961 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2962 * bug 147 fixed: The compiler preserves its block link/count
2963 invariants more correctly now so that it doesn't crash. (thanks
2965 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2967 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2968 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2969 no-value-for-this-character as NIL (as specified by ANSI).
2970 * HOST-NAMESTRING on physical pathnames now returns a string that is
2971 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2972 (thanks to Christophe Rhodes)
2973 * The Alpha port handles icache flushing more correctly. (thanks to
2975 * More progress has been made toward bootstrapping under CLISP. (thanks
2976 to Christophe Rhodes)
2977 * The fasl file format has changed again, because dynamic loading
2978 on OpenBSD (which has non-ELF object files) motivated some cleanups
2979 in the way that foreign symbols are transformed and passed around.
2980 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2981 is no longer treated as whitespace by the reader, but instead as
2982 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2983 |A
\7fB|, instead of A as it used to.
2985 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2986 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2987 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2988 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2989 this as of 0.7.1.45, but was left out of the previous news section
2990 (thanks to Dan Barlow)
2991 * SBCL now runs on the Solaris operating system on SPARC architectures
2992 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2993 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2994 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2995 * SPARC backend cleanups, allowing builds of cores optimized for V8
2996 and V9 SPARCS, and also emission of code targeted to a particular
2997 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2999 * SBCL is closer to bootstrapping under CLISP, thanks to various
3000 fixes by Christophe Rhodes.
3001 * The fasl file format has changed again, to allow the compiler's
3002 INFO database to support symbol macros.
3003 * The user manual (in doc/) is formatted into HTML more nicely.
3004 (thanks to coreythomas)
3005 * The system is smarter about SUBTYPEP relationships, especially
3006 those involving NOT types (including types such as ATOM which are
3007 represented internally using NOT types). Thus SUBTYPEP is less
3008 likely to return (VALUES NIL NIL) in general, and in particular
3009 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3011 * The fasl file format has changed again, because the internal
3012 representation of types now includes a new slot to support the new
3013 SUBTYPEP-of-NOT-types logic.
3014 * (not a change in the main branch of SBCL, but a related prototype
3015 which can hopefully be merged into the main branch of SBCL in the
3016 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3017 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3018 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3019 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3020 dispatch macro character combinations. (thanks to Alexey Dejneka)
3021 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3022 that superficially look like logical namestrings correctly.
3023 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3025 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3026 default supplied-p) for &optional and &key arguments. (thanks to
3029 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3030 * incompatible change: The compiler is now less aggressive about
3031 tail call optimization, doing it only when (> SPACE DEBUG) or
3032 (> SPEED DEBUG). (This is an incompatible change because there are
3033 programs which relied on the old CMU-CL-style behavior to optimize
3034 away their unbounded recursion which will now die of stack overflow.)
3035 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3036 for non-GENCGC systems has been increased to 20M (since that
3037 seems much closer to the likely performance optimum for modern
3038 systems than the old 4M value was)
3039 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3040 and *DEBUG-PRINT-LEVEL*
3041 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3042 of CMU CL's support for SPARC, and various endianness and other
3043 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3044 * new syntactic sugar for the Unix command line: --load foo.bar is now
3045 an alternate notation for --eval '(load "foo.bar")'.
3047 ** The system now detects stack overflow and handles it gracefully,
3048 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3049 optimization settings. (This is a good thing in general, and
3050 its introduction in this version should be particularly timely
3051 for anyone whose code fails because of suppression of tail
3053 ** The system now hunts for the C variable "environ" in a more
3054 devious way, to avoid segfaults when the C library version differs
3055 between compile time and run time. (thanks to Christophe Rhodes)
3056 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3057 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3058 ** The compiler no longer issues bogus style warnings for undefined
3059 classes in the same source file as the DEFCLASSes which defined
3060 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3062 ** fixes in CONDITION class precedence list for undefined function
3063 errors (thanks to Alexei Dejneka)
3064 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3065 correctly. (thanks to Dan Barlow)
3066 ** portability fixes aiming at bootstrapping under CLISP (thanks
3067 to Dave McDonald and Christophe Rhodes)
3068 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3069 ** fixes in type translation and and type inference (thanks to
3071 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3072 ** various fixes in the new ports (thanks to Dan Barlow)
3073 * several changes related to debugging:
3074 ** suppression of tail recursion, as noted above
3075 ** stack overflow detection, as noted above
3076 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3077 is now the default. (For some time encapsulation has been more
3078 reliable than the breakpoint-based :ENCAPSULATE NIL
3079 implementation, at least on X86 systems; and I just noticed that
3080 encapsulation also seems closer to the spirit of the ANSI
3083 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3085 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3086 up properly again. (There was a packaging bug in 0.7.0 which
3087 left their definitions in SB-SYS::LOAD-FOREIGN and
3088 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3089 vital for most things which interface to C-level interfaces,
3090 like extensions working with sockets or databases or
3091 Perl-compatible regexes or whatever, and the need to fix
3092 this bug is the main reason that 0.7.1 was released so
3094 ** DEFGENERIC is now choosier about the methods it redefines, so that
3095 reLOADing a previously-LOADed file containing DEFGENERICs does
3096 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3097 cycle now works as it should. (thanks to Alexey Dejneka)
3098 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3099 fixed by Christophe Rhodes. (He actually submitted this patch
3100 months ago, and I delayed until after 0.7.0.)
3101 ** Bug 111 (internal compiler confusion about runtime checks on
3102 FUNCTION types) was fixed by Alexey Dejneka.
3103 * Some internal cleanups (getting rid of variables which aren't
3104 needed now that the byte interpreter is gone) caused the fasl
3105 file format number to change again.
3107 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3108 * major incompatible change: The default fasl file extension, i.e. the
3109 default extension for files produced by COMPILE-FILE, has changed
3110 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3112 ** There are many changes in the implementation of the compiler.
3113 SBCL is now essentially a compiler-only implementation of ANSI
3114 Common Lisp. EVAL still "interprets" a few special cases, but
3115 almost all the interesting cases are handled by creating
3116 a LAMBDA expression, calling COMPILE on it, then calling
3117 FUNCALL on the result.
3118 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3119 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3120 Since the code is newer, there might still be some new bugs
3121 (though not as many as before Martin Atzmueller's fixes:-). But
3122 the new code is substantially simpler and clearer, and hopefully
3123 any remaining bugs will be simpler, less fundamental, and more
3124 fixable then the bugs in the old code.
3125 ** The revised compiler is still a little unsteady on its feet.
3127 *** The debugging information it produces (particularly the names
3128 of FUNCTION objects) is sometimes much less useful than what
3129 the old compiler produced.
3130 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3131 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3132 within a MACROLET) has been temporarily weakened.
3133 ** There are new compiler optimizations for various functions:
3134 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3135 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3136 *** the math functions TRUNCATE, FLOOR, and CEILING
3137 *** the function-of-all-trades COERCE
3138 Mostly these should be transparent, but there's one
3139 potentially-annoying problem (bug 117): when the compiler
3140 inline-expands a function and does type analysis on the result,
3141 it can create control paths which have type mismatches, and
3142 when it can't prove that those control paths aren't taken,
3143 it will issue WARNINGs about the type mismatches. This is
3144 a particular problem in practice for the new sequence functions.
3145 It's not clear how this should be fixed, and for now, a
3146 workaround is given in the entry for 117 in the BUGS file.
3147 ** (Because of the interaction between the two previous items --
3148 occasional inlining problems and new inline expansions -- some
3149 of the new sequence function optimizations won't really kick in
3150 completely until debugging information, and then inlining, are
3151 straightened out in some future version.)
3152 * minor incompatible changes:
3153 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3154 DIRECTORY no longer implicitly promotes NIL slots of its
3155 pathname argument to :WILD. In particular, when you ask for the
3156 contents of a directory (which you used to be able to do without
3157 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3158 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3159 ** changes in behavior that ANSI explicitly defines to be
3160 implementation dependent:
3161 *** The new compiler-only implementation still conforms with ANSI,
3162 but acts a little different than before. Besides the obvious
3163 changes in performance tradeoffs (that the cost per form passed
3164 to EVAL has gone up, and the cost per form executed by EVAL
3165 has gone down), the behavior of the system changes a little
3166 because there are no longer any interpreted function objects.
3167 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3168 e.g. doing COMPILE on the output of interactive DEFUN is
3170 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3172 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3173 has changed from (:CL) to NIL.
3174 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3175 e.g. #\Newline and #\Tab, don't have names specified in the
3176 ANSI Common Lisp standard, is now based on their ASCII symbolic
3177 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3178 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3179 are no longer used for output.
3180 ** changes in internal implementation constants:
3181 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3182 4 million. (If your application spends a lot of time GCing and
3183 you have a lot of RAM, you might want to experiment with
3184 increasing it even more.)
3185 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3186 However, almost all old code should still continue to work without
3187 immediate update, as SB-C-CALL is now a (deprecated) nickname
3189 ** Old operator names in the style DEF-FOO are now deprecated in
3190 favor of new corresponding names DEFINE-FOO, for consistency with
3191 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3192 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3193 internal symbols, but a few supported extensions like
3194 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3195 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3196 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3197 etc. as you get deeper into recursive calls to the debugger
3198 command loop, instead of the old "5]", "5]]", "5]]]"
3199 sequence. (I was motivated to do this when squabbles between
3200 ILISP and SBCL left me very deeply nested in the debugger. In the
3201 short term, this change will probably provoke more ILISP/SBCL
3202 squabbles, but hopefully it will be an improvement in the long run.)
3203 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3204 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3205 function, since declaiming return types involving aliens is
3206 (1) annoyingly messy to do by hand and (2) vital to efficient
3207 compilation of code which calls such functions.
3208 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3209 longer reexported by the SB-EXT package. They're solely useful
3210 for alien code, so it seems more logical that you should get
3211 them from the SB-ALIEN package, not in SB-EXT.
3212 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3213 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3214 features. Instead, the code that they used to control is always
3215 built into the system.
3216 * many other bug fixes
3217 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3218 advantage of the new EVAL-WHEN stuff and to clean them up in
3219 general, and they are now more ANSI-compliant in a number of
3220 ways. Martin Atzmueller is responsible for a lot of this.
3221 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3223 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3224 *** correct ERROR type for various file operations
3225 *** some fixes for Lisp streams
3226 *** DEFMETHOD syntax checking
3227 *** changing old weird representation of debug information as
3228 strings (which, among their other deficiencies, don't transform
3229 correctly when you rename packages, and don't change their
3230 print representation when you change things like *PACKAGE*
3231 and *PRINT-LENGTH*) to symbols and lists of symbols
3232 He also made several improvements and fixed several bugs in DESCRIBE.
3233 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3235 *** misbehavior of WRITE-STRING/WRITE-LINE
3236 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3237 and several other LOOP problems as well
3238 *** DIRECTORY when similar filenames are present
3239 *** DEFGENERIC with :METHOD options
3240 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3241 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3242 *** argument ordering in FIND with :TEST option
3243 *** mishandled package designator argument in APROPOS-LIST
3244 *** various problems in the backquote readmacro
3245 *** a bug in APROPOS
3246 *** probably some others that I'm not describing very well here,
3247 since the CVS log documents them by reference to sbcl-devel
3248 messages, and the SourceForge archives aren't working well.:-(
3249 ** Dan Barlow improved the Alpha port (and is making progress on the
3250 PPC port, for those of you who think different).
3251 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3252 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3253 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3254 continues to work on the SPARC port (for those of you in a position
3255 to look down upon our little PC-compatible boxes from a great height).
3256 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3257 and uses that copy, rather than the current dynamic value, when
3258 it's trying to decide whether to truncate output. Thus e.g.
3259 (let ((*print-lines* 50))
3260 (pprint-logical-block (stream nil)
3262 (let ((*print-lines* 8))
3263 (print (aref possiblybigthings i) stream)))))
3264 should now truncate the logical block only at 50 lines, instead of
3265 often truncating it at 8 lines, as it did before.
3266 * The doc/cmucl/ directory, containing old CMU CL documentation
3267 from the time of the fork, is no longer part of the base system.
3268 SourceForge has shut down its anonymous FTP service, and with it
3269 my original plan for distributing the old CMU CL documentation
3270 there. For now, if you need these files you can download an old
3271 SBCL source release and extract them from it.
3272 * The fasl file version number changed again, for dozens of reasons,
3273 some of which are apparent above.
3275 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3276 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3277 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3278 patch, so that DISASSEMBLE output is much nicer.
3279 * The code in the SB-PROFILE package now seems reasonably stable.
3280 I still haven't decided what the final interface should look like
3281 (I'd like PROFILE to interact cleanly with TRACE, since both
3282 facilities use function encapsulation) but if you have a need
3283 for profiling now, you can probably use it successfully with
3284 the current CMU-CL-style interface.
3285 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3286 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3287 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3288 link now returns the dangling link itself, and for similar
3289 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3290 link itself. (In these cases the old code signalled an error and
3291 looped endlessly, respectively.) Thus, DIRECTORY now works even
3292 in the presence of dangling and cyclic symbolic links.
3293 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3294 is now a supported extension again, since the consensus on
3295 sbcl-devel was that it can be useful for ordinary development
3296 work, not just for debugging SBCL itself.
3297 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3298 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3299 that currently-defined functions might be redefined later with
3300 different return types.
3301 * Hash tables can be printed readably, as inspired by CMU CL code
3302 of Eric Marsden and SBCL code of Martin Atzmueller.
3303 * better error handling in CLOS method combination, thanks to
3304 Martin Atzmueller porting Pierre Mai's CMU CL patches
3305 * more overflow fixes for >16Mbyte I/O buffers
3306 * A bug in READ has been fixed, so that now a single Ctrl-D
3307 character suffices to cause end-of-file on character streams.
3308 In particular, now you only need one Ctrl-D at the command
3309 line (not two) to exit SBCL.
3310 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3312 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3313 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3314 does the right thing.
3315 * The implementation of some type tests, especially for CONDITION
3316 types, is now tidier and maybe faster, due to CMU CL code
3317 originally by Douglas Crosher, ported by Martin Atzmueller.
3318 * Some math functions have been fixed, and there are new
3319 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3320 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3321 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3322 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3323 likely to be built into sbcl-0.7.0.)
3324 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3325 CL, ported to SBCL by Martin Atzmueller)
3326 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3327 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3328 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3329 * There's a new slam.sh hack to shorten the edit/compile/debug
3330 cycle for low-level changes to SBCL itself, and a new
3331 :SB-AFTER-XC-CORE target feature to control the generation of
3332 the after-xc.core file needed by slam.sh.
3333 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3334 COMPILE-FILE is no longer supported, so that now every function
3335 gets an entry point, so that block compilation looks a little
3336 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3337 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3338 returns the number of bytes consed since the system started,
3339 rather than the number consed since the first time the function
3340 was called. (The new definition parallels ANSI functions like
3341 CL:GET-INTERNAL-RUN-TIME.)
3342 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3343 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3344 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3345 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3346 * Fasl file version numbers are now independent of the target CPU,
3347 since historically most system changes which required version
3348 number changes have affected all CPUs equally. Similarly,
3349 the byte fasl file version is now equal to the ordinary
3352 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3353 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3354 is no longer recognized. I apologize for this, because it was
3355 listed in SB-EXT as a supported extension, but I found that
3356 its existing behavior was poorly specified, as well as incorrectly
3357 specified, and it looked like too much of a mess to straighten it
3358 out. I have enough on my hands trying to get ANSI stuff to work..
3359 * many patches ported from CMU CL by Martin Atzmueller, with
3360 half a dozen bug fixes in pretty-printing and the debugger, and
3361 half a dozen others elsewhere
3362 * fixed bug 13: Floating point infinities are now supported again.
3363 They might still be a little bit flaky, but thanks to bug reports
3364 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3365 as flaky as they were.
3366 * The --noprogrammer command line option is now supported. (Its
3367 behavior is slightly different in detail from what the old man
3368 page claimed it would do, but it's still appropriate under the
3369 same circumstances that the man page talks about.)
3370 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3371 are now supported, and enabled by default. Thus, the compiler can
3372 handle many floating point and complex operations much less
3373 inefficiently. (Thus e.g. you can implement a complex FFT
3375 * The compiler now detects type mismatches between DECLAIM FTYPE
3376 and DEFUN better, and implements CHECK-TYPE more correctly, and
3377 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3378 thanks to patches from Martin Atzmueller.
3379 * various fixes to make the cross-compiler more portable to
3380 ANSI-conforming-but-different cross-compilation hosts (notably
3381 Lispworks for Windows, following bug reports from Arthur Lemmens)
3382 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3383 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3384 fixed thanks to Pierre Mai's CMU CL patches.
3385 * fixes to keep the system from overflowing internal counters when
3386 it tries to use i/o buffers larger than 16M bytes
3387 * fixed bug 45a: Various internal functions required to support
3388 complex special functions have been merged from CMU CL sources.
3389 (When I was first setting up SBCL, I misunderstood a compile-time
3390 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3391 * improved support for type intersection and union, fixing bug 12
3392 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3393 more obscure bugs as well
3394 * some steps toward byte-compiling non-performance-critical
3395 parts of the system, courtesy of patches from Martin Atzmueller
3396 * Christophe Rhodes has made some debian packages of sbcl at
3397 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3398 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3399 stable, but are nonetheless usable. When he's ready, I'd be happy
3400 to add them to the SourceForge "File Releases" section. (And if
3401 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3402 * new fasl file format version number (because of changes in
3403 internal representation of (OR ..) types to accommodate the new
3404 support for (AND ..) types, among other things)
3406 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3407 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3409 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3411 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3412 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3413 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3414 * DESCRIBE now works on structure objects again.
3415 * Most function call argument type mismatches are now handled as
3416 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3417 know whether the function will be redefined before the call is
3418 executed. (The compiler could flag local calls with full WARNINGs,
3419 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3420 it doesn't keep track of enough information to know whether calls
3421 are local in this sense.)
3422 * Compiler output is now more verbose, with messages truncated
3423 later than before. (There should be some supported way for users
3424 to override the default verbosity, but I haven't decided how to
3425 provide it yet, so this behavior is still controlled by the internal
3426 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3427 src/compiler/ir1util.lisp.)
3428 * Fasl file format version numbers have increased again, because
3429 support for the Gray streams extension changes the layout of the
3430 system's STREAM objects.
3431 * The Gray subclassable streams extension now works, thanks to a
3432 patch from Martin Atzmueller.
3433 * The full LOAD-FOREIGN extension (not just the primitive
3434 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3435 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3436 but like most other programs, it defaults to copying the Unix
3437 environment from the original process instead of starting the
3438 new process in an empty environment.
3439 * Extensions which manipulate the Unix environment now support
3440 an :ENVIRONMENT keyword option which doesn't smash case or
3441 do other bad things. The CMU-CL-style :ENV option is retained
3442 for porting convenience.
3443 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3444 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3446 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3448 * A patch from Martin Atzmueller seems to have solved the SIGINT
3449 problem, and as far as we know, signal-handling now works cleanly.
3450 (If you find any new bugs, please report them!)
3451 * The system no longer defaults Lisp source file names to types
3452 ".l", ".cl", or ".lsp", but only to ".lisp".
3453 * The compiler no longer uses special default file extensions for
3454 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3455 seems to expect a single default extension for all compiled code,
3456 and there's no compelling reason to try to stretch the standard
3457 to allow two different extensions.) Instead, byte-compiled files
3458 default to the same extension as native-compiled files.
3459 * Fasl file format version numbers have increased again, because
3460 a rearrangement of internal implementation packages made some
3461 dumped symbols in old fasl files unreadable in new cores.
3462 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3463 many fewer weird special cases.
3464 * Bug #17 (differing COMPILE-FILE behavior between logical and
3465 physical pathnames) has been fixed, and some related misbehavior too,
3466 thanks to a patch from Martin Atzmueller.
3467 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3468 by Tim Moore, ported to SBCL by Martin Atzmueller.
3469 * Martin Atzmueller fixed several filesystem-related problems,
3470 including bug #36, in part by porting CMU CL patches, which were
3471 written in part by Paul Werkowski.
3472 * More compiler warnings in src/runtime/ are gone, thanks to
3473 more patches from Martin Atzmueller.
3474 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3477 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3479 * DESCRIBE now works on CONDITION objects.
3480 * The debugger now handles errors which arise when trying to print
3481 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3483 * The build system now uses an additional file, customize-target-features.lisp,
3484 to allow local modifications to the target *FEATURES* list. (The point of
3485 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3486 debugging features enabled, without having to worry about propagating it
3487 into everyone's system when I do a "cvs update".) When no
3488 customize-target-features.lisp file exists, the target *FEATURES* list
3489 should be constructed the same way as before.
3490 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3491 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3492 warn about behavior which is undefined under the ANSI spec. Note: This
3493 is specified by ANSI, but it's not very popular with programmers.
3494 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3495 macro in the SBCL sources for an example of a workaround which you
3496 might use to make portable ANSI-standard code which does what you want.
3497 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3498 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3499 when it's used as a non-toplevel form. (This is required in order
3500 to implement the DEFCONSTANT-EQX macro.)
3501 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3502 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3503 funny things, due to bugs in the implementation of EVAL-WHEN
3504 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3505 non-toplevel DEFCONSTANTs are uncommon.)
3506 * The core file version number and fasl file version number have been
3507 incremented, because the old noncompliant DEFCONSTANT behavior involved
3508 calling functions which no longer exist, and because I also took the
3509 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3510 * fixed bug 1 (error handling before read-eval-print loop starts), and
3511 redid debugger restarts and related debugger commands somewhat while
3513 ** The QUIT debugger command is gone, since it did something
3514 rather different than the SB-EXT:QUIT command, and since it never
3515 worked properly outside the main toplevel read/eval/print loop.
3516 Invoking the new TOPLEVEL restart provides the same functionality.
3517 ** The GO debugger command is also gone, since you can just invoke
3518 the CONTINUE restart directly instead.
3519 ** The TOP debugger command is also gone, since it's redundant with the
3520 FRAME 0 command, and since it interfered with abbreviations for the
3522 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3524 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3525 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3526 some porting work by Martin Atzmueller)
3527 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3528 with a boatload of patches which he ported to SBCL. Now that those
3530 ** The system tries to make sure that its low-priority messages
3531 are prefixed by semicolons, to help people who like to use
3532 syntax highlighting in their ILISP buffer. (This patch
3533 was originally due to Raymond Toy.)
3534 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3535 patches originally written by Raymond Toy.
3536 ** The compiler understands coercion between single-value and
3537 multiple-VALUES type expressions better, getting rid of some very
3538 weird behavior, thanks to patches originally by Robert MacLachlan
3539 and Douglas Crosher.
3540 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3541 lambda lists, thanks to a patch originally by Pierre Mai.
3542 ** The system no longer bogusly warns about "abbreviated type
3544 ** The compiler gets less confused by inlining and RETURN-FROM,
3545 thanks to some patches originally by Tim Moore.
3546 ** The system no longer hangs when dumping circular lists to fasl
3547 files, thanks to a patch originally from Douglas Crosher.
3548 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3549 undefined function error.
3550 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3551 in structure slot initforms) for now, documented workaround instead:-|
3552 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3553 * fixed bug 5: added stubs for various Gray stream functions called
3554 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3555 installed, at least appropriate type errors are generated
3556 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3557 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3558 consistently in DEFMETHOD forms.
3559 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3560 it doesn't seem to affect SBCL after all
3561 * The C runtime system now builds with better optimization and many
3562 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3564 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3566 * The system is now under CVS at SourceForge (instead of the
3567 CVS repository on my home machine).
3568 * The new signal handling code has been tweaked to treat register
3569 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3570 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3571 downstream, e.g. in debug-int.lisp, has implicit dependencies
3572 on the unsignedness of integer representation of machine words,
3573 and that caused the system to bomb out with infinite regress
3574 when trying to recover from type errors involving signed values,
3575 e.g. (BUTLAST '(1 2 3) -1).)
3576 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3577 (This was one of the bugs Peter Van Eynde reported back in July.)
3578 * The system now uses code inspired by Colin Walters' O(N)
3579 implementation of MAP (from the cmucl-imp@cons.org mailing
3580 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3581 inline the MAP operation, and there is more than one
3582 sequence argument to the MAP call (so that it can't just
3583 do ETYPECASE once and for all based on the type of the
3584 single sequence argument). (The old non-inline implementation
3585 of the general M-argument sequence-of-length-N case required
3586 O(M*N*N) time when any of the sequence arguments were LISTs.)
3587 * The QUIT :UNIX-CODE keyword argument has been renamed to
3588 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3589 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3591 * Raymond Wiker's patches to port dynamic loading from Linux to
3592 FreeBSD have been added.
3593 * The BUGS file is now more nearly up to date, thanks in large part
3594 to Martin Atzmueller's review of it.
3595 * The debugger now flushes standard output streams before it begins
3596 its output ("debugger invoked" and so forth).
3597 * The core version number and fasl file version number have both
3598 been incremented, because of incompatible changes in the layout
3600 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3601 used to not be called for a saved Lisp image.)
3602 * Martin Atzmueller's version of a patch to fix a compiler crash,
3603 as posted on sbcl-devel 13 September 2000, has been installed.
3604 * Instead of installing Martin Atzmueller's patch for the
3605 compiler transform for SUBSEQ, I deleted the compiler transform,
3606 and transforms for some similar consing operations.
3607 * A bug in signal handling which kept TRACE from working on OpenBSD
3609 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3610 away by constant folding
3611 * The system now defines its address space constants in one place
3612 (in the Lisp sources), and propagates them automatically elsewhere
3613 (through GENESIS and the sbcl.h file). Therefore, patching the
3614 address map is less unnecessarily tedious and error-prone. The
3615 Lisp names of address space constants have also been systematized.
3616 * CVS tags like dollar-Header-dollar have been removed from
3617 the sources, because they have never saved me trouble and
3618 they've been source of trouble working with patches and other
3619 diff-related operations.
3620 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3621 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3623 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3625 * The system has been ported to OpenBSD.
3626 * The system now compiles with a simple "sh make.sh" on the systems
3627 that it's supported on. I.e., now you no longer need to tweak
3628 text in the target-features.lisp-expr and symlinks in src/runtime/
3629 by hand, the make.sh takes care of it for you.
3630 * The system is no longer so grossly inefficient when compiling code
3631 involving vectors implemented as general (not simple) vectors (VECTOR T),
3632 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3633 which dares to use the various sequence functions on non-simple
3634 vectors, takes less of a performance hit.
3635 * There is now a primitive type predicate VECTOR-T-P
3636 to test for the (VECTOR T) type, so that e.g.
3637 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3638 can now be compiled with some semblance of efficiency. (The old code
3639 turned the type declaration into a full call to %TYPEP at runtime!)
3640 * AREF on (VECTOR T) is still not fast, since it's still compiled
3641 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3642 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3643 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3644 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3645 without making full calls to SUBTYPEP at runtime.
3646 (Some analogous efficiency issues for non-simple vectors specialized to
3647 element types other than T, or for non-simple multidimensional arrays,
3648 have not been addressed. They could almost certainly be handled the
3649 same way if anyone is motivated to do so.)
3650 * The changes in array handling break binary compatibility, so
3651 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3652 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3653 instead of (VALUES T T).
3654 * By following the instructions that Dan Barlow posted to sbcl-devel
3655 on 2 July 2000, I was able to enable primitive dynamic object
3656 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3657 functionality is not implemented (since it calls ld to resolve
3658 library references automatically, requiring RUN-PROGRAM for its
3659 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3660 try to resolve library references) is now supported.
3661 * The system now flushes the standard output streams when it terminates,
3662 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3663 them at several other probably-convenient times, e.g. in each pass of
3664 the toplevel read-eval-print loop, and after evaluating a form given
3665 as an "--eval" command-line option. (These changes were motivated by a
3666 discussion of stream flushing issues on cmucl-imp in August 2000.)
3667 * The source transform for TYPEP of array types no longer assumes
3668 that an array whose element type is a not-yet-defined type
3669 is implemented as an array of T, but instead punts, so that the
3670 type will be interpreted at runtime.
3671 * There is now some support for cross-compiling in make.sh: each of
3672 the phases of make.sh has its own script. (This should be transparent
3673 to people doing ordinary, non-cross-compile builds.)
3674 * Since my laptop doesn't have hundreds of megabytes of memory like
3675 my desktop machine, I became more motivated to do some items on
3676 my to-do list in order to reduce the size of the system a little:
3677 ** Arrange for various needed-only-at-cold-init things to be
3678 uninterned after cold init. To support this, those things have
3679 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3680 symbols with such names are now uninterned after cold init).
3681 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3682 fasl files for cold load.
3683 ** Remove the old compiler structure pooling code (which used to
3684 be conditional on the target feature :SB-ALLOC) completely.
3685 ** Redo the representation of some data in cold init to be more compact.
3686 (I also looked into supporting byte compiled code at bootstrap time,
3687 which would probably reduce the size of the system a lot, but that
3688 looked too complicated, so I punted for now.)
3689 * The maximum signal nesting depth in the src/runtime/ support code has
3690 been reduced from 4096 to 256. (I don't know any reason for the very
3691 large old value. If the new smaller value turns out to break something,
3692 I'll probably just bump it back up.)
3693 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3695 * Many, many bugs reported by Peter Van Eynde have been added to
3696 the BUGS list; some have even been fixed.
3697 * While enabling dynamic object file loading, I tried to make the
3698 code easier to understand, renaming various functions and variables
3699 with less ambiguous names, and changing some function calling
3700 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3701 * While trying to figure out how to do the OpenBSD port, I tried to
3702 clean up some of the code in src/runtime/. In particular, I dropped
3703 support for non-POSIX signal handling, added various comments,
3704 tweaked the code to reduce the number of compilation warnings, and
3705 renamed some files to increase consistency.
3706 * To support the new automatic configuration functionality in make.sh,
3707 the source file target-features.lisp-expr has been replaced with the
3708 source file base-target-features.lisp-expr and the machine-generated
3709 file local-target-features.lisp-expr.
3710 * fixed a stupid quoting error in make.sh so that using CMU CL
3711 "lisp -batch" as cross-compilation host works again
3713 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3715 * DESCRIBE no longer tries to call itself recursively to describe
3716 bound/fbound values, so that it no longer fails on symbols which are
3717 bound to themselves (like keywords, T, and NIL).
3718 * DESCRIBE now works on generic functions.
3719 * The printer now prints less-screwed-up representations of closures
3720 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3721 * A private symbol is used instead of the :EMPTY keyword previously
3722 used to mark empty slots in hash tables. Thus
3723 (DEFVAR *HT* (MAKE-HASH-TABLE))
3724 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3725 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3726 now does what ANSI says that it should. (You can still get
3727 similar noncompliant behavior if bang on the hash table
3728 implementation with all the symbols you get back from
3729 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3730 This breaks binary compatibility, since tests for equality to
3731 :EMPTY are wired into things like the macroexpansion of
3732 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3734 * There's now a minimal placeholder implementation for CL:STEP,
3735 as required by ANSI.
3736 * An obscure bug in the interaction of the normal compiler, the byte
3737 compiler, inlining, and structure predicates has been patched
3738 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3739 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3741 * Missing ordinary arguments in a macro call are now detected even
3742 when the macro lambda list contains &KEY or &REST.
3743 * The debugger no longer complains about encountering the top of the
3744 stack when you type "FRAME 0" to explicitly instruct it to go to
3745 the top of the stack. And it now prints the frame you request even
3746 if it's the current frame (instead of saying "You are here.").
3747 * As specified by ANSI, the system now always prints keywords
3748 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3750 * The default initial SIZE of HASH-TABLEs is now smaller.
3751 * Type information from CLOS class dispatch is now propagated
3752 into DEFMETHOD bodies, so that e.g.
3753 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3755 is now basically equivalent to
3756 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3757 (DECLARE (TYPE SINGLE-FLOAT X))
3759 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3760 operation, without having to do run-time type dispatch.
3761 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3762 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3763 * Since I don't seem to be making big file reorganizations very often
3764 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3765 is overflowing my ability to conveniently back them up), I've finally
3766 checked the system into CVS. (The CVS repository is on my home system,
3767 not at SourceForge -- putting it on SourceForge might come later.)
3768 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3769 high-level GC-NOTIFY-FOO functions send their output. (There's
3770 still very little control of where low-level verbose GC functions
3771 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3772 less than it used to -- the GC-NOTIFY-FOO functions are now under
3773 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3774 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3775 in only one place in the source code, and propagates it automatically
3776 everywhere that it's needed. Thus e.g. when I bump the version from
3777 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3778 * The C source files now include boilerplate legalese and documentation
3779 at the head of each file (just as the Lisp source files already did).
3780 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3781 to his page will be replaced with a link to his new CLiki service.
3783 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3785 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3786 * The build process now looks for GNU make under the default name "gmake",
3787 instead of "make" as it used to. If GNU make is not available as "gmake"
3788 on your system, you can change this default behavior by setting the
3789 GNUMAKE environment variable.
3790 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3791 can build without error under CMU CL.
3793 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3795 * There is now a partial SBCL user manual (with some new text and some
3796 text cribbed from the CMU CL manual).
3797 * The beginnings of a profiler have been added (starting with the
3798 CMU CL profiler and simplifying and cleaning up). Eventually the
3799 main interface should be through the TRACE macro, but for now,
3800 it's still accessed through vaguely CMU-CL-style functions and macros
3801 exported from the package SB-PROFILE.
3802 * Some problems left over from porting CMU CL to the new
3803 cross-compilation bootstrap process have been cleaned up:
3804 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3805 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3806 part when getting rid of the compiler *BACKEND* stuff.)
3807 ** The value of *NULL-TYPE* was screwed up, because it was
3808 being initialized before the type system knew the final
3809 definition of the 'NULL type. This screwed up several key
3810 optimizations in the compiler, causing inefficiency in all sorts
3811 of places. (I found it because I wanted to understand why
3812 GET-INTERNAL-RUN-TIME was consing.)
3813 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3814 PROCLAIM FTYPE information. Unfortunately this broke binary
3815 compatibility again, since now the forms output by DEFGENERIC
3816 to refer to functions which didn't exist in 0.6.3.
3817 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3818 can use the new (as of 0.6.3) transform for SOME into MAP into
3820 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3821 values to (INTEGER 0 1000000), so that the time code will no longer
3822 occasionally get blown up by Linux returning 1000000 microseconds
3823 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3824 its output conform to the ANSI spec. (Alas, this makes its output
3825 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3826 * A full call to MAP NIL with a single sequence argument no longer conses.
3827 * fixes to problems pointed out by Martin Atzmueller:
3828 * The manual page no longer talks about multiprocessing as though
3829 it were currently supported.
3830 * The ILISP support patches have been removed from the distribution,
3831 because as of version 5.10.1, ILISP now supports SBCL without us
3832 having to maintain patches.
3833 * added a modified version of Raymond Toy's recent CMU CL patch for
3834 EQUALP comparison of HASH-TABLE
3836 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3838 * The system still can't cross-compile itself with
3839 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3840 entails), but at least it can compile more of itself that way
3841 than it used to be able to, and various buglets which were uncovered
3842 by trying to cross-compile itself that way have now been fixed.
3843 * This release breaks binary compatibility again. This time
3844 at least I've incremented the FASL file format version to 2, so that the
3845 problem can be detected reliably instead of just causing weird errors.
3846 * various new style warnings:
3847 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3848 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3849 ** using the lexical binding of a variable named in the *FOO* style
3850 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3851 as specified by ANSI.
3852 * *RANDOM-STATE* is no longer automatically initialized from
3853 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3854 default behavior of the system is to repeat its behavior every time
3855 it's run. If you'd like to change this behavior, you can always
3856 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3857 old convention there was no comparably easy way to get the system to
3858 repeat its behavior every time it was run.
3859 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3860 FTYPE declarations has been removed, in favor of their ANSI
3861 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3862 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3863 the types of their sequence arguments can be determined at compile time.
3864 This is done through a new open code expansion for MAP which eliminates
3865 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3866 argument types can be determined at compile time.
3867 * The optimizer now transforms COERCE into an identity operation when it
3868 can prove that the coerced object is already of the correct type. (This
3869 can be a win for machine generated code, including the output of other
3870 optimization transforms, such as the MAP transform above.)
3871 * Credit information has been moved from source file headers into CREDITS.
3872 * Source file headers have been made more standard.
3873 * The CASE macro now compiles without complaining even when it has
3876 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3878 * (Note that the way that the PCL macroexpansions were rewritten
3879 to accommodate the change in DEFGENERIC below breaks binary
3880 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3881 not run under sbcl-0.6.2. Once we get out of alpha releases,
3882 i.e. hit release 1.0.0, we'll probably try to maintain binary
3883 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3884 and sbcl-1.4.4. Until then, however, it might be fairly common
3885 for maintenance releases to break binary compatibility.)
3886 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3888 * The reporting of unhandled signals has been changed to print some
3889 explanatory text as well as the report form. (Previously only
3890 the report form was printed.)
3891 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3892 it defines, so that the compiler no longer issues undefined function
3893 warnings for compiled-but-not-yet-loaded generic functions.
3894 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3895 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3896 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3897 section of the ANSI spec are supported.
3898 * The "" nickname for the "KEYWORD" package has been removed.
3899 The reader still handles symbol tokens which begin with a package marker
3900 as keywords, but it doesn't expose its mechanism for doing so in the
3901 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3902 * The system now issues STYLE-WARNINGs for contradictory TYPE
3903 proclamations. (Warnings for contradictory FTYPE proclamations would
3904 be nice too, but those can't be done usefully unless the type system
3905 is made smarter about FUNCTION types.)
3906 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3907 systematized, so that "*target-*.lisp is supposed to exist only on the
3908 target and imply that there's a related file which exists on the
3909 host, and *host-*.lisp is supposed to exist only on the host and imply
3910 that there's a related file which exists on the target. This involves a
3911 lot of renaming. Hopefully the acute confusion caused by the renaming
3912 will be justified by the reduction in chronic confusion..
3913 ** runtime-type.lisp -> early-target-type.lisp
3914 ** target-type.lisp -> late-target-type.lisp
3915 ** early-host-format.lisp -> early-format.lisp
3916 ** late-host-format.lisp -> late-format.lisp
3917 ** host-error.lisp -> misc-error.lisp
3918 ** early-error.lisp -> early-target-error.lisp
3919 ** late-error.lisp -> late-target-error.lisp
3920 ** host-defboot.lisp -> early-defboot.lisp
3921 ** code/misc.lisp -> code/target-misc.lisp
3922 ** code/host-misc.lisp -> code/misc.lisp
3923 ** code/numbers.lisp -> code/target-numbers.lisp
3924 ** code/early-numbers.lisp -> numbers.lisp
3925 ** early-host-type.lisp -> early-type.lisp
3926 ** late-host-type.lisp -> late-type.lisp
3927 ** host-typep.lisp -> typep.lisp
3928 ** load.lisp -> target-load.lisp
3929 ** host-load.lisp -> load.lisp
3930 ** host-disassem.lisp -> disassem.lisp
3931 ** host-insts.lisp -> insts.lisp
3932 ** byte-comp.lisp -> target-byte-comp.lisp
3933 ** host-byte-comp.lisp -> byte-comp.lisp
3934 ** host-signal.lisp -> signal.lisp
3935 ** host-defstruct.lisp -> defstruct.lisp
3936 ** late-target-type.lisp -> deftypes-for-target.lisp
3937 Furthermore, several other previously target-only files foo.lisp (e.g.
3938 hash-table.lisp and random.lisp) have been split into a target-and-host
3939 foo.lisp file and a target-only target-foo.lisp file, with their key type
3940 definitions in the target-and-host part, so that the cross-compiler will
3941 know more about target types.
3942 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3943 gone away. In their place are various *BACKEND-FOO* variables
3944 corresponding to the slots of the old structure.
3945 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3946 into the target SBCL has been fixed.
3947 * The chill.lisp system for loading cold code into a running SBCL
3949 * Support for the CMU CL "scavenger hook" extension has been removed.
3950 (It was undocumented and unused in the CMU CL sources that SBCL was
3951 derived from, and stale in sbcl-0.6.1.)
3952 * Various errors in the cross-compiler type system were detected
3953 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3954 (enabling various consistency checks). Many of them were fixed,
3955 but some hard problems remain, so the compiler is back to
3956 running without *TYPE-SYSTEM-INITIALIZED* for now.
3957 * As part of the cross-compiler type system cleanup, I implemented
3958 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3959 * I have started adding UNCROSS calls throughout the type system
3960 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3961 on cross-compiler input files will be able to go away, and various
3963 * CONSTANTP now returns true for quoted forms (as explicitly required
3966 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3968 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3969 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3970 (MAKE-LIST -1). In the longer term, it ought to become true
3971 that declarations are assertions even at SAFETY 1. For now, it's not
3972 quite true even at SAFETY 3, but it's at least more nearly true..
3973 (Note that this change seems to increases the size of the system by
3974 O(5%) and to decrease the speed of the compiler by 20% or more.)
3975 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3976 for the problem of printing dozens of lines of distracting information
3977 about low-level system machinery as part of the top stack frame
3978 on entry to the debugger when an undefined function was called.
3979 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3980 is not reset to COMMON-LISP-USER.
3981 * Compilation of stuff related to dyncount.lisp has been made conditional
3982 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3983 smaller. The various dyncount-related symbols have been moved into
3984 a new "SB-DYNCOUNT" package.
3985 * tty-inspect.lisp has been renamed to inspect.lisp.
3986 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3987 feature has gone away. (When we eventually port to other flavors of
3988 libc and/or Unix, we'll try to make the differences between flavors
3989 invisible at the user level.)
3990 * Various other *FEATURES* tags, and/or their associated conditionals,
3991 have been removed if obsolescent, or given better documentation, or
3992 sometimes given more-mnemonic names.
3994 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3996 * tidied up "make.sh" script
3997 * tidied up system directory structure
3998 * better "clean.sh" behavior
3999 * added doc/FOR-CMUCL-DEVELOPERS
4000 * many many small tweaks to output format, e.g. removing possibly-confusing
4001 trailing #\. character in DESCRIBE-INSTANCE
4002 * (EQUALP #\A 'A) no longer signals an error.
4003 * new hashing code, including EQUALP hashing
4004 * tidied up Lisp initialization and toplevel
4005 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4006 * command line argument processing
4007 * added POSIX-GETENV function to deal with Unix-ish environment variables
4008 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4009 terminating SBCL on EOF
4010 * non-verbose GC by default
4011 * There is no more "sbcl" shell script; the sbcl file is now the C
4012 runtime executable (just like CMU CL).
4013 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4015 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4016 symbol and package information as Lisp native symbol and package objects
4017 instead of strings naming symbols and strings naming packages. This way,
4018 whenever packages are renamed (as in warm init), debug information is
4019 transformed along with everything else.
4020 * tweaked the optimization policy declarations which control the building
4021 of SBCL itself. Now, among other things, the system no longer saves
4022 source location debugging information. (This helps two problems at once
4023 by reducing SBCL size and by keeping SBCL from trying to look for its
4024 sources -- which may not exist -- when reporting errors.)
4025 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4026 debugging and testing purposes
4027 * cleaned up printing, making the printer call PRINT-OBJECT for
4028 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4029 methods, giving nearly-ANSI behavior
4030 * converted almost all special variables to use *FOO* naming convention
4031 * deleted PARSE-TIME functionality, since it can be done portably
4032 * moved some files out of cold init into warm init
4033 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4035 * regularized formatting of source files
4036 * added an install.sh script
4037 * fixed ridiculous memory usage of cross-compiler by making
4038 compiler/alloc.lisp not try to do pooling unless it can hook
4039 itself into the GC of the cross-compilation host. Now the system
4040 builds nicely on my old laptop.
4041 * added :SB-ALLOC in target-features.lisp-expr
4042 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4043 was not implemented)
4044 * re-did condition handling and note reporting in the compiler. Notes
4045 are no longer handled by signalling conditions. Style warnings
4046 and warnings are handled more correctly and reported in such a way
4047 that it's easy to find one or the other in your output (so that you
4048 can e.g. figure out which of many problems caused COMPILE-FILE to
4050 * changed the severity of several compiler warnings from full WARNING
4051 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4052 compiler note reporting so that it doesn't use the condition system
4053 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4054 COMPILE-FILE command)
4055 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4056 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4057 no longer supported, since their implementation was deeply intertwingled
4058 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4059 * removed bogus "support" for compiler macros named (SETF FOO), and
4060 removed the compiler macro for SETF INFO (but only after making a fool
4061 of myself on the cmucl-imp mailing list by posting a bogus patch for
4062 DEFINE-COMPILER-MACRO..)
4063 * Compiled files containing forms which have side effects on the Lisp
4064 reader (such as DEFPACKAGE forms) are now handled more correctly.
4065 (Compiler queuing of top level lambdas has been suppressed by setting
4066 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4067 * deleted various currently-unused source files, e.g. gengc.lisp. They
4068 may be added back at some point e.g. when porting to other architectures,
4069 but until they are it's distracting to distribute them and to try to
4071 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4072 there were so many of them they're just distractions, and UNCROSS is
4073 known to be able to handle the current sources
4074 * moved PROFILE functionality into TRACE, so that it will be clear
4075 how the wrapping and unwrapping of functions when you profile them
4076 interacts with the wrapping and unwrapping of functions when you
4077 trace them. (Actually, the functionality isn't there yet, but at least
4078 the interface specification is there. Hopefully, the functionality will
4079 arrive with some future maintenance release.)
4080 * removed host-oops.lisp
4081 * changed signature of QUIT function to allow UNIX-CODE argument
4082 * fixed READ-SEQUENCE bug
4083 * tweaked verbose GC output so that it looks more like the progress
4084 output that ANSI specifies for functions like LOAD
4085 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4086 * added <http://sbcl.sourceforge.com> to the banner information printed by