1 changes in sbcl-0.8.22 relative to sbcl-0.8.21:
2 * incompatible change: the --noprogrammer option, deprecated since
3 version 0.7.5, has been removed. Please use the equivalent
4 --disable-debugger option instead.
5 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
7 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
8 inoperational for a while now) has been completely removed.
9 * null lexical environments are now printed as #<NULL-LEXENV>,
10 significantly reducing the amount of clutter in typical backtraces.
11 * documentation on weak pointers, finalization, and after GC hooks
12 has been added to the manual.
13 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
14 as other specialized array types, is much faster. SUBSEQ and
15 COPY-SEQ on such arrays have also been sped up.
16 * optimization: EQL is now more efficient when at least other argument
17 is known to be of type (OR FIXNUM (NOT NUMBER)).
18 * fixed bug: the runtime is now less vulnerable to changes in the
19 size of the SBCL object on OS X, and virtual memory is reserved for
20 all spaces that need to be at a fixed address.
21 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
22 * fixed bug: finalizers and after GC hooks that cause consing are now
24 * fixed bug: compiler error messages and summaries are now printed to
25 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
26 * fixed inference of the upper bound of an iteration variable.
27 (reported by Rajat Datta).
28 * fixed bug 211e: calling local functions with duplicated constant
29 keyword argument no longer causes a bogus style warning about an
31 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
32 ftype declaration to be disregarded. (reported by Dave Roberts)
33 * fixed bug 373: caused by erronous compilation of references to alien
34 variables in the runtime on ppc/darwin.
35 * fixed bug 376: CONJUGATE type deriver.
36 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
37 (reported by Baughn on #lisp)
38 * a cross-compiler bug on non-x86 platforms has been identified and
39 fixed. (thanks to Bruno Haible)
40 * improvements to the MIPS runtime code for increased stability.
41 (thanks to Thiemo Seufer)
42 * increased the maximimum compact environment size to allow
43 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
44 * partial workaround for bug 135: don't name the function we're
45 calling for hairy cases of EVAL, so as not to accumulate one environment
46 entry per call to EVAL. (reported by Kevin Reid)
47 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
48 * optimization: type testing for non-vector arrays should be faster.
49 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
51 * bug fix: setting 31st element of a bit vector to zero did not work
53 * bug fix: redefining a class definition which failed due to a
54 previous accessor / function clash now works (but see BUGS entry
55 #380 for more problems in this area). (thanks to Zach Beane)
56 * on x86 compiler supports stack allocation of results of simple
57 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
58 * fixed some bugs related to Unicode integration:
59 ** the restarts for recovering from input and output encoding
60 errors only appear when there is in fact such an error to
62 * fixed some bugs revealed by Paul Dietz' test suite:
63 ** MISC.549 and similar: late transformation of unsafe type
64 assertions into derived types caused unexpected code
66 ** SCALE-FLOAT type deriver is less wrong.
67 ** type derivers for EXP, LOG and similar functions compute result
68 types for complex arguments better.
69 ** (MISC.563) CONJUGATE type deriver works for very restricted
71 ** out-of-line type testers for character strings are available.
72 ** EQUAL compiler transform understands specialness of objects
74 ** accessing double-floats stored on the stack now works on x86-64.
75 ** debugger internals could sometimes create invalid lispobjs,
76 resulting in GC crashes.
77 ** MISC.548: type check weakening can convert required type into
79 ** initialization forms for bindings are not in scope of free special
82 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
83 * incompatible change: thread support for non-NPTL systems has
84 been removed - locking is buggy and unreliable. A threaded
85 SBCL build will now warn at startup and refuse to create
86 new threads, unless futex support is detected in the kernel
87 * incompatible change: the top level REPL now has only an ABORT
88 restart associated with it, not TOPLEVEL and ABORT as it used to.
89 TOP and TOPLEVEL are now available as debugger commands for
90 returning to the top level.
91 * incompatible change: forms evaluated in the REPL now use the
92 global optimization policy.
93 * incompatible change: user- and system-initialization files are
94 no longer processed with LOAD, but by READ and EVAL; hence the
95 global optimization policy, startup package, readtable, etc,
97 * The .fasl file format number has been incremented because of
98 various incompatible changes.
99 * internal entry point details and argument counts no longer appear
100 in backtraces unless explicitly requested by setting
101 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
102 * built-in and standard functions no longer have names like "top
103 level local call to FOO".
104 * fixed bug 32: functions defined in non-null lexical environments
105 now have more legible printed representation
106 * fixed bug 33: functions defined in non-null lexical environemnts
107 are now more amenable to inspection by INSPECT.
108 * workaround for bug 354: XEPs no longer appear in backtraces unless
109 explicitly requested.
110 * fixed bug: receiving the signal which results from attempting to
111 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
112 SIGBUS on other *BSDs) on architectures where the C stack is also
113 the Lisp stack (x86 and x86-64) from foreign code no longer leads
114 to debugger confusion or wild instability. (reported by Cheuksan
116 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
117 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
118 (reported by Lutz Euler)
119 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
120 compiled code. (reported by Paul Dietz)
121 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
122 specialized parameter to a method to something that is not TYPEP
123 the specializer is now possible.
124 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
125 face of package deletion.
126 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
127 pathname. (reported by David Tolpin on c.l.l)
128 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
129 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
130 STRING-STREAM and FILE-STREAM, respectively.
131 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
132 than previously; in addition, its test suite is now run on build.
133 (thanks to Luke Gorrie)
134 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
135 * optimization: added a immediate representation for single-floats
137 * fixed some bugs revealed by Paul Dietz' test suite:
138 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
139 ** attempting to create a package with a colliding nickname causes
140 correctable errors to be signalled.
141 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
142 ** several x86-64 backend bugs related to sign-extension of immediate
145 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
146 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
147 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
149 * fixed loading of multiply forward-referenced layouts.
150 (thanks to Cheuksan Wang)
151 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
152 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
153 (see COMPILE-FILE documentation for details of :PRINT option)
154 * fixed bugs 19 and 317: fixed-format floating point printing is
155 more accurate. This also fixes a bug reported by Adam Warner
156 related to the ~@F format directive.
157 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
159 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
160 dispatch table if the from-char is a dispatch macro character.
161 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
162 by the wordsize no longer ignore the last word. (reported by Lutz
164 * fixed bug in type checking of dynamic-extent variables. (reported
166 * optimization: sequence traversal functions use their freedom to
167 coerce function designators to functions.
168 * optimization: code with many calls to CLOS methods specialized on
169 CLOS classes has had redundant type checks removed.
170 * optimization: added declarations to speed up operations that access
171 the internal character database (for example STRING-UPCASE)
172 * optimization: comparison operations between floats and sufficiently small
173 fixnums no longer create extra rationals
174 * fixed some bugs related to Unicode integration:
175 ** portions of multibyte characters at the end of buffers for
176 character-based file input are correctly transferred to the
177 start of the buffer at the next read.
178 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
179 passing it through to OPEN.
180 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
181 argument given to internal calls to OPEN.
182 ** the built-in comment readers (introduced by character sequences
183 ";" and "#|") are more forgiving to encoding errors; they will
184 STYLE-WARN and then attempt to resync the stream at a character
185 boundary. (thanks to Teemu Kalvas)
186 * fixed some bugs in the x86-64 port:
187 ** Negative short int return values from c-calls are sign-extended
189 ** The stack is aligned to 16-bytes for c-calls, as required by
190 the ABI. (thanks to Cheuksan Wang)
191 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
192 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
193 for enums. (thanks to Vincent Arkesteijn)
194 ** Multiple small optimizations and bugfixes for floating point
196 * fixed some bugs revealed by Paul Dietz' test suite:
197 ** Space, Tab, Linefeed, Return and Page have the invalid
198 secondary constituent character trait.
199 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
201 ** WITH-INPUT-FROM-STRING should only update the index place on
203 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
205 ** Bit-array operations (BIT-AND and similar) worked incorrectly
206 with one-dimensional arrays with fill pointers.
207 ** TYPE-OF failed on a complex with an integer realpart and a
209 ** compiler failure during type inference for the code of form
210 (IF (EQL X (THE ...)) ...) (MISC.535).
212 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
213 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
214 platform. The port supports 61-bit fixnums, large memory spaces
215 and reloading shared object files.
216 * enhancement: saving cores with foreign code loaded is now
217 supported on ppc/Darwin in addition to the previously supported
219 * enhancement: the statistical profiler now walks deeper into the
220 call stack for more meaningful call-graphs and accrued time
221 reports (x86/x86-64 only). It also now reports time spent in
223 * enhancement: it is now possible to trace most individual methods
224 of a generic function in addition to tracing the generic function
226 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
228 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
229 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
230 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
231 produces an error. (thanks to Vincent Arkesteijn)
232 * bug fix: NAMESTRING on pathnames with :WILD components in their
233 directories works correctly. (thanks to Artem V. Andreev)
234 * fixed bug 125: compiler preserves identity of closures. (reported
236 * bug fixed: functions with &REST arguments sometimes failed with
237 "Undefined function" when compiled with (DEBUG 3). (reported by
239 * bug fix: overflow during compiling of setting element of a bit
240 vector with constant index and value. (reported by Timmy Douglas)
241 * build fix: fixed the dependence on *LOAD-PATHNAME* and
242 *COMPILE-FILE-PATHNAME* being absolute pathnames.
243 * on x86 compiler partially supports stack allocation of dynamic-extent
245 * GO and RETURN-FROM do not check the extent of their exit points
246 when compiled with SAFETY 0.
247 * fixed some bugs related to Unicode integration:
248 ** encoding and decoding errors are now much more robustly
249 handled; it should now be possible to recover even from invalid
250 input or output to the terminal. (thanks to Teemu Kalvas)
251 ** provided a first cut at implementing STRING-TO-OCTETS and
252 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
253 ** altered the SB-MD5 contributed module slightly, changing the
254 interface just enough for it to be supportable for builds where
255 lisp characters are not eight bits.
256 * fixed some bugs revealed by Paul Dietz' test suite:
257 ** the FORMATTER-generated functions for ~V[ conditionals require
258 the correct number of arguments.
259 ** READ-FROM-STRING returns the mandated second value when applied
260 to displaced strings.
261 ** the #\Rubout and #\Backspace characters are treated as invalid
262 constituent characters by the tokenizer.
264 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
265 * new feature: reloading changed shared object files with
266 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
267 * new feature: references to foreign variables and functions
268 can now be compiled and loaded before the corresponding shared
269 object file is loaded, as long as the foreign definitions are
270 available at runtime.
271 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
272 supported. (thanks to Dan Debertin)
273 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
274 just keywords. (thanks to Vincent Arkesteijn)
275 * fixed bug #331: structure-class instances corresponding to
276 DEFSTRUCT forms are now created eagerly.
277 * fixed bug #345: backtraces from calls to undefined functions work
278 on x86 as well. Related bug #61 is now also partially fixed on x86
279 (backtraces from throws to unknown catch tags.)
280 * bug fix: lambda-list parsing is now stricter vrt. order and number
281 of lambda-list keywords.
282 * bug fix: as specified by AMOP, an error is signalled if a
283 class-option appears multiple times in a DEFCLASS form. (reported
285 * bug fix: RANDOM can be compiled when the compiler derives the type
286 of its numeric argument as a disjoint set of small integers.
287 (reported by Paul Dietz)
288 * bug fix: starting a core saved with shared objects loaded when
289 those objects are not available no longer causes threaded SBCL to
290 hang. (reported by Sean Ross)
291 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
293 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
294 argument, which must always be NIL. (reported by Kalle Niemitalo)
295 * bug fix: printing 1.0d+23 no longer results in an error.
296 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
297 * bug fix: structure slot setters preserve evaluation order. (thanks
299 * fixed some bugs related to Unicode integration:
300 ** RUN-PROGRAM can allow its child to take input from a Lisp
301 stream. (reported by Stefan Scholl)
302 ** ASDF-INSTALL successfully downloads .tar.gz files in all
303 locales. (reported by Ken Causey)
304 * fixed some bugs revealed by Paul Dietz' test suite:
305 ** INCF, DECF and REMF evaluate their place form as specified in
307 ** FORMATTER expands ~{ ~} iteration directives with V or #
308 parameters correctly.
309 ** FORMATTER deals with the ~@[ ~] conditional directive where the
310 consequent uses no arguments correctly.
311 ** the system has a partial understanding of the (COMPLEX RATIO)
313 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
316 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
317 * new feature: a build-time option (controlled by the :SB-UNICODE
318 keyword feature, enabled by default) for building the system with
319 support for the entire 21-bit character space defined by the
321 * new feature: the system now has rudimentary external-format
322 support; the primary user-visible change at this time is that
323 characters with the high bit set (such as lower-case-e-acute) will
324 print correctly to a terminal in a UTF-8 environment.
325 * minor incompatible change: BASE-CHAR no longer names a class;
326 however, CHARACTER continues to do so, as required by ANSI.
327 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
328 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
329 has been moved to the SB-EXT package (temporarily re-exported from
331 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
332 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
334 * fixed bug #351: better error-handlind and reporting for malformed
336 * fixed bug #350: bignum-printing is now more memory-efficient,
337 allowing printing of very large bignums, eg. (expt 2 10000000).
338 (reported by Bruno Haible)
339 * fixed bug #302: better primitive-type selection for intersection
341 * fixed bug #308: non-graphic characters now all have names, as
342 required. (reported by Bruno Haible)
343 * bug fix: redefining a class with different superclasses now correctly
344 removes it from the direct-subclasses of its previous superclasses.
345 (reported by David Morse)
346 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
347 new value now works. (reported by Bruno Haible)
348 * bug fix: correct canonicalization of multiple non-standard slot
349 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
350 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
351 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
352 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
353 now exists, an signals an error.
354 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
355 during macroexpansion for non-iteration variables as well. (reported
356 by Bruno Haible for CMUCL)
357 * bug fix: Cyclic structures and unprintable objects in compiler
358 messages no longer cause errors. (reported by Bruno Haible)
359 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
360 and READ-FROM-STRING all now return a primary value of NIL if
361 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
362 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
363 (reported by Bruno Haible for CMUCL)
364 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
365 do not designate a condition. (reported by Bruno Haible for
367 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
368 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
369 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
370 permitting resolution in favour of any of the conflicting symbols.
371 (reported by Bruno Haible for CMUCL)
372 * bug fix: EQUAL compiler optimizations is less aggressive on
373 strings which can potentially compare true despite having distinct
374 specialized array element types.
375 * bug fix: unit enumerations can be defined without dividing by
376 zero. (thanks to Vincent Arkesteijn)
377 * FORMAT compile-time argument count checking has been enhanced.
378 (report from Bruno Haible for CMUCL)
379 * a partial workaround for the bug 262: the compiler does not try to
380 inline-expand a local function doing RETURN-FROM from a deleted
381 BLOCK. (thanks to Peter Denno for the bug report and to David
382 Wragg for the simple test case)
383 * fixed some bugs revealed by Paul Dietz' test suite:
384 ** RENAME-PACKAGE allows all package designators as new package
386 ** constraint propagation and lambda variable substitution are
387 more cautious in dealing with partially deleted code.
388 ** compiler sometimes lost reoptimization passes.
389 ** CERROR, given a condition as condition designator, passes its
390 remaining arguments to the continue format control without
392 ** Case-altering FORMAT directives work correctly on non-ASCII
394 ** The REAL type specifier handles bounds outside the floating
395 point ranges without signalling FLOATING-POINT-OVERFLOW.
396 ** Functions with IR1-transformations can create intercomponent
397 references to global functions.
398 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
400 ** FORMAT ~R treats a nil value for its first parameter correctly.
402 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
403 * enhancement: saving cores with foreign code loaded is now
404 supported on x86/NetBSD and sparc/Linux in addition to the previously
406 * bug fix: on some platforms repeated installations caused multiple
407 copies of HTML documentation to be installed -- should not happen
408 any more. (reported by Stefan Scholl)
409 * bug fix: parsing self-recursive alien record types multiple times
410 no longer causes infinite recursion. (reported by Thomas F. Burdick,
411 original patch by Helmut Eller for CMUCL)
412 * bug fix: stack-exhaustion detection works now on NetBSD as well.
413 (thanks to Richard Kreuter)
414 * bug fix: defining classes whose accessors are methods on existing
415 generic functions in other (locked) packages no longer signals
416 bogus package lock violations. (reported by François-René Rideau)
417 * bug fix: special variables as DEFMETHOD parameters no longer have
418 associated bogus type declarations. (reported by David Wragg and
420 * bug fix: read-write consistency on streams of element-type
421 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
422 * bug fix: redefiniton of the only method of a generic function with
423 no DEFGENERIC no longer emits a full WARNING. In addition,
424 redefinition of generic functions with no DEFGENERIC to an
425 incompatible lambda list now signals an error. (thanks to Zach
427 * bug fix: DEFGENERIC now works even when there's a function of the
428 same name in an enclosing lexical environment. (thanks to Zach
430 * fixed compiler failure, caused by instrumenting code during
431 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
432 * optimization: added loop analysis and improved register allocation
433 * fixed some bugs revealed by Paul Dietz' test suite:
434 ** POSITION on displaced vectors with non-zero displacement
435 returns the right answer.
436 ** (SIMPLE-STRING) is a valid type specifier for sequence
438 ** *PRINT-LEVEL* handling for slotless structures is pedantically
440 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
441 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
443 ** The justification version of the FORMAT ~< directive treats
444 non-zero minpad parameter correctly.
446 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
447 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
448 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
449 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
450 the supported interface.
451 * new feature: Single-stepping of code compiled with DEBUG 2 or
452 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
453 * new feature: saving cores with foreign code loaded is now
454 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
455 Timothy Moore's work for CMUCL)
456 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
457 parameters to * instead of NIL if no initform is supplied.
458 (reported by Johan Bockgaard)
459 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
460 correctly when FUNCALL appears as the car of the form. Note:
461 despite this FUNCALL forms are not currently subject to
462 compiler-macro expansion. (port of Raymond Toy's fix for the
463 same from CMUCL, reported by Johan Bockgaard)
464 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
465 (thanks for Teemu Kalvas)
466 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
467 case instead of terminating the iteration (thanks for Julian
468 Squires, Sean Champ and Raymond Toy)
469 * bug fix: incorrect expansion of defgeneric that caused a style
470 warning. (thanks for Zach Beane)
471 * bug fix: slot accessor effective method computation works properly
472 for classes with multiple non-standard applicable methods on
473 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
474 * on x86 compiler supports stack allocation of results of LIST and
475 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
476 CMUCL implementation by Gerd Moellmann)
477 * fixed some bugs revealed by Paul Dietz' test suite:
478 ** FORMAT strings with both the ~<~:;~> form of the justification
479 directive and pretty-printing directives cause an error.
481 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
482 * incompatible change: the internal functions
483 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
484 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
485 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
486 instead of the old functions.
487 * new feature: on platforms where "dladdr" is available foreign
488 function names now appear in backtraces. (based on Helmut Eller's
490 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
491 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
493 * documentation: ASDF manual is now distributed with SBCL.
494 * bug fix: SBCL can now load its contributed modules with REQUIRE
495 even if the system-provided entries have been removed from
496 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
498 * bug fix: backtraces involving undefined functions or assembly
499 routines are more informative. (thanks to Brian Downing)
500 * bug fix: mutually referent alien structures now work correctly.
501 (reported by Rick Taube)
502 * bug fix: structures defined by WITH-ALIEN can be referred to
503 within other definitions in the same WITH-ALIEN.
504 * bug fix: division operators (MOD, TRUNCATE and the like) with
505 constant zero divisors and integer dividends no longer generate
507 * bug fix: provide default methods for INPUT-STREAM-P and
508 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
509 * bug fix: improve the detection and resolution of MOP metacycles
510 (where effective method computation depends on the generic function
511 having its effective method computed). (reported by Bruno Haible)
512 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
513 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
514 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
515 * bug fix: pathnames with relative directory components can be
516 represented relative to default pathnames.
517 * optimization: in taking the GCD of bignums, reduce the two bignums
518 to approximately the same size (using Euclid's algorithm) before
519 applying the more sophisticated binary GCD. (thanks to Juho
521 * optimization: COUNT on bitvectors now operates word-at-a-time.
522 * optimization: ASH with a positive, but not necessarily constant,
523 (leftwards) shift, when in a modular context, is compiled to a
525 * fixed some bugs revealed by Paul Dietz' test suite:
526 ** FORMAT variable parameters ("~V<char>") are defaulted properly
527 if the corresponding argument is NIL.
528 ** FORMAT directives accepting a minpad parameter treat negative
530 ** literal commas as character parameters to FORMAT directives are
532 ** literal spaces directly after ~<Newline> directives within a
533 format-logical-block (~:< ~@:>) do not induce :FILL-style
534 conditional newlines.
535 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
536 as specified: it no longer includes conditional newlines).
537 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
539 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
540 :PER-LINE-PREFIX argument does not evaluate to a string.
542 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
543 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
544 the manual for details; for now, package locks can be disabled by
545 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
546 if no major problems are found then it is likely that they will be
547 compiled in unconditionally.
548 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
549 unsupported operators on all platforms. To load a shared library
550 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
551 object file, link it into a shared library outside of SBCL and
552 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
554 * bug workaround: ROOM T and the GC were not working together
555 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
556 reporting facility aren't preserved. That reporting has been
557 disabled, so now until and unless someone figures out how to make it
558 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
559 * minor incompatible change: as threatened around sbcl-0.8.0, the
560 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
561 The symbols are also exported from SB-PCL for backwards
562 compatibility, but more so than before SB-PCL should be treated as
563 an implementation-internal package.
564 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
566 * the SB-SPROF contrib now works on (most) non-x86 architectures.
567 It is known as of this release not to work on the Alpha, however.
568 * fixed bug #167: errors signalled due to illegal syntax in method
569 bodies are now more legible.
570 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
571 designators and can hence be used with TYPEP.
572 * fixed bug #333: CHECK-TYPE now ensures that the type error
573 signalled, if any, has the right object to be accessed by
574 TYPE-ERROR-DATUM. (reported by Tony Martinez)
575 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
576 5.1.2.3 for multiple-value place subforms. (reported by Kalle
578 * fixed bug #334: programmatic addition of slots using specialized
579 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
580 and :ALLOCATION :CLASS slots.
581 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
582 float exponents. (rereported by Peter Seibel)
583 * fixed a bug: #\Space (and other whitespace characters) are no
584 longer considered to be macro characters in standard syntax by
586 * fixed bug: initialization of condition class metaobjects no longer
587 causes an instance of the condition to be created. (reported by
589 * fixed bug: it is now possible to have slots such that
590 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
591 description is neither :INSTANCE nor :CLASS.
592 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
593 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
594 SLOT-VALUE-USING-CLASS), and no longer causes errors with
595 non-standard SLOT-DEFINITION-ALLOCATION values.
596 * fixed bugs: various race conditions handling exiting threads.
597 CL-PPCRE's thread torture test now passes.
598 * fixed arguably-a-bug: GC time proportional to number of threads in
599 system even when most of them are idle
600 * optimization: improved performance of BIT and SBIT on bit-vectors.
601 * .fasl file incompatibility: The fasl file version number has
602 been incremented because of changes associated with package locks.
604 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
605 * minor incompatible change: the system no longer provides
606 optimization hints (conditions which are TYPEP
607 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
608 mode; these hints are still emitted when higher SPEED optimization
610 * new contrib module: a sampling profiler (profiling by statistical
611 sampling, rather than by instrumenting functions) is available as
612 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
613 * the behaviour of the standard function ED is now customizeable by
614 third parties through a hook variable: see ED's documentation
615 string for information on the protocol.
616 * the compiler no longer emits efficiency notes for (FUNCALL X)
617 when the type of X is uncertain under default optimization
619 * fixed bug 276: mutating a binding of a specialized parameter to a
620 method to something that is not TYPEP the specializer is now
622 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
623 exhibit the required behaviour on the x86 platform. (thanks to
624 Peter van Eynde, Eric Marsden and Bruno Haible)
625 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
626 even for difficult arguments. (reported by Peter Graves)
627 * fixed bug 141a: the backquote printer now descends quoted
629 * fixed another bug in backquote printing: no more destructive
630 modification of the form's list structure. (reported by Brian
632 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
633 move between its address being taken and the call to
634 interrupt_thread, fixing a crashing race condition.
635 * the SB-POSIX contrib implementation has been adjusted so that it
636 no longer exhibits ridiculously poor performance when constructing
637 instances corresponding to C structs.
639 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
640 * minor incompatible change: the sb-grovel contrib now treats C
641 structures as alien (in the sense of SB-ALIEN) objects rather than
642 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
643 has implications for memory management of client code
644 (sb-grovel-returned objects must now be manually managed) and for
645 type safety (alien objects now have full types).
646 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
647 used to control emission of compiler diagnostics, rather than the
648 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
649 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
650 quality should be considered deprecated.
651 * install.sh now installs the user manual as well
652 * (not quite a new documentable feature, but worth considering in
653 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
654 beginnings of a semantically meaningful condition hierarchy is
655 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
657 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
658 designator as the defaults argument.
659 * fixed bug: Displaced arrays whose displaced-to array has become
660 too small now cause ARRAY-DIMENSION to signal an error, providing
661 for safer bounds-checking. (reported by Bruno Haible)
662 * fixed bug: DEFCLASS slot definitions with identical :READER and
663 :WRITER names now signal a reasonable error. (reported by Thomas
665 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
667 * fixed bug: Invalid slot specification errors now print correctly.
668 (thanks to Zach Beane)
669 * fixed bug 320: Shared to local slot value transfers in class
670 redefinitions now happen corrently from superclasses as
671 well. (reported by Bruno Haible)
672 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
674 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
675 improper lists correctly. (reported by Bruno Haible)
676 * fixed bug 313: source-transform for <fun-name> was erroneously
677 applied to a call of a value of a variable with name <fun-name>.
678 (reported by Antonio Menezes Leitao)
679 * fixed bug 307: The obsolete instance protocol ensures that
680 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
681 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
682 remove dead unknown-values globs from the middle of the stack.
683 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
684 easier to resume long computations after using *BREAK-ON-SIGNALS*
685 to diagnose and fix failures. (thanks to Nikodemus Siivola)
686 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
687 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
688 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
689 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
690 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
692 * fixed a (fairly theoretical) bug in string printing: if
693 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
694 does not have array-element-type equal to the most general string
696 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
697 function name. (thanks to Nikodemus Siivola)
698 * fixed bug: compile-time format string checker failed on
699 non-closed ~{. (reported by Thomas F Burdick)
700 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
701 don't warn on legal loop code involving more than one aggregate
702 boolean. (thanks to Nikodemus Siivola)
703 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
704 in keyword position to MAKE-PACKAGE was not being reported as
706 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
707 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
709 * fixed bugs 280 and 312: the checking for multiple definitions in a
710 file is less likely to become confused by uses of inline
712 * fixed bug: the #S reader macro performs the keyword coercion
713 specified for slot names. (reported by Kalle Niemitalo)
714 * fixed bug: lambda lists may contain symbols whose names start with
715 & but are not lambda-list-keywords; their occurrence triggers a
717 * fixed bug 321: define-method-combination argument lambda lists do
718 not cause invalid code to be generated when &optional or &aux
719 variables have default values. (reported by Bruno Haible)
720 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
721 have CLOS classes; this appears also to have fixed reported
722 instability in stack exhaustion detection.
723 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
724 MUFFLE-WARNING when no associated restart is present is now
726 * optimization: rearranged the expansion of various defining macros
727 so that each expands into only one top-level form in a
728 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
730 * optimization: used a previously little-used slot in symbols to
731 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
733 * fixed some bugs revealed by Paul Dietz' test suite:
734 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
736 ** Fixed an optimization bug related to inheritance of initforms
737 from local to shared slots.
738 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
739 ** CAST optimizer forgot to flush argument derived type.
740 ** print/read consistency on floats is now orders of magnitude
741 more likely. (thanks also to Bruno Haible for a similar report
743 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
744 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
745 for objects of type REAL. Make it so.
746 ** SXHASH is specified (infelicitously) to respect similarity,
747 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
748 it so. (thanks to Markus Ziegler)
749 ** on the Alpha, the compiler succeeds in compiling functions
750 returning a known number of arguments greater than 63.
751 ** fixed handling of invalid NIL arguments in keyword position
753 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
755 ** provided a readably-printable representation for RANDOM-STATE
757 ** ensured that pathnames, where they have a namestring, always
758 print using #P"..." syntax.
760 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
761 * Support for the forthcoming 2.0 version of the NetBSD kernel
762 running on x86 hardware has been added. (thanks to Perry
763 E. Metzger most immediately, and others for their past work)
764 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
765 current release version; SBCL's OpenBSD support had been broken
766 since about the time of OpenBSD's switch to ELF binary format.)
767 * [placeholder for DX summary]
768 ** user code with &REST lists declared dynamic-extent, under high
769 speed or space and low safety and debug optimization policy.
770 * The manual has been converted to Texinfo format and the debugger
771 chapter from the cmucl manual has been added.
772 * A facility has been added to extract documentation strings from
773 sbcl and store them as Texinfo-formatted snippets for inclusion in
774 the manual (via Texinfo's @include directive)
775 * bug fix: compiler emitted division in optimized DEREF. (thanks for
776 the test case to Dave Roberts)
777 * bug fix: multidimensional simple arrays loaded from FASLs had fill
778 pointers. (reported by Sean Ross)
779 * bug fix: PROFILE output is printed nicely even for large numerical
780 values. (thanks to Zach Beane)
781 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
782 greater than 32 handle EOF correctly.
783 * bug fix: on X86 an immediate argument of the IMUL instruction is
784 correctly printed in disassembly. (thanks to Lutz Euler)
785 * bug fix: class slots in redefined classes preserve their old
786 values. (thanks to Bruno Haible and Nikodemus Siivola)
787 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
788 succeeds. (reported by Marco Baringer)
789 * fixed some bugs revealed by Paul Dietz' test suite:
790 ** READ-SEQUENCE now works on ECHO-STREAMs.
791 ** RATIONALIZE works more according to its specification. (thanks
794 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
795 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
796 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
797 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
798 should matter to you only if you rebind the printer control
799 variables and then find you want different bindings in the
800 debugger than in the ordinary execution of your program.)
801 * The runtime build system has been tweaked to support building
802 (on SPARC/SunOS) using a C compiler which invokes Sun's own
803 assembler and linker. (thanks to Nikodemus Siivola)
804 * Unbound, undefined, undeclared variables now trigger full
805 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
806 more often programmer error than deliberate exploitation of undefined
808 * optimization: the hash algorithm for strings has changed to one
809 that is less vulnerable to spurious collisions. (thanks to Juho
811 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
812 needless bounds checking. (thanks to Juho Snellman)
813 * optimization: implemented multiplication as a modular
814 (UNSIGNED-BYTE 32) operation on the PPC backend.
815 * fixed some bugs revealed by Paul Dietz' test suite:
816 ** ADJUST-ARRAY now copies the datum in a zero rank array if
818 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
819 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
820 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
822 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
824 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
826 ** ECHO-STREAMs no longer attempt to echo the end of file value to
827 their output stream on EOF from read.
828 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
829 have been read to end-of-file.
830 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
832 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
833 description of determination of which consecutive characters
835 ** Printing the "Space" character with escaping on now yields "#\\ ",
836 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
837 ** Reading floating-point numbers with *READ-BASE* set to a number
838 less than 10 works correctly.
839 ** Reading floating-point numbers with *READ-BASE* set to a number
840 more than 10 works correctly.
841 ** Printing with *PRINT-READABLY* targets the standard readtable, not
842 the readtable currently in effect.
844 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
845 * minor incompatible change: parsing of namestrings on a physical
846 (Unix) host has changed; numbers after the final #\. in a
847 namestring are no longer interpreted as a version field. This is
848 intented to be largely invisible to the user, except that the
849 meaning of the namestring "*.*.*" has changed: it now refers to a
850 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
851 should usually be replaced by
852 (make-pathname :name :wild :type :wild :version :wild)
853 with the added benefit that this is more likely to be portable.
854 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
855 option to OPEN now signals an error if the file being opened
856 exists; this may have an impact on existing code.
857 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
858 signals received, on the PowerPC platforms (both Linux and
859 Darwin). (thanks to Pierre Mai for pointing out the location of
861 * several fixes on OS X: The system now builds and runs cleanly on
862 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
863 (thanks to Brian Mastenbrook)
864 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
865 second-resolution: integer multiples of 1/3600 between -24 and 24.
866 (thanks to Vincent Arkesteijn)
867 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
868 type when called with 1 argument; PEEK-CHAR checked type of
869 PEEK-TYPE only after having read first character from a
870 stream. (reported by Peter Graves)
871 * bug fix: the garbage collector now has much better locality
872 behaviour, and in particular no longer treats all memory as being
873 exhausted when many small objects point to each other in a deeply
875 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
876 recognized as being TYPEP their class.
877 * bug fix: the PUSHNEW documentation string has been corrected.
878 (thanks to Vincent Arkesteijn)
879 * bug fix: defaulting of the value for the last of an atypically
880 large number of multiple values being bound was not being
881 performed correctly on the Alpha or PPC platforms
882 * optimization: implemented multiplication as a modular
883 (UNSIGNED-BYTE 32) operation on the x86 backend.
884 * optimization: SEARCH on simple-base-strings can now be open-coded.
885 (see also contrib/compiler-extras.lisp for inspiration for
886 teaching the compiler about the Boyer-Moore algorithm).
887 * value, returned by MAX (and MIN) called with several EQUALP, but
888 not EQL, arguments now does not depend on compiler settings.
889 * fixed some bugs revealed by Paul Dietz' test suite:
890 ** in stack analysis liveness information is propagated from
891 non-local entry points.
892 ** pathwise CAST removing failed when the CAST node did not start
894 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
895 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
897 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
898 host is already defined.
899 ** RENAME-FILE works on streams instead of signalling an internal
901 ** PEEK-CHAR uses the current readtable when determining whether
902 or not a character is whitespace.
903 ** MERGE-PATHNAMES handles the case when the pathname does not
904 specify a name while the default-pathname specifies a version.
905 ** Pathnames now stand a better chance of respecting print/read
907 ** Attempting to use standardized file system operators with a
908 pathname with invalid :DIRECTORY components signals a
910 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
912 ** DIRECTORY on logical pathnames is more correct.
913 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
914 signal a TYPE-ERROR if their argument is not a stream.
915 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
916 designator argument does not designate a stream.
917 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
918 examining the synonym.
919 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
921 ** OPEN and WITH-OPEN-STREAM allow opening streams with
922 element-type larger than ([UN]SIGNED-BYTE 32).
924 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
925 * When built with the :SB-FUTEX feature, threaded builds now take
926 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
927 for faster/more reliable mutex and condition variable support.
928 * Incompatible change (but one you probably shouldn't have been using
929 anyway): the interface and code for arbitrating between multiple
930 threads in the same user session has been redesigned.
931 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
932 error when called without an explicit environment argument.
933 (thanks to Nikodemus Siivola)
934 * bug fix: buffered :DIRECTION :IO streams are less likely to become
935 confused about their position. (thanks to Adam Warner and Gerd
937 * bug fix: Pretty printing backquoted forms with unquotations in the
938 argument list position of various code constructs such as LAMBDA
939 now works correctly. (reported by Paul Dietz)
940 * bug fix: Pretty printing unquotations no longer loses all
941 stream position information.
942 * optimization: performance of string output streams is now less
943 poor for multiple small sequence writes.
944 * optimization: performance of CSUBTYPEP in the presence of complex
945 expressions involving CONS and NOT many times has been improved.
946 (reported by Paul Dietz)
947 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
949 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
950 streams for the REPL, for the debugger, and so on. (thanks to
952 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
953 * fixed some bugs revealed by Paul Dietz' test suite:
954 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
955 is ignored if it is too small, rather than propagating through
956 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
957 ** extremely complex negations of CONS types were not being
958 sufficiently canonicalized, leading to inconsistencies in
960 ** VALUES tranformer lost derived type.
962 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
963 * fixed a bootstrapping bug: the build process no longer assumes
964 that the various BOOLE-related constants have the same value in
965 host and target lisps. (noted by Paul Dietz' test suite on an
966 SBCL binary built from CLISP)
967 * The system can now be dynamically linked on the MIPS platform,
968 which enables dynamic loading of foreign code from Lisp. (thanks
969 to Ralf Baechle for discussions on the MIPS ABI)
970 * The system now records debugging information for its own source
971 files in a filesystem-position-independent manner, relative to
972 the "SYS" logical host.
973 * fixed a compiler bug: MV-LET convertion did not check references
974 to the "max args" entry point. (reported by Brian Downing)
975 * tweaked disassembly notes to be less confident about proclaiming
976 some instruction as an LRA. (thanks to Brian Downing)
977 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
978 now each have their own history, command character, and other
979 characteristics. (thanks to David Lichteblau)
980 * fixed some bugs revealed by Paul Dietz' test suite:
981 ** compiler failure in compiling LOGAND expressions including a
983 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
984 shift greater than 32.
985 ** FLUSH-DEST did not mark blocks for type check regeneration.
986 ** HANDLER-CASE failed to accept declarations in handler clauses
987 in some circumstances.
989 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
990 * New code in contrib/sb-introspect (still probably not entirely
991 stable yet) provides some support for smart Lisp development
992 environments like SLIME.
993 * The conditions signalled for errors occurring when loading .fasl
994 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
995 in a way which should help ASDF recover gracefully.
996 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
997 stuff has been cleaned up. If you code contrib/ stuff, this might
998 affect you, and you can look at contrib/README, contrib/STANDARDS,
999 and/or the 0.8.4.27 diff to check.
1000 * In full calls the compiler now does not generate checks for declared
1001 argument types for all arguments.
1002 * various threading fixes
1003 ** and some experimental patches which didn't make it into
1004 the main tree for this release, but which are shipped in
1005 contrib/experimental-thread.patch as a possible fix for some
1006 failures (deadlock, spinning...) in GC-intensive multithreaded
1008 * fixed PPC build problem (source code incompatibility of different
1009 library versions): added offsetof() hackery which attempts to divine
1010 where glibc maintainers put uc_mcontext today
1011 * fixed bug 282: compiler does not trust type assertions while passing
1012 arguments to a full call.
1013 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1014 &OPTIONAL VALUES type parameter.
1015 * fix bug 214: algorithm for noting rejected templates is now more
1016 similar to that of template seletion. (also reported by rydis on
1018 * fixed bug 141b: printing backquoted information readably and prettily
1019 inserts a space where necessary.
1020 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1021 (CHARACTER 10) are now reported as errors, rather than propagated
1022 as unknown types. (reported by piso on #lisp)
1023 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1024 with values NIL and :ERROR. (thanks to Milan Zamazal)
1025 * fixed bug 191c: CLOS now does proper keyword argument checking as
1026 described in CLHS 7.6.5 and 7.6.5.1.
1027 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1028 longer raise an error; further, using a list as a for-as-arithmetic
1029 counter now raises a meaningful error.
1030 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1031 now understood by sequence creation functions such as MAKE-SEQUENCE
1033 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1034 a STRING-INPUT-STREAM.
1035 * compiler enhancement: SIGNUM is now better able to derive the type
1037 * type declarations inside WITH-SLOTS are checked. (reported by
1039 * fixed some bugs revealed by Paul Dietz' test suite:
1040 ** incorrect optimization of TRUNCATE for a positive first
1041 argument and negative second.
1042 ** compiler failure in let-convertion during flushing dead code.
1043 ** compiler failure while deriving type of TRUNCATE on an
1044 interval, containing 0.
1045 ** ASH of a negative bignum by a negative bignum count now returns
1047 ** intersection of CONS types now canonicalizes properly, fixing
1048 inconsistencies in SUBTYPEP.
1050 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1051 * incompatible change: The --disable-debugger command line
1052 option now clobbers the debugger at a more fundamental
1053 level, by redefining #'INVOKE-DEBUGGER instead of by
1054 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1055 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1056 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1057 enter the debugger REPL and then suffer recursive errors
1058 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1059 while under the new system, BREAK in --disable-debugger mode
1060 terminates the system just as an unhandled error would.
1061 * fixed compiler performance when processing loops with a step >1;
1062 * bug fix: DOCUMENTATION now retrieves generic function
1063 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1064 support has been systematized, and now supports the methods
1065 specified by ANSI, along with a default method and a method for
1066 slot documentation. (reported by Nathan Froyd)
1067 * bug fix: effective methods associated with a generic function are
1068 no longer cached over a change of that generic function's method
1069 combination. (reported by Andreas Fuchs)
1070 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1071 names a non-existent directory. (thanks to Andreas Fuchs)
1072 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1073 the PPC platform, lead to stack corruption; this has been fixed.
1074 (reported by Rainer Joswig)
1075 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1076 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1077 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1078 arithmetic in general, is now much more reliable on the PPC
1080 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1081 platform now returns the right answer.
1082 * optimization: restored some effective method precomputation in
1083 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1084 precomputation is now tunable.
1085 * optimization: compiler-internal data structure use has been
1086 reviewed, and changes have been made that should improve the
1087 performance of the compiler by about 20%.
1088 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1089 simple-base-strings and simple-bit-vectors is improved.
1090 * optimization: the optimization of 32-bit logical and arithmetic
1091 functions introduced in version 0.8.3 on the x86 has been
1092 implemented on the mips, ppc and sparc platforms; an
1093 implementation of the same facility, but for 64-bit arithmetic,
1094 has been added for the alpha.
1095 * microoptimization: the compiler is better able to make use of the
1096 x86 LEA instruction for multiplication by constants.
1097 * bug fix: in some situations compiler did not report usage of
1098 generic arithmetic in (SPEED 3) policy.
1099 * bug 145b fix: compiler used wrong type specifier while converting
1100 MEMBER-types to numeric.
1101 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1103 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1104 index argument now works properly on the Alpha platform.
1105 * bug fix: floating point exception treatment on the Alpha platform
1107 * bug fix: FILE-POSITION works much better on string input and
1108 output streams. (thanks to Nikodemus Siivola)
1109 * bug fix: many threading/garbage collection symptoms sorted.
1110 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1111 might be pseudo-atomic.
1112 * internal change: Stopping for GC is now done with signals not
1113 ptrace. GC is now done in whichever thread wanted it, instead of
1115 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1117 * bug fix: result form in DO is not contained in the implicit
1119 * incompatible change: ICR structure is changed; the value part of
1120 CONTINUATION is now called LVAR; corresponding functions are
1121 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1122 * added type deriver for ISQRT (thanks to Robert E. Brown).
1123 * bug fix: better support for loading from the command line when an
1124 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1126 * fixed some bugs revealed by Paul Dietz' test suite:
1127 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1128 ** ROUND and FROUND now give the right answer when given very
1129 small float arguments.
1130 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1132 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1133 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1134 ** LCM should return a non-negative integer.
1135 ** PARSE-INTEGER returned the index of a terminator instead of the
1136 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1137 ** PARSE-INTEGER returned an incorrect index being applied to a
1139 ** LCM with two arguments of 0 returns 0 rather than signalling
1141 ** unsigned addition of a 32-bit constant with the high bit set no
1142 longer causes an internal compiler error on the x86.
1143 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1144 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1145 with negative last argument.
1146 ** byte specifiers with zero size and position no longer cause
1147 an error during type derivation.
1148 ** bignum multiplication on the Alpha platform now returns the
1150 * porting: The system now builds on SuSE AMD64, although it still
1151 generates a 32-bit binary.
1152 * .fasl file incompatibility: The fasl file version number has
1153 been incremented (because of the changes to internal compiler
1154 data structures referred to above).
1156 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1157 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1158 more accurately, on the Darwin kernel running on PowerPC hardware.
1159 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1160 * Compiler code deletion notes now signal a condition of type
1161 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1162 an associated MUFFLE-WARNING restart.
1163 * The compiler now performs limited argument count validation of
1164 constant format strings in FORMAT, and where appropriate in ERROR,
1165 CERROR and WARN. (thanks to Gerd Moellmann)
1166 * New ASDF-INSTALL contrib can be used for automatic download and
1167 installation of third-party Lisp code from CCLAN or other sites
1169 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1170 which forces another thread to execute a function supplied by the
1172 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1173 now accept and act upon their :ELEMENT-TYPE keyword argument.
1174 (reported by Martin Atzmueller, Edi Weitz)
1175 * bug fix: FILE-POSITION now accepts position designators up to
1176 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1177 is the greater. (thanks to Patrik Nordebo)
1178 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1179 Kalle Olavi Niemitalo)
1180 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1181 printable. (reported by Eric Marsden)
1182 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1183 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1184 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1185 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1186 (thanks to Stig Sandoe)
1187 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1188 UNREAD-CHAR now correctly includes the unread character in the
1189 target sequence. (thanks to Gerd Moellmann)
1190 * bug fix in threaded builds: the system can now be suspended and
1191 resumed by shell job control with minimal disruption.
1192 * bug fixes in times and timezones >2038AD
1193 * better handling of "where is GNU make?" problem in build scripts
1194 (thanks to Nikodemus Siivola)
1195 * new optimization: inside a named function any reference to a
1196 function with the same name is considered to be a self-reference;
1197 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1198 optimization quality.
1199 * new optimization on x86: logical functions and + now have
1200 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1201 used when the result is truncated to 32 bits.
1202 * VALUES declaration is partially enabled.
1203 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1204 * fixed some bugs revealed by Paul Dietz' test suite:
1205 ** The system now obeys the constraint imposed by
1206 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1207 types form a lattice under type intersection.
1208 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1209 ** ASSOC now ignores NIL elements in an alist.
1210 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1211 and (1+ MOST-POSITIVE-FIXNUM) answers.
1212 ** The addition of a method with invalid qualifiers to a generic
1213 function does not cause an error to be signalled immediately;
1214 a warning is signalled, and the error is generated only on
1215 calling the generic function.
1216 * changed .fasl file version number, in part to add type codes for
1217 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1218 obscure ANSI requirements
1220 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1221 * fixed bug 148: failure to inline-expand a local function left
1222 garbage, confusing the compiler.
1223 * fixed bugs 3cd: structure slot readers perform type check if the
1224 slot can have an invalid value (i.e. it is either not initialized
1225 or can be written with a less specific slot writer).
1226 * bug fix: the compiler now traps array references to elements off
1227 the end of an array; previously, the bounds checking in some
1228 circumstances could go off-by-one.
1229 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1231 * type declarations for array element types now obey the description
1232 on the CLHS page "Declaration TYPE", as per discussions on
1233 sbcl-help around 2003-05-08. This means that a declaration
1234 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1235 declaration, all references to BAR will be asserted or assumed
1236 (with THE, so dependent on compiler policy) to involve objects of
1237 type FOO. Note that no such declaration is implied in
1238 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1239 * declared types of functions from the "Conditions"
1240 chapter. (reported by Paul Dietz)
1241 * bug fix: CERROR accepts a function as its first argument.
1242 * bug fix: NTH an NTHCDR accept a bignum as index
1243 arguments. (reported by Adam Warner)
1244 * optimization: character compare routines now optimize comparing
1245 against a constant character. (reported by Gilbert Baumann)
1246 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1247 argument now works properly on the MIPS platform.
1248 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1249 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1250 declared to have a functional type, does not cause an error
1252 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1253 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1254 * fixed bug: if last continuation of a deleted block has a
1255 destination, this destination should be deleted too. (reported by
1257 * fixed a bug in the bootstrap process: the host compiler's values
1258 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1259 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1260 test case from Patrik Nordebo)
1261 * improved the ability of the disassembler on the PPC platform to
1262 provide helpful disassembly notes.
1263 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1264 the class in more cases than previously.
1265 * bug fix: FILE-POSITION now understands :START and :END for
1266 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1267 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1268 8) data. (thanks to David Lichteblau)
1269 * bug fix: it is possible to add a method to a generic function
1270 without lambda list.
1271 * bug fix: reader failed to signal END-OF-FILE inside an
1272 object representation. (reported by Nikodemus Siivola)
1273 * fixed some bugs revealed by Paul Dietz' test suite:
1274 ** LAST and [N]BUTLAST should accept a bignum.
1275 ** condition slot accessors are methods.
1276 ** (VECTOR NIL) is a subtype of STRING.
1278 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1279 * minor incompatible change: some nonsensical specialized lambda
1280 lists (used in DEFMETHOD) which were previously ignored now signal
1282 * minor incompatible change: the system is now aware of the types of
1283 variables in the COMMON-LISP package, and will signal errors for
1284 most violations of these type constraints (where previously they
1285 were silently accepted).
1286 * minor incompatible change: COMPILE-FILE now uses the freedom
1287 afforded (ANSI 3.2.2.3) to use derived function types for
1288 functions defined in the same file. This also permits the system
1289 to warn on static type mismatches and function
1290 redefinition. (Currently it does not work with high DEBUG level.)
1291 * minor incompatible change: VALUES declaration is disabled.
1292 * When issuing notes, the compiler now signals a condition of type
1293 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1294 restart for use in user handlers. It is expected that the
1295 COMPILER-NOTE condition will eventually become a condition
1296 supertype to a hierarchy of note types, which will then be
1297 handleable in a similar fashion. However, at the moment, no such
1298 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1299 but it's an implementation detail, not a classification for the
1301 * Changes in type checking closed the following bugs:
1302 ** type checking of unused values (192b, 194d, 203);
1303 ** template selection based on unsafe type assertions (192c, 236);
1304 ** type checking in branches (194bc).
1305 * A short form of VALUES type specifier has ANSI meaning (it has
1306 increased the number of situations when SBCL cannot perform type
1308 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1309 keywords or constants is permissible.
1310 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1311 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1312 classes. (thanks to Antonio Martinez)
1313 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1314 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1315 argument to a pretty-print stream. Instead, it leaves any such
1316 operations to DESCRIBE-OBJECT methods.
1317 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1319 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1320 lambda lists. (thanks to David Lichteblau)
1321 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1322 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1323 (thanks to Gerd Moellmann)
1324 * fixed bug 246: increased compilation speed of long
1325 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1327 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1329 * DEFCONSTANT now throws a condition of type
1330 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1331 constant to a non-EQL value; CONTINUE and ABORT restarts
1332 respectively change and preserve the value.
1333 * fixed bug 63: The code walker, part of the implementation of CLOS,
1334 is now better at handling symbol macros.
1335 * bug fix: There is no longer an internal implementation type named
1336 CL:LENGTH. (reported by Raymond Toy)
1337 * bug fix: In macro-like defining macros/special operators the
1338 implicit block does not enclose the lambda list.
1339 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1340 enclosed in parentheses) are not suitable as type specifiers, and
1341 their use properly signals an error now.
1342 * bug fix: An argument count mismatch for a type specifier in code
1343 being compiled no longer causes an unhandled error at compile
1344 time, but signals a compile-time warning.
1345 * fixed simple vector readable printing
1346 * bug fix: DESCRIBE takes more care over whether the class
1347 precedence list slot of a class is bound before accessing it.
1348 (reported by Markus Krummenacker)
1349 * bug fix: FORMATTER can successfully compile pretty-printer format
1350 strings which use variants of the ~* directive inside.
1351 * bug fix: SEARCH now applies its TEST predicate to the elements of
1352 the arguments in the correct order. (thanks to Wolfhard Buss)
1353 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1354 in type inference. (thanks to Robert E. Brown)
1355 * bug fix: Reading in symbols with an explicit package name of ""
1356 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1357 * fixed some bugs revealed by Paul Dietz' test suite:
1358 ** NIL is now allowed as a structure slot name.
1359 ** Arbitrary numbers, not just REALs, are allowed in certain
1360 circumstances in LOOP for-as-arithmetic clauses.
1361 ** Multiple class redefinitions before slot access no longer
1362 causes a type error.
1363 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1364 association between the name and a class.
1365 ** Generic functions with non-standard method-combination and over
1366 five methods all of which return constants no longer return NIL
1367 after the first few invocations. (thanks to Gerd Moellmann)
1368 ** CALL-NEXT-METHOD with no arguments now passes the original
1369 values of the arguments, even in the presence of assignment.
1370 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1372 ** DISASSEMBLE works with closures and funcallable instances.
1373 ** ADD-METHOD now returns the generic function, not the new method.
1374 ** FIND-METHOD signals an error if the lengths of the specializers
1375 is incompatible with the generic function, even if the ERRORP
1377 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1378 which its argument is a member.
1379 ** DEFCLASS only redefines the class named by its class-name
1380 argument if that name is the proper name of the class;
1381 otherwise, it creates a new class.
1382 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1383 of the UNBOUND-SLOT condition to the name of the slot.
1384 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1385 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1386 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1387 SLOT-MAKUNBOUND in the specified fashion.
1389 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1390 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1391 cross-compilation host. As a consequence, we can now bootstrap our
1392 way up to SBCL starting with a bare gcc toolchain and human-readable
1393 source code (first the source to CLISP, then the source to SBCL).
1394 * A contributed module containing a partial implementation of the
1395 simple-streams interface has been included. (thanks to Rudi
1397 * A contributed module implementing the RFC1321 Message Digest
1398 Algorithm, known as MD5, has been included.
1399 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1400 no longer has any effect, as the code controlled by this feature
1401 has been deleted. (As far as we know, no-one has ever built using
1402 this feature, and its semantics were confused in any case).
1403 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1404 work on conditions (as required by the ANSI specification),
1405 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1406 expected behaviour on conditions. Users should note, however,
1407 that such behaviour is not required by the ANSI specification,
1408 and so use of this behaviour may render their code unportable.
1409 * fixed some bugs revealed by Paul Dietz' test suite:
1410 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1412 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1413 ** SXHASH on condition objects no longer returns NIL.
1414 ** :ALLOCATION :CLASS slots are better treated; their values are
1415 updated on class redefinition, and initforms inherited from
1416 superclasses are applied.
1417 ** REMOVE-METHOD returns its generic function argument even when
1418 no method was removed.
1419 ** SHARED-INITIALIZE now initializes the values of the requested
1420 slots, including those with :ALLOCATION :CLASS.
1421 ** ALLOCATE-INSTANCE now works on structure classes defined via
1422 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1424 ** SLOT-EXISTS-P now works on conditions, as well as structures
1426 ** MAKE-LOAD-FORM now has the required methods on
1427 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1428 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1429 keyword, and now implements the SLOT-NAMES argument.
1430 ** methods with &OPTIONAL arguments no longer allow too many
1431 arguments to be passed in the call without error.
1432 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1433 option is consistent with the required arguments of the generic
1434 function lambda list.
1435 * bug fix: REQUIRE accepts a string designator. (Thanks to
1437 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1438 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1439 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1440 * bug fix: The compiler now checks for duplicated variables in macro
1442 * bug fix: SETQ on globals returns the correct value.
1443 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1444 not a COMPILER-ERROR (followed by some other strange error on
1445 choosing the CONTINUE restart).
1446 * bug fix: make.sh and friends are now more consistent in the way that
1447 they look for GNU "make".
1449 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1450 * experimental native threads support (on x86 Linux >=2.4 only).
1451 This is not compiled in by default: you need to add :SB-THREAD to
1452 the target features. See the "Beyond ANSI" chapter of the manual
1454 * fix for longstanding nonANSIism: The old distinction between
1455 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1456 The return value from CL:FIND-CLASS is now a CLOS class, and
1457 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1458 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1459 * An interface to the MetaObject Protocol, as described in Kiczales,
1460 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1461 MIT Press, 1991, is available from the SB-MOP package.
1462 * incompatible change: the SB-PCL package should now be considered
1463 a private implementation detail, and no longer a semi-private MOP
1465 * minor incompatible change: due to rearrangement for threads, the
1466 control stack and binding stack are now allocated at arbitrary
1467 addresses instead of being hardcoded per-port. Users affected by
1468 this probably have to be doing advanced things with shared
1469 libraries, and will know who they are.
1470 * minor incompatible change: Previously, all --eval forms used were
1471 processed with READ before any of them were processed with EVAL.
1472 Now each --eval form is processed with both READ and EVAL before
1473 the next --eval form is processed. (Thus package operations like
1474 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1475 work as the user might reasonably expect.)
1476 * minor incompatible change: *STANDARD-INPUT* is now only an
1477 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1479 * minor incompatible change: Y-OR-N-P is now character-oriented,
1480 not line oriented. Also, YES-OR-NO-P now works without errors.
1481 (thanks to Antonio Martinez)
1482 * sb-aclrepl module improvements: an integrated inspector, added
1483 repl features, and a bug fix to :trace command.
1484 * Known functions, which cannot be open coded by the backend, are
1485 considered to be able to check types of their arguments. (fixing
1486 a bug report by Nathan J. Froyd)
1487 * fixed a bug in computing method discriminating functions: It is
1488 now possible to define methods specialized on classes which have
1489 forward-referenced superclasses. (thanks to Gerd Moellmann)
1490 * fixed evaluation order in optional entries (reported by Gilbert
1492 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1493 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1494 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1495 required-by-AMOP NAME argument, as well as CLASS and
1496 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1497 * fixed bug 20: DEFMETHOD can define methods using names that are
1498 not the proper names of classes to designate class specializers.
1499 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1500 Unix file descriptors, instead of blowing up. (thanks to Antonio
1502 * Garbage collection refactoring: user-visible change is that a
1503 call to the GC function during WITHOUT-GCING will not do garbage
1504 collection until the end of the WITHOUT-GCING. If you were doing
1505 this you were probably losing anyway.
1506 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1507 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1508 TYPEP the latter but not the former.
1509 * The compiler issues a full WARNING for calls to undefined functions
1510 with names from the CL package.
1511 * MAP-INTO for a vector destination is open coded. (reported by
1512 Brian Downing on c.l.l)
1513 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1514 documentation string.
1515 * fixed some bugs revealed by Paul Dietz' test suite:
1516 ** COPY-ALIST now signals an error if its argument is a dotted
1518 ** Condition slots are now accessed more correctly in the presence
1519 of multiple initargs for a given slot.
1520 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1521 exclude restarts of the same name associated with a different
1523 ** DEFCLASS of forward-referenced classes with another
1524 forward-referenced class in the superclasses list no longer
1526 ** Condition slots are now initialized once each, not multiple
1527 times. (thanks to Gerd Moellmann)
1528 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1530 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1531 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1532 arguments contain duplicated elements.
1533 ** RESTART-CASE understands local macros.
1534 ** RESTART-CASE associates exactly its own restarts with a condition.
1535 ** ENDP in safe mode checks its argument to be of type LIST.
1536 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1537 in question is unbound.
1538 ** Optimized MAKE-INSTANCE functions no longer cause internal
1539 assertion failures in the presence of duplicate initargs.
1540 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1541 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1542 propagated even in the ctor optimized implementation of
1544 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1546 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1547 * a better implementation of SXHASH on (simple) bit vectors,
1548 measured both in execution speed and in distribution of results
1549 over the positive fixnums, has been installed. Likewise, a better
1550 implementation of EQUAL for simple bit vectors is now available.
1551 * fixed CEILING optimization for a divisor of form 2^k.
1552 * fixed bug 240 (emitting extra style warnings "using the lexical
1553 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1554 by Antonio Martinez)
1555 * fixed SXHASH, giving different results for NIL depending on type
1556 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1557 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1558 variable. (found by Rolf Wester)
1559 * SBCL does not ignore type declarations for special
1560 variables. (reported by rif on c.l.l 2003-03-05)
1561 * some bug fixes in contrib/sb-aclrepl/
1562 * fixed some bugs revealed by Paul Dietz' test suite:
1563 ** a bug in the CONS type specifier, whereby the CAR and CDR
1564 types got intertwined, has been fixed;
1565 ** the type system is now able to reason about the interaction
1566 between INTEGER and RATIO types more completely;
1567 ** APPEND, [N]REVERSE and NRECONC check that those their
1568 arguments, which must be proper lists, are really so;
1569 ** An array specialized to be unable to hold elements has been
1570 implemented, as required -- yes, really -- by ANSI;
1571 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1572 on malformed property lists;
1574 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1575 * incompatible packaging change: in line with Unix convention,
1576 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1577 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1578 * REQUIRE and PROVIDE are now optionally capable of doing something
1579 useful. See the documentation string for REQUIRE.
1580 * infrastructure for a managed SBCL contrib system: contributed
1581 modules in this release include:
1582 ** the ASDF system definition facility;
1583 ** an interface to the BSD Sockets API;
1584 ** an ACL-like convenience interface to the repl;
1585 (thanks to Kevin Rosenberg)
1586 ** an implementation of ROTATE-BYTE, with efficient implementation
1588 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1589 gives a closer approximation to the right answer than previously.
1590 (thanks to Raymond Toy)
1591 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1592 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1593 optional environment arguments, as required by ANSI.
1594 * fixed bugs in other functions taking environment objects, allowing
1595 calls with an explicit NIL environment argument to be compiled
1597 * fixed bug 228: primary return values from
1598 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1599 COMPILE or FUNCTION.
1600 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1601 :TYPE will no longer signal errors on innocuous objects.
1602 * fixed bug 231b: SETQ is better at respecting type declarations in
1603 the lexical environment.
1604 * fixed a bug in DEFCLASS: classes named by symbols with no or
1605 unprintable packages can now be defined.
1606 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1607 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1608 * fixed compiler failure related to checking types of functions.
1609 (reported by Robert E. Brown)
1610 * the compiler is now much more consistent in its error-checking
1611 treatment of bounding index arguments to sequence functions: in
1612 (SAFETY 3) code, errors will be signalled in almost all cases if
1613 invalid sequence bounding indices are passed to functions defined
1614 by ANSI to operate on sequences.
1615 * fixed a bug in the build procedure: documentation of SBCL-specific
1616 packages is now preserved and available in the final Lisp image.
1617 * lifted FDEFINITION lookup out of loops in the implementation of
1618 many list operations. (thanks to Robert E. Brown)
1619 * fixed a bug in the reader: the #n# reader macro now works for
1620 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1621 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1622 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1623 E. Brown and Tony Martinez respectively)
1624 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1625 SLOT-BOUNDP: in particular, it is now able to optimize them much
1626 better, and is now not vulnerable to having packages renamed.
1627 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1628 always called when a slot is not present in an instance. (thanks
1630 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1631 longer signal an error on unreading a character following EOF on
1632 the previous constituent stream. (thanks to Tony Martinez)
1633 * fixed some bugs revealed by Paul Dietz' test suite:
1634 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1635 not just nonnegative fixnums;
1636 ** the logical bit-array operators such as BIT-AND now accept an
1637 explicit NIL for their "opt-arg" argument (to indicate a
1638 freshly-consed result bit-array);
1639 ** ELT now signals an error on an invalid sequence index in safe
1641 ** the type system is now cleverer about negations of numeric
1642 types, and consequently understands the BIGNUM and RATIO types
1644 ** the type system is now cleverer about the interaction between
1645 INTEGER and RATIO types: while bugs still remain, many more
1646 cases are accurately computed;
1647 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1648 if it is in the last clause;
1649 ** CONSTANTLY now correctly returns a side-effect-free function in
1651 ** DECLARE is no longer treated as a special-operator; in
1652 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1653 * incremented fasl file version number due to the change in the
1654 DEFSTRUCT-SLOT-DESCRIPTION structure.
1656 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1657 * minor incompatible change: code processed by the "interpreter" or
1658 EVAL now has a compilation optimization policy of (DEBUG 2)
1659 (changed from (DEBUG 1)) to improve debuggability of interactive
1660 development, and to allow the use of the debug RETURN command in
1662 * an experimental implementation of the RETURN command for the
1663 debugger has been included. (thanks to Frederik Kuivinen)
1664 * fixed bug 62: constraints were not propagated into a loop.
1665 * fixed bug in embedded calls of SORT (reported and investigated by
1667 * fixed some bugs revealed by Paul Dietz' test suite:
1668 ** printing and reading of arrays with some dimensions having
1669 length 0 (thanks to Gerd Moellmann);
1670 ** BOA constructor with &AUX argument without a default value does
1671 not cause a type error;
1672 ** CONSTANTP now returns true for all self-evaluating objects.
1674 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1675 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1676 accessors that are related by inheritance, as specified in the
1677 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1678 Valtteri Vuorikoski)
1679 * The compiler is now able to inline functions that were defined in
1680 a complex lexical environment (e.g. inside a MACROLET).
1681 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1683 * Support for the upcoming FreeBSD-5.0 release has been included.
1684 (thanks to Dag-Erling Smorgrav)
1685 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1686 effect when it is not in a toplevel context.
1687 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1688 stand a better chance of being correct. (thanks to Gerd
1690 * fixed bug in COERCE, which now signals an error on coercing a
1691 rational to a bounded real type which excludes the expected
1693 * The compiler is now able to derive types more accurately from the
1694 COERCE and COMPILE functions.
1695 * fixed bug 223: functional binding is considered to be constant
1696 only for symbols in the CL package.
1697 * fixed bug 231: SETQ did not check the type of a variable being set
1698 (reported by Robert E. Brown)
1699 * A new optimization for MAKE-INSTANCE has been included, fixing
1700 various bugs (including relating to :ALLOCATION :CLASS slots and
1701 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1703 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1704 ** As required by ANSI, LOOP now disallows anonymous collection
1705 clauses such as COLLECT I in conjunction with aggregate boolean
1706 clauses such as THEREIS (= I 1);
1707 ** LOOP now signals an error when any variable is reused in the
1708 same loop (including the potentially useful construct analogous
1709 to WITH A = 1 WITH A = (1+ A);
1710 ** IT is only a special loop symbol within the first clause of a
1711 conditional loop clause;
1712 ** LOOP with a typed iteration variable over a hashtable now
1713 signals a type error iff it should.
1714 * fixed some other bugs revealed by Paul Dietz' test suite:
1715 ** FILE-STREAM now names the class previously known as FD-STREAM;
1716 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1717 argument) no longer signals an error;
1718 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1719 of the slot symbol, rather than using the current package
1720 ((:CONC-NAME "") continues to intern the slot's name in the
1722 * incremented fasl file version number, because of the incompatible
1723 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1724 of the new implementation of DEFINE-COMPILER-MACRO.
1726 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1727 * Support for building SBCL for MIPS platforms running in
1728 little-endian mode has now been checked in, and basic
1729 functionality on said platforms verified.
1730 * minor incompatible change: PCL now records the pathname of a file
1731 in which methods and the like are defined, rather than its
1733 * minor incompatible change: TRUENAME now considers the truename of
1734 a file naming a directory to be the pathname with :DIRECTORY
1735 component indicating that directory.
1736 * minor incompatible change: a NAMED clause in the extended form of
1737 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1738 reason for the previous behaviour is unclear.
1739 * more systematization and improvement of CLOS and MOP conformance
1740 in PCL (thanks to Gerd Moellman and Pierre Mai):
1741 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1743 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1744 primary methods with no specializers;
1745 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1747 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1748 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1749 FORWARD-REFERENCED-CLASSes; error reporting on
1750 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1752 ** SXHASH on CLOS instances now uses a slot internal to the
1753 instance to return different numbers on distinct instances,
1754 while preserving the same return value through invocations of
1756 ** DEFMETHOD signals errors when methods with longer incongruent
1757 lambda lists are added to generic functions;
1758 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1759 CLASS, as specified in AMOP;
1760 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1761 on the order returned by the primary method for classes of
1762 class STANDARD-CLASS;
1763 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1764 * fixed some bugs shown by Paul Dietz' test suite:
1765 ** DOLIST puts its body in TAGBODY;
1766 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1768 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1769 value producing form;
1770 ** if more variables are given to PROGV than values, extra
1771 variables are bound and made to have no value;
1772 ** NSUBSTITUTE on list arguments gets the right answer with
1774 ** ELT signals an error of type TYPE-ERROR when the index argument
1775 is not a valid sequence index;
1776 ** LOOP signals (at macroexpansion time) an error of type
1777 PROGRAM-ERROR when duplicate variable names are found;
1778 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1779 ** FUNCALL of special-operators now cause an error of type
1781 ** PSETQ now works as required in the presence of side-effecting
1782 symbol-macro places;
1783 ** NCONC accepts any object as its last argument;
1784 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1786 ** loop-for-as-package does not require a package to be explicitely
1788 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1790 * fixed bug 166: compiler preserves "there is a way to go"
1791 invariant when deleting code.
1792 * fixed bug 172: macro lambda lists with required arguments after
1793 &REST arguments now cause an error to be signalled. (thanks to
1795 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1796 bug. (thanks to Matthew Danish)
1797 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1799 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1800 arguments are lexically rebound. (thanks to Gerd Moellmann and
1802 * fixed bug 194: error messages are now more informative when there
1803 is no primary method applicable in a call to a generic
1804 function. (thanks to Gerd Moellmann)
1805 * fixed bug in command line argument checking (thanks to Julian
1807 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1808 (thanks to Matthew Danish)
1809 * incremented fasl file version number, because of the
1810 SXHASH-related changes in the layout of CLOS data structures
1812 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1813 * minor incompatible change: The runtime (the Unix executable named
1814 "sbcl") is now much pickier about the .core files it will load.
1815 Essentially it now requires .core files to descend from the same
1816 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1817 as the runtime does. (The intent is to prevent the crashes which
1818 can occur, and which can even be reported as mysterious failures,
1819 when people patch the sources or change the build parameters
1820 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1821 sbcl and .core files.)
1822 * fixed bug: VALUES-LIST is no longer optimized away.
1823 * fixed bug 142: The FFI conversion of C string values to Lisp
1824 string values no longer conses excessively. (thanks to Nathan
1825 Froyd porting Raymond Toy's fix to CMU CL)
1826 * began to systematize and improve MOP conformance in PCL (thanks to
1827 Nathan Froyd, Gerd Moellman and Pierre Mai):
1828 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1830 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1831 ** FINALIZE-INHERITANCE is now called on class finalization;
1832 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1833 argument precedence order.
1834 * fixed bug 202: The compiler no longer fails on functions whose
1835 derived types contradict their declared type.
1836 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1837 so it can be non-toplevel.
1838 * The fasl file version number has changed (because of the new
1839 implementation of DEFMACRO).
1840 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1841 safe code, that any length requirement by their type-specifier
1842 argument is valid. The exceptions to this are described in bug
1844 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1845 functions now check the types of their inputs as required by ANSI.
1846 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1847 introduced symbol is DECLAREd to be SPECIAL.
1848 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1849 * fixed bug: :COUNT argument to sequence functions may be negative
1850 * fixed bug: body of DO-SYMBOLS may contain declarations
1851 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1852 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1853 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1854 symbol macro only once
1855 * fixed printing of call frame when argument list is unavailable
1856 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1857 * compiler no longer signals WARNING on unknown keyword
1860 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1861 * A beta-quality port to the mips architecture running Linux,
1862 based on the old CMUCL backend, has been made. It has been tested
1863 on a big-endian kernel, and works sufficiently well to be able to
1864 rebuild itself; it has not been tested in little-endian mode.
1865 * fixed an inconsistency between gencgc.c and purify.c which made
1866 dumping/loading .core files unreliable
1867 * fixed bug 120a: The compiler now deals correctly with IFs where
1868 the consequent is the same as the alternative, instead of
1869 misderiving the return type. (thanks to Alexey Dejneka)
1870 * fixed bug 113: Logical pathnames are now dumpable (the logical
1871 host is resolved at load-time, throwing an error if it is not
1873 * fixed bug 174: FORMAT's error message is slightly clearer when a
1874 non-printing character is used in a format directive.
1875 * fixed several bugs in compiler checking of type declarations, i.e.
1876 violations of the Python "declarations are assertions" principle
1877 (thanks to Alexey Dejneka)
1878 * fixed several bugs in PCL's error checking (thanks to Gerd
1880 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1882 * fixed bug in compilation of functions as first class values
1883 (thanks to Antonio Martinez-Shotton)
1884 * The compiler's handling TYPE-ERRORs which it can prove will
1885 inevitably happen at runtime has been cleaned up and corrected
1886 in several ways. (thanks to Alexey Dejneka)
1887 * improved argument type checking for various basic arithmetic
1888 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1889 lavished on them in the past that they can be compiled in many
1890 ways in different special cases
1891 * fixed bug 181: compiler checks validity of user supplied type
1893 * cleaned up code flushing in optimization: Function calls which
1894 should signal errors for safety purposes (e.g. which ANSI says
1895 should signal errors when their arguments are of incorrect type)
1896 are no longer optimized away.
1897 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1898 * incremented fasl file version number, because changes in the
1899 implementation of sequence functions like COERCE caused
1900 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1901 in old inline expansions) to become undefined. (Actually these
1902 changes were later undone, so we might very well be binary
1903 compatible with 0.7.7 after all, but leaving the version number
1904 incremented seemed like the simplest and most conservative
1907 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1908 * An alpha-quality port to the parisc architecture running Linux,
1909 based on the old CMUCL backend, has been made. This, even more so
1910 than the other backends, should be considered still a work in
1911 progress; known problems include that the Linux kernel in 64-bit
1912 mode does not propagate the correct sigcontext structure to
1913 userspace, and consequently SBCL on a parisc64 kernel will not
1915 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1916 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1917 suppressed.) Also now that the compiler is looking at declarations
1918 in the environment, it checks optimization declarations as well,
1919 and suppresses inlining when (> DEBUG SPEED).
1920 * More fixes have been made to treatment of floating point exception
1921 treatment and other Unix signals. In particular, floating point
1922 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1923 * The detection and handling of control stack exhaustion (infinite
1924 or very deeply nested recursion) has changed. Stack exhaustion
1925 detection is now done by write-protecting pages at the OS level
1926 and applies at all optimization settings; when found, a
1927 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1928 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1929 be caught using IGNORE-ERRORS.
1930 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1931 correctly when transforming recursive LABELS functions to LETs.
1932 (thanks to Alexey Dejneka)
1933 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1934 that are names of constants or global variables.
1935 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1936 alien routines with docstrings.
1937 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1938 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1940 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1941 (thanks to Eric Marsden)
1942 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1943 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1944 * Bug fix: The SPARC backend can now compile functions involving
1945 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1946 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1947 to a functions expecting a PATHNAME-DESIGNATOR.
1948 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1949 lambda lists. (thanks to Alexey Dejneka)
1950 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1951 functions so that EXPORTing the name of the class doesn't cause
1952 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1953 to Antonio Martinez for reporting this)
1954 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1955 constant-folded without failing an assertion. (thanks to Einar
1956 Floystad Dorum for reporting this)
1957 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1958 types are now dealt with more correctly by the compiler.
1959 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1960 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1961 behaviour analogous to RENAME-FILE. This puts its behaviour more
1962 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1964 * The fasl file version number has changed again. (because of the
1965 bug fix involving the names of PCL MAKE-INSTANCE functions)
1967 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1968 * bug fix: Floating point exceptions are treated much more
1969 consistently on the x86/Linux and PPC/Linux platforms.
1970 * Array initialization with :INITIAL-ELEMENT is now much faster for
1971 cases when the compiler cannot open code the array creation, but
1972 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1973 array accesses have also seen a speed increase.
1974 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1975 specified. (This was at the root of some bad interactions between
1976 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1978 * bug fix: Internal error arguments for undefined functions are now
1979 computed correctly on the PPC/Linux platform.
1980 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1981 Raymond Toy's patch for CMU CL)
1982 * Support for the Solaris 9 operating environment has been included
1983 (thanks to Daniel Merritt)
1984 * A very ugly but hopefully complete draft of the missing FFI chapter
1985 of the manual has been created by reformatting the corresponding
1986 CMU CL manual chapter into (currently very ugly and incoherent)
1987 DocBook and bringing it up to date for SBCL behavior. Thus, the
1988 manual is now essentially complete, at least by my extreme
1989 once-and-only-once standards, whereby it's acceptable to refer to
1990 the doc strings of SB-EXT functions as the primary documentation.
1991 * The fasl file version number has changed again, due to cleanup of
1992 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1993 is no longer a static symbol.)
1995 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1996 * SBCL now builds with OpenMCL (version 0.12) as the
1997 cross-compilation host; also, more progress has been made toward
1998 bootstrapping under CLISP.
1999 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2001 * bug 158 fixed: The compiler can now deal with integer loop
2002 increments different from 1; fixing this turned out also to fix
2004 * bug 169 fixed: no more bogus warnings about using lexical bindings
2005 despite the presence of perfectly good SPECIAL declarations (thanks
2006 to David Lichteblau)
2007 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2008 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2009 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2011 * bug 180 fixed: Method combination specifications no longer ignore
2012 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2013 * bug fix: Structure type predicate functions now check their argument
2014 count as they should.
2015 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2016 correctly. (thanks to Pierre Mai)
2017 * minor incompatible change: The --noprogrammer option is deprecated
2018 in favor of the new --disable-debugger option, which is very similar.
2019 (The major difference is that it takes effect at a slightly different
2020 time at startup, causing handling of errors in --sysinit and
2021 --userinit files will be affected differently.) The
2022 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2023 been added to allow this functionality to be controlled from ordinary
2024 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2025 maintainers, who might want to run non-interactive scripts to
2026 build SBCL cores which will later be used interactively.)
2027 * minor incompatible change: The LOAD function no longer, when given
2028 a wild pathname to load, loads all files matching that pathname.
2029 Instead, an error of type FILE-ERROR is signalled.
2031 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2032 * bug 147 fixed: The compiler preserves its block link/count
2033 invariants more correctly now so that it doesn't crash. (thanks
2035 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2037 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2038 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2039 no-value-for-this-character as NIL (as specified by ANSI).
2040 * HOST-NAMESTRING on physical pathnames now returns a string that is
2041 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2042 (thanks to Christophe Rhodes)
2043 * The Alpha port handles icache flushing more correctly. (thanks to
2045 * More progress has been made toward bootstrapping under CLISP. (thanks
2046 to Christophe Rhodes)
2047 * The fasl file format has changed again, because dynamic loading
2048 on OpenBSD (which has non-ELF object files) motivated some cleanups
2049 in the way that foreign symbols are transformed and passed around.
2050 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2051 is no longer treated as whitespace by the reader, but instead as
2052 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2053 |A
\7fB|, instead of A as it used to.
2055 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2056 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2057 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2058 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2059 this as of 0.7.1.45, but was left out of the previous news section
2060 (thanks to Dan Barlow)
2061 * SBCL now runs on the Solaris operating system on SPARC architectures
2062 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2063 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2064 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2065 * SPARC backend cleanups, allowing builds of cores optimized for V8
2066 and V9 SPARCS, and also emission of code targeted to a particular
2067 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2069 * SBCL is closer to bootstrapping under CLISP, thanks to various
2070 fixes by Christophe Rhodes.
2071 * The fasl file format has changed again, to allow the compiler's
2072 INFO database to support symbol macros.
2073 * The user manual (in doc/) is formatted into HTML more nicely.
2074 (thanks to coreythomas)
2075 * The system is smarter about SUBTYPEP relationships, especially
2076 those involving NOT types (including types such as ATOM which are
2077 represented internally using NOT types). Thus SUBTYPEP is less
2078 likely to return (VALUES NIL NIL) in general, and in particular
2079 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2081 * The fasl file format has changed again, because the internal
2082 representation of types now includes a new slot to support the new
2083 SUBTYPEP-of-NOT-types logic.
2084 * (not a change in the main branch of SBCL, but a related prototype
2085 which can hopefully be merged into the main branch of SBCL in the
2086 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2087 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2088 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2089 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2090 dispatch macro character combinations. (thanks to Alexey Dejneka)
2091 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2092 that superficially look like logical namestrings correctly.
2093 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2095 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2096 default supplied-p) for &optional and &key arguments. (thanks to
2099 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2100 * incompatible change: The compiler is now less aggressive about
2101 tail call optimization, doing it only when (> SPACE DEBUG) or
2102 (> SPEED DEBUG). (This is an incompatible change because there are
2103 programs which relied on the old CMU-CL-style behavior to optimize
2104 away their unbounded recursion which will now die of stack overflow.)
2105 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2106 for non-GENCGC systems has been increased to 20M (since that
2107 seems much closer to the likely performance optimum for modern
2108 systems than the old 4M value was)
2109 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2110 and *DEBUG-PRINT-LEVEL*
2111 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2112 of CMU CL's support for SPARC, and various endianness and other
2113 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2114 * new syntactic sugar for the Unix command line: --load foo.bar is now
2115 an alternate notation for --eval '(load "foo.bar")'.
2117 ** The system now detects stack overflow and handles it gracefully,
2118 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2119 optimization settings. (This is a good thing in general, and
2120 its introduction in this version should be particularly timely
2121 for anyone whose code fails because of suppression of tail
2123 ** The system now hunts for the C variable "environ" in a more
2124 devious way, to avoid segfaults when the C library version differs
2125 between compile time and run time. (thanks to Christophe Rhodes)
2126 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2127 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2128 ** The compiler no longer issues bogus style warnings for undefined
2129 classes in the same source file as the DEFCLASSes which defined
2130 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2132 ** fixes in CONDITION class precedence list for undefined function
2133 errors (thanks to Alexei Dejneka)
2134 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2135 correctly. (thanks to Dan Barlow)
2136 ** portability fixes aiming at bootstrapping under CLISP (thanks
2137 to Dave McDonald and Christophe Rhodes)
2138 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2139 ** fixes in type translation and and type inference (thanks to
2141 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2142 ** various fixes in the new ports (thanks to Dan Barlow)
2143 * several changes related to debugging:
2144 ** suppression of tail recursion, as noted above
2145 ** stack overflow detection, as noted above
2146 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2147 is now the default. (For some time encapsulation has been more
2148 reliable than the breakpoint-based :ENCAPSULATE NIL
2149 implementation, at least on X86 systems; and I just noticed that
2150 encapsulation also seems closer to the spirit of the ANSI
2153 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2155 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2156 up properly again. (There was a packaging bug in 0.7.0 which
2157 left their definitions in SB-SYS::LOAD-FOREIGN and
2158 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2159 vital for most things which interface to C-level interfaces,
2160 like extensions working with sockets or databases or
2161 Perl-compatible regexes or whatever, and the need to fix
2162 this bug is the main reason that 0.7.1 was released so
2164 ** DEFGENERIC is now choosier about the methods it redefines, so that
2165 reLOADing a previously-LOADed file containing DEFGENERICs does
2166 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2167 cycle now works as it should. (thanks to Alexey Dejneka)
2168 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2169 fixed by Christophe Rhodes. (He actually submitted this patch
2170 months ago, and I delayed until after 0.7.0.)
2171 ** Bug 111 (internal compiler confusion about runtime checks on
2172 FUNCTION types) was fixed by Alexey Dejneka.
2173 * Some internal cleanups (getting rid of variables which aren't
2174 needed now that the byte interpreter is gone) caused the fasl
2175 file format number to change again.
2177 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2178 * major incompatible change: The default fasl file extension, i.e. the
2179 default extension for files produced by COMPILE-FILE, has changed
2180 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2182 ** There are many changes in the implementation of the compiler.
2183 SBCL is now essentially a compiler-only implementation of ANSI
2184 Common Lisp. EVAL still "interprets" a few special cases, but
2185 almost all the interesting cases are handled by creating
2186 a LAMBDA expression, calling COMPILE on it, then calling
2187 FUNCALL on the result.
2188 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2189 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2190 Since the code is newer, there might still be some new bugs
2191 (though not as many as before Martin Atzmueller's fixes:-). But
2192 the new code is substantially simpler and clearer, and hopefully
2193 any remaining bugs will be simpler, less fundamental, and more
2194 fixable then the bugs in the old code.
2195 ** The revised compiler is still a little unsteady on its feet.
2197 *** The debugging information it produces (particularly the names
2198 of FUNCTION objects) is sometimes much less useful than what
2199 the old compiler produced.
2200 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2201 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2202 within a MACROLET) has been temporarily weakened.
2203 ** There are new compiler optimizations for various functions:
2204 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2205 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2206 *** the math functions TRUNCATE, FLOOR, and CEILING
2207 *** the function-of-all-trades COERCE
2208 Mostly these should be transparent, but there's one
2209 potentially-annoying problem (bug 117): when the compiler
2210 inline-expands a function and does type analysis on the result,
2211 it can create control paths which have type mismatches, and
2212 when it can't prove that those control paths aren't taken,
2213 it will issue WARNINGs about the type mismatches. This is
2214 a particular problem in practice for the new sequence functions.
2215 It's not clear how this should be fixed, and for now, a
2216 workaround is given in the entry for 117 in the BUGS file.
2217 ** (Because of the interaction between the two previous items --
2218 occasional inlining problems and new inline expansions -- some
2219 of the new sequence function optimizations won't really kick in
2220 completely until debugging information, and then inlining, are
2221 straightened out in some future version.)
2222 * minor incompatible changes:
2223 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2224 DIRECTORY no longer implicitly promotes NIL slots of its
2225 pathname argument to :WILD. In particular, when you ask for the
2226 contents of a directory (which you used to be able to do without
2227 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2228 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2229 ** changes in behavior that ANSI explicitly defines to be
2230 implementation dependent:
2231 *** The new compiler-only implementation still conforms with ANSI,
2232 but acts a little different than before. Besides the obvious
2233 changes in performance tradeoffs (that the cost per form passed
2234 to EVAL has gone up, and the cost per form executed by EVAL
2235 has gone down), the behavior of the system changes a little
2236 because there are no longer any interpreted function objects.
2237 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2238 e.g. doing COMPILE on the output of interactive DEFUN is
2240 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2242 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2243 has changed from (:CL) to NIL.
2244 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2245 e.g. #\Newline and #\Tab, don't have names specified in the
2246 ANSI Common Lisp standard, is now based on their ASCII symbolic
2247 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2248 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2249 are no longer used for output.
2250 ** changes in internal implementation constants:
2251 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2252 4 million. (If your application spends a lot of time GCing and
2253 you have a lot of RAM, you might want to experiment with
2254 increasing it even more.)
2255 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2256 However, almost all old code should still continue to work without
2257 immediate update, as SB-C-CALL is now a (deprecated) nickname
2259 ** Old operator names in the style DEF-FOO are now deprecated in
2260 favor of new corresponding names DEFINE-FOO, for consistency with
2261 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2262 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2263 internal symbols, but a few supported extensions like
2264 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2265 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2266 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2267 etc. as you get deeper into recursive calls to the debugger
2268 command loop, instead of the old "5]", "5]]", "5]]]"
2269 sequence. (I was motivated to do this when squabbles between
2270 ILISP and SBCL left me very deeply nested in the debugger. In the
2271 short term, this change will probably provoke more ILISP/SBCL
2272 squabbles, but hopefully it will be an improvement in the long run.)
2273 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2274 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2275 function, since declaiming return types involving aliens is
2276 (1) annoyingly messy to do by hand and (2) vital to efficient
2277 compilation of code which calls such functions.
2278 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2279 longer reexported by the SB-EXT package. They're solely useful
2280 for alien code, so it seems more logical that you should get
2281 them from the SB-ALIEN package, not in SB-EXT.
2282 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2283 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2284 features. Instead, the code that they used to control is always
2285 built into the system.
2286 * many other bug fixes
2287 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2288 advantage of the new EVAL-WHEN stuff and to clean them up in
2289 general, and they are now more ANSI-compliant in a number of
2290 ways. Martin Atzmueller is responsible for a lot of this.
2291 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2293 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2294 *** correct ERROR type for various file operations
2295 *** some fixes for Lisp streams
2296 *** DEFMETHOD syntax checking
2297 *** changing old weird representation of debug information as
2298 strings (which, among their other deficiencies, don't transform
2299 correctly when you rename packages, and don't change their
2300 print representation when you change things like *PACKAGE*
2301 and *PRINT-LENGTH*) to symbols and lists of symbols
2302 He also made several improvements and fixed several bugs in DESCRIBE.
2303 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2305 *** misbehavior of WRITE-STRING/WRITE-LINE
2306 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2307 and several other LOOP problems as well
2308 *** DIRECTORY when similar filenames are present
2309 *** DEFGENERIC with :METHOD options
2310 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2311 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2312 *** argument ordering in FIND with :TEST option
2313 *** mishandled package designator argument in APROPOS-LIST
2314 *** various problems in the backquote readmacro
2315 *** a bug in APROPOS
2316 *** probably some others that I'm not describing very well here,
2317 since the CVS log documents them by reference to sbcl-devel
2318 messages, and the SourceForge archives aren't working well.:-(
2319 ** Dan Barlow improved the Alpha port (and is making progress on the
2320 PPC port, for those of you who think different).
2321 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2322 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2323 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2324 continues to work on the SPARC port (for those of you in a position
2325 to look down upon our little PC-compatible boxes from a great height).
2326 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2327 and uses that copy, rather than the current dynamic value, when
2328 it's trying to decide whether to truncate output. Thus e.g.
2329 (let ((*print-lines* 50))
2330 (pprint-logical-block (stream nil)
2332 (let ((*print-lines* 8))
2333 (print (aref possiblybigthings i) stream)))))
2334 should now truncate the logical block only at 50 lines, instead of
2335 often truncating it at 8 lines, as it did before.
2336 * The doc/cmucl/ directory, containing old CMU CL documentation
2337 from the time of the fork, is no longer part of the base system.
2338 SourceForge has shut down its anonymous FTP service, and with it
2339 my original plan for distributing the old CMU CL documentation
2340 there. For now, if you need these files you can download an old
2341 SBCL source release and extract them from it.
2342 * The fasl file version number changed again, for dozens of reasons,
2343 some of which are apparent above.
2345 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2346 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2347 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2348 patch, so that DISASSEMBLE output is much nicer.
2349 * The code in the SB-PROFILE package now seems reasonably stable.
2350 I still haven't decided what the final interface should look like
2351 (I'd like PROFILE to interact cleanly with TRACE, since both
2352 facilities use function encapsulation) but if you have a need
2353 for profiling now, you can probably use it successfully with
2354 the current CMU-CL-style interface.
2355 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2356 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2357 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2358 link now returns the dangling link itself, and for similar
2359 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2360 link itself. (In these cases the old code signalled an error and
2361 looped endlessly, respectively.) Thus, DIRECTORY now works even
2362 in the presence of dangling and cyclic symbolic links.
2363 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2364 is now a supported extension again, since the consensus on
2365 sbcl-devel was that it can be useful for ordinary development
2366 work, not just for debugging SBCL itself.
2367 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2368 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2369 that currently-defined functions might be redefined later with
2370 different return types.
2371 * Hash tables can be printed readably, as inspired by CMU CL code
2372 of Eric Marsden and SBCL code of Martin Atzmueller.
2373 * better error handling in CLOS method combination, thanks to
2374 Martin Atzmueller porting Pierre Mai's CMU CL patches
2375 * more overflow fixes for >16Mbyte I/O buffers
2376 * A bug in READ has been fixed, so that now a single Ctrl-D
2377 character suffices to cause end-of-file on character streams.
2378 In particular, now you only need one Ctrl-D at the command
2379 line (not two) to exit SBCL.
2380 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2382 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2383 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2384 does the right thing.
2385 * The implementation of some type tests, especially for CONDITION
2386 types, is now tidier and maybe faster, due to CMU CL code
2387 originally by Douglas Crosher, ported by Martin Atzmueller.
2388 * Some math functions have been fixed, and there are new
2389 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2390 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2391 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2392 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2393 likely to be built into sbcl-0.7.0.)
2394 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2395 CL, ported to SBCL by Martin Atzmueller)
2396 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2397 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2398 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2399 * There's a new slam.sh hack to shorten the edit/compile/debug
2400 cycle for low-level changes to SBCL itself, and a new
2401 :SB-AFTER-XC-CORE target feature to control the generation of
2402 the after-xc.core file needed by slam.sh.
2403 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2404 COMPILE-FILE is no longer supported, so that now every function
2405 gets an entry point, so that block compilation looks a little
2406 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2407 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2408 returns the number of bytes consed since the system started,
2409 rather than the number consed since the first time the function
2410 was called. (The new definition parallels ANSI functions like
2411 CL:GET-INTERNAL-RUN-TIME.)
2412 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2413 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2414 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2415 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2416 * Fasl file version numbers are now independent of the target CPU,
2417 since historically most system changes which required version
2418 number changes have affected all CPUs equally. Similarly,
2419 the byte fasl file version is now equal to the ordinary
2422 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2423 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2424 is no longer recognized. I apologize for this, because it was
2425 listed in SB-EXT as a supported extension, but I found that
2426 its existing behavior was poorly specified, as well as incorrectly
2427 specified, and it looked like too much of a mess to straighten it
2428 out. I have enough on my hands trying to get ANSI stuff to work..
2429 * many patches ported from CMU CL by Martin Atzmueller, with
2430 half a dozen bug fixes in pretty-printing and the debugger, and
2431 half a dozen others elsewhere
2432 * fixed bug 13: Floating point infinities are now supported again.
2433 They might still be a little bit flaky, but thanks to bug reports
2434 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2435 as flaky as they were.
2436 * The --noprogrammer command line option is now supported. (Its
2437 behavior is slightly different in detail from what the old man
2438 page claimed it would do, but it's still appropriate under the
2439 same circumstances that the man page talks about.)
2440 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2441 are now supported, and enabled by default. Thus, the compiler can
2442 handle many floating point and complex operations much less
2443 inefficiently. (Thus e.g. you can implement a complex FFT
2445 * The compiler now detects type mismatches between DECLAIM FTYPE
2446 and DEFUN better, and implements CHECK-TYPE more correctly, and
2447 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2448 thanks to patches from Martin Atzmueller.
2449 * various fixes to make the cross-compiler more portable to
2450 ANSI-conforming-but-different cross-compilation hosts (notably
2451 Lispworks for Windows, following bug reports from Arthur Lemmens)
2452 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2453 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2454 fixed thanks to Pierre Mai's CMU CL patches.
2455 * fixes to keep the system from overflowing internal counters when
2456 it tries to use i/o buffers larger than 16M bytes
2457 * fixed bug 45a: Various internal functions required to support
2458 complex special functions have been merged from CMU CL sources.
2459 (When I was first setting up SBCL, I misunderstood a compile-time
2460 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2461 * improved support for type intersection and union, fixing bug 12
2462 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2463 more obscure bugs as well
2464 * some steps toward byte-compiling non-performance-critical
2465 parts of the system, courtesy of patches from Martin Atzmueller
2466 * Christophe Rhodes has made some debian packages of sbcl at
2467 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2468 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2469 stable, but are nonetheless usable. When he's ready, I'd be happy
2470 to add them to the SourceForge "File Releases" section. (And if
2471 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2472 * new fasl file format version number (because of changes in
2473 internal representation of (OR ..) types to accommodate the new
2474 support for (AND ..) types, among other things)
2476 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2477 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2479 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2481 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2482 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2483 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2484 * DESCRIBE now works on structure objects again.
2485 * Most function call argument type mismatches are now handled as
2486 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2487 know whether the function will be redefined before the call is
2488 executed. (The compiler could flag local calls with full WARNINGs,
2489 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2490 it doesn't keep track of enough information to know whether calls
2491 are local in this sense.)
2492 * Compiler output is now more verbose, with messages truncated
2493 later than before. (There should be some supported way for users
2494 to override the default verbosity, but I haven't decided how to
2495 provide it yet, so this behavior is still controlled by the internal
2496 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2497 src/compiler/ir1util.lisp.)
2498 * Fasl file format version numbers have increased again, because
2499 support for the Gray streams extension changes the layout of the
2500 system's STREAM objects.
2501 * The Gray subclassable streams extension now works, thanks to a
2502 patch from Martin Atzmueller.
2503 * The full LOAD-FOREIGN extension (not just the primitive
2504 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2505 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2506 but like most other programs, it defaults to copying the Unix
2507 environment from the original process instead of starting the
2508 new process in an empty environment.
2509 * Extensions which manipulate the Unix environment now support
2510 an :ENVIRONMENT keyword option which doesn't smash case or
2511 do other bad things. The CMU-CL-style :ENV option is retained
2512 for porting convenience.
2513 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2514 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2516 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2518 * A patch from Martin Atzmueller seems to have solved the SIGINT
2519 problem, and as far as we know, signal-handling now works cleanly.
2520 (If you find any new bugs, please report them!)
2521 * The system no longer defaults Lisp source file names to types
2522 ".l", ".cl", or ".lsp", but only to ".lisp".
2523 * The compiler no longer uses special default file extensions for
2524 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2525 seems to expect a single default extension for all compiled code,
2526 and there's no compelling reason to try to stretch the standard
2527 to allow two different extensions.) Instead, byte-compiled files
2528 default to the same extension as native-compiled files.
2529 * Fasl file format version numbers have increased again, because
2530 a rearrangement of internal implementation packages made some
2531 dumped symbols in old fasl files unreadable in new cores.
2532 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2533 many fewer weird special cases.
2534 * Bug #17 (differing COMPILE-FILE behavior between logical and
2535 physical pathnames) has been fixed, and some related misbehavior too,
2536 thanks to a patch from Martin Atzmueller.
2537 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2538 by Tim Moore, ported to SBCL by Martin Atzmueller.
2539 * Martin Atzmueller fixed several filesystem-related problems,
2540 including bug #36, in part by porting CMU CL patches, which were
2541 written in part by Paul Werkowski.
2542 * More compiler warnings in src/runtime/ are gone, thanks to
2543 more patches from Martin Atzmueller.
2544 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2547 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2549 * DESCRIBE now works on CONDITION objects.
2550 * The debugger now handles errors which arise when trying to print
2551 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2553 * The build system now uses an additional file, customize-target-features.lisp,
2554 to allow local modifications to the target *FEATURES* list. (The point of
2555 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2556 debugging features enabled, without having to worry about propagating it
2557 into everyone's system when I do a "cvs update".) When no
2558 customize-target-features.lisp file exists, the target *FEATURES* list
2559 should be constructed the same way as before.
2560 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2561 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2562 warn about behavior which is undefined under the ANSI spec. Note: This
2563 is specified by ANSI, but it's not very popular with programmers.
2564 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2565 macro in the SBCL sources for an example of a workaround which you
2566 might use to make portable ANSI-standard code which does what you want.
2567 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2568 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2569 when it's used as a non-toplevel form. (This is required in order
2570 to implement the DEFCONSTANT-EQX macro.)
2571 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2572 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2573 funny things, due to bugs in the implementation of EVAL-WHEN
2574 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2575 non-toplevel DEFCONSTANTs are uncommon.)
2576 * The core file version number and fasl file version number have been
2577 incremented, because the old noncompliant DEFCONSTANT behavior involved
2578 calling functions which no longer exist, and because I also took the
2579 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2580 * fixed bug 1 (error handling before read-eval-print loop starts), and
2581 redid debugger restarts and related debugger commands somewhat while
2583 ** The QUIT debugger command is gone, since it did something
2584 rather different than the SB-EXT:QUIT command, and since it never
2585 worked properly outside the main toplevel read/eval/print loop.
2586 Invoking the new TOPLEVEL restart provides the same functionality.
2587 ** The GO debugger command is also gone, since you can just invoke
2588 the CONTINUE restart directly instead.
2589 ** The TOP debugger command is also gone, since it's redundant with the
2590 FRAME 0 command, and since it interfered with abbreviations for the
2592 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2594 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2595 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2596 some porting work by Martin Atzmueller)
2597 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2598 with a boatload of patches which he ported to SBCL. Now that those
2600 ** The system tries to make sure that its low-priority messages
2601 are prefixed by semicolons, to help people who like to use
2602 syntax highlighting in their ILISP buffer. (This patch
2603 was originally due to Raymond Toy.)
2604 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2605 patches originally written by Raymond Toy.
2606 ** The compiler understands coercion between single-value and
2607 multiple-VALUES type expressions better, getting rid of some very
2608 weird behavior, thanks to patches originally by Robert MacLachlan
2609 and Douglas Crosher.
2610 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2611 lambda lists, thanks to a patch originally by Pierre Mai.
2612 ** The system no longer bogusly warns about "abbreviated type
2614 ** The compiler gets less confused by inlining and RETURN-FROM,
2615 thanks to some patches originally by Tim Moore.
2616 ** The system no longer hangs when dumping circular lists to fasl
2617 files, thanks to a patch originally from Douglas Crosher.
2618 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2619 undefined function error.
2620 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2621 in structure slot initforms) for now, documented workaround instead:-|
2622 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2623 * fixed bug 5: added stubs for various Gray stream functions called
2624 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2625 installed, at least appropriate type errors are generated
2626 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2627 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2628 consistently in DEFMETHOD forms.
2629 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2630 it doesn't seem to affect SBCL after all
2631 * The C runtime system now builds with better optimization and many
2632 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2634 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2636 * The system is now under CVS at SourceForge (instead of the
2637 CVS repository on my home machine).
2638 * The new signal handling code has been tweaked to treat register
2639 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2640 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2641 downstream, e.g. in debug-int.lisp, has implicit dependencies
2642 on the unsignedness of integer representation of machine words,
2643 and that caused the system to bomb out with infinite regress
2644 when trying to recover from type errors involving signed values,
2645 e.g. (BUTLAST '(1 2 3) -1).)
2646 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2647 (This was one of the bugs Peter Van Eynde reported back in July.)
2648 * The system now uses code inspired by Colin Walters' O(N)
2649 implementation of MAP (from the cmucl-imp@cons.org mailing
2650 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2651 inline the MAP operation, and there is more than one
2652 sequence argument to the MAP call (so that it can't just
2653 do ETYPECASE once and for all based on the type of the
2654 single sequence argument). (The old non-inline implementation
2655 of the general M-argument sequence-of-length-N case required
2656 O(M*N*N) time when any of the sequence arguments were LISTs.)
2657 * The QUIT :UNIX-CODE keyword argument has been renamed to
2658 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2659 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2661 * Raymond Wiker's patches to port dynamic loading from Linux to
2662 FreeBSD have been added.
2663 * The BUGS file is now more nearly up to date, thanks in large part
2664 to Martin Atzmueller's review of it.
2665 * The debugger now flushes standard output streams before it begins
2666 its output ("debugger invoked" and so forth).
2667 * The core version number and fasl file version number have both
2668 been incremented, because of incompatible changes in the layout
2670 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2671 used to not be called for a saved Lisp image.)
2672 * Martin Atzmueller's version of a patch to fix a compiler crash,
2673 as posted on sbcl-devel 13 September 2000, has been installed.
2674 * Instead of installing Martin Atzmueller's patch for the
2675 compiler transform for SUBSEQ, I deleted the compiler transform,
2676 and transforms for some similar consing operations.
2677 * A bug in signal handling which kept TRACE from working on OpenBSD
2679 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2680 away by constant folding
2681 * The system now defines its address space constants in one place
2682 (in the Lisp sources), and propagates them automatically elsewhere
2683 (through GENESIS and the sbcl.h file). Therefore, patching the
2684 address map is less unnecessarily tedious and error-prone. The
2685 Lisp names of address space constants have also been systematized.
2686 * CVS tags like dollar-Header-dollar have been removed from
2687 the sources, because they have never saved me trouble and
2688 they've been source of trouble working with patches and other
2689 diff-related operations.
2690 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2691 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2693 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2695 * The system has been ported to OpenBSD.
2696 * The system now compiles with a simple "sh make.sh" on the systems
2697 that it's supported on. I.e., now you no longer need to tweak
2698 text in the target-features.lisp-expr and symlinks in src/runtime/
2699 by hand, the make.sh takes care of it for you.
2700 * The system is no longer so grossly inefficient when compiling code
2701 involving vectors implemented as general (not simple) vectors (VECTOR T),
2702 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2703 which dares to use the various sequence functions on non-simple
2704 vectors, takes less of a performance hit.
2705 * There is now a primitive type predicate VECTOR-T-P
2706 to test for the (VECTOR T) type, so that e.g.
2707 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2708 can now be compiled with some semblance of efficiency. (The old code
2709 turned the type declaration into a full call to %TYPEP at runtime!)
2710 * AREF on (VECTOR T) is still not fast, since it's still compiled
2711 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2712 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2713 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2714 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2715 without making full calls to SUBTYPEP at runtime.
2716 (Some analogous efficiency issues for non-simple vectors specialized to
2717 element types other than T, or for non-simple multidimensional arrays,
2718 have not been addressed. They could almost certainly be handled the
2719 same way if anyone is motivated to do so.)
2720 * The changes in array handling break binary compatibility, so
2721 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2722 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2723 instead of (VALUES T T).
2724 * By following the instructions that Dan Barlow posted to sbcl-devel
2725 on 2 July 2000, I was able to enable primitive dynamic object
2726 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2727 functionality is not implemented (since it calls ld to resolve
2728 library references automatically, requiring RUN-PROGRAM for its
2729 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2730 try to resolve library references) is now supported.
2731 * The system now flushes the standard output streams when it terminates,
2732 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2733 them at several other probably-convenient times, e.g. in each pass of
2734 the toplevel read-eval-print loop, and after evaluating a form given
2735 as an "--eval" command-line option. (These changes were motivated by a
2736 discussion of stream flushing issues on cmucl-imp in August 2000.)
2737 * The source transform for TYPEP of array types no longer assumes
2738 that an array whose element type is a not-yet-defined type
2739 is implemented as an array of T, but instead punts, so that the
2740 type will be interpreted at runtime.
2741 * There is now some support for cross-compiling in make.sh: each of
2742 the phases of make.sh has its own script. (This should be transparent
2743 to people doing ordinary, non-cross-compile builds.)
2744 * Since my laptop doesn't have hundreds of megabytes of memory like
2745 my desktop machine, I became more motivated to do some items on
2746 my to-do list in order to reduce the size of the system a little:
2747 ** Arrange for various needed-only-at-cold-init things to be
2748 uninterned after cold init. To support this, those things have
2749 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2750 symbols with such names are now uninterned after cold init).
2751 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2752 fasl files for cold load.
2753 ** Remove the old compiler structure pooling code (which used to
2754 be conditional on the target feature :SB-ALLOC) completely.
2755 ** Redo the representation of some data in cold init to be more compact.
2756 (I also looked into supporting byte compiled code at bootstrap time,
2757 which would probably reduce the size of the system a lot, but that
2758 looked too complicated, so I punted for now.)
2759 * The maximum signal nesting depth in the src/runtime/ support code has
2760 been reduced from 4096 to 256. (I don't know any reason for the very
2761 large old value. If the new smaller value turns out to break something,
2762 I'll probably just bump it back up.)
2763 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2765 * Many, many bugs reported by Peter Van Eynde have been added to
2766 the BUGS list; some have even been fixed.
2767 * While enabling dynamic object file loading, I tried to make the
2768 code easier to understand, renaming various functions and variables
2769 with less ambiguous names, and changing some function calling
2770 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2771 * While trying to figure out how to do the OpenBSD port, I tried to
2772 clean up some of the code in src/runtime/. In particular, I dropped
2773 support for non-POSIX signal handling, added various comments,
2774 tweaked the code to reduce the number of compilation warnings, and
2775 renamed some files to increase consistency.
2776 * To support the new automatic configuration functionality in make.sh,
2777 the source file target-features.lisp-expr has been replaced with the
2778 source file base-target-features.lisp-expr and the machine-generated
2779 file local-target-features.lisp-expr.
2780 * fixed a stupid quoting error in make.sh so that using CMU CL
2781 "lisp -batch" as cross-compilation host works again
2783 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2785 * DESCRIBE no longer tries to call itself recursively to describe
2786 bound/fbound values, so that it no longer fails on symbols which are
2787 bound to themselves (like keywords, T, and NIL).
2788 * DESCRIBE now works on generic functions.
2789 * The printer now prints less-screwed-up representations of closures
2790 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2791 * A private symbol is used instead of the :EMPTY keyword previously
2792 used to mark empty slots in hash tables. Thus
2793 (DEFVAR *HT* (MAKE-HASH-TABLE))
2794 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2795 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2796 now does what ANSI says that it should. (You can still get
2797 similar noncompliant behavior if bang on the hash table
2798 implementation with all the symbols you get back from
2799 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2800 This breaks binary compatibility, since tests for equality to
2801 :EMPTY are wired into things like the macroexpansion of
2802 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2804 * There's now a minimal placeholder implementation for CL:STEP,
2805 as required by ANSI.
2806 * An obscure bug in the interaction of the normal compiler, the byte
2807 compiler, inlining, and structure predicates has been patched
2808 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2809 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2811 * Missing ordinary arguments in a macro call are now detected even
2812 when the macro lambda list contains &KEY or &REST.
2813 * The debugger no longer complains about encountering the top of the
2814 stack when you type "FRAME 0" to explicitly instruct it to go to
2815 the top of the stack. And it now prints the frame you request even
2816 if it's the current frame (instead of saying "You are here.").
2817 * As specified by ANSI, the system now always prints keywords
2818 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2820 * The default initial SIZE of HASH-TABLEs is now smaller.
2821 * Type information from CLOS class dispatch is now propagated
2822 into DEFMETHOD bodies, so that e.g.
2823 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2825 is now basically equivalent to
2826 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2827 (DECLARE (TYPE SINGLE-FLOAT X))
2829 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2830 operation, without having to do run-time type dispatch.
2831 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2832 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2833 * Since I don't seem to be making big file reorganizations very often
2834 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2835 is overflowing my ability to conveniently back them up), I've finally
2836 checked the system into CVS. (The CVS repository is on my home system,
2837 not at SourceForge -- putting it on SourceForge might come later.)
2838 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2839 high-level GC-NOTIFY-FOO functions send their output. (There's
2840 still very little control of where low-level verbose GC functions
2841 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2842 less than it used to -- the GC-NOTIFY-FOO functions are now under
2843 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2844 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2845 in only one place in the source code, and propagates it automatically
2846 everywhere that it's needed. Thus e.g. when I bump the version from
2847 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2848 * The C source files now include boilerplate legalese and documentation
2849 at the head of each file (just as the Lisp source files already did).
2850 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2851 to his page will be replaced with a link to his new CLiki service.
2853 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2855 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2856 * The build process now looks for GNU make under the default name "gmake",
2857 instead of "make" as it used to. If GNU make is not available as "gmake"
2858 on your system, you can change this default behavior by setting the
2859 GNUMAKE environment variable.
2860 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2861 can build without error under CMU CL.
2863 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2865 * There is now a partial SBCL user manual (with some new text and some
2866 text cribbed from the CMU CL manual).
2867 * The beginnings of a profiler have been added (starting with the
2868 CMU CL profiler and simplifying and cleaning up). Eventually the
2869 main interface should be through the TRACE macro, but for now,
2870 it's still accessed through vaguely CMU-CL-style functions and macros
2871 exported from the package SB-PROFILE.
2872 * Some problems left over from porting CMU CL to the new
2873 cross-compilation bootstrap process have been cleaned up:
2874 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2875 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2876 part when getting rid of the compiler *BACKEND* stuff.)
2877 ** The value of *NULL-TYPE* was screwed up, because it was
2878 being initialized before the type system knew the final
2879 definition of the 'NULL type. This screwed up several key
2880 optimizations in the compiler, causing inefficiency in all sorts
2881 of places. (I found it because I wanted to understand why
2882 GET-INTERNAL-RUN-TIME was consing.)
2883 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2884 PROCLAIM FTYPE information. Unfortunately this broke binary
2885 compatibility again, since now the forms output by DEFGENERIC
2886 to refer to functions which didn't exist in 0.6.3.
2887 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2888 can use the new (as of 0.6.3) transform for SOME into MAP into
2890 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2891 values to (INTEGER 0 1000000), so that the time code will no longer
2892 occasionally get blown up by Linux returning 1000000 microseconds
2893 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2894 its output conform to the ANSI spec. (Alas, this makes its output
2895 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2896 * A full call to MAP NIL with a single sequence argument no longer conses.
2897 * fixes to problems pointed out by Martin Atzmueller:
2898 * The manual page no longer talks about multiprocessing as though
2899 it were currently supported.
2900 * The ILISP support patches have been removed from the distribution,
2901 because as of version 5.10.1, ILISP now supports SBCL without us
2902 having to maintain patches.
2903 * added a modified version of Raymond Toy's recent CMU CL patch for
2904 EQUALP comparison of HASH-TABLE
2906 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2908 * The system still can't cross-compile itself with
2909 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2910 entails), but at least it can compile more of itself that way
2911 than it used to be able to, and various buglets which were uncovered
2912 by trying to cross-compile itself that way have now been fixed.
2913 * This release breaks binary compatibility again. This time
2914 at least I've incremented the FASL file format version to 2, so that the
2915 problem can be detected reliably instead of just causing weird errors.
2916 * various new style warnings:
2917 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2918 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2919 ** using the lexical binding of a variable named in the *FOO* style
2920 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2921 as specified by ANSI.
2922 * *RANDOM-STATE* is no longer automatically initialized from
2923 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2924 default behavior of the system is to repeat its behavior every time
2925 it's run. If you'd like to change this behavior, you can always
2926 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2927 old convention there was no comparably easy way to get the system to
2928 repeat its behavior every time it was run.
2929 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2930 FTYPE declarations has been removed, in favor of their ANSI
2931 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2932 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2933 the types of their sequence arguments can be determined at compile time.
2934 This is done through a new open code expansion for MAP which eliminates
2935 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2936 argument types can be determined at compile time.
2937 * The optimizer now transforms COERCE into an identity operation when it
2938 can prove that the coerced object is already of the correct type. (This
2939 can be a win for machine generated code, including the output of other
2940 optimization transforms, such as the MAP transform above.)
2941 * Credit information has been moved from source file headers into CREDITS.
2942 * Source file headers have been made more standard.
2943 * The CASE macro now compiles without complaining even when it has
2946 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2948 * (Note that the way that the PCL macroexpansions were rewritten
2949 to accommodate the change in DEFGENERIC below breaks binary
2950 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2951 not run under sbcl-0.6.2. Once we get out of alpha releases,
2952 i.e. hit release 1.0.0, we'll probably try to maintain binary
2953 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2954 and sbcl-1.4.4. Until then, however, it might be fairly common
2955 for maintenance releases to break binary compatibility.)
2956 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2958 * The reporting of unhandled signals has been changed to print some
2959 explanatory text as well as the report form. (Previously only
2960 the report form was printed.)
2961 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2962 it defines, so that the compiler no longer issues undefined function
2963 warnings for compiled-but-not-yet-loaded generic functions.
2964 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2965 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2966 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2967 section of the ANSI spec are supported.
2968 * The "" nickname for the "KEYWORD" package has been removed.
2969 The reader still handles symbol tokens which begin with a package marker
2970 as keywords, but it doesn't expose its mechanism for doing so in the
2971 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2972 * The system now issues STYLE-WARNINGs for contradictory TYPE
2973 proclamations. (Warnings for contradictory FTYPE proclamations would
2974 be nice too, but those can't be done usefully unless the type system
2975 is made smarter about FUNCTION types.)
2976 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2977 systematized, so that "*target-*.lisp is supposed to exist only on the
2978 target and imply that there's a related file which exists on the
2979 host, and *host-*.lisp is supposed to exist only on the host and imply
2980 that there's a related file which exists on the target. This involves a
2981 lot of renaming. Hopefully the acute confusion caused by the renaming
2982 will be justified by the reduction in chronic confusion..
2983 ** runtime-type.lisp -> early-target-type.lisp
2984 ** target-type.lisp -> late-target-type.lisp
2985 ** early-host-format.lisp -> early-format.lisp
2986 ** late-host-format.lisp -> late-format.lisp
2987 ** host-error.lisp -> misc-error.lisp
2988 ** early-error.lisp -> early-target-error.lisp
2989 ** late-error.lisp -> late-target-error.lisp
2990 ** host-defboot.lisp -> early-defboot.lisp
2991 ** code/misc.lisp -> code/target-misc.lisp
2992 ** code/host-misc.lisp -> code/misc.lisp
2993 ** code/numbers.lisp -> code/target-numbers.lisp
2994 ** code/early-numbers.lisp -> numbers.lisp
2995 ** early-host-type.lisp -> early-type.lisp
2996 ** late-host-type.lisp -> late-type.lisp
2997 ** host-typep.lisp -> typep.lisp
2998 ** load.lisp -> target-load.lisp
2999 ** host-load.lisp -> load.lisp
3000 ** host-disassem.lisp -> disassem.lisp
3001 ** host-insts.lisp -> insts.lisp
3002 ** byte-comp.lisp -> target-byte-comp.lisp
3003 ** host-byte-comp.lisp -> byte-comp.lisp
3004 ** host-signal.lisp -> signal.lisp
3005 ** host-defstruct.lisp -> defstruct.lisp
3006 ** late-target-type.lisp -> deftypes-for-target.lisp
3007 Furthermore, several other previously target-only files foo.lisp (e.g.
3008 hash-table.lisp and random.lisp) have been split into a target-and-host
3009 foo.lisp file and a target-only target-foo.lisp file, with their key type
3010 definitions in the target-and-host part, so that the cross-compiler will
3011 know more about target types.
3012 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3013 gone away. In their place are various *BACKEND-FOO* variables
3014 corresponding to the slots of the old structure.
3015 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3016 into the target SBCL has been fixed.
3017 * The chill.lisp system for loading cold code into a running SBCL
3019 * Support for the CMU CL "scavenger hook" extension has been removed.
3020 (It was undocumented and unused in the CMU CL sources that SBCL was
3021 derived from, and stale in sbcl-0.6.1.)
3022 * Various errors in the cross-compiler type system were detected
3023 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3024 (enabling various consistency checks). Many of them were fixed,
3025 but some hard problems remain, so the compiler is back to
3026 running without *TYPE-SYSTEM-INITIALIZED* for now.
3027 * As part of the cross-compiler type system cleanup, I implemented
3028 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3029 * I have started adding UNCROSS calls throughout the type system
3030 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3031 on cross-compiler input files will be able to go away, and various
3033 * CONSTANTP now returns true for quoted forms (as explicitly required
3036 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3038 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3039 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3040 (MAKE-LIST -1). In the longer term, it ought to become true
3041 that declarations are assertions even at SAFETY 1. For now, it's not
3042 quite true even at SAFETY 3, but it's at least more nearly true..
3043 (Note that this change seems to increases the size of the system by
3044 O(5%) and to decrease the speed of the compiler by 20% or more.)
3045 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3046 for the problem of printing dozens of lines of distracting information
3047 about low-level system machinery as part of the top stack frame
3048 on entry to the debugger when an undefined function was called.
3049 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3050 is not reset to COMMON-LISP-USER.
3051 * Compilation of stuff related to dyncount.lisp has been made conditional
3052 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3053 smaller. The various dyncount-related symbols have been moved into
3054 a new "SB-DYNCOUNT" package.
3055 * tty-inspect.lisp has been renamed to inspect.lisp.
3056 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3057 feature has gone away. (When we eventually port to other flavors of
3058 libc and/or Unix, we'll try to make the differences between flavors
3059 invisible at the user level.)
3060 * Various other *FEATURES* tags, and/or their associated conditionals,
3061 have been removed if obsolescent, or given better documentation, or
3062 sometimes given more-mnemonic names.
3064 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3066 * tidied up "make.sh" script
3067 * tidied up system directory structure
3068 * better "clean.sh" behavior
3069 * added doc/FOR-CMUCL-DEVELOPERS
3070 * many many small tweaks to output format, e.g. removing possibly-confusing
3071 trailing #\. character in DESCRIBE-INSTANCE
3072 * (EQUALP #\A 'A) no longer signals an error.
3073 * new hashing code, including EQUALP hashing
3074 * tidied up Lisp initialization and toplevel
3075 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3076 * command line argument processing
3077 * added POSIX-GETENV function to deal with Unix-ish environment variables
3078 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3079 terminating SBCL on EOF
3080 * non-verbose GC by default
3081 * There is no more "sbcl" shell script; the sbcl file is now the C
3082 runtime executable (just like CMU CL).
3083 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3085 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3086 symbol and package information as Lisp native symbol and package objects
3087 instead of strings naming symbols and strings naming packages. This way,
3088 whenever packages are renamed (as in warm init), debug information is
3089 transformed along with everything else.
3090 * tweaked the optimization policy declarations which control the building
3091 of SBCL itself. Now, among other things, the system no longer saves
3092 source location debugging information. (This helps two problems at once
3093 by reducing SBCL size and by keeping SBCL from trying to look for its
3094 sources -- which may not exist -- when reporting errors.)
3095 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3096 debugging and testing purposes
3097 * cleaned up printing, making the printer call PRINT-OBJECT for
3098 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3099 methods, giving nearly-ANSI behavior
3100 * converted almost all special variables to use *FOO* naming convention
3101 * deleted PARSE-TIME functionality, since it can be done portably
3102 * moved some files out of cold init into warm init
3103 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3105 * regularized formatting of source files
3106 * added an install.sh script
3107 * fixed ridiculous memory usage of cross-compiler by making
3108 compiler/alloc.lisp not try to do pooling unless it can hook
3109 itself into the GC of the cross-compilation host. Now the system
3110 builds nicely on my old laptop.
3111 * added :SB-ALLOC in target-features.lisp-expr
3112 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3113 was not implemented)
3114 * re-did condition handling and note reporting in the compiler. Notes
3115 are no longer handled by signalling conditions. Style warnings
3116 and warnings are handled more correctly and reported in such a way
3117 that it's easy to find one or the other in your output (so that you
3118 can e.g. figure out which of many problems caused COMPILE-FILE to
3120 * changed the severity of several compiler warnings from full WARNING
3121 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3122 compiler note reporting so that it doesn't use the condition system
3123 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3124 COMPILE-FILE command)
3125 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3126 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3127 no longer supported, since their implementation was deeply intertwingled
3128 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3129 * removed bogus "support" for compiler macros named (SETF FOO), and
3130 removed the compiler macro for SETF INFO (but only after making a fool
3131 of myself on the cmucl-imp mailing list by posting a bogus patch for
3132 DEFINE-COMPILER-MACRO..)
3133 * Compiled files containing forms which have side effects on the Lisp
3134 reader (such as DEFPACKAGE forms) are now handled more correctly.
3135 (Compiler queuing of top level lambdas has been suppressed by setting
3136 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3137 * deleted various currently-unused source files, e.g. gengc.lisp. They
3138 may be added back at some point e.g. when porting to other architectures,
3139 but until they are it's distracting to distribute them and to try to
3141 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3142 there were so many of them they're just distractions, and UNCROSS is
3143 known to be able to handle the current sources
3144 * moved PROFILE functionality into TRACE, so that it will be clear
3145 how the wrapping and unwrapping of functions when you profile them
3146 interacts with the wrapping and unwrapping of functions when you
3147 trace them. (Actually, the functionality isn't there yet, but at least
3148 the interface specification is there. Hopefully, the functionality will
3149 arrive with some future maintenance release.)
3150 * removed host-oops.lisp
3151 * changed signature of QUIT function to allow UNIX-CODE argument
3152 * fixed READ-SEQUENCE bug
3153 * tweaked verbose GC output so that it looks more like the progress
3154 output that ANSI specifies for functions like LOAD
3155 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3156 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob