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 * the long form of DEFINE-METHOD-COMBINATION disables method group
57 checking when given a single method group with pattern *. (thanks
59 * on x86 compiler supports stack allocation of results of simple
60 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
61 * fixed some bugs related to Unicode integration:
62 ** the restarts for recovering from input and output encoding
63 errors only appear when there is in fact such an error to
65 * fixed some bugs revealed by Paul Dietz' test suite:
66 ** MISC.549 and similar: late transformation of unsafe type
67 assertions into derived types caused unexpected code
69 ** SCALE-FLOAT type deriver is less wrong.
70 ** type derivers for EXP, LOG and similar functions compute result
71 types for complex arguments better.
72 ** (MISC.563) CONJUGATE type deriver works for very restricted
74 ** out-of-line type testers for character strings are available.
75 ** EQUAL compiler transform understands specialness of objects
77 ** accessing double-floats stored on the stack now works on x86-64.
78 ** debugger internals could sometimes create invalid lispobjs,
79 resulting in GC crashes.
80 ** MISC.548: type check weakening can convert required type into
82 ** initialization forms for bindings are not in scope of free special
85 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
86 * incompatible change: thread support for non-NPTL systems has
87 been removed - locking is buggy and unreliable. A threaded
88 SBCL build will now warn at startup and refuse to create
89 new threads, unless futex support is detected in the kernel
90 * incompatible change: the top level REPL now has only an ABORT
91 restart associated with it, not TOPLEVEL and ABORT as it used to.
92 TOP and TOPLEVEL are now available as debugger commands for
93 returning to the top level.
94 * incompatible change: forms evaluated in the REPL now use the
95 global optimization policy.
96 * incompatible change: user- and system-initialization files are
97 no longer processed with LOAD, but by READ and EVAL; hence the
98 global optimization policy, startup package, readtable, etc,
100 * The .fasl file format number has been incremented because of
101 various incompatible changes.
102 * internal entry point details and argument counts no longer appear
103 in backtraces unless explicitly requested by setting
104 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
105 * built-in and standard functions no longer have names like "top
106 level local call to FOO".
107 * fixed bug 32: functions defined in non-null lexical environments
108 now have more legible printed representation
109 * fixed bug 33: functions defined in non-null lexical environemnts
110 are now more amenable to inspection by INSPECT.
111 * workaround for bug 354: XEPs no longer appear in backtraces unless
112 explicitly requested.
113 * fixed bug: receiving the signal which results from attempting to
114 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
115 SIGBUS on other *BSDs) on architectures where the C stack is also
116 the Lisp stack (x86 and x86-64) from foreign code no longer leads
117 to debugger confusion or wild instability. (reported by Cheuksan
119 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
120 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
121 (reported by Lutz Euler)
122 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
123 compiled code. (reported by Paul Dietz)
124 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
125 specialized parameter to a method to something that is not TYPEP
126 the specializer is now possible.
127 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
128 face of package deletion.
129 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
130 pathname. (reported by David Tolpin on c.l.l)
131 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
132 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
133 STRING-STREAM and FILE-STREAM, respectively.
134 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
135 than previously; in addition, its test suite is now run on build.
136 (thanks to Luke Gorrie)
137 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
138 * optimization: added a immediate representation for single-floats
140 * fixed some bugs revealed by Paul Dietz' test suite:
141 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
142 ** attempting to create a package with a colliding nickname causes
143 correctable errors to be signalled.
144 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
145 ** several x86-64 backend bugs related to sign-extension of immediate
148 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
149 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
150 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
152 * fixed loading of multiply forward-referenced layouts.
153 (thanks to Cheuksan Wang)
154 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
155 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
156 (see COMPILE-FILE documentation for details of :PRINT option)
157 * fixed bugs 19 and 317: fixed-format floating point printing is
158 more accurate. This also fixes a bug reported by Adam Warner
159 related to the ~@F format directive.
160 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
162 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
163 dispatch table if the from-char is a dispatch macro character.
164 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
165 by the wordsize no longer ignore the last word. (reported by Lutz
167 * fixed bug in type checking of dynamic-extent variables. (reported
169 * optimization: sequence traversal functions use their freedom to
170 coerce function designators to functions.
171 * optimization: code with many calls to CLOS methods specialized on
172 CLOS classes has had redundant type checks removed.
173 * optimization: added declarations to speed up operations that access
174 the internal character database (for example STRING-UPCASE)
175 * optimization: comparison operations between floats and sufficiently small
176 fixnums no longer create extra rationals
177 * fixed some bugs related to Unicode integration:
178 ** portions of multibyte characters at the end of buffers for
179 character-based file input are correctly transferred to the
180 start of the buffer at the next read.
181 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
182 passing it through to OPEN.
183 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
184 argument given to internal calls to OPEN.
185 ** the built-in comment readers (introduced by character sequences
186 ";" and "#|") are more forgiving to encoding errors; they will
187 STYLE-WARN and then attempt to resync the stream at a character
188 boundary. (thanks to Teemu Kalvas)
189 * fixed some bugs in the x86-64 port:
190 ** Negative short int return values from c-calls are sign-extended
192 ** The stack is aligned to 16-bytes for c-calls, as required by
193 the ABI. (thanks to Cheuksan Wang)
194 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
195 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
196 for enums. (thanks to Vincent Arkesteijn)
197 ** Multiple small optimizations and bugfixes for floating point
199 * fixed some bugs revealed by Paul Dietz' test suite:
200 ** Space, Tab, Linefeed, Return and Page have the invalid
201 secondary constituent character trait.
202 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
204 ** WITH-INPUT-FROM-STRING should only update the index place on
206 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
208 ** Bit-array operations (BIT-AND and similar) worked incorrectly
209 with one-dimensional arrays with fill pointers.
210 ** TYPE-OF failed on a complex with an integer realpart and a
212 ** compiler failure during type inference for the code of form
213 (IF (EQL X (THE ...)) ...) (MISC.535).
215 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
216 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
217 platform. The port supports 61-bit fixnums, large memory spaces
218 and reloading shared object files.
219 * enhancement: saving cores with foreign code loaded is now
220 supported on ppc/Darwin in addition to the previously supported
222 * enhancement: the statistical profiler now walks deeper into the
223 call stack for more meaningful call-graphs and accrued time
224 reports (x86/x86-64 only). It also now reports time spent in
226 * enhancement: it is now possible to trace most individual methods
227 of a generic function in addition to tracing the generic function
229 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
231 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
232 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
233 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
234 produces an error. (thanks to Vincent Arkesteijn)
235 * bug fix: NAMESTRING on pathnames with :WILD components in their
236 directories works correctly. (thanks to Artem V. Andreev)
237 * fixed bug 125: compiler preserves identity of closures. (reported
239 * bug fixed: functions with &REST arguments sometimes failed with
240 "Undefined function" when compiled with (DEBUG 3). (reported by
242 * bug fix: overflow during compiling of setting element of a bit
243 vector with constant index and value. (reported by Timmy Douglas)
244 * build fix: fixed the dependence on *LOAD-PATHNAME* and
245 *COMPILE-FILE-PATHNAME* being absolute pathnames.
246 * on x86 compiler partially supports stack allocation of dynamic-extent
248 * GO and RETURN-FROM do not check the extent of their exit points
249 when compiled with SAFETY 0.
250 * fixed some bugs related to Unicode integration:
251 ** encoding and decoding errors are now much more robustly
252 handled; it should now be possible to recover even from invalid
253 input or output to the terminal. (thanks to Teemu Kalvas)
254 ** provided a first cut at implementing STRING-TO-OCTETS and
255 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
256 ** altered the SB-MD5 contributed module slightly, changing the
257 interface just enough for it to be supportable for builds where
258 lisp characters are not eight bits.
259 * fixed some bugs revealed by Paul Dietz' test suite:
260 ** the FORMATTER-generated functions for ~V[ conditionals require
261 the correct number of arguments.
262 ** READ-FROM-STRING returns the mandated second value when applied
263 to displaced strings.
264 ** the #\Rubout and #\Backspace characters are treated as invalid
265 constituent characters by the tokenizer.
267 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
268 * new feature: reloading changed shared object files with
269 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
270 * new feature: references to foreign variables and functions
271 can now be compiled and loaded before the corresponding shared
272 object file is loaded, as long as the foreign definitions are
273 available at runtime.
274 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
275 supported. (thanks to Dan Debertin)
276 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
277 just keywords. (thanks to Vincent Arkesteijn)
278 * fixed bug #331: structure-class instances corresponding to
279 DEFSTRUCT forms are now created eagerly.
280 * fixed bug #345: backtraces from calls to undefined functions work
281 on x86 as well. Related bug #61 is now also partially fixed on x86
282 (backtraces from throws to unknown catch tags.)
283 * bug fix: lambda-list parsing is now stricter vrt. order and number
284 of lambda-list keywords.
285 * bug fix: as specified by AMOP, an error is signalled if a
286 class-option appears multiple times in a DEFCLASS form. (reported
288 * bug fix: RANDOM can be compiled when the compiler derives the type
289 of its numeric argument as a disjoint set of small integers.
290 (reported by Paul Dietz)
291 * bug fix: starting a core saved with shared objects loaded when
292 those objects are not available no longer causes threaded SBCL to
293 hang. (reported by Sean Ross)
294 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
296 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
297 argument, which must always be NIL. (reported by Kalle Niemitalo)
298 * bug fix: printing 1.0d+23 no longer results in an error.
299 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
300 * bug fix: structure slot setters preserve evaluation order. (thanks
302 * fixed some bugs related to Unicode integration:
303 ** RUN-PROGRAM can allow its child to take input from a Lisp
304 stream. (reported by Stefan Scholl)
305 ** ASDF-INSTALL successfully downloads .tar.gz files in all
306 locales. (reported by Ken Causey)
307 * fixed some bugs revealed by Paul Dietz' test suite:
308 ** INCF, DECF and REMF evaluate their place form as specified in
310 ** FORMATTER expands ~{ ~} iteration directives with V or #
311 parameters correctly.
312 ** FORMATTER deals with the ~@[ ~] conditional directive where the
313 consequent uses no arguments correctly.
314 ** the system has a partial understanding of the (COMPLEX RATIO)
316 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
319 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
320 * new feature: a build-time option (controlled by the :SB-UNICODE
321 keyword feature, enabled by default) for building the system with
322 support for the entire 21-bit character space defined by the
324 * new feature: the system now has rudimentary external-format
325 support; the primary user-visible change at this time is that
326 characters with the high bit set (such as lower-case-e-acute) will
327 print correctly to a terminal in a UTF-8 environment.
328 * minor incompatible change: BASE-CHAR no longer names a class;
329 however, CHARACTER continues to do so, as required by ANSI.
330 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
331 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
332 has been moved to the SB-EXT package (temporarily re-exported from
334 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
335 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
337 * fixed bug #351: better error-handlind and reporting for malformed
339 * fixed bug #350: bignum-printing is now more memory-efficient,
340 allowing printing of very large bignums, eg. (expt 2 10000000).
341 (reported by Bruno Haible)
342 * fixed bug #302: better primitive-type selection for intersection
344 * fixed bug #308: non-graphic characters now all have names, as
345 required. (reported by Bruno Haible)
346 * bug fix: redefining a class with different superclasses now correctly
347 removes it from the direct-subclasses of its previous superclasses.
348 (reported by David Morse)
349 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
350 new value now works. (reported by Bruno Haible)
351 * bug fix: correct canonicalization of multiple non-standard slot
352 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
353 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
354 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
355 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
356 now exists, an signals an error.
357 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
358 during macroexpansion for non-iteration variables as well. (reported
359 by Bruno Haible for CMUCL)
360 * bug fix: Cyclic structures and unprintable objects in compiler
361 messages no longer cause errors. (reported by Bruno Haible)
362 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
363 and READ-FROM-STRING all now return a primary value of NIL if
364 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
365 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
366 (reported by Bruno Haible for CMUCL)
367 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
368 do not designate a condition. (reported by Bruno Haible for
370 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
371 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
372 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
373 permitting resolution in favour of any of the conflicting symbols.
374 (reported by Bruno Haible for CMUCL)
375 * bug fix: EQUAL compiler optimizations is less aggressive on
376 strings which can potentially compare true despite having distinct
377 specialized array element types.
378 * bug fix: unit enumerations can be defined without dividing by
379 zero. (thanks to Vincent Arkesteijn)
380 * FORMAT compile-time argument count checking has been enhanced.
381 (report from Bruno Haible for CMUCL)
382 * a partial workaround for the bug 262: the compiler does not try to
383 inline-expand a local function doing RETURN-FROM from a deleted
384 BLOCK. (thanks to Peter Denno for the bug report and to David
385 Wragg for the simple test case)
386 * fixed some bugs revealed by Paul Dietz' test suite:
387 ** RENAME-PACKAGE allows all package designators as new package
389 ** constraint propagation and lambda variable substitution are
390 more cautious in dealing with partially deleted code.
391 ** compiler sometimes lost reoptimization passes.
392 ** CERROR, given a condition as condition designator, passes its
393 remaining arguments to the continue format control without
395 ** Case-altering FORMAT directives work correctly on non-ASCII
397 ** The REAL type specifier handles bounds outside the floating
398 point ranges without signalling FLOATING-POINT-OVERFLOW.
399 ** Functions with IR1-transformations can create intercomponent
400 references to global functions.
401 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
403 ** FORMAT ~R treats a nil value for its first parameter correctly.
405 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
406 * enhancement: saving cores with foreign code loaded is now
407 supported on x86/NetBSD and sparc/Linux in addition to the previously
409 * bug fix: on some platforms repeated installations caused multiple
410 copies of HTML documentation to be installed -- should not happen
411 any more. (reported by Stefan Scholl)
412 * bug fix: parsing self-recursive alien record types multiple times
413 no longer causes infinite recursion. (reported by Thomas F. Burdick,
414 original patch by Helmut Eller for CMUCL)
415 * bug fix: stack-exhaustion detection works now on NetBSD as well.
416 (thanks to Richard Kreuter)
417 * bug fix: defining classes whose accessors are methods on existing
418 generic functions in other (locked) packages no longer signals
419 bogus package lock violations. (reported by François-René Rideau)
420 * bug fix: special variables as DEFMETHOD parameters no longer have
421 associated bogus type declarations. (reported by David Wragg and
423 * bug fix: read-write consistency on streams of element-type
424 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
425 * bug fix: redefiniton of the only method of a generic function with
426 no DEFGENERIC no longer emits a full WARNING. In addition,
427 redefinition of generic functions with no DEFGENERIC to an
428 incompatible lambda list now signals an error. (thanks to Zach
430 * bug fix: DEFGENERIC now works even when there's a function of the
431 same name in an enclosing lexical environment. (thanks to Zach
433 * fixed compiler failure, caused by instrumenting code during
434 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
435 * optimization: added loop analysis and improved register allocation
436 * fixed some bugs revealed by Paul Dietz' test suite:
437 ** POSITION on displaced vectors with non-zero displacement
438 returns the right answer.
439 ** (SIMPLE-STRING) is a valid type specifier for sequence
441 ** *PRINT-LEVEL* handling for slotless structures is pedantically
443 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
444 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
446 ** The justification version of the FORMAT ~< directive treats
447 non-zero minpad parameter correctly.
449 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
450 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
451 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
452 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
453 the supported interface.
454 * new feature: Single-stepping of code compiled with DEBUG 2 or
455 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
456 * new feature: saving cores with foreign code loaded is now
457 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
458 Timothy Moore's work for CMUCL)
459 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
460 parameters to * instead of NIL if no initform is supplied.
461 (reported by Johan Bockgaard)
462 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
463 correctly when FUNCALL appears as the car of the form. Note:
464 despite this FUNCALL forms are not currently subject to
465 compiler-macro expansion. (port of Raymond Toy's fix for the
466 same from CMUCL, reported by Johan Bockgaard)
467 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
468 (thanks for Teemu Kalvas)
469 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
470 case instead of terminating the iteration (thanks for Julian
471 Squires, Sean Champ and Raymond Toy)
472 * bug fix: incorrect expansion of defgeneric that caused a style
473 warning. (thanks for Zach Beane)
474 * bug fix: slot accessor effective method computation works properly
475 for classes with multiple non-standard applicable methods on
476 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
477 * on x86 compiler supports stack allocation of results of LIST and
478 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
479 CMUCL implementation by Gerd Moellmann)
480 * fixed some bugs revealed by Paul Dietz' test suite:
481 ** FORMAT strings with both the ~<~:;~> form of the justification
482 directive and pretty-printing directives cause an error.
484 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
485 * incompatible change: the internal functions
486 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
487 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
488 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
489 instead of the old functions.
490 * new feature: on platforms where "dladdr" is available foreign
491 function names now appear in backtraces. (based on Helmut Eller's
493 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
494 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
496 * documentation: ASDF manual is now distributed with SBCL.
497 * bug fix: SBCL can now load its contributed modules with REQUIRE
498 even if the system-provided entries have been removed from
499 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
501 * bug fix: backtraces involving undefined functions or assembly
502 routines are more informative. (thanks to Brian Downing)
503 * bug fix: mutually referent alien structures now work correctly.
504 (reported by Rick Taube)
505 * bug fix: structures defined by WITH-ALIEN can be referred to
506 within other definitions in the same WITH-ALIEN.
507 * bug fix: division operators (MOD, TRUNCATE and the like) with
508 constant zero divisors and integer dividends no longer generate
510 * bug fix: provide default methods for INPUT-STREAM-P and
511 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
512 * bug fix: improve the detection and resolution of MOP metacycles
513 (where effective method computation depends on the generic function
514 having its effective method computed). (reported by Bruno Haible)
515 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
516 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
517 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
518 * bug fix: pathnames with relative directory components can be
519 represented relative to default pathnames.
520 * optimization: in taking the GCD of bignums, reduce the two bignums
521 to approximately the same size (using Euclid's algorithm) before
522 applying the more sophisticated binary GCD. (thanks to Juho
524 * optimization: COUNT on bitvectors now operates word-at-a-time.
525 * optimization: ASH with a positive, but not necessarily constant,
526 (leftwards) shift, when in a modular context, is compiled to a
528 * fixed some bugs revealed by Paul Dietz' test suite:
529 ** FORMAT variable parameters ("~V<char>") are defaulted properly
530 if the corresponding argument is NIL.
531 ** FORMAT directives accepting a minpad parameter treat negative
533 ** literal commas as character parameters to FORMAT directives are
535 ** literal spaces directly after ~<Newline> directives within a
536 format-logical-block (~:< ~@:>) do not induce :FILL-style
537 conditional newlines.
538 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
539 as specified: it no longer includes conditional newlines).
540 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
542 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
543 :PER-LINE-PREFIX argument does not evaluate to a string.
545 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
546 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
547 the manual for details; for now, package locks can be disabled by
548 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
549 if no major problems are found then it is likely that they will be
550 compiled in unconditionally.
551 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
552 unsupported operators on all platforms. To load a shared library
553 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
554 object file, link it into a shared library outside of SBCL and
555 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
557 * bug workaround: ROOM T and the GC were not working together
558 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
559 reporting facility aren't preserved. That reporting has been
560 disabled, so now until and unless someone figures out how to make it
561 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
562 * minor incompatible change: as threatened around sbcl-0.8.0, the
563 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
564 The symbols are also exported from SB-PCL for backwards
565 compatibility, but more so than before SB-PCL should be treated as
566 an implementation-internal package.
567 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
569 * the SB-SPROF contrib now works on (most) non-x86 architectures.
570 It is known as of this release not to work on the Alpha, however.
571 * fixed bug #167: errors signalled due to illegal syntax in method
572 bodies are now more legible.
573 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
574 designators and can hence be used with TYPEP.
575 * fixed bug #333: CHECK-TYPE now ensures that the type error
576 signalled, if any, has the right object to be accessed by
577 TYPE-ERROR-DATUM. (reported by Tony Martinez)
578 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
579 5.1.2.3 for multiple-value place subforms. (reported by Kalle
581 * fixed bug #334: programmatic addition of slots using specialized
582 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
583 and :ALLOCATION :CLASS slots.
584 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
585 float exponents. (rereported by Peter Seibel)
586 * fixed a bug: #\Space (and other whitespace characters) are no
587 longer considered to be macro characters in standard syntax by
589 * fixed bug: initialization of condition class metaobjects no longer
590 causes an instance of the condition to be created. (reported by
592 * fixed bug: it is now possible to have slots such that
593 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
594 description is neither :INSTANCE nor :CLASS.
595 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
596 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
597 SLOT-VALUE-USING-CLASS), and no longer causes errors with
598 non-standard SLOT-DEFINITION-ALLOCATION values.
599 * fixed bugs: various race conditions handling exiting threads.
600 CL-PPCRE's thread torture test now passes.
601 * fixed arguably-a-bug: GC time proportional to number of threads in
602 system even when most of them are idle
603 * optimization: improved performance of BIT and SBIT on bit-vectors.
604 * .fasl file incompatibility: The fasl file version number has
605 been incremented because of changes associated with package locks.
607 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
608 * minor incompatible change: the system no longer provides
609 optimization hints (conditions which are TYPEP
610 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
611 mode; these hints are still emitted when higher SPEED optimization
613 * new contrib module: a sampling profiler (profiling by statistical
614 sampling, rather than by instrumenting functions) is available as
615 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
616 * the behaviour of the standard function ED is now customizeable by
617 third parties through a hook variable: see ED's documentation
618 string for information on the protocol.
619 * the compiler no longer emits efficiency notes for (FUNCALL X)
620 when the type of X is uncertain under default optimization
622 * fixed bug 276: mutating a binding of a specialized parameter to a
623 method to something that is not TYPEP the specializer is now
625 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
626 exhibit the required behaviour on the x86 platform. (thanks to
627 Peter van Eynde, Eric Marsden and Bruno Haible)
628 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
629 even for difficult arguments. (reported by Peter Graves)
630 * fixed bug 141a: the backquote printer now descends quoted
632 * fixed another bug in backquote printing: no more destructive
633 modification of the form's list structure. (reported by Brian
635 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
636 move between its address being taken and the call to
637 interrupt_thread, fixing a crashing race condition.
638 * the SB-POSIX contrib implementation has been adjusted so that it
639 no longer exhibits ridiculously poor performance when constructing
640 instances corresponding to C structs.
642 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
643 * minor incompatible change: the sb-grovel contrib now treats C
644 structures as alien (in the sense of SB-ALIEN) objects rather than
645 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
646 has implications for memory management of client code
647 (sb-grovel-returned objects must now be manually managed) and for
648 type safety (alien objects now have full types).
649 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
650 used to control emission of compiler diagnostics, rather than the
651 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
652 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
653 quality should be considered deprecated.
654 * install.sh now installs the user manual as well
655 * (not quite a new documentable feature, but worth considering in
656 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
657 beginnings of a semantically meaningful condition hierarchy is
658 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
660 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
661 designator as the defaults argument.
662 * fixed bug: Displaced arrays whose displaced-to array has become
663 too small now cause ARRAY-DIMENSION to signal an error, providing
664 for safer bounds-checking. (reported by Bruno Haible)
665 * fixed bug: DEFCLASS slot definitions with identical :READER and
666 :WRITER names now signal a reasonable error. (reported by Thomas
668 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
670 * fixed bug: Invalid slot specification errors now print correctly.
671 (thanks to Zach Beane)
672 * fixed bug 320: Shared to local slot value transfers in class
673 redefinitions now happen corrently from superclasses as
674 well. (reported by Bruno Haible)
675 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
677 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
678 improper lists correctly. (reported by Bruno Haible)
679 * fixed bug 313: source-transform for <fun-name> was erroneously
680 applied to a call of a value of a variable with name <fun-name>.
681 (reported by Antonio Menezes Leitao)
682 * fixed bug 307: The obsolete instance protocol ensures that
683 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
684 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
685 remove dead unknown-values globs from the middle of the stack.
686 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
687 easier to resume long computations after using *BREAK-ON-SIGNALS*
688 to diagnose and fix failures. (thanks to Nikodemus Siivola)
689 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
690 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
691 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
692 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
693 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
695 * fixed a (fairly theoretical) bug in string printing: if
696 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
697 does not have array-element-type equal to the most general string
699 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
700 function name. (thanks to Nikodemus Siivola)
701 * fixed bug: compile-time format string checker failed on
702 non-closed ~{. (reported by Thomas F Burdick)
703 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
704 don't warn on legal loop code involving more than one aggregate
705 boolean. (thanks to Nikodemus Siivola)
706 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
707 in keyword position to MAKE-PACKAGE was not being reported as
709 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
710 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
712 * fixed bugs 280 and 312: the checking for multiple definitions in a
713 file is less likely to become confused by uses of inline
715 * fixed bug: the #S reader macro performs the keyword coercion
716 specified for slot names. (reported by Kalle Niemitalo)
717 * fixed bug: lambda lists may contain symbols whose names start with
718 & but are not lambda-list-keywords; their occurrence triggers a
720 * fixed bug 321: define-method-combination argument lambda lists do
721 not cause invalid code to be generated when &optional or &aux
722 variables have default values. (reported by Bruno Haible)
723 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
724 have CLOS classes; this appears also to have fixed reported
725 instability in stack exhaustion detection.
726 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
727 MUFFLE-WARNING when no associated restart is present is now
729 * optimization: rearranged the expansion of various defining macros
730 so that each expands into only one top-level form in a
731 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
733 * optimization: used a previously little-used slot in symbols to
734 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
736 * fixed some bugs revealed by Paul Dietz' test suite:
737 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
739 ** Fixed an optimization bug related to inheritance of initforms
740 from local to shared slots.
741 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
742 ** CAST optimizer forgot to flush argument derived type.
743 ** print/read consistency on floats is now orders of magnitude
744 more likely. (thanks also to Bruno Haible for a similar report
746 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
747 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
748 for objects of type REAL. Make it so.
749 ** SXHASH is specified (infelicitously) to respect similarity,
750 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
751 it so. (thanks to Markus Ziegler)
752 ** on the Alpha, the compiler succeeds in compiling functions
753 returning a known number of arguments greater than 63.
754 ** fixed handling of invalid NIL arguments in keyword position
756 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
758 ** provided a readably-printable representation for RANDOM-STATE
760 ** ensured that pathnames, where they have a namestring, always
761 print using #P"..." syntax.
763 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
764 * Support for the forthcoming 2.0 version of the NetBSD kernel
765 running on x86 hardware has been added. (thanks to Perry
766 E. Metzger most immediately, and others for their past work)
767 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
768 current release version; SBCL's OpenBSD support had been broken
769 since about the time of OpenBSD's switch to ELF binary format.)
770 * [placeholder for DX summary]
771 ** user code with &REST lists declared dynamic-extent, under high
772 speed or space and low safety and debug optimization policy.
773 * The manual has been converted to Texinfo format and the debugger
774 chapter from the cmucl manual has been added.
775 * A facility has been added to extract documentation strings from
776 sbcl and store them as Texinfo-formatted snippets for inclusion in
777 the manual (via Texinfo's @include directive)
778 * bug fix: compiler emitted division in optimized DEREF. (thanks for
779 the test case to Dave Roberts)
780 * bug fix: multidimensional simple arrays loaded from FASLs had fill
781 pointers. (reported by Sean Ross)
782 * bug fix: PROFILE output is printed nicely even for large numerical
783 values. (thanks to Zach Beane)
784 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
785 greater than 32 handle EOF correctly.
786 * bug fix: on X86 an immediate argument of the IMUL instruction is
787 correctly printed in disassembly. (thanks to Lutz Euler)
788 * bug fix: class slots in redefined classes preserve their old
789 values. (thanks to Bruno Haible and Nikodemus Siivola)
790 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
791 succeeds. (reported by Marco Baringer)
792 * fixed some bugs revealed by Paul Dietz' test suite:
793 ** READ-SEQUENCE now works on ECHO-STREAMs.
794 ** RATIONALIZE works more according to its specification. (thanks
797 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
798 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
799 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
800 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
801 should matter to you only if you rebind the printer control
802 variables and then find you want different bindings in the
803 debugger than in the ordinary execution of your program.)
804 * The runtime build system has been tweaked to support building
805 (on SPARC/SunOS) using a C compiler which invokes Sun's own
806 assembler and linker. (thanks to Nikodemus Siivola)
807 * Unbound, undefined, undeclared variables now trigger full
808 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
809 more often programmer error than deliberate exploitation of undefined
811 * optimization: the hash algorithm for strings has changed to one
812 that is less vulnerable to spurious collisions. (thanks to Juho
814 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
815 needless bounds checking. (thanks to Juho Snellman)
816 * optimization: implemented multiplication as a modular
817 (UNSIGNED-BYTE 32) operation on the PPC backend.
818 * fixed some bugs revealed by Paul Dietz' test suite:
819 ** ADJUST-ARRAY now copies the datum in a zero rank array if
821 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
822 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
823 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
825 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
827 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
829 ** ECHO-STREAMs no longer attempt to echo the end of file value to
830 their output stream on EOF from read.
831 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
832 have been read to end-of-file.
833 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
835 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
836 description of determination of which consecutive characters
838 ** Printing the "Space" character with escaping on now yields "#\\ ",
839 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
840 ** Reading floating-point numbers with *READ-BASE* set to a number
841 less than 10 works correctly.
842 ** Reading floating-point numbers with *READ-BASE* set to a number
843 more than 10 works correctly.
844 ** Printing with *PRINT-READABLY* targets the standard readtable, not
845 the readtable currently in effect.
847 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
848 * minor incompatible change: parsing of namestrings on a physical
849 (Unix) host has changed; numbers after the final #\. in a
850 namestring are no longer interpreted as a version field. This is
851 intented to be largely invisible to the user, except that the
852 meaning of the namestring "*.*.*" has changed: it now refers to a
853 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
854 should usually be replaced by
855 (make-pathname :name :wild :type :wild :version :wild)
856 with the added benefit that this is more likely to be portable.
857 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
858 option to OPEN now signals an error if the file being opened
859 exists; this may have an impact on existing code.
860 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
861 signals received, on the PowerPC platforms (both Linux and
862 Darwin). (thanks to Pierre Mai for pointing out the location of
864 * several fixes on OS X: The system now builds and runs cleanly on
865 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
866 (thanks to Brian Mastenbrook)
867 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
868 second-resolution: integer multiples of 1/3600 between -24 and 24.
869 (thanks to Vincent Arkesteijn)
870 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
871 type when called with 1 argument; PEEK-CHAR checked type of
872 PEEK-TYPE only after having read first character from a
873 stream. (reported by Peter Graves)
874 * bug fix: the garbage collector now has much better locality
875 behaviour, and in particular no longer treats all memory as being
876 exhausted when many small objects point to each other in a deeply
878 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
879 recognized as being TYPEP their class.
880 * bug fix: the PUSHNEW documentation string has been corrected.
881 (thanks to Vincent Arkesteijn)
882 * bug fix: defaulting of the value for the last of an atypically
883 large number of multiple values being bound was not being
884 performed correctly on the Alpha or PPC platforms
885 * optimization: implemented multiplication as a modular
886 (UNSIGNED-BYTE 32) operation on the x86 backend.
887 * optimization: SEARCH on simple-base-strings can now be open-coded.
888 (see also contrib/compiler-extras.lisp for inspiration for
889 teaching the compiler about the Boyer-Moore algorithm).
890 * value, returned by MAX (and MIN) called with several EQUALP, but
891 not EQL, arguments now does not depend on compiler settings.
892 * fixed some bugs revealed by Paul Dietz' test suite:
893 ** in stack analysis liveness information is propagated from
894 non-local entry points.
895 ** pathwise CAST removing failed when the CAST node did not start
897 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
898 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
900 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
901 host is already defined.
902 ** RENAME-FILE works on streams instead of signalling an internal
904 ** PEEK-CHAR uses the current readtable when determining whether
905 or not a character is whitespace.
906 ** MERGE-PATHNAMES handles the case when the pathname does not
907 specify a name while the default-pathname specifies a version.
908 ** Pathnames now stand a better chance of respecting print/read
910 ** Attempting to use standardized file system operators with a
911 pathname with invalid :DIRECTORY components signals a
913 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
915 ** DIRECTORY on logical pathnames is more correct.
916 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
917 signal a TYPE-ERROR if their argument is not a stream.
918 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
919 designator argument does not designate a stream.
920 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
921 examining the synonym.
922 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
924 ** OPEN and WITH-OPEN-STREAM allow opening streams with
925 element-type larger than ([UN]SIGNED-BYTE 32).
927 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
928 * When built with the :SB-FUTEX feature, threaded builds now take
929 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
930 for faster/more reliable mutex and condition variable support.
931 * Incompatible change (but one you probably shouldn't have been using
932 anyway): the interface and code for arbitrating between multiple
933 threads in the same user session has been redesigned.
934 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
935 error when called without an explicit environment argument.
936 (thanks to Nikodemus Siivola)
937 * bug fix: buffered :DIRECTION :IO streams are less likely to become
938 confused about their position. (thanks to Adam Warner and Gerd
940 * bug fix: Pretty printing backquoted forms with unquotations in the
941 argument list position of various code constructs such as LAMBDA
942 now works correctly. (reported by Paul Dietz)
943 * bug fix: Pretty printing unquotations no longer loses all
944 stream position information.
945 * optimization: performance of string output streams is now less
946 poor for multiple small sequence writes.
947 * optimization: performance of CSUBTYPEP in the presence of complex
948 expressions involving CONS and NOT many times has been improved.
949 (reported by Paul Dietz)
950 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
952 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
953 streams for the REPL, for the debugger, and so on. (thanks to
955 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
956 * fixed some bugs revealed by Paul Dietz' test suite:
957 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
958 is ignored if it is too small, rather than propagating through
959 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
960 ** extremely complex negations of CONS types were not being
961 sufficiently canonicalized, leading to inconsistencies in
963 ** VALUES tranformer lost derived type.
965 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
966 * fixed a bootstrapping bug: the build process no longer assumes
967 that the various BOOLE-related constants have the same value in
968 host and target lisps. (noted by Paul Dietz' test suite on an
969 SBCL binary built from CLISP)
970 * The system can now be dynamically linked on the MIPS platform,
971 which enables dynamic loading of foreign code from Lisp. (thanks
972 to Ralf Baechle for discussions on the MIPS ABI)
973 * The system now records debugging information for its own source
974 files in a filesystem-position-independent manner, relative to
975 the "SYS" logical host.
976 * fixed a compiler bug: MV-LET convertion did not check references
977 to the "max args" entry point. (reported by Brian Downing)
978 * tweaked disassembly notes to be less confident about proclaiming
979 some instruction as an LRA. (thanks to Brian Downing)
980 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
981 now each have their own history, command character, and other
982 characteristics. (thanks to David Lichteblau)
983 * fixed some bugs revealed by Paul Dietz' test suite:
984 ** compiler failure in compiling LOGAND expressions including a
986 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
987 shift greater than 32.
988 ** FLUSH-DEST did not mark blocks for type check regeneration.
989 ** HANDLER-CASE failed to accept declarations in handler clauses
990 in some circumstances.
992 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
993 * New code in contrib/sb-introspect (still probably not entirely
994 stable yet) provides some support for smart Lisp development
995 environments like SLIME.
996 * The conditions signalled for errors occurring when loading .fasl
997 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
998 in a way which should help ASDF recover gracefully.
999 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1000 stuff has been cleaned up. If you code contrib/ stuff, this might
1001 affect you, and you can look at contrib/README, contrib/STANDARDS,
1002 and/or the 0.8.4.27 diff to check.
1003 * In full calls the compiler now does not generate checks for declared
1004 argument types for all arguments.
1005 * various threading fixes
1006 ** and some experimental patches which didn't make it into
1007 the main tree for this release, but which are shipped in
1008 contrib/experimental-thread.patch as a possible fix for some
1009 failures (deadlock, spinning...) in GC-intensive multithreaded
1011 * fixed PPC build problem (source code incompatibility of different
1012 library versions): added offsetof() hackery which attempts to divine
1013 where glibc maintainers put uc_mcontext today
1014 * fixed bug 282: compiler does not trust type assertions while passing
1015 arguments to a full call.
1016 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1017 &OPTIONAL VALUES type parameter.
1018 * fix bug 214: algorithm for noting rejected templates is now more
1019 similar to that of template seletion. (also reported by rydis on
1021 * fixed bug 141b: printing backquoted information readably and prettily
1022 inserts a space where necessary.
1023 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1024 (CHARACTER 10) are now reported as errors, rather than propagated
1025 as unknown types. (reported by piso on #lisp)
1026 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1027 with values NIL and :ERROR. (thanks to Milan Zamazal)
1028 * fixed bug 191c: CLOS now does proper keyword argument checking as
1029 described in CLHS 7.6.5 and 7.6.5.1.
1030 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1031 longer raise an error; further, using a list as a for-as-arithmetic
1032 counter now raises a meaningful error.
1033 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1034 now understood by sequence creation functions such as MAKE-SEQUENCE
1036 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1037 a STRING-INPUT-STREAM.
1038 * compiler enhancement: SIGNUM is now better able to derive the type
1040 * type declarations inside WITH-SLOTS are checked. (reported by
1042 * fixed some bugs revealed by Paul Dietz' test suite:
1043 ** incorrect optimization of TRUNCATE for a positive first
1044 argument and negative second.
1045 ** compiler failure in let-convertion during flushing dead code.
1046 ** compiler failure while deriving type of TRUNCATE on an
1047 interval, containing 0.
1048 ** ASH of a negative bignum by a negative bignum count now returns
1050 ** intersection of CONS types now canonicalizes properly, fixing
1051 inconsistencies in SUBTYPEP.
1053 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1054 * incompatible change: The --disable-debugger command line
1055 option now clobbers the debugger at a more fundamental
1056 level, by redefining #'INVOKE-DEBUGGER instead of by
1057 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1058 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1059 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1060 enter the debugger REPL and then suffer recursive errors
1061 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1062 while under the new system, BREAK in --disable-debugger mode
1063 terminates the system just as an unhandled error would.
1064 * fixed compiler performance when processing loops with a step >1;
1065 * bug fix: DOCUMENTATION now retrieves generic function
1066 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1067 support has been systematized, and now supports the methods
1068 specified by ANSI, along with a default method and a method for
1069 slot documentation. (reported by Nathan Froyd)
1070 * bug fix: effective methods associated with a generic function are
1071 no longer cached over a change of that generic function's method
1072 combination. (reported by Andreas Fuchs)
1073 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1074 names a non-existent directory. (thanks to Andreas Fuchs)
1075 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1076 the PPC platform, lead to stack corruption; this has been fixed.
1077 (reported by Rainer Joswig)
1078 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1079 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1080 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1081 arithmetic in general, is now much more reliable on the PPC
1083 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1084 platform now returns the right answer.
1085 * optimization: restored some effective method precomputation in
1086 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1087 precomputation is now tunable.
1088 * optimization: compiler-internal data structure use has been
1089 reviewed, and changes have been made that should improve the
1090 performance of the compiler by about 20%.
1091 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1092 simple-base-strings and simple-bit-vectors is improved.
1093 * optimization: the optimization of 32-bit logical and arithmetic
1094 functions introduced in version 0.8.3 on the x86 has been
1095 implemented on the mips, ppc and sparc platforms; an
1096 implementation of the same facility, but for 64-bit arithmetic,
1097 has been added for the alpha.
1098 * microoptimization: the compiler is better able to make use of the
1099 x86 LEA instruction for multiplication by constants.
1100 * bug fix: in some situations compiler did not report usage of
1101 generic arithmetic in (SPEED 3) policy.
1102 * bug 145b fix: compiler used wrong type specifier while converting
1103 MEMBER-types to numeric.
1104 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1106 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1107 index argument now works properly on the Alpha platform.
1108 * bug fix: floating point exception treatment on the Alpha platform
1110 * bug fix: FILE-POSITION works much better on string input and
1111 output streams. (thanks to Nikodemus Siivola)
1112 * bug fix: many threading/garbage collection symptoms sorted.
1113 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1114 might be pseudo-atomic.
1115 * internal change: Stopping for GC is now done with signals not
1116 ptrace. GC is now done in whichever thread wanted it, instead of
1118 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1120 * bug fix: result form in DO is not contained in the implicit
1122 * incompatible change: ICR structure is changed; the value part of
1123 CONTINUATION is now called LVAR; corresponding functions are
1124 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1125 * added type deriver for ISQRT (thanks to Robert E. Brown).
1126 * bug fix: better support for loading from the command line when an
1127 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1129 * fixed some bugs revealed by Paul Dietz' test suite:
1130 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1131 ** ROUND and FROUND now give the right answer when given very
1132 small float arguments.
1133 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1135 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1136 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1137 ** LCM should return a non-negative integer.
1138 ** PARSE-INTEGER returned the index of a terminator instead of the
1139 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1140 ** PARSE-INTEGER returned an incorrect index being applied to a
1142 ** LCM with two arguments of 0 returns 0 rather than signalling
1144 ** unsigned addition of a 32-bit constant with the high bit set no
1145 longer causes an internal compiler error on the x86.
1146 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1147 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1148 with negative last argument.
1149 ** byte specifiers with zero size and position no longer cause
1150 an error during type derivation.
1151 ** bignum multiplication on the Alpha platform now returns the
1153 * porting: The system now builds on SuSE AMD64, although it still
1154 generates a 32-bit binary.
1155 * .fasl file incompatibility: The fasl file version number has
1156 been incremented (because of the changes to internal compiler
1157 data structures referred to above).
1159 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1160 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1161 more accurately, on the Darwin kernel running on PowerPC hardware.
1162 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1163 * Compiler code deletion notes now signal a condition of type
1164 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1165 an associated MUFFLE-WARNING restart.
1166 * The compiler now performs limited argument count validation of
1167 constant format strings in FORMAT, and where appropriate in ERROR,
1168 CERROR and WARN. (thanks to Gerd Moellmann)
1169 * New ASDF-INSTALL contrib can be used for automatic download and
1170 installation of third-party Lisp code from CCLAN or other sites
1172 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1173 which forces another thread to execute a function supplied by the
1175 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1176 now accept and act upon their :ELEMENT-TYPE keyword argument.
1177 (reported by Martin Atzmueller, Edi Weitz)
1178 * bug fix: FILE-POSITION now accepts position designators up to
1179 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1180 is the greater. (thanks to Patrik Nordebo)
1181 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1182 Kalle Olavi Niemitalo)
1183 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1184 printable. (reported by Eric Marsden)
1185 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1186 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1187 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1188 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1189 (thanks to Stig Sandoe)
1190 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1191 UNREAD-CHAR now correctly includes the unread character in the
1192 target sequence. (thanks to Gerd Moellmann)
1193 * bug fix in threaded builds: the system can now be suspended and
1194 resumed by shell job control with minimal disruption.
1195 * bug fixes in times and timezones >2038AD
1196 * better handling of "where is GNU make?" problem in build scripts
1197 (thanks to Nikodemus Siivola)
1198 * new optimization: inside a named function any reference to a
1199 function with the same name is considered to be a self-reference;
1200 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1201 optimization quality.
1202 * new optimization on x86: logical functions and + now have
1203 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1204 used when the result is truncated to 32 bits.
1205 * VALUES declaration is partially enabled.
1206 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1207 * fixed some bugs revealed by Paul Dietz' test suite:
1208 ** The system now obeys the constraint imposed by
1209 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1210 types form a lattice under type intersection.
1211 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1212 ** ASSOC now ignores NIL elements in an alist.
1213 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1214 and (1+ MOST-POSITIVE-FIXNUM) answers.
1215 ** The addition of a method with invalid qualifiers to a generic
1216 function does not cause an error to be signalled immediately;
1217 a warning is signalled, and the error is generated only on
1218 calling the generic function.
1219 * changed .fasl file version number, in part to add type codes for
1220 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1221 obscure ANSI requirements
1223 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1224 * fixed bug 148: failure to inline-expand a local function left
1225 garbage, confusing the compiler.
1226 * fixed bugs 3cd: structure slot readers perform type check if the
1227 slot can have an invalid value (i.e. it is either not initialized
1228 or can be written with a less specific slot writer).
1229 * bug fix: the compiler now traps array references to elements off
1230 the end of an array; previously, the bounds checking in some
1231 circumstances could go off-by-one.
1232 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1234 * type declarations for array element types now obey the description
1235 on the CLHS page "Declaration TYPE", as per discussions on
1236 sbcl-help around 2003-05-08. This means that a declaration
1237 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1238 declaration, all references to BAR will be asserted or assumed
1239 (with THE, so dependent on compiler policy) to involve objects of
1240 type FOO. Note that no such declaration is implied in
1241 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1242 * declared types of functions from the "Conditions"
1243 chapter. (reported by Paul Dietz)
1244 * bug fix: CERROR accepts a function as its first argument.
1245 * bug fix: NTH an NTHCDR accept a bignum as index
1246 arguments. (reported by Adam Warner)
1247 * optimization: character compare routines now optimize comparing
1248 against a constant character. (reported by Gilbert Baumann)
1249 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1250 argument now works properly on the MIPS platform.
1251 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1252 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1253 declared to have a functional type, does not cause an error
1255 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1256 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1257 * fixed bug: if last continuation of a deleted block has a
1258 destination, this destination should be deleted too. (reported by
1260 * fixed a bug in the bootstrap process: the host compiler's values
1261 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1262 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1263 test case from Patrik Nordebo)
1264 * improved the ability of the disassembler on the PPC platform to
1265 provide helpful disassembly notes.
1266 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1267 the class in more cases than previously.
1268 * bug fix: FILE-POSITION now understands :START and :END for
1269 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1270 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1271 8) data. (thanks to David Lichteblau)
1272 * bug fix: it is possible to add a method to a generic function
1273 without lambda list.
1274 * bug fix: reader failed to signal END-OF-FILE inside an
1275 object representation. (reported by Nikodemus Siivola)
1276 * fixed some bugs revealed by Paul Dietz' test suite:
1277 ** LAST and [N]BUTLAST should accept a bignum.
1278 ** condition slot accessors are methods.
1279 ** (VECTOR NIL) is a subtype of STRING.
1281 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1282 * minor incompatible change: some nonsensical specialized lambda
1283 lists (used in DEFMETHOD) which were previously ignored now signal
1285 * minor incompatible change: the system is now aware of the types of
1286 variables in the COMMON-LISP package, and will signal errors for
1287 most violations of these type constraints (where previously they
1288 were silently accepted).
1289 * minor incompatible change: COMPILE-FILE now uses the freedom
1290 afforded (ANSI 3.2.2.3) to use derived function types for
1291 functions defined in the same file. This also permits the system
1292 to warn on static type mismatches and function
1293 redefinition. (Currently it does not work with high DEBUG level.)
1294 * minor incompatible change: VALUES declaration is disabled.
1295 * When issuing notes, the compiler now signals a condition of type
1296 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1297 restart for use in user handlers. It is expected that the
1298 COMPILER-NOTE condition will eventually become a condition
1299 supertype to a hierarchy of note types, which will then be
1300 handleable in a similar fashion. However, at the moment, no such
1301 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1302 but it's an implementation detail, not a classification for the
1304 * Changes in type checking closed the following bugs:
1305 ** type checking of unused values (192b, 194d, 203);
1306 ** template selection based on unsafe type assertions (192c, 236);
1307 ** type checking in branches (194bc).
1308 * A short form of VALUES type specifier has ANSI meaning (it has
1309 increased the number of situations when SBCL cannot perform type
1311 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1312 keywords or constants is permissible.
1313 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1314 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1315 classes. (thanks to Antonio Martinez)
1316 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1317 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1318 argument to a pretty-print stream. Instead, it leaves any such
1319 operations to DESCRIBE-OBJECT methods.
1320 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1322 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1323 lambda lists. (thanks to David Lichteblau)
1324 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1325 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1326 (thanks to Gerd Moellmann)
1327 * fixed bug 246: increased compilation speed of long
1328 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1330 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1332 * DEFCONSTANT now throws a condition of type
1333 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1334 constant to a non-EQL value; CONTINUE and ABORT restarts
1335 respectively change and preserve the value.
1336 * fixed bug 63: The code walker, part of the implementation of CLOS,
1337 is now better at handling symbol macros.
1338 * bug fix: There is no longer an internal implementation type named
1339 CL:LENGTH. (reported by Raymond Toy)
1340 * bug fix: In macro-like defining macros/special operators the
1341 implicit block does not enclose the lambda list.
1342 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1343 enclosed in parentheses) are not suitable as type specifiers, and
1344 their use properly signals an error now.
1345 * bug fix: An argument count mismatch for a type specifier in code
1346 being compiled no longer causes an unhandled error at compile
1347 time, but signals a compile-time warning.
1348 * fixed simple vector readable printing
1349 * bug fix: DESCRIBE takes more care over whether the class
1350 precedence list slot of a class is bound before accessing it.
1351 (reported by Markus Krummenacker)
1352 * bug fix: FORMATTER can successfully compile pretty-printer format
1353 strings which use variants of the ~* directive inside.
1354 * bug fix: SEARCH now applies its TEST predicate to the elements of
1355 the arguments in the correct order. (thanks to Wolfhard Buss)
1356 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1357 in type inference. (thanks to Robert E. Brown)
1358 * bug fix: Reading in symbols with an explicit package name of ""
1359 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1360 * fixed some bugs revealed by Paul Dietz' test suite:
1361 ** NIL is now allowed as a structure slot name.
1362 ** Arbitrary numbers, not just REALs, are allowed in certain
1363 circumstances in LOOP for-as-arithmetic clauses.
1364 ** Multiple class redefinitions before slot access no longer
1365 causes a type error.
1366 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1367 association between the name and a class.
1368 ** Generic functions with non-standard method-combination and over
1369 five methods all of which return constants no longer return NIL
1370 after the first few invocations. (thanks to Gerd Moellmann)
1371 ** CALL-NEXT-METHOD with no arguments now passes the original
1372 values of the arguments, even in the presence of assignment.
1373 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1375 ** DISASSEMBLE works with closures and funcallable instances.
1376 ** ADD-METHOD now returns the generic function, not the new method.
1377 ** FIND-METHOD signals an error if the lengths of the specializers
1378 is incompatible with the generic function, even if the ERRORP
1380 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1381 which its argument is a member.
1382 ** DEFCLASS only redefines the class named by its class-name
1383 argument if that name is the proper name of the class;
1384 otherwise, it creates a new class.
1385 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1386 of the UNBOUND-SLOT condition to the name of the slot.
1387 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1388 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1389 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1390 SLOT-MAKUNBOUND in the specified fashion.
1392 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1393 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1394 cross-compilation host. As a consequence, we can now bootstrap our
1395 way up to SBCL starting with a bare gcc toolchain and human-readable
1396 source code (first the source to CLISP, then the source to SBCL).
1397 * A contributed module containing a partial implementation of the
1398 simple-streams interface has been included. (thanks to Rudi
1400 * A contributed module implementing the RFC1321 Message Digest
1401 Algorithm, known as MD5, has been included.
1402 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1403 no longer has any effect, as the code controlled by this feature
1404 has been deleted. (As far as we know, no-one has ever built using
1405 this feature, and its semantics were confused in any case).
1406 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1407 work on conditions (as required by the ANSI specification),
1408 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1409 expected behaviour on conditions. Users should note, however,
1410 that such behaviour is not required by the ANSI specification,
1411 and so use of this behaviour may render their code unportable.
1412 * fixed some bugs revealed by Paul Dietz' test suite:
1413 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1415 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1416 ** SXHASH on condition objects no longer returns NIL.
1417 ** :ALLOCATION :CLASS slots are better treated; their values are
1418 updated on class redefinition, and initforms inherited from
1419 superclasses are applied.
1420 ** REMOVE-METHOD returns its generic function argument even when
1421 no method was removed.
1422 ** SHARED-INITIALIZE now initializes the values of the requested
1423 slots, including those with :ALLOCATION :CLASS.
1424 ** ALLOCATE-INSTANCE now works on structure classes defined via
1425 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1427 ** SLOT-EXISTS-P now works on conditions, as well as structures
1429 ** MAKE-LOAD-FORM now has the required methods on
1430 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1431 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1432 keyword, and now implements the SLOT-NAMES argument.
1433 ** methods with &OPTIONAL arguments no longer allow too many
1434 arguments to be passed in the call without error.
1435 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1436 option is consistent with the required arguments of the generic
1437 function lambda list.
1438 * bug fix: REQUIRE accepts a string designator. (Thanks to
1440 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1441 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1442 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1443 * bug fix: The compiler now checks for duplicated variables in macro
1445 * bug fix: SETQ on globals returns the correct value.
1446 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1447 not a COMPILER-ERROR (followed by some other strange error on
1448 choosing the CONTINUE restart).
1449 * bug fix: make.sh and friends are now more consistent in the way that
1450 they look for GNU "make".
1452 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1453 * experimental native threads support (on x86 Linux >=2.4 only).
1454 This is not compiled in by default: you need to add :SB-THREAD to
1455 the target features. See the "Beyond ANSI" chapter of the manual
1457 * fix for longstanding nonANSIism: The old distinction between
1458 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1459 The return value from CL:FIND-CLASS is now a CLOS class, and
1460 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1461 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1462 * An interface to the MetaObject Protocol, as described in Kiczales,
1463 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1464 MIT Press, 1991, is available from the SB-MOP package.
1465 * incompatible change: the SB-PCL package should now be considered
1466 a private implementation detail, and no longer a semi-private MOP
1468 * minor incompatible change: due to rearrangement for threads, the
1469 control stack and binding stack are now allocated at arbitrary
1470 addresses instead of being hardcoded per-port. Users affected by
1471 this probably have to be doing advanced things with shared
1472 libraries, and will know who they are.
1473 * minor incompatible change: Previously, all --eval forms used were
1474 processed with READ before any of them were processed with EVAL.
1475 Now each --eval form is processed with both READ and EVAL before
1476 the next --eval form is processed. (Thus package operations like
1477 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1478 work as the user might reasonably expect.)
1479 * minor incompatible change: *STANDARD-INPUT* is now only an
1480 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1482 * minor incompatible change: Y-OR-N-P is now character-oriented,
1483 not line oriented. Also, YES-OR-NO-P now works without errors.
1484 (thanks to Antonio Martinez)
1485 * sb-aclrepl module improvements: an integrated inspector, added
1486 repl features, and a bug fix to :trace command.
1487 * Known functions, which cannot be open coded by the backend, are
1488 considered to be able to check types of their arguments. (fixing
1489 a bug report by Nathan J. Froyd)
1490 * fixed a bug in computing method discriminating functions: It is
1491 now possible to define methods specialized on classes which have
1492 forward-referenced superclasses. (thanks to Gerd Moellmann)
1493 * fixed evaluation order in optional entries (reported by Gilbert
1495 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1496 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1497 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1498 required-by-AMOP NAME argument, as well as CLASS and
1499 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1500 * fixed bug 20: DEFMETHOD can define methods using names that are
1501 not the proper names of classes to designate class specializers.
1502 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1503 Unix file descriptors, instead of blowing up. (thanks to Antonio
1505 * Garbage collection refactoring: user-visible change is that a
1506 call to the GC function during WITHOUT-GCING will not do garbage
1507 collection until the end of the WITHOUT-GCING. If you were doing
1508 this you were probably losing anyway.
1509 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1510 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1511 TYPEP the latter but not the former.
1512 * The compiler issues a full WARNING for calls to undefined functions
1513 with names from the CL package.
1514 * MAP-INTO for a vector destination is open coded. (reported by
1515 Brian Downing on c.l.l)
1516 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1517 documentation string.
1518 * fixed some bugs revealed by Paul Dietz' test suite:
1519 ** COPY-ALIST now signals an error if its argument is a dotted
1521 ** Condition slots are now accessed more correctly in the presence
1522 of multiple initargs for a given slot.
1523 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1524 exclude restarts of the same name associated with a different
1526 ** DEFCLASS of forward-referenced classes with another
1527 forward-referenced class in the superclasses list no longer
1529 ** Condition slots are now initialized once each, not multiple
1530 times. (thanks to Gerd Moellmann)
1531 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1533 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1534 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1535 arguments contain duplicated elements.
1536 ** RESTART-CASE understands local macros.
1537 ** RESTART-CASE associates exactly its own restarts with a condition.
1538 ** ENDP in safe mode checks its argument to be of type LIST.
1539 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1540 in question is unbound.
1541 ** Optimized MAKE-INSTANCE functions no longer cause internal
1542 assertion failures in the presence of duplicate initargs.
1543 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1544 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1545 propagated even in the ctor optimized implementation of
1547 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1549 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1550 * a better implementation of SXHASH on (simple) bit vectors,
1551 measured both in execution speed and in distribution of results
1552 over the positive fixnums, has been installed. Likewise, a better
1553 implementation of EQUAL for simple bit vectors is now available.
1554 * fixed CEILING optimization for a divisor of form 2^k.
1555 * fixed bug 240 (emitting extra style warnings "using the lexical
1556 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1557 by Antonio Martinez)
1558 * fixed SXHASH, giving different results for NIL depending on type
1559 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1560 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1561 variable. (found by Rolf Wester)
1562 * SBCL does not ignore type declarations for special
1563 variables. (reported by rif on c.l.l 2003-03-05)
1564 * some bug fixes in contrib/sb-aclrepl/
1565 * fixed some bugs revealed by Paul Dietz' test suite:
1566 ** a bug in the CONS type specifier, whereby the CAR and CDR
1567 types got intertwined, has been fixed;
1568 ** the type system is now able to reason about the interaction
1569 between INTEGER and RATIO types more completely;
1570 ** APPEND, [N]REVERSE and NRECONC check that those their
1571 arguments, which must be proper lists, are really so;
1572 ** An array specialized to be unable to hold elements has been
1573 implemented, as required -- yes, really -- by ANSI;
1574 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1575 on malformed property lists;
1577 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1578 * incompatible packaging change: in line with Unix convention,
1579 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1580 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1581 * REQUIRE and PROVIDE are now optionally capable of doing something
1582 useful. See the documentation string for REQUIRE.
1583 * infrastructure for a managed SBCL contrib system: contributed
1584 modules in this release include:
1585 ** the ASDF system definition facility;
1586 ** an interface to the BSD Sockets API;
1587 ** an ACL-like convenience interface to the repl;
1588 (thanks to Kevin Rosenberg)
1589 ** an implementation of ROTATE-BYTE, with efficient implementation
1591 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1592 gives a closer approximation to the right answer than previously.
1593 (thanks to Raymond Toy)
1594 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1595 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1596 optional environment arguments, as required by ANSI.
1597 * fixed bugs in other functions taking environment objects, allowing
1598 calls with an explicit NIL environment argument to be compiled
1600 * fixed bug 228: primary return values from
1601 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1602 COMPILE or FUNCTION.
1603 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1604 :TYPE will no longer signal errors on innocuous objects.
1605 * fixed bug 231b: SETQ is better at respecting type declarations in
1606 the lexical environment.
1607 * fixed a bug in DEFCLASS: classes named by symbols with no or
1608 unprintable packages can now be defined.
1609 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1610 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1611 * fixed compiler failure related to checking types of functions.
1612 (reported by Robert E. Brown)
1613 * the compiler is now much more consistent in its error-checking
1614 treatment of bounding index arguments to sequence functions: in
1615 (SAFETY 3) code, errors will be signalled in almost all cases if
1616 invalid sequence bounding indices are passed to functions defined
1617 by ANSI to operate on sequences.
1618 * fixed a bug in the build procedure: documentation of SBCL-specific
1619 packages is now preserved and available in the final Lisp image.
1620 * lifted FDEFINITION lookup out of loops in the implementation of
1621 many list operations. (thanks to Robert E. Brown)
1622 * fixed a bug in the reader: the #n# reader macro now works for
1623 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1624 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1625 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1626 E. Brown and Tony Martinez respectively)
1627 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1628 SLOT-BOUNDP: in particular, it is now able to optimize them much
1629 better, and is now not vulnerable to having packages renamed.
1630 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1631 always called when a slot is not present in an instance. (thanks
1633 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1634 longer signal an error on unreading a character following EOF on
1635 the previous constituent stream. (thanks to Tony Martinez)
1636 * fixed some bugs revealed by Paul Dietz' test suite:
1637 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1638 not just nonnegative fixnums;
1639 ** the logical bit-array operators such as BIT-AND now accept an
1640 explicit NIL for their "opt-arg" argument (to indicate a
1641 freshly-consed result bit-array);
1642 ** ELT now signals an error on an invalid sequence index in safe
1644 ** the type system is now cleverer about negations of numeric
1645 types, and consequently understands the BIGNUM and RATIO types
1647 ** the type system is now cleverer about the interaction between
1648 INTEGER and RATIO types: while bugs still remain, many more
1649 cases are accurately computed;
1650 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1651 if it is in the last clause;
1652 ** CONSTANTLY now correctly returns a side-effect-free function in
1654 ** DECLARE is no longer treated as a special-operator; in
1655 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1656 * incremented fasl file version number due to the change in the
1657 DEFSTRUCT-SLOT-DESCRIPTION structure.
1659 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1660 * minor incompatible change: code processed by the "interpreter" or
1661 EVAL now has a compilation optimization policy of (DEBUG 2)
1662 (changed from (DEBUG 1)) to improve debuggability of interactive
1663 development, and to allow the use of the debug RETURN command in
1665 * an experimental implementation of the RETURN command for the
1666 debugger has been included. (thanks to Frederik Kuivinen)
1667 * fixed bug 62: constraints were not propagated into a loop.
1668 * fixed bug in embedded calls of SORT (reported and investigated by
1670 * fixed some bugs revealed by Paul Dietz' test suite:
1671 ** printing and reading of arrays with some dimensions having
1672 length 0 (thanks to Gerd Moellmann);
1673 ** BOA constructor with &AUX argument without a default value does
1674 not cause a type error;
1675 ** CONSTANTP now returns true for all self-evaluating objects.
1677 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1678 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1679 accessors that are related by inheritance, as specified in the
1680 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1681 Valtteri Vuorikoski)
1682 * The compiler is now able to inline functions that were defined in
1683 a complex lexical environment (e.g. inside a MACROLET).
1684 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1686 * Support for the upcoming FreeBSD-5.0 release has been included.
1687 (thanks to Dag-Erling Smorgrav)
1688 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1689 effect when it is not in a toplevel context.
1690 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1691 stand a better chance of being correct. (thanks to Gerd
1693 * fixed bug in COERCE, which now signals an error on coercing a
1694 rational to a bounded real type which excludes the expected
1696 * The compiler is now able to derive types more accurately from the
1697 COERCE and COMPILE functions.
1698 * fixed bug 223: functional binding is considered to be constant
1699 only for symbols in the CL package.
1700 * fixed bug 231: SETQ did not check the type of a variable being set
1701 (reported by Robert E. Brown)
1702 * A new optimization for MAKE-INSTANCE has been included, fixing
1703 various bugs (including relating to :ALLOCATION :CLASS slots and
1704 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1706 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1707 ** As required by ANSI, LOOP now disallows anonymous collection
1708 clauses such as COLLECT I in conjunction with aggregate boolean
1709 clauses such as THEREIS (= I 1);
1710 ** LOOP now signals an error when any variable is reused in the
1711 same loop (including the potentially useful construct analogous
1712 to WITH A = 1 WITH A = (1+ A);
1713 ** IT is only a special loop symbol within the first clause of a
1714 conditional loop clause;
1715 ** LOOP with a typed iteration variable over a hashtable now
1716 signals a type error iff it should.
1717 * fixed some other bugs revealed by Paul Dietz' test suite:
1718 ** FILE-STREAM now names the class previously known as FD-STREAM;
1719 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1720 argument) no longer signals an error;
1721 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1722 of the slot symbol, rather than using the current package
1723 ((:CONC-NAME "") continues to intern the slot's name in the
1725 * incremented fasl file version number, because of the incompatible
1726 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1727 of the new implementation of DEFINE-COMPILER-MACRO.
1729 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1730 * Support for building SBCL for MIPS platforms running in
1731 little-endian mode has now been checked in, and basic
1732 functionality on said platforms verified.
1733 * minor incompatible change: PCL now records the pathname of a file
1734 in which methods and the like are defined, rather than its
1736 * minor incompatible change: TRUENAME now considers the truename of
1737 a file naming a directory to be the pathname with :DIRECTORY
1738 component indicating that directory.
1739 * minor incompatible change: a NAMED clause in the extended form of
1740 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1741 reason for the previous behaviour is unclear.
1742 * more systematization and improvement of CLOS and MOP conformance
1743 in PCL (thanks to Gerd Moellman and Pierre Mai):
1744 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1746 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1747 primary methods with no specializers;
1748 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1750 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1751 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1752 FORWARD-REFERENCED-CLASSes; error reporting on
1753 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1755 ** SXHASH on CLOS instances now uses a slot internal to the
1756 instance to return different numbers on distinct instances,
1757 while preserving the same return value through invocations of
1759 ** DEFMETHOD signals errors when methods with longer incongruent
1760 lambda lists are added to generic functions;
1761 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1762 CLASS, as specified in AMOP;
1763 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1764 on the order returned by the primary method for classes of
1765 class STANDARD-CLASS;
1766 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1767 * fixed some bugs shown by Paul Dietz' test suite:
1768 ** DOLIST puts its body in TAGBODY;
1769 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1771 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1772 value producing form;
1773 ** if more variables are given to PROGV than values, extra
1774 variables are bound and made to have no value;
1775 ** NSUBSTITUTE on list arguments gets the right answer with
1777 ** ELT signals an error of type TYPE-ERROR when the index argument
1778 is not a valid sequence index;
1779 ** LOOP signals (at macroexpansion time) an error of type
1780 PROGRAM-ERROR when duplicate variable names are found;
1781 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1782 ** FUNCALL of special-operators now cause an error of type
1784 ** PSETQ now works as required in the presence of side-effecting
1785 symbol-macro places;
1786 ** NCONC accepts any object as its last argument;
1787 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1789 ** loop-for-as-package does not require a package to be explicitely
1791 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1793 * fixed bug 166: compiler preserves "there is a way to go"
1794 invariant when deleting code.
1795 * fixed bug 172: macro lambda lists with required arguments after
1796 &REST arguments now cause an error to be signalled. (thanks to
1798 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1799 bug. (thanks to Matthew Danish)
1800 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1802 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1803 arguments are lexically rebound. (thanks to Gerd Moellmann and
1805 * fixed bug 194: error messages are now more informative when there
1806 is no primary method applicable in a call to a generic
1807 function. (thanks to Gerd Moellmann)
1808 * fixed bug in command line argument checking (thanks to Julian
1810 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1811 (thanks to Matthew Danish)
1812 * incremented fasl file version number, because of the
1813 SXHASH-related changes in the layout of CLOS data structures
1815 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1816 * minor incompatible change: The runtime (the Unix executable named
1817 "sbcl") is now much pickier about the .core files it will load.
1818 Essentially it now requires .core files to descend from the same
1819 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1820 as the runtime does. (The intent is to prevent the crashes which
1821 can occur, and which can even be reported as mysterious failures,
1822 when people patch the sources or change the build parameters
1823 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1824 sbcl and .core files.)
1825 * fixed bug: VALUES-LIST is no longer optimized away.
1826 * fixed bug 142: The FFI conversion of C string values to Lisp
1827 string values no longer conses excessively. (thanks to Nathan
1828 Froyd porting Raymond Toy's fix to CMU CL)
1829 * began to systematize and improve MOP conformance in PCL (thanks to
1830 Nathan Froyd, Gerd Moellman and Pierre Mai):
1831 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1833 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1834 ** FINALIZE-INHERITANCE is now called on class finalization;
1835 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1836 argument precedence order.
1837 * fixed bug 202: The compiler no longer fails on functions whose
1838 derived types contradict their declared type.
1839 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1840 so it can be non-toplevel.
1841 * The fasl file version number has changed (because of the new
1842 implementation of DEFMACRO).
1843 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1844 safe code, that any length requirement by their type-specifier
1845 argument is valid. The exceptions to this are described in bug
1847 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1848 functions now check the types of their inputs as required by ANSI.
1849 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1850 introduced symbol is DECLAREd to be SPECIAL.
1851 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1852 * fixed bug: :COUNT argument to sequence functions may be negative
1853 * fixed bug: body of DO-SYMBOLS may contain declarations
1854 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1855 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1856 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1857 symbol macro only once
1858 * fixed printing of call frame when argument list is unavailable
1859 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1860 * compiler no longer signals WARNING on unknown keyword
1863 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1864 * A beta-quality port to the mips architecture running Linux,
1865 based on the old CMUCL backend, has been made. It has been tested
1866 on a big-endian kernel, and works sufficiently well to be able to
1867 rebuild itself; it has not been tested in little-endian mode.
1868 * fixed an inconsistency between gencgc.c and purify.c which made
1869 dumping/loading .core files unreliable
1870 * fixed bug 120a: The compiler now deals correctly with IFs where
1871 the consequent is the same as the alternative, instead of
1872 misderiving the return type. (thanks to Alexey Dejneka)
1873 * fixed bug 113: Logical pathnames are now dumpable (the logical
1874 host is resolved at load-time, throwing an error if it is not
1876 * fixed bug 174: FORMAT's error message is slightly clearer when a
1877 non-printing character is used in a format directive.
1878 * fixed several bugs in compiler checking of type declarations, i.e.
1879 violations of the Python "declarations are assertions" principle
1880 (thanks to Alexey Dejneka)
1881 * fixed several bugs in PCL's error checking (thanks to Gerd
1883 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1885 * fixed bug in compilation of functions as first class values
1886 (thanks to Antonio Martinez-Shotton)
1887 * The compiler's handling TYPE-ERRORs which it can prove will
1888 inevitably happen at runtime has been cleaned up and corrected
1889 in several ways. (thanks to Alexey Dejneka)
1890 * improved argument type checking for various basic arithmetic
1891 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1892 lavished on them in the past that they can be compiled in many
1893 ways in different special cases
1894 * fixed bug 181: compiler checks validity of user supplied type
1896 * cleaned up code flushing in optimization: Function calls which
1897 should signal errors for safety purposes (e.g. which ANSI says
1898 should signal errors when their arguments are of incorrect type)
1899 are no longer optimized away.
1900 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1901 * incremented fasl file version number, because changes in the
1902 implementation of sequence functions like COERCE caused
1903 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1904 in old inline expansions) to become undefined. (Actually these
1905 changes were later undone, so we might very well be binary
1906 compatible with 0.7.7 after all, but leaving the version number
1907 incremented seemed like the simplest and most conservative
1910 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1911 * An alpha-quality port to the parisc architecture running Linux,
1912 based on the old CMUCL backend, has been made. This, even more so
1913 than the other backends, should be considered still a work in
1914 progress; known problems include that the Linux kernel in 64-bit
1915 mode does not propagate the correct sigcontext structure to
1916 userspace, and consequently SBCL on a parisc64 kernel will not
1918 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1919 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1920 suppressed.) Also now that the compiler is looking at declarations
1921 in the environment, it checks optimization declarations as well,
1922 and suppresses inlining when (> DEBUG SPEED).
1923 * More fixes have been made to treatment of floating point exception
1924 treatment and other Unix signals. In particular, floating point
1925 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1926 * The detection and handling of control stack exhaustion (infinite
1927 or very deeply nested recursion) has changed. Stack exhaustion
1928 detection is now done by write-protecting pages at the OS level
1929 and applies at all optimization settings; when found, a
1930 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1931 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1932 be caught using IGNORE-ERRORS.
1933 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1934 correctly when transforming recursive LABELS functions to LETs.
1935 (thanks to Alexey Dejneka)
1936 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1937 that are names of constants or global variables.
1938 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1939 alien routines with docstrings.
1940 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1941 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1943 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1944 (thanks to Eric Marsden)
1945 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1946 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1947 * Bug fix: The SPARC backend can now compile functions involving
1948 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1949 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1950 to a functions expecting a PATHNAME-DESIGNATOR.
1951 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1952 lambda lists. (thanks to Alexey Dejneka)
1953 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1954 functions so that EXPORTing the name of the class doesn't cause
1955 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1956 to Antonio Martinez for reporting this)
1957 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1958 constant-folded without failing an assertion. (thanks to Einar
1959 Floystad Dorum for reporting this)
1960 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1961 types are now dealt with more correctly by the compiler.
1962 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1963 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1964 behaviour analogous to RENAME-FILE. This puts its behaviour more
1965 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1967 * The fasl file version number has changed again. (because of the
1968 bug fix involving the names of PCL MAKE-INSTANCE functions)
1970 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1971 * bug fix: Floating point exceptions are treated much more
1972 consistently on the x86/Linux and PPC/Linux platforms.
1973 * Array initialization with :INITIAL-ELEMENT is now much faster for
1974 cases when the compiler cannot open code the array creation, but
1975 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1976 array accesses have also seen a speed increase.
1977 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1978 specified. (This was at the root of some bad interactions between
1979 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1981 * bug fix: Internal error arguments for undefined functions are now
1982 computed correctly on the PPC/Linux platform.
1983 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1984 Raymond Toy's patch for CMU CL)
1985 * Support for the Solaris 9 operating environment has been included
1986 (thanks to Daniel Merritt)
1987 * A very ugly but hopefully complete draft of the missing FFI chapter
1988 of the manual has been created by reformatting the corresponding
1989 CMU CL manual chapter into (currently very ugly and incoherent)
1990 DocBook and bringing it up to date for SBCL behavior. Thus, the
1991 manual is now essentially complete, at least by my extreme
1992 once-and-only-once standards, whereby it's acceptable to refer to
1993 the doc strings of SB-EXT functions as the primary documentation.
1994 * The fasl file version number has changed again, due to cleanup of
1995 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1996 is no longer a static symbol.)
1998 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1999 * SBCL now builds with OpenMCL (version 0.12) as the
2000 cross-compilation host; also, more progress has been made toward
2001 bootstrapping under CLISP.
2002 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2004 * bug 158 fixed: The compiler can now deal with integer loop
2005 increments different from 1; fixing this turned out also to fix
2007 * bug 169 fixed: no more bogus warnings about using lexical bindings
2008 despite the presence of perfectly good SPECIAL declarations (thanks
2009 to David Lichteblau)
2010 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2011 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2012 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2014 * bug 180 fixed: Method combination specifications no longer ignore
2015 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2016 * bug fix: Structure type predicate functions now check their argument
2017 count as they should.
2018 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2019 correctly. (thanks to Pierre Mai)
2020 * minor incompatible change: The --noprogrammer option is deprecated
2021 in favor of the new --disable-debugger option, which is very similar.
2022 (The major difference is that it takes effect at a slightly different
2023 time at startup, causing handling of errors in --sysinit and
2024 --userinit files will be affected differently.) The
2025 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2026 been added to allow this functionality to be controlled from ordinary
2027 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2028 maintainers, who might want to run non-interactive scripts to
2029 build SBCL cores which will later be used interactively.)
2030 * minor incompatible change: The LOAD function no longer, when given
2031 a wild pathname to load, loads all files matching that pathname.
2032 Instead, an error of type FILE-ERROR is signalled.
2034 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2035 * bug 147 fixed: The compiler preserves its block link/count
2036 invariants more correctly now so that it doesn't crash. (thanks
2038 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2040 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2041 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2042 no-value-for-this-character as NIL (as specified by ANSI).
2043 * HOST-NAMESTRING on physical pathnames now returns a string that is
2044 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2045 (thanks to Christophe Rhodes)
2046 * The Alpha port handles icache flushing more correctly. (thanks to
2048 * More progress has been made toward bootstrapping under CLISP. (thanks
2049 to Christophe Rhodes)
2050 * The fasl file format has changed again, because dynamic loading
2051 on OpenBSD (which has non-ELF object files) motivated some cleanups
2052 in the way that foreign symbols are transformed and passed around.
2053 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2054 is no longer treated as whitespace by the reader, but instead as
2055 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2056 |A
\7fB|, instead of A as it used to.
2058 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2059 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2060 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2061 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2062 this as of 0.7.1.45, but was left out of the previous news section
2063 (thanks to Dan Barlow)
2064 * SBCL now runs on the Solaris operating system on SPARC architectures
2065 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2066 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2067 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2068 * SPARC backend cleanups, allowing builds of cores optimized for V8
2069 and V9 SPARCS, and also emission of code targeted to a particular
2070 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2072 * SBCL is closer to bootstrapping under CLISP, thanks to various
2073 fixes by Christophe Rhodes.
2074 * The fasl file format has changed again, to allow the compiler's
2075 INFO database to support symbol macros.
2076 * The user manual (in doc/) is formatted into HTML more nicely.
2077 (thanks to coreythomas)
2078 * The system is smarter about SUBTYPEP relationships, especially
2079 those involving NOT types (including types such as ATOM which are
2080 represented internally using NOT types). Thus SUBTYPEP is less
2081 likely to return (VALUES NIL NIL) in general, and in particular
2082 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2084 * The fasl file format has changed again, because the internal
2085 representation of types now includes a new slot to support the new
2086 SUBTYPEP-of-NOT-types logic.
2087 * (not a change in the main branch of SBCL, but a related prototype
2088 which can hopefully be merged into the main branch of SBCL in the
2089 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2090 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2091 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2092 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2093 dispatch macro character combinations. (thanks to Alexey Dejneka)
2094 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2095 that superficially look like logical namestrings correctly.
2096 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2098 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2099 default supplied-p) for &optional and &key arguments. (thanks to
2102 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2103 * incompatible change: The compiler is now less aggressive about
2104 tail call optimization, doing it only when (> SPACE DEBUG) or
2105 (> SPEED DEBUG). (This is an incompatible change because there are
2106 programs which relied on the old CMU-CL-style behavior to optimize
2107 away their unbounded recursion which will now die of stack overflow.)
2108 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2109 for non-GENCGC systems has been increased to 20M (since that
2110 seems much closer to the likely performance optimum for modern
2111 systems than the old 4M value was)
2112 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2113 and *DEBUG-PRINT-LEVEL*
2114 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2115 of CMU CL's support for SPARC, and various endianness and other
2116 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2117 * new syntactic sugar for the Unix command line: --load foo.bar is now
2118 an alternate notation for --eval '(load "foo.bar")'.
2120 ** The system now detects stack overflow and handles it gracefully,
2121 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2122 optimization settings. (This is a good thing in general, and
2123 its introduction in this version should be particularly timely
2124 for anyone whose code fails because of suppression of tail
2126 ** The system now hunts for the C variable "environ" in a more
2127 devious way, to avoid segfaults when the C library version differs
2128 between compile time and run time. (thanks to Christophe Rhodes)
2129 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2130 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2131 ** The compiler no longer issues bogus style warnings for undefined
2132 classes in the same source file as the DEFCLASSes which defined
2133 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2135 ** fixes in CONDITION class precedence list for undefined function
2136 errors (thanks to Alexei Dejneka)
2137 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2138 correctly. (thanks to Dan Barlow)
2139 ** portability fixes aiming at bootstrapping under CLISP (thanks
2140 to Dave McDonald and Christophe Rhodes)
2141 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2142 ** fixes in type translation and and type inference (thanks to
2144 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2145 ** various fixes in the new ports (thanks to Dan Barlow)
2146 * several changes related to debugging:
2147 ** suppression of tail recursion, as noted above
2148 ** stack overflow detection, as noted above
2149 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2150 is now the default. (For some time encapsulation has been more
2151 reliable than the breakpoint-based :ENCAPSULATE NIL
2152 implementation, at least on X86 systems; and I just noticed that
2153 encapsulation also seems closer to the spirit of the ANSI
2156 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2158 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2159 up properly again. (There was a packaging bug in 0.7.0 which
2160 left their definitions in SB-SYS::LOAD-FOREIGN and
2161 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2162 vital for most things which interface to C-level interfaces,
2163 like extensions working with sockets or databases or
2164 Perl-compatible regexes or whatever, and the need to fix
2165 this bug is the main reason that 0.7.1 was released so
2167 ** DEFGENERIC is now choosier about the methods it redefines, so that
2168 reLOADing a previously-LOADed file containing DEFGENERICs does
2169 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2170 cycle now works as it should. (thanks to Alexey Dejneka)
2171 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2172 fixed by Christophe Rhodes. (He actually submitted this patch
2173 months ago, and I delayed until after 0.7.0.)
2174 ** Bug 111 (internal compiler confusion about runtime checks on
2175 FUNCTION types) was fixed by Alexey Dejneka.
2176 * Some internal cleanups (getting rid of variables which aren't
2177 needed now that the byte interpreter is gone) caused the fasl
2178 file format number to change again.
2180 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2181 * major incompatible change: The default fasl file extension, i.e. the
2182 default extension for files produced by COMPILE-FILE, has changed
2183 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2185 ** There are many changes in the implementation of the compiler.
2186 SBCL is now essentially a compiler-only implementation of ANSI
2187 Common Lisp. EVAL still "interprets" a few special cases, but
2188 almost all the interesting cases are handled by creating
2189 a LAMBDA expression, calling COMPILE on it, then calling
2190 FUNCALL on the result.
2191 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2192 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2193 Since the code is newer, there might still be some new bugs
2194 (though not as many as before Martin Atzmueller's fixes:-). But
2195 the new code is substantially simpler and clearer, and hopefully
2196 any remaining bugs will be simpler, less fundamental, and more
2197 fixable then the bugs in the old code.
2198 ** The revised compiler is still a little unsteady on its feet.
2200 *** The debugging information it produces (particularly the names
2201 of FUNCTION objects) is sometimes much less useful than what
2202 the old compiler produced.
2203 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2204 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2205 within a MACROLET) has been temporarily weakened.
2206 ** There are new compiler optimizations for various functions:
2207 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2208 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2209 *** the math functions TRUNCATE, FLOOR, and CEILING
2210 *** the function-of-all-trades COERCE
2211 Mostly these should be transparent, but there's one
2212 potentially-annoying problem (bug 117): when the compiler
2213 inline-expands a function and does type analysis on the result,
2214 it can create control paths which have type mismatches, and
2215 when it can't prove that those control paths aren't taken,
2216 it will issue WARNINGs about the type mismatches. This is
2217 a particular problem in practice for the new sequence functions.
2218 It's not clear how this should be fixed, and for now, a
2219 workaround is given in the entry for 117 in the BUGS file.
2220 ** (Because of the interaction between the two previous items --
2221 occasional inlining problems and new inline expansions -- some
2222 of the new sequence function optimizations won't really kick in
2223 completely until debugging information, and then inlining, are
2224 straightened out in some future version.)
2225 * minor incompatible changes:
2226 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2227 DIRECTORY no longer implicitly promotes NIL slots of its
2228 pathname argument to :WILD. In particular, when you ask for the
2229 contents of a directory (which you used to be able to do without
2230 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2231 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2232 ** changes in behavior that ANSI explicitly defines to be
2233 implementation dependent:
2234 *** The new compiler-only implementation still conforms with ANSI,
2235 but acts a little different than before. Besides the obvious
2236 changes in performance tradeoffs (that the cost per form passed
2237 to EVAL has gone up, and the cost per form executed by EVAL
2238 has gone down), the behavior of the system changes a little
2239 because there are no longer any interpreted function objects.
2240 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2241 e.g. doing COMPILE on the output of interactive DEFUN is
2243 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2245 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2246 has changed from (:CL) to NIL.
2247 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2248 e.g. #\Newline and #\Tab, don't have names specified in the
2249 ANSI Common Lisp standard, is now based on their ASCII symbolic
2250 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2251 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2252 are no longer used for output.
2253 ** changes in internal implementation constants:
2254 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2255 4 million. (If your application spends a lot of time GCing and
2256 you have a lot of RAM, you might want to experiment with
2257 increasing it even more.)
2258 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2259 However, almost all old code should still continue to work without
2260 immediate update, as SB-C-CALL is now a (deprecated) nickname
2262 ** Old operator names in the style DEF-FOO are now deprecated in
2263 favor of new corresponding names DEFINE-FOO, for consistency with
2264 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2265 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2266 internal symbols, but a few supported extensions like
2267 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2268 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2269 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2270 etc. as you get deeper into recursive calls to the debugger
2271 command loop, instead of the old "5]", "5]]", "5]]]"
2272 sequence. (I was motivated to do this when squabbles between
2273 ILISP and SBCL left me very deeply nested in the debugger. In the
2274 short term, this change will probably provoke more ILISP/SBCL
2275 squabbles, but hopefully it will be an improvement in the long run.)
2276 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2277 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2278 function, since declaiming return types involving aliens is
2279 (1) annoyingly messy to do by hand and (2) vital to efficient
2280 compilation of code which calls such functions.
2281 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2282 longer reexported by the SB-EXT package. They're solely useful
2283 for alien code, so it seems more logical that you should get
2284 them from the SB-ALIEN package, not in SB-EXT.
2285 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2286 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2287 features. Instead, the code that they used to control is always
2288 built into the system.
2289 * many other bug fixes
2290 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2291 advantage of the new EVAL-WHEN stuff and to clean them up in
2292 general, and they are now more ANSI-compliant in a number of
2293 ways. Martin Atzmueller is responsible for a lot of this.
2294 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2296 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2297 *** correct ERROR type for various file operations
2298 *** some fixes for Lisp streams
2299 *** DEFMETHOD syntax checking
2300 *** changing old weird representation of debug information as
2301 strings (which, among their other deficiencies, don't transform
2302 correctly when you rename packages, and don't change their
2303 print representation when you change things like *PACKAGE*
2304 and *PRINT-LENGTH*) to symbols and lists of symbols
2305 He also made several improvements and fixed several bugs in DESCRIBE.
2306 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2308 *** misbehavior of WRITE-STRING/WRITE-LINE
2309 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2310 and several other LOOP problems as well
2311 *** DIRECTORY when similar filenames are present
2312 *** DEFGENERIC with :METHOD options
2313 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2314 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2315 *** argument ordering in FIND with :TEST option
2316 *** mishandled package designator argument in APROPOS-LIST
2317 *** various problems in the backquote readmacro
2318 *** a bug in APROPOS
2319 *** probably some others that I'm not describing very well here,
2320 since the CVS log documents them by reference to sbcl-devel
2321 messages, and the SourceForge archives aren't working well.:-(
2322 ** Dan Barlow improved the Alpha port (and is making progress on the
2323 PPC port, for those of you who think different).
2324 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2325 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2326 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2327 continues to work on the SPARC port (for those of you in a position
2328 to look down upon our little PC-compatible boxes from a great height).
2329 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2330 and uses that copy, rather than the current dynamic value, when
2331 it's trying to decide whether to truncate output. Thus e.g.
2332 (let ((*print-lines* 50))
2333 (pprint-logical-block (stream nil)
2335 (let ((*print-lines* 8))
2336 (print (aref possiblybigthings i) stream)))))
2337 should now truncate the logical block only at 50 lines, instead of
2338 often truncating it at 8 lines, as it did before.
2339 * The doc/cmucl/ directory, containing old CMU CL documentation
2340 from the time of the fork, is no longer part of the base system.
2341 SourceForge has shut down its anonymous FTP service, and with it
2342 my original plan for distributing the old CMU CL documentation
2343 there. For now, if you need these files you can download an old
2344 SBCL source release and extract them from it.
2345 * The fasl file version number changed again, for dozens of reasons,
2346 some of which are apparent above.
2348 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2349 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2350 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2351 patch, so that DISASSEMBLE output is much nicer.
2352 * The code in the SB-PROFILE package now seems reasonably stable.
2353 I still haven't decided what the final interface should look like
2354 (I'd like PROFILE to interact cleanly with TRACE, since both
2355 facilities use function encapsulation) but if you have a need
2356 for profiling now, you can probably use it successfully with
2357 the current CMU-CL-style interface.
2358 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2359 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2360 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2361 link now returns the dangling link itself, and for similar
2362 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2363 link itself. (In these cases the old code signalled an error and
2364 looped endlessly, respectively.) Thus, DIRECTORY now works even
2365 in the presence of dangling and cyclic symbolic links.
2366 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2367 is now a supported extension again, since the consensus on
2368 sbcl-devel was that it can be useful for ordinary development
2369 work, not just for debugging SBCL itself.
2370 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2371 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2372 that currently-defined functions might be redefined later with
2373 different return types.
2374 * Hash tables can be printed readably, as inspired by CMU CL code
2375 of Eric Marsden and SBCL code of Martin Atzmueller.
2376 * better error handling in CLOS method combination, thanks to
2377 Martin Atzmueller porting Pierre Mai's CMU CL patches
2378 * more overflow fixes for >16Mbyte I/O buffers
2379 * A bug in READ has been fixed, so that now a single Ctrl-D
2380 character suffices to cause end-of-file on character streams.
2381 In particular, now you only need one Ctrl-D at the command
2382 line (not two) to exit SBCL.
2383 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2385 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2386 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2387 does the right thing.
2388 * The implementation of some type tests, especially for CONDITION
2389 types, is now tidier and maybe faster, due to CMU CL code
2390 originally by Douglas Crosher, ported by Martin Atzmueller.
2391 * Some math functions have been fixed, and there are new
2392 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2393 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2394 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2395 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2396 likely to be built into sbcl-0.7.0.)
2397 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2398 CL, ported to SBCL by Martin Atzmueller)
2399 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2400 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2401 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2402 * There's a new slam.sh hack to shorten the edit/compile/debug
2403 cycle for low-level changes to SBCL itself, and a new
2404 :SB-AFTER-XC-CORE target feature to control the generation of
2405 the after-xc.core file needed by slam.sh.
2406 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2407 COMPILE-FILE is no longer supported, so that now every function
2408 gets an entry point, so that block compilation looks a little
2409 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2410 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2411 returns the number of bytes consed since the system started,
2412 rather than the number consed since the first time the function
2413 was called. (The new definition parallels ANSI functions like
2414 CL:GET-INTERNAL-RUN-TIME.)
2415 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2416 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2417 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2418 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2419 * Fasl file version numbers are now independent of the target CPU,
2420 since historically most system changes which required version
2421 number changes have affected all CPUs equally. Similarly,
2422 the byte fasl file version is now equal to the ordinary
2425 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2426 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2427 is no longer recognized. I apologize for this, because it was
2428 listed in SB-EXT as a supported extension, but I found that
2429 its existing behavior was poorly specified, as well as incorrectly
2430 specified, and it looked like too much of a mess to straighten it
2431 out. I have enough on my hands trying to get ANSI stuff to work..
2432 * many patches ported from CMU CL by Martin Atzmueller, with
2433 half a dozen bug fixes in pretty-printing and the debugger, and
2434 half a dozen others elsewhere
2435 * fixed bug 13: Floating point infinities are now supported again.
2436 They might still be a little bit flaky, but thanks to bug reports
2437 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2438 as flaky as they were.
2439 * The --noprogrammer command line option is now supported. (Its
2440 behavior is slightly different in detail from what the old man
2441 page claimed it would do, but it's still appropriate under the
2442 same circumstances that the man page talks about.)
2443 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2444 are now supported, and enabled by default. Thus, the compiler can
2445 handle many floating point and complex operations much less
2446 inefficiently. (Thus e.g. you can implement a complex FFT
2448 * The compiler now detects type mismatches between DECLAIM FTYPE
2449 and DEFUN better, and implements CHECK-TYPE more correctly, and
2450 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2451 thanks to patches from Martin Atzmueller.
2452 * various fixes to make the cross-compiler more portable to
2453 ANSI-conforming-but-different cross-compilation hosts (notably
2454 Lispworks for Windows, following bug reports from Arthur Lemmens)
2455 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2456 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2457 fixed thanks to Pierre Mai's CMU CL patches.
2458 * fixes to keep the system from overflowing internal counters when
2459 it tries to use i/o buffers larger than 16M bytes
2460 * fixed bug 45a: Various internal functions required to support
2461 complex special functions have been merged from CMU CL sources.
2462 (When I was first setting up SBCL, I misunderstood a compile-time
2463 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2464 * improved support for type intersection and union, fixing bug 12
2465 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2466 more obscure bugs as well
2467 * some steps toward byte-compiling non-performance-critical
2468 parts of the system, courtesy of patches from Martin Atzmueller
2469 * Christophe Rhodes has made some debian packages of sbcl at
2470 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2471 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2472 stable, but are nonetheless usable. When he's ready, I'd be happy
2473 to add them to the SourceForge "File Releases" section. (And if
2474 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2475 * new fasl file format version number (because of changes in
2476 internal representation of (OR ..) types to accommodate the new
2477 support for (AND ..) types, among other things)
2479 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2480 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2482 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2484 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2485 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2486 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2487 * DESCRIBE now works on structure objects again.
2488 * Most function call argument type mismatches are now handled as
2489 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2490 know whether the function will be redefined before the call is
2491 executed. (The compiler could flag local calls with full WARNINGs,
2492 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2493 it doesn't keep track of enough information to know whether calls
2494 are local in this sense.)
2495 * Compiler output is now more verbose, with messages truncated
2496 later than before. (There should be some supported way for users
2497 to override the default verbosity, but I haven't decided how to
2498 provide it yet, so this behavior is still controlled by the internal
2499 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2500 src/compiler/ir1util.lisp.)
2501 * Fasl file format version numbers have increased again, because
2502 support for the Gray streams extension changes the layout of the
2503 system's STREAM objects.
2504 * The Gray subclassable streams extension now works, thanks to a
2505 patch from Martin Atzmueller.
2506 * The full LOAD-FOREIGN extension (not just the primitive
2507 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2508 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2509 but like most other programs, it defaults to copying the Unix
2510 environment from the original process instead of starting the
2511 new process in an empty environment.
2512 * Extensions which manipulate the Unix environment now support
2513 an :ENVIRONMENT keyword option which doesn't smash case or
2514 do other bad things. The CMU-CL-style :ENV option is retained
2515 for porting convenience.
2516 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2517 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2519 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2521 * A patch from Martin Atzmueller seems to have solved the SIGINT
2522 problem, and as far as we know, signal-handling now works cleanly.
2523 (If you find any new bugs, please report them!)
2524 * The system no longer defaults Lisp source file names to types
2525 ".l", ".cl", or ".lsp", but only to ".lisp".
2526 * The compiler no longer uses special default file extensions for
2527 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2528 seems to expect a single default extension for all compiled code,
2529 and there's no compelling reason to try to stretch the standard
2530 to allow two different extensions.) Instead, byte-compiled files
2531 default to the same extension as native-compiled files.
2532 * Fasl file format version numbers have increased again, because
2533 a rearrangement of internal implementation packages made some
2534 dumped symbols in old fasl files unreadable in new cores.
2535 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2536 many fewer weird special cases.
2537 * Bug #17 (differing COMPILE-FILE behavior between logical and
2538 physical pathnames) has been fixed, and some related misbehavior too,
2539 thanks to a patch from Martin Atzmueller.
2540 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2541 by Tim Moore, ported to SBCL by Martin Atzmueller.
2542 * Martin Atzmueller fixed several filesystem-related problems,
2543 including bug #36, in part by porting CMU CL patches, which were
2544 written in part by Paul Werkowski.
2545 * More compiler warnings in src/runtime/ are gone, thanks to
2546 more patches from Martin Atzmueller.
2547 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2550 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2552 * DESCRIBE now works on CONDITION objects.
2553 * The debugger now handles errors which arise when trying to print
2554 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2556 * The build system now uses an additional file, customize-target-features.lisp,
2557 to allow local modifications to the target *FEATURES* list. (The point of
2558 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2559 debugging features enabled, without having to worry about propagating it
2560 into everyone's system when I do a "cvs update".) When no
2561 customize-target-features.lisp file exists, the target *FEATURES* list
2562 should be constructed the same way as before.
2563 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2564 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2565 warn about behavior which is undefined under the ANSI spec. Note: This
2566 is specified by ANSI, but it's not very popular with programmers.
2567 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2568 macro in the SBCL sources for an example of a workaround which you
2569 might use to make portable ANSI-standard code which does what you want.
2570 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2571 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2572 when it's used as a non-toplevel form. (This is required in order
2573 to implement the DEFCONSTANT-EQX macro.)
2574 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2575 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2576 funny things, due to bugs in the implementation of EVAL-WHEN
2577 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2578 non-toplevel DEFCONSTANTs are uncommon.)
2579 * The core file version number and fasl file version number have been
2580 incremented, because the old noncompliant DEFCONSTANT behavior involved
2581 calling functions which no longer exist, and because I also took the
2582 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2583 * fixed bug 1 (error handling before read-eval-print loop starts), and
2584 redid debugger restarts and related debugger commands somewhat while
2586 ** The QUIT debugger command is gone, since it did something
2587 rather different than the SB-EXT:QUIT command, and since it never
2588 worked properly outside the main toplevel read/eval/print loop.
2589 Invoking the new TOPLEVEL restart provides the same functionality.
2590 ** The GO debugger command is also gone, since you can just invoke
2591 the CONTINUE restart directly instead.
2592 ** The TOP debugger command is also gone, since it's redundant with the
2593 FRAME 0 command, and since it interfered with abbreviations for the
2595 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2597 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2598 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2599 some porting work by Martin Atzmueller)
2600 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2601 with a boatload of patches which he ported to SBCL. Now that those
2603 ** The system tries to make sure that its low-priority messages
2604 are prefixed by semicolons, to help people who like to use
2605 syntax highlighting in their ILISP buffer. (This patch
2606 was originally due to Raymond Toy.)
2607 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2608 patches originally written by Raymond Toy.
2609 ** The compiler understands coercion between single-value and
2610 multiple-VALUES type expressions better, getting rid of some very
2611 weird behavior, thanks to patches originally by Robert MacLachlan
2612 and Douglas Crosher.
2613 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2614 lambda lists, thanks to a patch originally by Pierre Mai.
2615 ** The system no longer bogusly warns about "abbreviated type
2617 ** The compiler gets less confused by inlining and RETURN-FROM,
2618 thanks to some patches originally by Tim Moore.
2619 ** The system no longer hangs when dumping circular lists to fasl
2620 files, thanks to a patch originally from Douglas Crosher.
2621 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2622 undefined function error.
2623 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2624 in structure slot initforms) for now, documented workaround instead:-|
2625 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2626 * fixed bug 5: added stubs for various Gray stream functions called
2627 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2628 installed, at least appropriate type errors are generated
2629 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2630 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2631 consistently in DEFMETHOD forms.
2632 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2633 it doesn't seem to affect SBCL after all
2634 * The C runtime system now builds with better optimization and many
2635 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2637 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2639 * The system is now under CVS at SourceForge (instead of the
2640 CVS repository on my home machine).
2641 * The new signal handling code has been tweaked to treat register
2642 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2643 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2644 downstream, e.g. in debug-int.lisp, has implicit dependencies
2645 on the unsignedness of integer representation of machine words,
2646 and that caused the system to bomb out with infinite regress
2647 when trying to recover from type errors involving signed values,
2648 e.g. (BUTLAST '(1 2 3) -1).)
2649 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2650 (This was one of the bugs Peter Van Eynde reported back in July.)
2651 * The system now uses code inspired by Colin Walters' O(N)
2652 implementation of MAP (from the cmucl-imp@cons.org mailing
2653 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2654 inline the MAP operation, and there is more than one
2655 sequence argument to the MAP call (so that it can't just
2656 do ETYPECASE once and for all based on the type of the
2657 single sequence argument). (The old non-inline implementation
2658 of the general M-argument sequence-of-length-N case required
2659 O(M*N*N) time when any of the sequence arguments were LISTs.)
2660 * The QUIT :UNIX-CODE keyword argument has been renamed to
2661 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2662 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2664 * Raymond Wiker's patches to port dynamic loading from Linux to
2665 FreeBSD have been added.
2666 * The BUGS file is now more nearly up to date, thanks in large part
2667 to Martin Atzmueller's review of it.
2668 * The debugger now flushes standard output streams before it begins
2669 its output ("debugger invoked" and so forth).
2670 * The core version number and fasl file version number have both
2671 been incremented, because of incompatible changes in the layout
2673 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2674 used to not be called for a saved Lisp image.)
2675 * Martin Atzmueller's version of a patch to fix a compiler crash,
2676 as posted on sbcl-devel 13 September 2000, has been installed.
2677 * Instead of installing Martin Atzmueller's patch for the
2678 compiler transform for SUBSEQ, I deleted the compiler transform,
2679 and transforms for some similar consing operations.
2680 * A bug in signal handling which kept TRACE from working on OpenBSD
2682 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2683 away by constant folding
2684 * The system now defines its address space constants in one place
2685 (in the Lisp sources), and propagates them automatically elsewhere
2686 (through GENESIS and the sbcl.h file). Therefore, patching the
2687 address map is less unnecessarily tedious and error-prone. The
2688 Lisp names of address space constants have also been systematized.
2689 * CVS tags like dollar-Header-dollar have been removed from
2690 the sources, because they have never saved me trouble and
2691 they've been source of trouble working with patches and other
2692 diff-related operations.
2693 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2694 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2696 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2698 * The system has been ported to OpenBSD.
2699 * The system now compiles with a simple "sh make.sh" on the systems
2700 that it's supported on. I.e., now you no longer need to tweak
2701 text in the target-features.lisp-expr and symlinks in src/runtime/
2702 by hand, the make.sh takes care of it for you.
2703 * The system is no longer so grossly inefficient when compiling code
2704 involving vectors implemented as general (not simple) vectors (VECTOR T),
2705 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2706 which dares to use the various sequence functions on non-simple
2707 vectors, takes less of a performance hit.
2708 * There is now a primitive type predicate VECTOR-T-P
2709 to test for the (VECTOR T) type, so that e.g.
2710 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2711 can now be compiled with some semblance of efficiency. (The old code
2712 turned the type declaration into a full call to %TYPEP at runtime!)
2713 * AREF on (VECTOR T) is still not fast, since it's still compiled
2714 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2715 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2716 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2717 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2718 without making full calls to SUBTYPEP at runtime.
2719 (Some analogous efficiency issues for non-simple vectors specialized to
2720 element types other than T, or for non-simple multidimensional arrays,
2721 have not been addressed. They could almost certainly be handled the
2722 same way if anyone is motivated to do so.)
2723 * The changes in array handling break binary compatibility, so
2724 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2725 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2726 instead of (VALUES T T).
2727 * By following the instructions that Dan Barlow posted to sbcl-devel
2728 on 2 July 2000, I was able to enable primitive dynamic object
2729 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2730 functionality is not implemented (since it calls ld to resolve
2731 library references automatically, requiring RUN-PROGRAM for its
2732 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2733 try to resolve library references) is now supported.
2734 * The system now flushes the standard output streams when it terminates,
2735 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2736 them at several other probably-convenient times, e.g. in each pass of
2737 the toplevel read-eval-print loop, and after evaluating a form given
2738 as an "--eval" command-line option. (These changes were motivated by a
2739 discussion of stream flushing issues on cmucl-imp in August 2000.)
2740 * The source transform for TYPEP of array types no longer assumes
2741 that an array whose element type is a not-yet-defined type
2742 is implemented as an array of T, but instead punts, so that the
2743 type will be interpreted at runtime.
2744 * There is now some support for cross-compiling in make.sh: each of
2745 the phases of make.sh has its own script. (This should be transparent
2746 to people doing ordinary, non-cross-compile builds.)
2747 * Since my laptop doesn't have hundreds of megabytes of memory like
2748 my desktop machine, I became more motivated to do some items on
2749 my to-do list in order to reduce the size of the system a little:
2750 ** Arrange for various needed-only-at-cold-init things to be
2751 uninterned after cold init. To support this, those things have
2752 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2753 symbols with such names are now uninterned after cold init).
2754 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2755 fasl files for cold load.
2756 ** Remove the old compiler structure pooling code (which used to
2757 be conditional on the target feature :SB-ALLOC) completely.
2758 ** Redo the representation of some data in cold init to be more compact.
2759 (I also looked into supporting byte compiled code at bootstrap time,
2760 which would probably reduce the size of the system a lot, but that
2761 looked too complicated, so I punted for now.)
2762 * The maximum signal nesting depth in the src/runtime/ support code has
2763 been reduced from 4096 to 256. (I don't know any reason for the very
2764 large old value. If the new smaller value turns out to break something,
2765 I'll probably just bump it back up.)
2766 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2768 * Many, many bugs reported by Peter Van Eynde have been added to
2769 the BUGS list; some have even been fixed.
2770 * While enabling dynamic object file loading, I tried to make the
2771 code easier to understand, renaming various functions and variables
2772 with less ambiguous names, and changing some function calling
2773 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2774 * While trying to figure out how to do the OpenBSD port, I tried to
2775 clean up some of the code in src/runtime/. In particular, I dropped
2776 support for non-POSIX signal handling, added various comments,
2777 tweaked the code to reduce the number of compilation warnings, and
2778 renamed some files to increase consistency.
2779 * To support the new automatic configuration functionality in make.sh,
2780 the source file target-features.lisp-expr has been replaced with the
2781 source file base-target-features.lisp-expr and the machine-generated
2782 file local-target-features.lisp-expr.
2783 * fixed a stupid quoting error in make.sh so that using CMU CL
2784 "lisp -batch" as cross-compilation host works again
2786 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2788 * DESCRIBE no longer tries to call itself recursively to describe
2789 bound/fbound values, so that it no longer fails on symbols which are
2790 bound to themselves (like keywords, T, and NIL).
2791 * DESCRIBE now works on generic functions.
2792 * The printer now prints less-screwed-up representations of closures
2793 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2794 * A private symbol is used instead of the :EMPTY keyword previously
2795 used to mark empty slots in hash tables. Thus
2796 (DEFVAR *HT* (MAKE-HASH-TABLE))
2797 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2798 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2799 now does what ANSI says that it should. (You can still get
2800 similar noncompliant behavior if bang on the hash table
2801 implementation with all the symbols you get back from
2802 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2803 This breaks binary compatibility, since tests for equality to
2804 :EMPTY are wired into things like the macroexpansion of
2805 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2807 * There's now a minimal placeholder implementation for CL:STEP,
2808 as required by ANSI.
2809 * An obscure bug in the interaction of the normal compiler, the byte
2810 compiler, inlining, and structure predicates has been patched
2811 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2812 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2814 * Missing ordinary arguments in a macro call are now detected even
2815 when the macro lambda list contains &KEY or &REST.
2816 * The debugger no longer complains about encountering the top of the
2817 stack when you type "FRAME 0" to explicitly instruct it to go to
2818 the top of the stack. And it now prints the frame you request even
2819 if it's the current frame (instead of saying "You are here.").
2820 * As specified by ANSI, the system now always prints keywords
2821 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2823 * The default initial SIZE of HASH-TABLEs is now smaller.
2824 * Type information from CLOS class dispatch is now propagated
2825 into DEFMETHOD bodies, so that e.g.
2826 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2828 is now basically equivalent to
2829 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2830 (DECLARE (TYPE SINGLE-FLOAT X))
2832 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2833 operation, without having to do run-time type dispatch.
2834 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2835 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2836 * Since I don't seem to be making big file reorganizations very often
2837 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2838 is overflowing my ability to conveniently back them up), I've finally
2839 checked the system into CVS. (The CVS repository is on my home system,
2840 not at SourceForge -- putting it on SourceForge might come later.)
2841 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2842 high-level GC-NOTIFY-FOO functions send their output. (There's
2843 still very little control of where low-level verbose GC functions
2844 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2845 less than it used to -- the GC-NOTIFY-FOO functions are now under
2846 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2847 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2848 in only one place in the source code, and propagates it automatically
2849 everywhere that it's needed. Thus e.g. when I bump the version from
2850 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2851 * The C source files now include boilerplate legalese and documentation
2852 at the head of each file (just as the Lisp source files already did).
2853 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2854 to his page will be replaced with a link to his new CLiki service.
2856 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2858 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2859 * The build process now looks for GNU make under the default name "gmake",
2860 instead of "make" as it used to. If GNU make is not available as "gmake"
2861 on your system, you can change this default behavior by setting the
2862 GNUMAKE environment variable.
2863 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2864 can build without error under CMU CL.
2866 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2868 * There is now a partial SBCL user manual (with some new text and some
2869 text cribbed from the CMU CL manual).
2870 * The beginnings of a profiler have been added (starting with the
2871 CMU CL profiler and simplifying and cleaning up). Eventually the
2872 main interface should be through the TRACE macro, but for now,
2873 it's still accessed through vaguely CMU-CL-style functions and macros
2874 exported from the package SB-PROFILE.
2875 * Some problems left over from porting CMU CL to the new
2876 cross-compilation bootstrap process have been cleaned up:
2877 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2878 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2879 part when getting rid of the compiler *BACKEND* stuff.)
2880 ** The value of *NULL-TYPE* was screwed up, because it was
2881 being initialized before the type system knew the final
2882 definition of the 'NULL type. This screwed up several key
2883 optimizations in the compiler, causing inefficiency in all sorts
2884 of places. (I found it because I wanted to understand why
2885 GET-INTERNAL-RUN-TIME was consing.)
2886 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2887 PROCLAIM FTYPE information. Unfortunately this broke binary
2888 compatibility again, since now the forms output by DEFGENERIC
2889 to refer to functions which didn't exist in 0.6.3.
2890 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2891 can use the new (as of 0.6.3) transform for SOME into MAP into
2893 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2894 values to (INTEGER 0 1000000), so that the time code will no longer
2895 occasionally get blown up by Linux returning 1000000 microseconds
2896 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2897 its output conform to the ANSI spec. (Alas, this makes its output
2898 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2899 * A full call to MAP NIL with a single sequence argument no longer conses.
2900 * fixes to problems pointed out by Martin Atzmueller:
2901 * The manual page no longer talks about multiprocessing as though
2902 it were currently supported.
2903 * The ILISP support patches have been removed from the distribution,
2904 because as of version 5.10.1, ILISP now supports SBCL without us
2905 having to maintain patches.
2906 * added a modified version of Raymond Toy's recent CMU CL patch for
2907 EQUALP comparison of HASH-TABLE
2909 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2911 * The system still can't cross-compile itself with
2912 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2913 entails), but at least it can compile more of itself that way
2914 than it used to be able to, and various buglets which were uncovered
2915 by trying to cross-compile itself that way have now been fixed.
2916 * This release breaks binary compatibility again. This time
2917 at least I've incremented the FASL file format version to 2, so that the
2918 problem can be detected reliably instead of just causing weird errors.
2919 * various new style warnings:
2920 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2921 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2922 ** using the lexical binding of a variable named in the *FOO* style
2923 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2924 as specified by ANSI.
2925 * *RANDOM-STATE* is no longer automatically initialized from
2926 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2927 default behavior of the system is to repeat its behavior every time
2928 it's run. If you'd like to change this behavior, you can always
2929 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2930 old convention there was no comparably easy way to get the system to
2931 repeat its behavior every time it was run.
2932 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2933 FTYPE declarations has been removed, in favor of their ANSI
2934 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2935 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2936 the types of their sequence arguments can be determined at compile time.
2937 This is done through a new open code expansion for MAP which eliminates
2938 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2939 argument types can be determined at compile time.
2940 * The optimizer now transforms COERCE into an identity operation when it
2941 can prove that the coerced object is already of the correct type. (This
2942 can be a win for machine generated code, including the output of other
2943 optimization transforms, such as the MAP transform above.)
2944 * Credit information has been moved from source file headers into CREDITS.
2945 * Source file headers have been made more standard.
2946 * The CASE macro now compiles without complaining even when it has
2949 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2951 * (Note that the way that the PCL macroexpansions were rewritten
2952 to accommodate the change in DEFGENERIC below breaks binary
2953 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2954 not run under sbcl-0.6.2. Once we get out of alpha releases,
2955 i.e. hit release 1.0.0, we'll probably try to maintain binary
2956 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2957 and sbcl-1.4.4. Until then, however, it might be fairly common
2958 for maintenance releases to break binary compatibility.)
2959 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2961 * The reporting of unhandled signals has been changed to print some
2962 explanatory text as well as the report form. (Previously only
2963 the report form was printed.)
2964 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2965 it defines, so that the compiler no longer issues undefined function
2966 warnings for compiled-but-not-yet-loaded generic functions.
2967 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2968 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2969 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2970 section of the ANSI spec are supported.
2971 * The "" nickname for the "KEYWORD" package has been removed.
2972 The reader still handles symbol tokens which begin with a package marker
2973 as keywords, but it doesn't expose its mechanism for doing so in the
2974 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2975 * The system now issues STYLE-WARNINGs for contradictory TYPE
2976 proclamations. (Warnings for contradictory FTYPE proclamations would
2977 be nice too, but those can't be done usefully unless the type system
2978 is made smarter about FUNCTION types.)
2979 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2980 systematized, so that "*target-*.lisp is supposed to exist only on the
2981 target and imply that there's a related file which exists on the
2982 host, and *host-*.lisp is supposed to exist only on the host and imply
2983 that there's a related file which exists on the target. This involves a
2984 lot of renaming. Hopefully the acute confusion caused by the renaming
2985 will be justified by the reduction in chronic confusion..
2986 ** runtime-type.lisp -> early-target-type.lisp
2987 ** target-type.lisp -> late-target-type.lisp
2988 ** early-host-format.lisp -> early-format.lisp
2989 ** late-host-format.lisp -> late-format.lisp
2990 ** host-error.lisp -> misc-error.lisp
2991 ** early-error.lisp -> early-target-error.lisp
2992 ** late-error.lisp -> late-target-error.lisp
2993 ** host-defboot.lisp -> early-defboot.lisp
2994 ** code/misc.lisp -> code/target-misc.lisp
2995 ** code/host-misc.lisp -> code/misc.lisp
2996 ** code/numbers.lisp -> code/target-numbers.lisp
2997 ** code/early-numbers.lisp -> numbers.lisp
2998 ** early-host-type.lisp -> early-type.lisp
2999 ** late-host-type.lisp -> late-type.lisp
3000 ** host-typep.lisp -> typep.lisp
3001 ** load.lisp -> target-load.lisp
3002 ** host-load.lisp -> load.lisp
3003 ** host-disassem.lisp -> disassem.lisp
3004 ** host-insts.lisp -> insts.lisp
3005 ** byte-comp.lisp -> target-byte-comp.lisp
3006 ** host-byte-comp.lisp -> byte-comp.lisp
3007 ** host-signal.lisp -> signal.lisp
3008 ** host-defstruct.lisp -> defstruct.lisp
3009 ** late-target-type.lisp -> deftypes-for-target.lisp
3010 Furthermore, several other previously target-only files foo.lisp (e.g.
3011 hash-table.lisp and random.lisp) have been split into a target-and-host
3012 foo.lisp file and a target-only target-foo.lisp file, with their key type
3013 definitions in the target-and-host part, so that the cross-compiler will
3014 know more about target types.
3015 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3016 gone away. In their place are various *BACKEND-FOO* variables
3017 corresponding to the slots of the old structure.
3018 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3019 into the target SBCL has been fixed.
3020 * The chill.lisp system for loading cold code into a running SBCL
3022 * Support for the CMU CL "scavenger hook" extension has been removed.
3023 (It was undocumented and unused in the CMU CL sources that SBCL was
3024 derived from, and stale in sbcl-0.6.1.)
3025 * Various errors in the cross-compiler type system were detected
3026 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3027 (enabling various consistency checks). Many of them were fixed,
3028 but some hard problems remain, so the compiler is back to
3029 running without *TYPE-SYSTEM-INITIALIZED* for now.
3030 * As part of the cross-compiler type system cleanup, I implemented
3031 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3032 * I have started adding UNCROSS calls throughout the type system
3033 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3034 on cross-compiler input files will be able to go away, and various
3036 * CONSTANTP now returns true for quoted forms (as explicitly required
3039 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3041 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3042 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3043 (MAKE-LIST -1). In the longer term, it ought to become true
3044 that declarations are assertions even at SAFETY 1. For now, it's not
3045 quite true even at SAFETY 3, but it's at least more nearly true..
3046 (Note that this change seems to increases the size of the system by
3047 O(5%) and to decrease the speed of the compiler by 20% or more.)
3048 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3049 for the problem of printing dozens of lines of distracting information
3050 about low-level system machinery as part of the top stack frame
3051 on entry to the debugger when an undefined function was called.
3052 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3053 is not reset to COMMON-LISP-USER.
3054 * Compilation of stuff related to dyncount.lisp has been made conditional
3055 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3056 smaller. The various dyncount-related symbols have been moved into
3057 a new "SB-DYNCOUNT" package.
3058 * tty-inspect.lisp has been renamed to inspect.lisp.
3059 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3060 feature has gone away. (When we eventually port to other flavors of
3061 libc and/or Unix, we'll try to make the differences between flavors
3062 invisible at the user level.)
3063 * Various other *FEATURES* tags, and/or their associated conditionals,
3064 have been removed if obsolescent, or given better documentation, or
3065 sometimes given more-mnemonic names.
3067 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3069 * tidied up "make.sh" script
3070 * tidied up system directory structure
3071 * better "clean.sh" behavior
3072 * added doc/FOR-CMUCL-DEVELOPERS
3073 * many many small tweaks to output format, e.g. removing possibly-confusing
3074 trailing #\. character in DESCRIBE-INSTANCE
3075 * (EQUALP #\A 'A) no longer signals an error.
3076 * new hashing code, including EQUALP hashing
3077 * tidied up Lisp initialization and toplevel
3078 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3079 * command line argument processing
3080 * added POSIX-GETENV function to deal with Unix-ish environment variables
3081 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3082 terminating SBCL on EOF
3083 * non-verbose GC by default
3084 * There is no more "sbcl" shell script; the sbcl file is now the C
3085 runtime executable (just like CMU CL).
3086 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3088 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3089 symbol and package information as Lisp native symbol and package objects
3090 instead of strings naming symbols and strings naming packages. This way,
3091 whenever packages are renamed (as in warm init), debug information is
3092 transformed along with everything else.
3093 * tweaked the optimization policy declarations which control the building
3094 of SBCL itself. Now, among other things, the system no longer saves
3095 source location debugging information. (This helps two problems at once
3096 by reducing SBCL size and by keeping SBCL from trying to look for its
3097 sources -- which may not exist -- when reporting errors.)
3098 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3099 debugging and testing purposes
3100 * cleaned up printing, making the printer call PRINT-OBJECT for
3101 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3102 methods, giving nearly-ANSI behavior
3103 * converted almost all special variables to use *FOO* naming convention
3104 * deleted PARSE-TIME functionality, since it can be done portably
3105 * moved some files out of cold init into warm init
3106 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3108 * regularized formatting of source files
3109 * added an install.sh script
3110 * fixed ridiculous memory usage of cross-compiler by making
3111 compiler/alloc.lisp not try to do pooling unless it can hook
3112 itself into the GC of the cross-compilation host. Now the system
3113 builds nicely on my old laptop.
3114 * added :SB-ALLOC in target-features.lisp-expr
3115 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3116 was not implemented)
3117 * re-did condition handling and note reporting in the compiler. Notes
3118 are no longer handled by signalling conditions. Style warnings
3119 and warnings are handled more correctly and reported in such a way
3120 that it's easy to find one or the other in your output (so that you
3121 can e.g. figure out which of many problems caused COMPILE-FILE to
3123 * changed the severity of several compiler warnings from full WARNING
3124 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3125 compiler note reporting so that it doesn't use the condition system
3126 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3127 COMPILE-FILE command)
3128 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3129 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3130 no longer supported, since their implementation was deeply intertwingled
3131 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3132 * removed bogus "support" for compiler macros named (SETF FOO), and
3133 removed the compiler macro for SETF INFO (but only after making a fool
3134 of myself on the cmucl-imp mailing list by posting a bogus patch for
3135 DEFINE-COMPILER-MACRO..)
3136 * Compiled files containing forms which have side effects on the Lisp
3137 reader (such as DEFPACKAGE forms) are now handled more correctly.
3138 (Compiler queuing of top level lambdas has been suppressed by setting
3139 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3140 * deleted various currently-unused source files, e.g. gengc.lisp. They
3141 may be added back at some point e.g. when porting to other architectures,
3142 but until they are it's distracting to distribute them and to try to
3144 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3145 there were so many of them they're just distractions, and UNCROSS is
3146 known to be able to handle the current sources
3147 * moved PROFILE functionality into TRACE, so that it will be clear
3148 how the wrapping and unwrapping of functions when you profile them
3149 interacts with the wrapping and unwrapping of functions when you
3150 trace them. (Actually, the functionality isn't there yet, but at least
3151 the interface specification is there. Hopefully, the functionality will
3152 arrive with some future maintenance release.)
3153 * removed host-oops.lisp
3154 * changed signature of QUIT function to allow UNIX-CODE argument
3155 * fixed READ-SEQUENCE bug
3156 * tweaked verbose GC output so that it looks more like the progress
3157 output that ANSI specifies for functions like LOAD
3158 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3159 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob