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 * fixed some bugs related to Unicode integration:
57 ** the restarts for recovering from input and output encoding
58 errors only appear when there is in fact such an error to
60 * fixed some bugs revealed by Paul Dietz' test suite:
61 ** MISC.549 and similar: late transformation of unsafe type
62 assertions into derived types caused unexpected code
64 ** SCALE-FLOAT type deriver is less wrong.
65 ** type derivers for EXP, LOG and similar functions compute result
66 types for complex arguments better.
67 ** (MISC.563) CONJUGATE type deriver works for very restricted
69 ** out-of-line type testers for character strings are available.
70 ** EQUAL compiler transform understands specialness of objects
72 ** accessing double-floats stored on the stack now works on x86-64.
73 ** debugger internals could sometimes create invalid lispobjs,
74 resulting in GC crashes.
75 ** MISC.548: type check weakening can convert required type into
77 ** initialization forms for bindings are not in scope of free special
80 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
81 * incompatible change: thread support for non-NPTL systems has
82 been removed - locking is buggy and unreliable. A threaded
83 SBCL build will now warn at startup and refuse to create
84 new threads, unless futex support is detected in the kernel
85 * incompatible change: the top level REPL now has only an ABORT
86 restart associated with it, not TOPLEVEL and ABORT as it used to.
87 TOP and TOPLEVEL are now available as debugger commands for
88 returning to the top level.
89 * incompatible change: forms evaluated in the REPL now use the
90 global optimization policy.
91 * incompatible change: user- and system-initialization files are
92 no longer processed with LOAD, but by READ and EVAL; hence the
93 global optimization policy, startup package, readtable, etc,
95 * The .fasl file format number has been incremented because of
96 various incompatible changes.
97 * internal entry point details and argument counts no longer appear
98 in backtraces unless explicitly requested by setting
99 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
100 * built-in and standard functions no longer have names like "top
101 level local call to FOO".
102 * fixed bug 32: functions defined in non-null lexical environments
103 now have more legible printed representation
104 * fixed bug 33: functions defined in non-null lexical environemnts
105 are now more amenable to inspection by INSPECT.
106 * workaround for bug 354: XEPs no longer appear in backtraces unless
107 explicitly requested.
108 * fixed bug: receiving the signal which results from attempting to
109 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
110 SIGBUS on other *BSDs) on architectures where the C stack is also
111 the Lisp stack (x86 and x86-64) from foreign code no longer leads
112 to debugger confusion or wild instability. (reported by Cheuksan
114 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
115 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
116 (reported by Lutz Euler)
117 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
118 compiled code. (reported by Paul Dietz)
119 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
120 specialized parameter to a method to something that is not TYPEP
121 the specializer is now possible.
122 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
123 face of package deletion.
124 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
125 pathname. (reported by David Tolpin on c.l.l)
126 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
127 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
128 STRING-STREAM and FILE-STREAM, respectively.
129 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
130 than previously; in addition, its test suite is now run on build.
131 (thanks to Luke Gorrie)
132 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
133 * optimization: added a immediate representation for single-floats
135 * fixed some bugs revealed by Paul Dietz' test suite:
136 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
137 ** attempting to create a package with a colliding nickname causes
138 correctable errors to be signalled.
139 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
140 ** several x86-64 backend bugs related to sign-extension of immediate
143 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
144 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
145 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
147 * fixed loading of multiply forward-referenced layouts.
148 (thanks to Cheuksan Wang)
149 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
150 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
151 (see COMPILE-FILE documentation for details of :PRINT option)
152 * fixed bugs 19 and 317: fixed-format floating point printing is
153 more accurate. This also fixes a bug reported by Adam Warner
154 related to the ~@F format directive.
155 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
157 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
158 dispatch table if the from-char is a dispatch macro character.
159 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
160 by the wordsize no longer ignore the last word. (reported by Lutz
162 * fixed bug in type checking of dynamic-extent variables. (reported
164 * optimization: sequence traversal functions use their freedom to
165 coerce function designators to functions.
166 * optimization: code with many calls to CLOS methods specialized on
167 CLOS classes has had redundant type checks removed.
168 * optimization: added declarations to speed up operations that access
169 the internal character database (for example STRING-UPCASE)
170 * optimization: comparison operations between floats and sufficiently small
171 fixnums no longer create extra rationals
172 * fixed some bugs related to Unicode integration:
173 ** portions of multibyte characters at the end of buffers for
174 character-based file input are correctly transferred to the
175 start of the buffer at the next read.
176 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
177 passing it through to OPEN.
178 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
179 argument given to internal calls to OPEN.
180 ** the built-in comment readers (introduced by character sequences
181 ";" and "#|") are more forgiving to encoding errors; they will
182 STYLE-WARN and then attempt to resync the stream at a character
183 boundary. (thanks to Teemu Kalvas)
184 * fixed some bugs in the x86-64 port:
185 ** Negative short int return values from c-calls are sign-extended
187 ** The stack is aligned to 16-bytes for c-calls, as required by
188 the ABI. (thanks to Cheuksan Wang)
189 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
190 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
191 for enums. (thanks to Vincent Arkesteijn)
192 ** Multiple small optimizations and bugfixes for floating point
194 * fixed some bugs revealed by Paul Dietz' test suite:
195 ** Space, Tab, Linefeed, Return and Page have the invalid
196 secondary constituent character trait.
197 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
199 ** WITH-INPUT-FROM-STRING should only update the index place on
201 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
203 ** Bit-array operations (BIT-AND and similar) worked incorrectly
204 with one-dimensional arrays with fill pointers.
205 ** TYPE-OF failed on a complex with an integer realpart and a
207 ** compiler failure during type inference for the code of form
208 (IF (EQL X (THE ...)) ...) (MISC.535).
210 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
211 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
212 platform. The port supports 61-bit fixnums, large memory spaces
213 and reloading shared object files.
214 * enhancement: saving cores with foreign code loaded is now
215 supported on ppc/Darwin in addition to the previously supported
217 * enhancement: the statistical profiler now walks deeper into the
218 call stack for more meaningful call-graphs and accrued time
219 reports (x86/x86-64 only). It also now reports time spent in
221 * enhancement: it is now possible to trace most individual methods
222 of a generic function in addition to tracing the generic function
224 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
226 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
227 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
228 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
229 produces an error. (thanks to Vincent Arkesteijn)
230 * bug fix: NAMESTRING on pathnames with :WILD components in their
231 directories works correctly. (thanks to Artem V. Andreev)
232 * fixed bug 125: compiler preserves identity of closures. (reported
234 * bug fixed: functions with &REST arguments sometimes failed with
235 "Undefined function" when compiled with (DEBUG 3). (reported by
237 * bug fix: overflow during compiling of setting element of a bit
238 vector with constant index and value. (reported by Timmy Douglas)
239 * build fix: fixed the dependence on *LOAD-PATHNAME* and
240 *COMPILE-FILE-PATHNAME* being absolute pathnames.
241 * on x86 compiler partially supports stack allocation of dynamic-extent
243 * GO and RETURN-FROM do not check the extent of their exit points
244 when compiled with SAFETY 0.
245 * fixed some bugs related to Unicode integration:
246 ** encoding and decoding errors are now much more robustly
247 handled; it should now be possible to recover even from invalid
248 input or output to the terminal. (thanks to Teemu Kalvas)
249 ** provided a first cut at implementing STRING-TO-OCTETS and
250 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
251 ** altered the SB-MD5 contributed module slightly, changing the
252 interface just enough for it to be supportable for builds where
253 lisp characters are not eight bits.
254 * fixed some bugs revealed by Paul Dietz' test suite:
255 ** the FORMATTER-generated functions for ~V[ conditionals require
256 the correct number of arguments.
257 ** READ-FROM-STRING returns the mandated second value when applied
258 to displaced strings.
259 ** the #\Rubout and #\Backspace characters are treated as invalid
260 constituent characters by the tokenizer.
262 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
263 * new feature: reloading changed shared object files with
264 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
265 * new feature: references to foreign variables and functions
266 can now be compiled and loaded before the corresponding shared
267 object file is loaded, as long as the foreign definitions are
268 available at runtime.
269 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
270 supported. (thanks to Dan Debertin)
271 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
272 just keywords. (thanks to Vincent Arkesteijn)
273 * fixed bug #331: structure-class instances corresponding to
274 DEFSTRUCT forms are now created eagerly.
275 * fixed bug #345: backtraces from calls to undefined functions work
276 on x86 as well. Related bug #61 is now also partially fixed on x86
277 (backtraces from throws to unknown catch tags.)
278 * bug fix: lambda-list parsing is now stricter vrt. order and number
279 of lambda-list keywords.
280 * bug fix: as specified by AMOP, an error is signalled if a
281 class-option appears multiple times in a DEFCLASS form. (reported
283 * bug fix: RANDOM can be compiled when the compiler derives the type
284 of its numeric argument as a disjoint set of small integers.
285 (reported by Paul Dietz)
286 * bug fix: starting a core saved with shared objects loaded when
287 those objects are not available no longer causes threaded SBCL to
288 hang. (reported by Sean Ross)
289 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
291 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
292 argument, which must always be NIL. (reported by Kalle Niemitalo)
293 * bug fix: printing 1.0d+23 no longer results in an error.
294 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
295 * bug fix: structure slot setters preserve evaluation order. (thanks
297 * fixed some bugs related to Unicode integration:
298 ** RUN-PROGRAM can allow its child to take input from a Lisp
299 stream. (reported by Stefan Scholl)
300 ** ASDF-INSTALL successfully downloads .tar.gz files in all
301 locales. (reported by Ken Causey)
302 * fixed some bugs revealed by Paul Dietz' test suite:
303 ** INCF, DECF and REMF evaluate their place form as specified in
305 ** FORMATTER expands ~{ ~} iteration directives with V or #
306 parameters correctly.
307 ** FORMATTER deals with the ~@[ ~] conditional directive where the
308 consequent uses no arguments correctly.
309 ** the system has a partial understanding of the (COMPLEX RATIO)
311 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
314 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
315 * new feature: a build-time option (controlled by the :SB-UNICODE
316 keyword feature, enabled by default) for building the system with
317 support for the entire 21-bit character space defined by the
319 * new feature: the system now has rudimentary external-format
320 support; the primary user-visible change at this time is that
321 characters with the high bit set (such as lower-case-e-acute) will
322 print correctly to a terminal in a UTF-8 environment.
323 * minor incompatible change: BASE-CHAR no longer names a class;
324 however, CHARACTER continues to do so, as required by ANSI.
325 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
326 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
327 has been moved to the SB-EXT package (temporarily re-exported from
329 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
330 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
332 * fixed bug #351: better error-handlind and reporting for malformed
334 * fixed bug #350: bignum-printing is now more memory-efficient,
335 allowing printing of very large bignums, eg. (expt 2 10000000).
336 (reported by Bruno Haible)
337 * fixed bug #302: better primitive-type selection for intersection
339 * fixed bug #308: non-graphic characters now all have names, as
340 required. (reported by Bruno Haible)
341 * bug fix: redefining a class with different superclasses now correctly
342 removes it from the direct-subclasses of its previous superclasses.
343 (reported by David Morse)
344 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
345 new value now works. (reported by Bruno Haible)
346 * bug fix: correct canonicalization of multiple non-standard slot
347 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
348 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
349 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
350 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
351 now exists, an signals an error.
352 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
353 during macroexpansion for non-iteration variables as well. (reported
354 by Bruno Haible for CMUCL)
355 * bug fix: Cyclic structures and unprintable objects in compiler
356 messages no longer cause errors. (reported by Bruno Haible)
357 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
358 and READ-FROM-STRING all now return a primary value of NIL if
359 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
360 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
361 (reported by Bruno Haible for CMUCL)
362 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
363 do not designate a condition. (reported by Bruno Haible for
365 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
366 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
367 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
368 permitting resolution in favour of any of the conflicting symbols.
369 (reported by Bruno Haible for CMUCL)
370 * bug fix: EQUAL compiler optimizations is less aggressive on
371 strings which can potentially compare true despite having distinct
372 specialized array element types.
373 * bug fix: unit enumerations can be defined without dividing by
374 zero. (thanks to Vincent Arkesteijn)
375 * FORMAT compile-time argument count checking has been enhanced.
376 (report from Bruno Haible for CMUCL)
377 * a partial workaround for the bug 262: the compiler does not try to
378 inline-expand a local function doing RETURN-FROM from a deleted
379 BLOCK. (thanks to Peter Denno for the bug report and to David
380 Wragg for the simple test case)
381 * fixed some bugs revealed by Paul Dietz' test suite:
382 ** RENAME-PACKAGE allows all package designators as new package
384 ** constraint propagation and lambda variable substitution are
385 more cautious in dealing with partially deleted code.
386 ** compiler sometimes lost reoptimization passes.
387 ** CERROR, given a condition as condition designator, passes its
388 remaining arguments to the continue format control without
390 ** Case-altering FORMAT directives work correctly on non-ASCII
392 ** The REAL type specifier handles bounds outside the floating
393 point ranges without signalling FLOATING-POINT-OVERFLOW.
394 ** Functions with IR1-transformations can create intercomponent
395 references to global functions.
396 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
398 ** FORMAT ~R treats a nil value for its first parameter correctly.
400 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
401 * enhancement: saving cores with foreign code loaded is now
402 supported on x86/NetBSD and sparc/Linux in addition to the previously
404 * bug fix: on some platforms repeated installations caused multiple
405 copies of HTML documentation to be installed -- should not happen
406 any more. (reported by Stefan Scholl)
407 * bug fix: parsing self-recursive alien record types multiple times
408 no longer causes infinite recursion. (reported by Thomas F. Burdick,
409 original patch by Helmut Eller for CMUCL)
410 * bug fix: stack-exhaustion detection works now on NetBSD as well.
411 (thanks to Richard Kreuter)
412 * bug fix: defining classes whose accessors are methods on existing
413 generic functions in other (locked) packages no longer signals
414 bogus package lock violations. (reported by François-René Rideau)
415 * bug fix: special variables as DEFMETHOD parameters no longer have
416 associated bogus type declarations. (reported by David Wragg and
418 * bug fix: read-write consistency on streams of element-type
419 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
420 * bug fix: redefiniton of the only method of a generic function with
421 no DEFGENERIC no longer emits a full WARNING. In addition,
422 redefinition of generic functions with no DEFGENERIC to an
423 incompatible lambda list now signals an error. (thanks to Zach
425 * bug fix: DEFGENERIC now works even when there's a function of the
426 same name in an enclosing lexical environment. (thanks to Zach
428 * fixed compiler failure, caused by instrumenting code during
429 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
430 * optimization: added loop analysis and improved register allocation
431 * fixed some bugs revealed by Paul Dietz' test suite:
432 ** POSITION on displaced vectors with non-zero displacement
433 returns the right answer.
434 ** (SIMPLE-STRING) is a valid type specifier for sequence
436 ** *PRINT-LEVEL* handling for slotless structures is pedantically
438 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
439 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
441 ** The justification version of the FORMAT ~< directive treats
442 non-zero minpad parameter correctly.
444 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
445 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
446 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
447 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
448 the supported interface.
449 * new feature: Single-stepping of code compiled with DEBUG 2 or
450 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
451 * new feature: saving cores with foreign code loaded is now
452 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
453 Timothy Moore's work for CMUCL)
454 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
455 parameters to * instead of NIL if no initform is supplied.
456 (reported by Johan Bockgaard)
457 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
458 correctly when FUNCALL appears as the car of the form. Note:
459 despite this FUNCALL forms are not currently subject to
460 compiler-macro expansion. (port of Raymond Toy's fix for the
461 same from CMUCL, reported by Johan Bockgaard)
462 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
463 (thanks for Teemu Kalvas)
464 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
465 case instead of terminating the iteration (thanks for Julian
466 Squires, Sean Champ and Raymond Toy)
467 * bug fix: incorrect expansion of defgeneric that caused a style
468 warning. (thanks for Zach Beane)
469 * bug fix: slot accessor effective method computation works properly
470 for classes with multiple non-standard applicable methods on
471 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
472 * on x86 compiler supports stack allocation of results of LIST and
473 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
474 CMUCL implementation by Gerd Moellmann)
475 * fixed some bugs revealed by Paul Dietz' test suite:
476 ** FORMAT strings with both the ~<~:;~> form of the justification
477 directive and pretty-printing directives cause an error.
479 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
480 * incompatible change: the internal functions
481 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
482 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
483 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
484 instead of the old functions.
485 * new feature: on platforms where "dladdr" is available foreign
486 function names now appear in backtraces. (based on Helmut Eller's
488 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
489 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
491 * documentation: ASDF manual is now distributed with SBCL.
492 * bug fix: SBCL can now load its contributed modules with REQUIRE
493 even if the system-provided entries have been removed from
494 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
496 * bug fix: backtraces involving undefined functions or assembly
497 routines are more informative. (thanks to Brian Downing)
498 * bug fix: mutually referent alien structures now work correctly.
499 (reported by Rick Taube)
500 * bug fix: structures defined by WITH-ALIEN can be referred to
501 within other definitions in the same WITH-ALIEN.
502 * bug fix: division operators (MOD, TRUNCATE and the like) with
503 constant zero divisors and integer dividends no longer generate
505 * bug fix: provide default methods for INPUT-STREAM-P and
506 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
507 * bug fix: improve the detection and resolution of MOP metacycles
508 (where effective method computation depends on the generic function
509 having its effective method computed). (reported by Bruno Haible)
510 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
511 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
512 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
513 * bug fix: pathnames with relative directory components can be
514 represented relative to default pathnames.
515 * optimization: in taking the GCD of bignums, reduce the two bignums
516 to approximately the same size (using Euclid's algorithm) before
517 applying the more sophisticated binary GCD. (thanks to Juho
519 * optimization: COUNT on bitvectors now operates word-at-a-time.
520 * optimization: ASH with a positive, but not necessarily constant,
521 (leftwards) shift, when in a modular context, is compiled to a
523 * fixed some bugs revealed by Paul Dietz' test suite:
524 ** FORMAT variable parameters ("~V<char>") are defaulted properly
525 if the corresponding argument is NIL.
526 ** FORMAT directives accepting a minpad parameter treat negative
528 ** literal commas as character parameters to FORMAT directives are
530 ** literal spaces directly after ~<Newline> directives within a
531 format-logical-block (~:< ~@:>) do not induce :FILL-style
532 conditional newlines.
533 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
534 as specified: it no longer includes conditional newlines).
535 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
537 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
538 :PER-LINE-PREFIX argument does not evaluate to a string.
540 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
541 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
542 the manual for details; for now, package locks can be disabled by
543 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
544 if no major problems are found then it is likely that they will be
545 compiled in unconditionally.
546 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
547 unsupported operators on all platforms. To load a shared library
548 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
549 object file, link it into a shared library outside of SBCL and
550 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
552 * bug workaround: ROOM T and the GC were not working together
553 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
554 reporting facility aren't preserved. That reporting has been
555 disabled, so now until and unless someone figures out how to make it
556 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
557 * minor incompatible change: as threatened around sbcl-0.8.0, the
558 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
559 The symbols are also exported from SB-PCL for backwards
560 compatibility, but more so than before SB-PCL should be treated as
561 an implementation-internal package.
562 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
564 * the SB-SPROF contrib now works on (most) non-x86 architectures.
565 It is known as of this release not to work on the Alpha, however.
566 * fixed bug #167: errors signalled due to illegal syntax in method
567 bodies are now more legible.
568 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
569 designators and can hence be used with TYPEP.
570 * fixed bug #333: CHECK-TYPE now ensures that the type error
571 signalled, if any, has the right object to be accessed by
572 TYPE-ERROR-DATUM. (reported by Tony Martinez)
573 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
574 5.1.2.3 for multiple-value place subforms. (reported by Kalle
576 * fixed bug #334: programmatic addition of slots using specialized
577 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
578 and :ALLOCATION :CLASS slots.
579 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
580 float exponents. (rereported by Peter Seibel)
581 * fixed a bug: #\Space (and other whitespace characters) are no
582 longer considered to be macro characters in standard syntax by
584 * fixed bug: initialization of condition class metaobjects no longer
585 causes an instance of the condition to be created. (reported by
587 * fixed bug: it is now possible to have slots such that
588 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
589 description is neither :INSTANCE nor :CLASS.
590 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
591 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
592 SLOT-VALUE-USING-CLASS), and no longer causes errors with
593 non-standard SLOT-DEFINITION-ALLOCATION values.
594 * fixed bugs: various race conditions handling exiting threads.
595 CL-PPCRE's thread torture test now passes.
596 * fixed arguably-a-bug: GC time proportional to number of threads in
597 system even when most of them are idle
598 * optimization: improved performance of BIT and SBIT on bit-vectors.
599 * .fasl file incompatibility: The fasl file version number has
600 been incremented because of changes associated with package locks.
602 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
603 * minor incompatible change: the system no longer provides
604 optimization hints (conditions which are TYPEP
605 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
606 mode; these hints are still emitted when higher SPEED optimization
608 * new contrib module: a sampling profiler (profiling by statistical
609 sampling, rather than by instrumenting functions) is available as
610 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
611 * the behaviour of the standard function ED is now customizeable by
612 third parties through a hook variable: see ED's documentation
613 string for information on the protocol.
614 * the compiler no longer emits efficiency notes for (FUNCALL X)
615 when the type of X is uncertain under default optimization
617 * fixed bug 276: mutating a binding of a specialized parameter to a
618 method to something that is not TYPEP the specializer is now
620 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
621 exhibit the required behaviour on the x86 platform. (thanks to
622 Peter van Eynde, Eric Marsden and Bruno Haible)
623 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
624 even for difficult arguments. (reported by Peter Graves)
625 * fixed bug 141a: the backquote printer now descends quoted
627 * fixed another bug in backquote printing: no more destructive
628 modification of the form's list structure. (reported by Brian
630 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
631 move between its address being taken and the call to
632 interrupt_thread, fixing a crashing race condition.
633 * the SB-POSIX contrib implementation has been adjusted so that it
634 no longer exhibits ridiculously poor performance when constructing
635 instances corresponding to C structs.
637 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
638 * minor incompatible change: the sb-grovel contrib now treats C
639 structures as alien (in the sense of SB-ALIEN) objects rather than
640 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
641 has implications for memory management of client code
642 (sb-grovel-returned objects must now be manually managed) and for
643 type safety (alien objects now have full types).
644 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
645 used to control emission of compiler diagnostics, rather than the
646 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
647 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
648 quality should be considered deprecated.
649 * install.sh now installs the user manual as well
650 * (not quite a new documentable feature, but worth considering in
651 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
652 beginnings of a semantically meaningful condition hierarchy is
653 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
655 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
656 designator as the defaults argument.
657 * fixed bug: Displaced arrays whose displaced-to array has become
658 too small now cause ARRAY-DIMENSION to signal an error, providing
659 for safer bounds-checking. (reported by Bruno Haible)
660 * fixed bug: DEFCLASS slot definitions with identical :READER and
661 :WRITER names now signal a reasonable error. (reported by Thomas
663 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
665 * fixed bug: Invalid slot specification errors now print correctly.
666 (thanks to Zach Beane)
667 * fixed bug 320: Shared to local slot value transfers in class
668 redefinitions now happen corrently from superclasses as
669 well. (reported by Bruno Haible)
670 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
672 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
673 improper lists correctly. (reported by Bruno Haible)
674 * fixed bug 313: source-transform for <fun-name> was erroneously
675 applied to a call of a value of a variable with name <fun-name>.
676 (reported by Antonio Menezes Leitao)
677 * fixed bug 307: The obsolete instance protocol ensures that
678 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
679 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
680 remove dead unknown-values globs from the middle of the stack.
681 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
682 easier to resume long computations after using *BREAK-ON-SIGNALS*
683 to diagnose and fix failures. (thanks to Nikodemus Siivola)
684 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
685 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
686 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
687 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
688 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
690 * fixed a (fairly theoretical) bug in string printing: if
691 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
692 does not have array-element-type equal to the most general string
694 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
695 function name. (thanks to Nikodemus Siivola)
696 * fixed bug: compile-time format string checker failed on
697 non-closed ~{. (reported by Thomas F Burdick)
698 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
699 don't warn on legal loop code involving more than one aggregate
700 boolean. (thanks to Nikodemus Siivola)
701 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
702 in keyword position to MAKE-PACKAGE was not being reported as
704 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
705 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
707 * fixed bugs 280 and 312: the checking for multiple definitions in a
708 file is less likely to become confused by uses of inline
710 * fixed bug: the #S reader macro performs the keyword coercion
711 specified for slot names. (reported by Kalle Niemitalo)
712 * fixed bug: lambda lists may contain symbols whose names start with
713 & but are not lambda-list-keywords; their occurrence triggers a
715 * fixed bug 321: define-method-combination argument lambda lists do
716 not cause invalid code to be generated when &optional or &aux
717 variables have default values. (reported by Bruno Haible)
718 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
719 have CLOS classes; this appears also to have fixed reported
720 instability in stack exhaustion detection.
721 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
722 MUFFLE-WARNING when no associated restart is present is now
724 * optimization: rearranged the expansion of various defining macros
725 so that each expands into only one top-level form in a
726 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
728 * optimization: used a previously little-used slot in symbols to
729 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
731 * fixed some bugs revealed by Paul Dietz' test suite:
732 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
734 ** Fixed an optimization bug related to inheritance of initforms
735 from local to shared slots.
736 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
737 ** CAST optimizer forgot to flush argument derived type.
738 ** print/read consistency on floats is now orders of magnitude
739 more likely. (thanks also to Bruno Haible for a similar report
741 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
742 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
743 for objects of type REAL. Make it so.
744 ** SXHASH is specified (infelicitously) to respect similarity,
745 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
746 it so. (thanks to Markus Ziegler)
747 ** on the Alpha, the compiler succeeds in compiling functions
748 returning a known number of arguments greater than 63.
749 ** fixed handling of invalid NIL arguments in keyword position
751 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
753 ** provided a readably-printable representation for RANDOM-STATE
755 ** ensured that pathnames, where they have a namestring, always
756 print using #P"..." syntax.
758 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
759 * Support for the forthcoming 2.0 version of the NetBSD kernel
760 running on x86 hardware has been added. (thanks to Perry
761 E. Metzger most immediately, and others for their past work)
762 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
763 current release version; SBCL's OpenBSD support had been broken
764 since about the time of OpenBSD's switch to ELF binary format.)
765 * [placeholder for DX summary]
766 ** user code with &REST lists declared dynamic-extent, under high
767 speed or space and low safety and debug optimization policy.
768 * The manual has been converted to Texinfo format and the debugger
769 chapter from the cmucl manual has been added.
770 * A facility has been added to extract documentation strings from
771 sbcl and store them as Texinfo-formatted snippets for inclusion in
772 the manual (via Texinfo's @include directive)
773 * bug fix: compiler emitted division in optimized DEREF. (thanks for
774 the test case to Dave Roberts)
775 * bug fix: multidimensional simple arrays loaded from FASLs had fill
776 pointers. (reported by Sean Ross)
777 * bug fix: PROFILE output is printed nicely even for large numerical
778 values. (thanks to Zach Beane)
779 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
780 greater than 32 handle EOF correctly.
781 * bug fix: on X86 an immediate argument of the IMUL instruction is
782 correctly printed in disassembly. (thanks to Lutz Euler)
783 * bug fix: class slots in redefined classes preserve their old
784 values. (thanks to Bruno Haible and Nikodemus Siivola)
785 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
786 succeeds. (reported by Marco Baringer)
787 * fixed some bugs revealed by Paul Dietz' test suite:
788 ** READ-SEQUENCE now works on ECHO-STREAMs.
789 ** RATIONALIZE works more according to its specification. (thanks
792 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
793 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
794 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
795 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
796 should matter to you only if you rebind the printer control
797 variables and then find you want different bindings in the
798 debugger than in the ordinary execution of your program.)
799 * The runtime build system has been tweaked to support building
800 (on SPARC/SunOS) using a C compiler which invokes Sun's own
801 assembler and linker. (thanks to Nikodemus Siivola)
802 * Unbound, undefined, undeclared variables now trigger full
803 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
804 more often programmer error than deliberate exploitation of undefined
806 * optimization: the hash algorithm for strings has changed to one
807 that is less vulnerable to spurious collisions. (thanks to Juho
809 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
810 needless bounds checking. (thanks to Juho Snellman)
811 * optimization: implemented multiplication as a modular
812 (UNSIGNED-BYTE 32) operation on the PPC backend.
813 * fixed some bugs revealed by Paul Dietz' test suite:
814 ** ADJUST-ARRAY now copies the datum in a zero rank array if
816 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
817 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
818 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
820 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
822 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
824 ** ECHO-STREAMs no longer attempt to echo the end of file value to
825 their output stream on EOF from read.
826 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
827 have been read to end-of-file.
828 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
830 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
831 description of determination of which consecutive characters
833 ** Printing the "Space" character with escaping on now yields "#\\ ",
834 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
835 ** Reading floating-point numbers with *READ-BASE* set to a number
836 less than 10 works correctly.
837 ** Reading floating-point numbers with *READ-BASE* set to a number
838 more than 10 works correctly.
839 ** Printing with *PRINT-READABLY* targets the standard readtable, not
840 the readtable currently in effect.
842 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
843 * minor incompatible change: parsing of namestrings on a physical
844 (Unix) host has changed; numbers after the final #\. in a
845 namestring are no longer interpreted as a version field. This is
846 intented to be largely invisible to the user, except that the
847 meaning of the namestring "*.*.*" has changed: it now refers to a
848 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
849 should usually be replaced by
850 (make-pathname :name :wild :type :wild :version :wild)
851 with the added benefit that this is more likely to be portable.
852 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
853 option to OPEN now signals an error if the file being opened
854 exists; this may have an impact on existing code.
855 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
856 signals received, on the PowerPC platforms (both Linux and
857 Darwin). (thanks to Pierre Mai for pointing out the location of
859 * several fixes on OS X: The system now builds and runs cleanly on
860 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
861 (thanks to Brian Mastenbrook)
862 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
863 second-resolution: integer multiples of 1/3600 between -24 and 24.
864 (thanks to Vincent Arkesteijn)
865 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
866 type when called with 1 argument; PEEK-CHAR checked type of
867 PEEK-TYPE only after having read first character from a
868 stream. (reported by Peter Graves)
869 * bug fix: the garbage collector now has much better locality
870 behaviour, and in particular no longer treats all memory as being
871 exhausted when many small objects point to each other in a deeply
873 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
874 recognized as being TYPEP their class.
875 * bug fix: the PUSHNEW documentation string has been corrected.
876 (thanks to Vincent Arkesteijn)
877 * bug fix: defaulting of the value for the last of an atypically
878 large number of multiple values being bound was not being
879 performed correctly on the Alpha or PPC platforms
880 * optimization: implemented multiplication as a modular
881 (UNSIGNED-BYTE 32) operation on the x86 backend.
882 * optimization: SEARCH on simple-base-strings can now be open-coded.
883 (see also contrib/compiler-extras.lisp for inspiration for
884 teaching the compiler about the Boyer-Moore algorithm).
885 * value, returned by MAX (and MIN) called with several EQUALP, but
886 not EQL, arguments now does not depend on compiler settings.
887 * fixed some bugs revealed by Paul Dietz' test suite:
888 ** in stack analysis liveness information is propagated from
889 non-local entry points.
890 ** pathwise CAST removing failed when the CAST node did not start
892 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
893 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
895 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
896 host is already defined.
897 ** RENAME-FILE works on streams instead of signalling an internal
899 ** PEEK-CHAR uses the current readtable when determining whether
900 or not a character is whitespace.
901 ** MERGE-PATHNAMES handles the case when the pathname does not
902 specify a name while the default-pathname specifies a version.
903 ** Pathnames now stand a better chance of respecting print/read
905 ** Attempting to use standardized file system operators with a
906 pathname with invalid :DIRECTORY components signals a
908 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
910 ** DIRECTORY on logical pathnames is more correct.
911 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
912 signal a TYPE-ERROR if their argument is not a stream.
913 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
914 designator argument does not designate a stream.
915 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
916 examining the synonym.
917 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
919 ** OPEN and WITH-OPEN-STREAM allow opening streams with
920 element-type larger than ([UN]SIGNED-BYTE 32).
922 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
923 * When built with the :SB-FUTEX feature, threaded builds now take
924 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
925 for faster/more reliable mutex and condition variable support.
926 * Incompatible change (but one you probably shouldn't have been using
927 anyway): the interface and code for arbitrating between multiple
928 threads in the same user session has been redesigned.
929 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
930 error when called without an explicit environment argument.
931 (thanks to Nikodemus Siivola)
932 * bug fix: buffered :DIRECTION :IO streams are less likely to become
933 confused about their position. (thanks to Adam Warner and Gerd
935 * bug fix: Pretty printing backquoted forms with unquotations in the
936 argument list position of various code constructs such as LAMBDA
937 now works correctly. (reported by Paul Dietz)
938 * bug fix: Pretty printing unquotations no longer loses all
939 stream position information.
940 * optimization: performance of string output streams is now less
941 poor for multiple small sequence writes.
942 * optimization: performance of CSUBTYPEP in the presence of complex
943 expressions involving CONS and NOT many times has been improved.
944 (reported by Paul Dietz)
945 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
947 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
948 streams for the REPL, for the debugger, and so on. (thanks to
950 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
951 * fixed some bugs revealed by Paul Dietz' test suite:
952 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
953 is ignored if it is too small, rather than propagating through
954 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
955 ** extremely complex negations of CONS types were not being
956 sufficiently canonicalized, leading to inconsistencies in
958 ** VALUES tranformer lost derived type.
960 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
961 * fixed a bootstrapping bug: the build process no longer assumes
962 that the various BOOLE-related constants have the same value in
963 host and target lisps. (noted by Paul Dietz' test suite on an
964 SBCL binary built from CLISP)
965 * The system can now be dynamically linked on the MIPS platform,
966 which enables dynamic loading of foreign code from Lisp. (thanks
967 to Ralf Baechle for discussions on the MIPS ABI)
968 * The system now records debugging information for its own source
969 files in a filesystem-position-independent manner, relative to
970 the "SYS" logical host.
971 * fixed a compiler bug: MV-LET convertion did not check references
972 to the "max args" entry point. (reported by Brian Downing)
973 * tweaked disassembly notes to be less confident about proclaiming
974 some instruction as an LRA. (thanks to Brian Downing)
975 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
976 now each have their own history, command character, and other
977 characteristics. (thanks to David Lichteblau)
978 * fixed some bugs revealed by Paul Dietz' test suite:
979 ** compiler failure in compiling LOGAND expressions including a
981 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
982 shift greater than 32.
983 ** FLUSH-DEST did not mark blocks for type check regeneration.
984 ** HANDLER-CASE failed to accept declarations in handler clauses
985 in some circumstances.
987 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
988 * New code in contrib/sb-introspect (still probably not entirely
989 stable yet) provides some support for smart Lisp development
990 environments like SLIME.
991 * The conditions signalled for errors occurring when loading .fasl
992 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
993 in a way which should help ASDF recover gracefully.
994 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
995 stuff has been cleaned up. If you code contrib/ stuff, this might
996 affect you, and you can look at contrib/README, contrib/STANDARDS,
997 and/or the 0.8.4.27 diff to check.
998 * In full calls the compiler now does not generate checks for declared
999 argument types for all arguments.
1000 * various threading fixes
1001 ** and some experimental patches which didn't make it into
1002 the main tree for this release, but which are shipped in
1003 contrib/experimental-thread.patch as a possible fix for some
1004 failures (deadlock, spinning...) in GC-intensive multithreaded
1006 * fixed PPC build problem (source code incompatibility of different
1007 library versions): added offsetof() hackery which attempts to divine
1008 where glibc maintainers put uc_mcontext today
1009 * fixed bug 282: compiler does not trust type assertions while passing
1010 arguments to a full call.
1011 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1012 &OPTIONAL VALUES type parameter.
1013 * fix bug 214: algorithm for noting rejected templates is now more
1014 similar to that of template seletion. (also reported by rydis on
1016 * fixed bug 141b: printing backquoted information readably and prettily
1017 inserts a space where necessary.
1018 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1019 (CHARACTER 10) are now reported as errors, rather than propagated
1020 as unknown types. (reported by piso on #lisp)
1021 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1022 with values NIL and :ERROR. (thanks to Milan Zamazal)
1023 * fixed bug 191c: CLOS now does proper keyword argument checking as
1024 described in CLHS 7.6.5 and 7.6.5.1.
1025 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1026 longer raise an error; further, using a list as a for-as-arithmetic
1027 counter now raises a meaningful error.
1028 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1029 now understood by sequence creation functions such as MAKE-SEQUENCE
1031 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1032 a STRING-INPUT-STREAM.
1033 * compiler enhancement: SIGNUM is now better able to derive the type
1035 * type declarations inside WITH-SLOTS are checked. (reported by
1037 * fixed some bugs revealed by Paul Dietz' test suite:
1038 ** incorrect optimization of TRUNCATE for a positive first
1039 argument and negative second.
1040 ** compiler failure in let-convertion during flushing dead code.
1041 ** compiler failure while deriving type of TRUNCATE on an
1042 interval, containing 0.
1043 ** ASH of a negative bignum by a negative bignum count now returns
1045 ** intersection of CONS types now canonicalizes properly, fixing
1046 inconsistencies in SUBTYPEP.
1048 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1049 * incompatible change: The --disable-debugger command line
1050 option now clobbers the debugger at a more fundamental
1051 level, by redefining #'INVOKE-DEBUGGER instead of by
1052 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1053 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1054 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1055 enter the debugger REPL and then suffer recursive errors
1056 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1057 while under the new system, BREAK in --disable-debugger mode
1058 terminates the system just as an unhandled error would.
1059 * fixed compiler performance when processing loops with a step >1;
1060 * bug fix: DOCUMENTATION now retrieves generic function
1061 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1062 support has been systematized, and now supports the methods
1063 specified by ANSI, along with a default method and a method for
1064 slot documentation. (reported by Nathan Froyd)
1065 * bug fix: effective methods associated with a generic function are
1066 no longer cached over a change of that generic function's method
1067 combination. (reported by Andreas Fuchs)
1068 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1069 names a non-existent directory. (thanks to Andreas Fuchs)
1070 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1071 the PPC platform, lead to stack corruption; this has been fixed.
1072 (reported by Rainer Joswig)
1073 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1074 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1075 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1076 arithmetic in general, is now much more reliable on the PPC
1078 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1079 platform now returns the right answer.
1080 * optimization: restored some effective method precomputation in
1081 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1082 precomputation is now tunable.
1083 * optimization: compiler-internal data structure use has been
1084 reviewed, and changes have been made that should improve the
1085 performance of the compiler by about 20%.
1086 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1087 simple-base-strings and simple-bit-vectors is improved.
1088 * optimization: the optimization of 32-bit logical and arithmetic
1089 functions introduced in version 0.8.3 on the x86 has been
1090 implemented on the mips, ppc and sparc platforms; an
1091 implementation of the same facility, but for 64-bit arithmetic,
1092 has been added for the alpha.
1093 * microoptimization: the compiler is better able to make use of the
1094 x86 LEA instruction for multiplication by constants.
1095 * bug fix: in some situations compiler did not report usage of
1096 generic arithmetic in (SPEED 3) policy.
1097 * bug 145b fix: compiler used wrong type specifier while converting
1098 MEMBER-types to numeric.
1099 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1101 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1102 index argument now works properly on the Alpha platform.
1103 * bug fix: floating point exception treatment on the Alpha platform
1105 * bug fix: FILE-POSITION works much better on string input and
1106 output streams. (thanks to Nikodemus Siivola)
1107 * bug fix: many threading/garbage collection symptoms sorted.
1108 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1109 might be pseudo-atomic.
1110 * internal change: Stopping for GC is now done with signals not
1111 ptrace. GC is now done in whichever thread wanted it, instead of
1113 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1115 * bug fix: result form in DO is not contained in the implicit
1117 * incompatible change: ICR structure is changed; the value part of
1118 CONTINUATION is now called LVAR; corresponding functions are
1119 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1120 * added type deriver for ISQRT (thanks to Robert E. Brown).
1121 * bug fix: better support for loading from the command line when an
1122 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1124 * fixed some bugs revealed by Paul Dietz' test suite:
1125 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1126 ** ROUND and FROUND now give the right answer when given very
1127 small float arguments.
1128 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1130 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1131 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1132 ** LCM should return a non-negative integer.
1133 ** PARSE-INTEGER returned the index of a terminator instead of the
1134 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1135 ** PARSE-INTEGER returned an incorrect index being applied to a
1137 ** LCM with two arguments of 0 returns 0 rather than signalling
1139 ** unsigned addition of a 32-bit constant with the high bit set no
1140 longer causes an internal compiler error on the x86.
1141 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1142 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1143 with negative last argument.
1144 ** byte specifiers with zero size and position no longer cause
1145 an error during type derivation.
1146 ** bignum multiplication on the Alpha platform now returns the
1148 * porting: The system now builds on SuSE AMD64, although it still
1149 generates a 32-bit binary.
1150 * .fasl file incompatibility: The fasl file version number has
1151 been incremented (because of the changes to internal compiler
1152 data structures referred to above).
1154 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1155 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1156 more accurately, on the Darwin kernel running on PowerPC hardware.
1157 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1158 * Compiler code deletion notes now signal a condition of type
1159 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1160 an associated MUFFLE-WARNING restart.
1161 * The compiler now performs limited argument count validation of
1162 constant format strings in FORMAT, and where appropriate in ERROR,
1163 CERROR and WARN. (thanks to Gerd Moellmann)
1164 * New ASDF-INSTALL contrib can be used for automatic download and
1165 installation of third-party Lisp code from CCLAN or other sites
1167 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1168 which forces another thread to execute a function supplied by the
1170 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1171 now accept and act upon their :ELEMENT-TYPE keyword argument.
1172 (reported by Martin Atzmueller, Edi Weitz)
1173 * bug fix: FILE-POSITION now accepts position designators up to
1174 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1175 is the greater. (thanks to Patrik Nordebo)
1176 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1177 Kalle Olavi Niemitalo)
1178 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1179 printable. (reported by Eric Marsden)
1180 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1181 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1182 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1183 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1184 (thanks to Stig Sandoe)
1185 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1186 UNREAD-CHAR now correctly includes the unread character in the
1187 target sequence. (thanks to Gerd Moellmann)
1188 * bug fix in threaded builds: the system can now be suspended and
1189 resumed by shell job control with minimal disruption.
1190 * bug fixes in times and timezones >2038AD
1191 * better handling of "where is GNU make?" problem in build scripts
1192 (thanks to Nikodemus Siivola)
1193 * new optimization: inside a named function any reference to a
1194 function with the same name is considered to be a self-reference;
1195 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1196 optimization quality.
1197 * new optimization on x86: logical functions and + now have
1198 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1199 used when the result is truncated to 32 bits.
1200 * VALUES declaration is partially enabled.
1201 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1202 * fixed some bugs revealed by Paul Dietz' test suite:
1203 ** The system now obeys the constraint imposed by
1204 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1205 types form a lattice under type intersection.
1206 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1207 ** ASSOC now ignores NIL elements in an alist.
1208 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1209 and (1+ MOST-POSITIVE-FIXNUM) answers.
1210 ** The addition of a method with invalid qualifiers to a generic
1211 function does not cause an error to be signalled immediately;
1212 a warning is signalled, and the error is generated only on
1213 calling the generic function.
1214 * changed .fasl file version number, in part to add type codes for
1215 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1216 obscure ANSI requirements
1218 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1219 * fixed bug 148: failure to inline-expand a local function left
1220 garbage, confusing the compiler.
1221 * fixed bugs 3cd: structure slot readers perform type check if the
1222 slot can have an invalid value (i.e. it is either not initialized
1223 or can be written with a less specific slot writer).
1224 * bug fix: the compiler now traps array references to elements off
1225 the end of an array; previously, the bounds checking in some
1226 circumstances could go off-by-one.
1227 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1229 * type declarations for array element types now obey the description
1230 on the CLHS page "Declaration TYPE", as per discussions on
1231 sbcl-help around 2003-05-08. This means that a declaration
1232 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1233 declaration, all references to BAR will be asserted or assumed
1234 (with THE, so dependent on compiler policy) to involve objects of
1235 type FOO. Note that no such declaration is implied in
1236 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1237 * declared types of functions from the "Conditions"
1238 chapter. (reported by Paul Dietz)
1239 * bug fix: CERROR accepts a function as its first argument.
1240 * bug fix: NTH an NTHCDR accept a bignum as index
1241 arguments. (reported by Adam Warner)
1242 * optimization: character compare routines now optimize comparing
1243 against a constant character. (reported by Gilbert Baumann)
1244 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1245 argument now works properly on the MIPS platform.
1246 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1247 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1248 declared to have a functional type, does not cause an error
1250 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1251 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1252 * fixed bug: if last continuation of a deleted block has a
1253 destination, this destination should be deleted too. (reported by
1255 * fixed a bug in the bootstrap process: the host compiler's values
1256 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1257 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1258 test case from Patrik Nordebo)
1259 * improved the ability of the disassembler on the PPC platform to
1260 provide helpful disassembly notes.
1261 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1262 the class in more cases than previously.
1263 * bug fix: FILE-POSITION now understands :START and :END for
1264 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1265 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1266 8) data. (thanks to David Lichteblau)
1267 * bug fix: it is possible to add a method to a generic function
1268 without lambda list.
1269 * bug fix: reader failed to signal END-OF-FILE inside an
1270 object representation. (reported by Nikodemus Siivola)
1271 * fixed some bugs revealed by Paul Dietz' test suite:
1272 ** LAST and [N]BUTLAST should accept a bignum.
1273 ** condition slot accessors are methods.
1274 ** (VECTOR NIL) is a subtype of STRING.
1276 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1277 * minor incompatible change: some nonsensical specialized lambda
1278 lists (used in DEFMETHOD) which were previously ignored now signal
1280 * minor incompatible change: the system is now aware of the types of
1281 variables in the COMMON-LISP package, and will signal errors for
1282 most violations of these type constraints (where previously they
1283 were silently accepted).
1284 * minor incompatible change: COMPILE-FILE now uses the freedom
1285 afforded (ANSI 3.2.2.3) to use derived function types for
1286 functions defined in the same file. This also permits the system
1287 to warn on static type mismatches and function
1288 redefinition. (Currently it does not work with high DEBUG level.)
1289 * minor incompatible change: VALUES declaration is disabled.
1290 * When issuing notes, the compiler now signals a condition of type
1291 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1292 restart for use in user handlers. It is expected that the
1293 COMPILER-NOTE condition will eventually become a condition
1294 supertype to a hierarchy of note types, which will then be
1295 handleable in a similar fashion. However, at the moment, no such
1296 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1297 but it's an implementation detail, not a classification for the
1299 * Changes in type checking closed the following bugs:
1300 ** type checking of unused values (192b, 194d, 203);
1301 ** template selection based on unsafe type assertions (192c, 236);
1302 ** type checking in branches (194bc).
1303 * A short form of VALUES type specifier has ANSI meaning (it has
1304 increased the number of situations when SBCL cannot perform type
1306 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1307 keywords or constants is permissible.
1308 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1309 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1310 classes. (thanks to Antonio Martinez)
1311 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1312 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1313 argument to a pretty-print stream. Instead, it leaves any such
1314 operations to DESCRIBE-OBJECT methods.
1315 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1317 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1318 lambda lists. (thanks to David Lichteblau)
1319 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1320 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1321 (thanks to Gerd Moellmann)
1322 * fixed bug 246: increased compilation speed of long
1323 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1325 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1327 * DEFCONSTANT now throws a condition of type
1328 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1329 constant to a non-EQL value; CONTINUE and ABORT restarts
1330 respectively change and preserve the value.
1331 * fixed bug 63: The code walker, part of the implementation of CLOS,
1332 is now better at handling symbol macros.
1333 * bug fix: There is no longer an internal implementation type named
1334 CL:LENGTH. (reported by Raymond Toy)
1335 * bug fix: In macro-like defining macros/special operators the
1336 implicit block does not enclose the lambda list.
1337 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1338 enclosed in parentheses) are not suitable as type specifiers, and
1339 their use properly signals an error now.
1340 * bug fix: An argument count mismatch for a type specifier in code
1341 being compiled no longer causes an unhandled error at compile
1342 time, but signals a compile-time warning.
1343 * fixed simple vector readable printing
1344 * bug fix: DESCRIBE takes more care over whether the class
1345 precedence list slot of a class is bound before accessing it.
1346 (reported by Markus Krummenacker)
1347 * bug fix: FORMATTER can successfully compile pretty-printer format
1348 strings which use variants of the ~* directive inside.
1349 * bug fix: SEARCH now applies its TEST predicate to the elements of
1350 the arguments in the correct order. (thanks to Wolfhard Buss)
1351 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1352 in type inference. (thanks to Robert E. Brown)
1353 * bug fix: Reading in symbols with an explicit package name of ""
1354 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1355 * fixed some bugs revealed by Paul Dietz' test suite:
1356 ** NIL is now allowed as a structure slot name.
1357 ** Arbitrary numbers, not just REALs, are allowed in certain
1358 circumstances in LOOP for-as-arithmetic clauses.
1359 ** Multiple class redefinitions before slot access no longer
1360 causes a type error.
1361 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1362 association between the name and a class.
1363 ** Generic functions with non-standard method-combination and over
1364 five methods all of which return constants no longer return NIL
1365 after the first few invocations. (thanks to Gerd Moellmann)
1366 ** CALL-NEXT-METHOD with no arguments now passes the original
1367 values of the arguments, even in the presence of assignment.
1368 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1370 ** DISASSEMBLE works with closures and funcallable instances.
1371 ** ADD-METHOD now returns the generic function, not the new method.
1372 ** FIND-METHOD signals an error if the lengths of the specializers
1373 is incompatible with the generic function, even if the ERRORP
1375 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1376 which its argument is a member.
1377 ** DEFCLASS only redefines the class named by its class-name
1378 argument if that name is the proper name of the class;
1379 otherwise, it creates a new class.
1380 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1381 of the UNBOUND-SLOT condition to the name of the slot.
1382 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1383 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1384 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1385 SLOT-MAKUNBOUND in the specified fashion.
1387 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1388 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1389 cross-compilation host. As a consequence, we can now bootstrap our
1390 way up to SBCL starting with a bare gcc toolchain and human-readable
1391 source code (first the source to CLISP, then the source to SBCL).
1392 * A contributed module containing a partial implementation of the
1393 simple-streams interface has been included. (thanks to Rudi
1395 * A contributed module implementing the RFC1321 Message Digest
1396 Algorithm, known as MD5, has been included.
1397 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1398 no longer has any effect, as the code controlled by this feature
1399 has been deleted. (As far as we know, no-one has ever built using
1400 this feature, and its semantics were confused in any case).
1401 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1402 work on conditions (as required by the ANSI specification),
1403 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1404 expected behaviour on conditions. Users should note, however,
1405 that such behaviour is not required by the ANSI specification,
1406 and so use of this behaviour may render their code unportable.
1407 * fixed some bugs revealed by Paul Dietz' test suite:
1408 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1410 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1411 ** SXHASH on condition objects no longer returns NIL.
1412 ** :ALLOCATION :CLASS slots are better treated; their values are
1413 updated on class redefinition, and initforms inherited from
1414 superclasses are applied.
1415 ** REMOVE-METHOD returns its generic function argument even when
1416 no method was removed.
1417 ** SHARED-INITIALIZE now initializes the values of the requested
1418 slots, including those with :ALLOCATION :CLASS.
1419 ** ALLOCATE-INSTANCE now works on structure classes defined via
1420 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1422 ** SLOT-EXISTS-P now works on conditions, as well as structures
1424 ** MAKE-LOAD-FORM now has the required methods on
1425 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1426 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1427 keyword, and now implements the SLOT-NAMES argument.
1428 ** methods with &OPTIONAL arguments no longer allow too many
1429 arguments to be passed in the call without error.
1430 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1431 option is consistent with the required arguments of the generic
1432 function lambda list.
1433 * bug fix: REQUIRE accepts a string designator. (Thanks to
1435 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1436 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1437 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1438 * bug fix: The compiler now checks for duplicated variables in macro
1440 * bug fix: SETQ on globals returns the correct value.
1441 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1442 not a COMPILER-ERROR (followed by some other strange error on
1443 choosing the CONTINUE restart).
1444 * bug fix: make.sh and friends are now more consistent in the way that
1445 they look for GNU "make".
1447 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1448 * experimental native threads support (on x86 Linux >=2.4 only).
1449 This is not compiled in by default: you need to add :SB-THREAD to
1450 the target features. See the "Beyond ANSI" chapter of the manual
1452 * fix for longstanding nonANSIism: The old distinction between
1453 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1454 The return value from CL:FIND-CLASS is now a CLOS class, and
1455 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1456 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1457 * An interface to the MetaObject Protocol, as described in Kiczales,
1458 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1459 MIT Press, 1991, is available from the SB-MOP package.
1460 * incompatible change: the SB-PCL package should now be considered
1461 a private implementation detail, and no longer a semi-private MOP
1463 * minor incompatible change: due to rearrangement for threads, the
1464 control stack and binding stack are now allocated at arbitrary
1465 addresses instead of being hardcoded per-port. Users affected by
1466 this probably have to be doing advanced things with shared
1467 libraries, and will know who they are.
1468 * minor incompatible change: Previously, all --eval forms used were
1469 processed with READ before any of them were processed with EVAL.
1470 Now each --eval form is processed with both READ and EVAL before
1471 the next --eval form is processed. (Thus package operations like
1472 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1473 work as the user might reasonably expect.)
1474 * minor incompatible change: *STANDARD-INPUT* is now only an
1475 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1477 * minor incompatible change: Y-OR-N-P is now character-oriented,
1478 not line oriented. Also, YES-OR-NO-P now works without errors.
1479 (thanks to Antonio Martinez)
1480 * sb-aclrepl module improvements: an integrated inspector, added
1481 repl features, and a bug fix to :trace command.
1482 * Known functions, which cannot be open coded by the backend, are
1483 considered to be able to check types of their arguments. (fixing
1484 a bug report by Nathan J. Froyd)
1485 * fixed a bug in computing method discriminating functions: It is
1486 now possible to define methods specialized on classes which have
1487 forward-referenced superclasses. (thanks to Gerd Moellmann)
1488 * fixed evaluation order in optional entries (reported by Gilbert
1490 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1491 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1492 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1493 required-by-AMOP NAME argument, as well as CLASS and
1494 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1495 * fixed bug 20: DEFMETHOD can define methods using names that are
1496 not the proper names of classes to designate class specializers.
1497 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1498 Unix file descriptors, instead of blowing up. (thanks to Antonio
1500 * Garbage collection refactoring: user-visible change is that a
1501 call to the GC function during WITHOUT-GCING will not do garbage
1502 collection until the end of the WITHOUT-GCING. If you were doing
1503 this you were probably losing anyway.
1504 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1505 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1506 TYPEP the latter but not the former.
1507 * The compiler issues a full WARNING for calls to undefined functions
1508 with names from the CL package.
1509 * MAP-INTO for a vector destination is open coded. (reported by
1510 Brian Downing on c.l.l)
1511 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1512 documentation string.
1513 * fixed some bugs revealed by Paul Dietz' test suite:
1514 ** COPY-ALIST now signals an error if its argument is a dotted
1516 ** Condition slots are now accessed more correctly in the presence
1517 of multiple initargs for a given slot.
1518 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1519 exclude restarts of the same name associated with a different
1521 ** DEFCLASS of forward-referenced classes with another
1522 forward-referenced class in the superclasses list no longer
1524 ** Condition slots are now initialized once each, not multiple
1525 times. (thanks to Gerd Moellmann)
1526 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1528 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1529 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1530 arguments contain duplicated elements.
1531 ** RESTART-CASE understands local macros.
1532 ** RESTART-CASE associates exactly its own restarts with a condition.
1533 ** ENDP in safe mode checks its argument to be of type LIST.
1534 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1535 in question is unbound.
1536 ** Optimized MAKE-INSTANCE functions no longer cause internal
1537 assertion failures in the presence of duplicate initargs.
1538 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1539 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1540 propagated even in the ctor optimized implementation of
1542 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1544 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1545 * a better implementation of SXHASH on (simple) bit vectors,
1546 measured both in execution speed and in distribution of results
1547 over the positive fixnums, has been installed. Likewise, a better
1548 implementation of EQUAL for simple bit vectors is now available.
1549 * fixed CEILING optimization for a divisor of form 2^k.
1550 * fixed bug 240 (emitting extra style warnings "using the lexical
1551 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1552 by Antonio Martinez)
1553 * fixed SXHASH, giving different results for NIL depending on type
1554 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1555 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1556 variable. (found by Rolf Wester)
1557 * SBCL does not ignore type declarations for special
1558 variables. (reported by rif on c.l.l 2003-03-05)
1559 * some bug fixes in contrib/sb-aclrepl/
1560 * fixed some bugs revealed by Paul Dietz' test suite:
1561 ** a bug in the CONS type specifier, whereby the CAR and CDR
1562 types got intertwined, has been fixed;
1563 ** the type system is now able to reason about the interaction
1564 between INTEGER and RATIO types more completely;
1565 ** APPEND, [N]REVERSE and NRECONC check that those their
1566 arguments, which must be proper lists, are really so;
1567 ** An array specialized to be unable to hold elements has been
1568 implemented, as required -- yes, really -- by ANSI;
1569 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1570 on malformed property lists;
1572 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1573 * incompatible packaging change: in line with Unix convention,
1574 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1575 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1576 * REQUIRE and PROVIDE are now optionally capable of doing something
1577 useful. See the documentation string for REQUIRE.
1578 * infrastructure for a managed SBCL contrib system: contributed
1579 modules in this release include:
1580 ** the ASDF system definition facility;
1581 ** an interface to the BSD Sockets API;
1582 ** an ACL-like convenience interface to the repl;
1583 (thanks to Kevin Rosenberg)
1584 ** an implementation of ROTATE-BYTE, with efficient implementation
1586 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1587 gives a closer approximation to the right answer than previously.
1588 (thanks to Raymond Toy)
1589 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1590 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1591 optional environment arguments, as required by ANSI.
1592 * fixed bugs in other functions taking environment objects, allowing
1593 calls with an explicit NIL environment argument to be compiled
1595 * fixed bug 228: primary return values from
1596 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1597 COMPILE or FUNCTION.
1598 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1599 :TYPE will no longer signal errors on innocuous objects.
1600 * fixed bug 231b: SETQ is better at respecting type declarations in
1601 the lexical environment.
1602 * fixed a bug in DEFCLASS: classes named by symbols with no or
1603 unprintable packages can now be defined.
1604 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1605 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1606 * fixed compiler failure related to checking types of functions.
1607 (reported by Robert E. Brown)
1608 * the compiler is now much more consistent in its error-checking
1609 treatment of bounding index arguments to sequence functions: in
1610 (SAFETY 3) code, errors will be signalled in almost all cases if
1611 invalid sequence bounding indices are passed to functions defined
1612 by ANSI to operate on sequences.
1613 * fixed a bug in the build procedure: documentation of SBCL-specific
1614 packages is now preserved and available in the final Lisp image.
1615 * lifted FDEFINITION lookup out of loops in the implementation of
1616 many list operations. (thanks to Robert E. Brown)
1617 * fixed a bug in the reader: the #n# reader macro now works for
1618 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1619 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1620 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1621 E. Brown and Tony Martinez respectively)
1622 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1623 SLOT-BOUNDP: in particular, it is now able to optimize them much
1624 better, and is now not vulnerable to having packages renamed.
1625 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1626 always called when a slot is not present in an instance. (thanks
1628 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1629 longer signal an error on unreading a character following EOF on
1630 the previous constituent stream. (thanks to Tony Martinez)
1631 * fixed some bugs revealed by Paul Dietz' test suite:
1632 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1633 not just nonnegative fixnums;
1634 ** the logical bit-array operators such as BIT-AND now accept an
1635 explicit NIL for their "opt-arg" argument (to indicate a
1636 freshly-consed result bit-array);
1637 ** ELT now signals an error on an invalid sequence index in safe
1639 ** the type system is now cleverer about negations of numeric
1640 types, and consequently understands the BIGNUM and RATIO types
1642 ** the type system is now cleverer about the interaction between
1643 INTEGER and RATIO types: while bugs still remain, many more
1644 cases are accurately computed;
1645 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1646 if it is in the last clause;
1647 ** CONSTANTLY now correctly returns a side-effect-free function in
1649 ** DECLARE is no longer treated as a special-operator; in
1650 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1651 * incremented fasl file version number due to the change in the
1652 DEFSTRUCT-SLOT-DESCRIPTION structure.
1654 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1655 * minor incompatible change: code processed by the "interpreter" or
1656 EVAL now has a compilation optimization policy of (DEBUG 2)
1657 (changed from (DEBUG 1)) to improve debuggability of interactive
1658 development, and to allow the use of the debug RETURN command in
1660 * an experimental implementation of the RETURN command for the
1661 debugger has been included. (thanks to Frederik Kuivinen)
1662 * fixed bug 62: constraints were not propagated into a loop.
1663 * fixed bug in embedded calls of SORT (reported and investigated by
1665 * fixed some bugs revealed by Paul Dietz' test suite:
1666 ** printing and reading of arrays with some dimensions having
1667 length 0 (thanks to Gerd Moellmann);
1668 ** BOA constructor with &AUX argument without a default value does
1669 not cause a type error;
1670 ** CONSTANTP now returns true for all self-evaluating objects.
1672 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1673 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1674 accessors that are related by inheritance, as specified in the
1675 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1676 Valtteri Vuorikoski)
1677 * The compiler is now able to inline functions that were defined in
1678 a complex lexical environment (e.g. inside a MACROLET).
1679 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1681 * Support for the upcoming FreeBSD-5.0 release has been included.
1682 (thanks to Dag-Erling Smorgrav)
1683 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1684 effect when it is not in a toplevel context.
1685 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1686 stand a better chance of being correct. (thanks to Gerd
1688 * fixed bug in COERCE, which now signals an error on coercing a
1689 rational to a bounded real type which excludes the expected
1691 * The compiler is now able to derive types more accurately from the
1692 COERCE and COMPILE functions.
1693 * fixed bug 223: functional binding is considered to be constant
1694 only for symbols in the CL package.
1695 * fixed bug 231: SETQ did not check the type of a variable being set
1696 (reported by Robert E. Brown)
1697 * A new optimization for MAKE-INSTANCE has been included, fixing
1698 various bugs (including relating to :ALLOCATION :CLASS slots and
1699 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1701 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1702 ** As required by ANSI, LOOP now disallows anonymous collection
1703 clauses such as COLLECT I in conjunction with aggregate boolean
1704 clauses such as THEREIS (= I 1);
1705 ** LOOP now signals an error when any variable is reused in the
1706 same loop (including the potentially useful construct analogous
1707 to WITH A = 1 WITH A = (1+ A);
1708 ** IT is only a special loop symbol within the first clause of a
1709 conditional loop clause;
1710 ** LOOP with a typed iteration variable over a hashtable now
1711 signals a type error iff it should.
1712 * fixed some other bugs revealed by Paul Dietz' test suite:
1713 ** FILE-STREAM now names the class previously known as FD-STREAM;
1714 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1715 argument) no longer signals an error;
1716 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1717 of the slot symbol, rather than using the current package
1718 ((:CONC-NAME "") continues to intern the slot's name in the
1720 * incremented fasl file version number, because of the incompatible
1721 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1722 of the new implementation of DEFINE-COMPILER-MACRO.
1724 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1725 * Support for building SBCL for MIPS platforms running in
1726 little-endian mode has now been checked in, and basic
1727 functionality on said platforms verified.
1728 * minor incompatible change: PCL now records the pathname of a file
1729 in which methods and the like are defined, rather than its
1731 * minor incompatible change: TRUENAME now considers the truename of
1732 a file naming a directory to be the pathname with :DIRECTORY
1733 component indicating that directory.
1734 * minor incompatible change: a NAMED clause in the extended form of
1735 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1736 reason for the previous behaviour is unclear.
1737 * more systematization and improvement of CLOS and MOP conformance
1738 in PCL (thanks to Gerd Moellman and Pierre Mai):
1739 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1741 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1742 primary methods with no specializers;
1743 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1745 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1746 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1747 FORWARD-REFERENCED-CLASSes; error reporting on
1748 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1750 ** SXHASH on CLOS instances now uses a slot internal to the
1751 instance to return different numbers on distinct instances,
1752 while preserving the same return value through invocations of
1754 ** DEFMETHOD signals errors when methods with longer incongruent
1755 lambda lists are added to generic functions;
1756 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1757 CLASS, as specified in AMOP;
1758 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1759 on the order returned by the primary method for classes of
1760 class STANDARD-CLASS;
1761 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1762 * fixed some bugs shown by Paul Dietz' test suite:
1763 ** DOLIST puts its body in TAGBODY;
1764 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1766 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1767 value producing form;
1768 ** if more variables are given to PROGV than values, extra
1769 variables are bound and made to have no value;
1770 ** NSUBSTITUTE on list arguments gets the right answer with
1772 ** ELT signals an error of type TYPE-ERROR when the index argument
1773 is not a valid sequence index;
1774 ** LOOP signals (at macroexpansion time) an error of type
1775 PROGRAM-ERROR when duplicate variable names are found;
1776 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1777 ** FUNCALL of special-operators now cause an error of type
1779 ** PSETQ now works as required in the presence of side-effecting
1780 symbol-macro places;
1781 ** NCONC accepts any object as its last argument;
1782 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1784 ** loop-for-as-package does not require a package to be explicitely
1786 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1788 * fixed bug 166: compiler preserves "there is a way to go"
1789 invariant when deleting code.
1790 * fixed bug 172: macro lambda lists with required arguments after
1791 &REST arguments now cause an error to be signalled. (thanks to
1793 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1794 bug. (thanks to Matthew Danish)
1795 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1797 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1798 arguments are lexically rebound. (thanks to Gerd Moellmann and
1800 * fixed bug 194: error messages are now more informative when there
1801 is no primary method applicable in a call to a generic
1802 function. (thanks to Gerd Moellmann)
1803 * fixed bug in command line argument checking (thanks to Julian
1805 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1806 (thanks to Matthew Danish)
1807 * incremented fasl file version number, because of the
1808 SXHASH-related changes in the layout of CLOS data structures
1810 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1811 * minor incompatible change: The runtime (the Unix executable named
1812 "sbcl") is now much pickier about the .core files it will load.
1813 Essentially it now requires .core files to descend from the same
1814 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1815 as the runtime does. (The intent is to prevent the crashes which
1816 can occur, and which can even be reported as mysterious failures,
1817 when people patch the sources or change the build parameters
1818 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1819 sbcl and .core files.)
1820 * fixed bug: VALUES-LIST is no longer optimized away.
1821 * fixed bug 142: The FFI conversion of C string values to Lisp
1822 string values no longer conses excessively. (thanks to Nathan
1823 Froyd porting Raymond Toy's fix to CMU CL)
1824 * began to systematize and improve MOP conformance in PCL (thanks to
1825 Nathan Froyd, Gerd Moellman and Pierre Mai):
1826 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1828 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1829 ** FINALIZE-INHERITANCE is now called on class finalization;
1830 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1831 argument precedence order.
1832 * fixed bug 202: The compiler no longer fails on functions whose
1833 derived types contradict their declared type.
1834 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1835 so it can be non-toplevel.
1836 * The fasl file version number has changed (because of the new
1837 implementation of DEFMACRO).
1838 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1839 safe code, that any length requirement by their type-specifier
1840 argument is valid. The exceptions to this are described in bug
1842 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1843 functions now check the types of their inputs as required by ANSI.
1844 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1845 introduced symbol is DECLAREd to be SPECIAL.
1846 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1847 * fixed bug: :COUNT argument to sequence functions may be negative
1848 * fixed bug: body of DO-SYMBOLS may contain declarations
1849 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1850 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1851 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1852 symbol macro only once
1853 * fixed printing of call frame when argument list is unavailable
1854 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1855 * compiler no longer signals WARNING on unknown keyword
1858 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1859 * A beta-quality port to the mips architecture running Linux,
1860 based on the old CMUCL backend, has been made. It has been tested
1861 on a big-endian kernel, and works sufficiently well to be able to
1862 rebuild itself; it has not been tested in little-endian mode.
1863 * fixed an inconsistency between gencgc.c and purify.c which made
1864 dumping/loading .core files unreliable
1865 * fixed bug 120a: The compiler now deals correctly with IFs where
1866 the consequent is the same as the alternative, instead of
1867 misderiving the return type. (thanks to Alexey Dejneka)
1868 * fixed bug 113: Logical pathnames are now dumpable (the logical
1869 host is resolved at load-time, throwing an error if it is not
1871 * fixed bug 174: FORMAT's error message is slightly clearer when a
1872 non-printing character is used in a format directive.
1873 * fixed several bugs in compiler checking of type declarations, i.e.
1874 violations of the Python "declarations are assertions" principle
1875 (thanks to Alexey Dejneka)
1876 * fixed several bugs in PCL's error checking (thanks to Gerd
1878 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1880 * fixed bug in compilation of functions as first class values
1881 (thanks to Antonio Martinez-Shotton)
1882 * The compiler's handling TYPE-ERRORs which it can prove will
1883 inevitably happen at runtime has been cleaned up and corrected
1884 in several ways. (thanks to Alexey Dejneka)
1885 * improved argument type checking for various basic arithmetic
1886 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1887 lavished on them in the past that they can be compiled in many
1888 ways in different special cases
1889 * fixed bug 181: compiler checks validity of user supplied type
1891 * cleaned up code flushing in optimization: Function calls which
1892 should signal errors for safety purposes (e.g. which ANSI says
1893 should signal errors when their arguments are of incorrect type)
1894 are no longer optimized away.
1895 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1896 * incremented fasl file version number, because changes in the
1897 implementation of sequence functions like COERCE caused
1898 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1899 in old inline expansions) to become undefined. (Actually these
1900 changes were later undone, so we might very well be binary
1901 compatible with 0.7.7 after all, but leaving the version number
1902 incremented seemed like the simplest and most conservative
1905 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1906 * An alpha-quality port to the parisc architecture running Linux,
1907 based on the old CMUCL backend, has been made. This, even more so
1908 than the other backends, should be considered still a work in
1909 progress; known problems include that the Linux kernel in 64-bit
1910 mode does not propagate the correct sigcontext structure to
1911 userspace, and consequently SBCL on a parisc64 kernel will not
1913 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1914 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1915 suppressed.) Also now that the compiler is looking at declarations
1916 in the environment, it checks optimization declarations as well,
1917 and suppresses inlining when (> DEBUG SPEED).
1918 * More fixes have been made to treatment of floating point exception
1919 treatment and other Unix signals. In particular, floating point
1920 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1921 * The detection and handling of control stack exhaustion (infinite
1922 or very deeply nested recursion) has changed. Stack exhaustion
1923 detection is now done by write-protecting pages at the OS level
1924 and applies at all optimization settings; when found, a
1925 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1926 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1927 be caught using IGNORE-ERRORS.
1928 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1929 correctly when transforming recursive LABELS functions to LETs.
1930 (thanks to Alexey Dejneka)
1931 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1932 that are names of constants or global variables.
1933 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1934 alien routines with docstrings.
1935 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1936 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1938 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1939 (thanks to Eric Marsden)
1940 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1941 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1942 * Bug fix: The SPARC backend can now compile functions involving
1943 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1944 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1945 to a functions expecting a PATHNAME-DESIGNATOR.
1946 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1947 lambda lists. (thanks to Alexey Dejneka)
1948 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1949 functions so that EXPORTing the name of the class doesn't cause
1950 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1951 to Antonio Martinez for reporting this)
1952 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1953 constant-folded without failing an assertion. (thanks to Einar
1954 Floystad Dorum for reporting this)
1955 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1956 types are now dealt with more correctly by the compiler.
1957 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1958 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1959 behaviour analogous to RENAME-FILE. This puts its behaviour more
1960 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1962 * The fasl file version number has changed again. (because of the
1963 bug fix involving the names of PCL MAKE-INSTANCE functions)
1965 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1966 * bug fix: Floating point exceptions are treated much more
1967 consistently on the x86/Linux and PPC/Linux platforms.
1968 * Array initialization with :INITIAL-ELEMENT is now much faster for
1969 cases when the compiler cannot open code the array creation, but
1970 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1971 array accesses have also seen a speed increase.
1972 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1973 specified. (This was at the root of some bad interactions between
1974 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1976 * bug fix: Internal error arguments for undefined functions are now
1977 computed correctly on the PPC/Linux platform.
1978 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1979 Raymond Toy's patch for CMU CL)
1980 * Support for the Solaris 9 operating environment has been included
1981 (thanks to Daniel Merritt)
1982 * A very ugly but hopefully complete draft of the missing FFI chapter
1983 of the manual has been created by reformatting the corresponding
1984 CMU CL manual chapter into (currently very ugly and incoherent)
1985 DocBook and bringing it up to date for SBCL behavior. Thus, the
1986 manual is now essentially complete, at least by my extreme
1987 once-and-only-once standards, whereby it's acceptable to refer to
1988 the doc strings of SB-EXT functions as the primary documentation.
1989 * The fasl file version number has changed again, due to cleanup of
1990 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1991 is no longer a static symbol.)
1993 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1994 * SBCL now builds with OpenMCL (version 0.12) as the
1995 cross-compilation host; also, more progress has been made toward
1996 bootstrapping under CLISP.
1997 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1999 * bug 158 fixed: The compiler can now deal with integer loop
2000 increments different from 1; fixing this turned out also to fix
2002 * bug 169 fixed: no more bogus warnings about using lexical bindings
2003 despite the presence of perfectly good SPECIAL declarations (thanks
2004 to David Lichteblau)
2005 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2006 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2007 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2009 * bug 180 fixed: Method combination specifications no longer ignore
2010 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2011 * bug fix: Structure type predicate functions now check their argument
2012 count as they should.
2013 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2014 correctly. (thanks to Pierre Mai)
2015 * minor incompatible change: The --noprogrammer option is deprecated
2016 in favor of the new --disable-debugger option, which is very similar.
2017 (The major difference is that it takes effect at a slightly different
2018 time at startup, causing handling of errors in --sysinit and
2019 --userinit files will be affected differently.) The
2020 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2021 been added to allow this functionality to be controlled from ordinary
2022 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2023 maintainers, who might want to run non-interactive scripts to
2024 build SBCL cores which will later be used interactively.)
2025 * minor incompatible change: The LOAD function no longer, when given
2026 a wild pathname to load, loads all files matching that pathname.
2027 Instead, an error of type FILE-ERROR is signalled.
2029 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2030 * bug 147 fixed: The compiler preserves its block link/count
2031 invariants more correctly now so that it doesn't crash. (thanks
2033 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2035 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2036 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2037 no-value-for-this-character as NIL (as specified by ANSI).
2038 * HOST-NAMESTRING on physical pathnames now returns a string that is
2039 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2040 (thanks to Christophe Rhodes)
2041 * The Alpha port handles icache flushing more correctly. (thanks to
2043 * More progress has been made toward bootstrapping under CLISP. (thanks
2044 to Christophe Rhodes)
2045 * The fasl file format has changed again, because dynamic loading
2046 on OpenBSD (which has non-ELF object files) motivated some cleanups
2047 in the way that foreign symbols are transformed and passed around.
2048 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2049 is no longer treated as whitespace by the reader, but instead as
2050 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2051 |A
\7fB|, instead of A as it used to.
2053 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2054 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2055 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2056 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2057 this as of 0.7.1.45, but was left out of the previous news section
2058 (thanks to Dan Barlow)
2059 * SBCL now runs on the Solaris operating system on SPARC architectures
2060 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2061 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2062 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2063 * SPARC backend cleanups, allowing builds of cores optimized for V8
2064 and V9 SPARCS, and also emission of code targeted to a particular
2065 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2067 * SBCL is closer to bootstrapping under CLISP, thanks to various
2068 fixes by Christophe Rhodes.
2069 * The fasl file format has changed again, to allow the compiler's
2070 INFO database to support symbol macros.
2071 * The user manual (in doc/) is formatted into HTML more nicely.
2072 (thanks to coreythomas)
2073 * The system is smarter about SUBTYPEP relationships, especially
2074 those involving NOT types (including types such as ATOM which are
2075 represented internally using NOT types). Thus SUBTYPEP is less
2076 likely to return (VALUES NIL NIL) in general, and in particular
2077 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2079 * The fasl file format has changed again, because the internal
2080 representation of types now includes a new slot to support the new
2081 SUBTYPEP-of-NOT-types logic.
2082 * (not a change in the main branch of SBCL, but a related prototype
2083 which can hopefully be merged into the main branch of SBCL in the
2084 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2085 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2086 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2087 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2088 dispatch macro character combinations. (thanks to Alexey Dejneka)
2089 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2090 that superficially look like logical namestrings correctly.
2091 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2093 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2094 default supplied-p) for &optional and &key arguments. (thanks to
2097 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2098 * incompatible change: The compiler is now less aggressive about
2099 tail call optimization, doing it only when (> SPACE DEBUG) or
2100 (> SPEED DEBUG). (This is an incompatible change because there are
2101 programs which relied on the old CMU-CL-style behavior to optimize
2102 away their unbounded recursion which will now die of stack overflow.)
2103 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2104 for non-GENCGC systems has been increased to 20M (since that
2105 seems much closer to the likely performance optimum for modern
2106 systems than the old 4M value was)
2107 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2108 and *DEBUG-PRINT-LEVEL*
2109 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2110 of CMU CL's support for SPARC, and various endianness and other
2111 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2112 * new syntactic sugar for the Unix command line: --load foo.bar is now
2113 an alternate notation for --eval '(load "foo.bar")'.
2115 ** The system now detects stack overflow and handles it gracefully,
2116 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2117 optimization settings. (This is a good thing in general, and
2118 its introduction in this version should be particularly timely
2119 for anyone whose code fails because of suppression of tail
2121 ** The system now hunts for the C variable "environ" in a more
2122 devious way, to avoid segfaults when the C library version differs
2123 between compile time and run time. (thanks to Christophe Rhodes)
2124 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2125 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2126 ** The compiler no longer issues bogus style warnings for undefined
2127 classes in the same source file as the DEFCLASSes which defined
2128 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2130 ** fixes in CONDITION class precedence list for undefined function
2131 errors (thanks to Alexei Dejneka)
2132 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2133 correctly. (thanks to Dan Barlow)
2134 ** portability fixes aiming at bootstrapping under CLISP (thanks
2135 to Dave McDonald and Christophe Rhodes)
2136 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2137 ** fixes in type translation and and type inference (thanks to
2139 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2140 ** various fixes in the new ports (thanks to Dan Barlow)
2141 * several changes related to debugging:
2142 ** suppression of tail recursion, as noted above
2143 ** stack overflow detection, as noted above
2144 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2145 is now the default. (For some time encapsulation has been more
2146 reliable than the breakpoint-based :ENCAPSULATE NIL
2147 implementation, at least on X86 systems; and I just noticed that
2148 encapsulation also seems closer to the spirit of the ANSI
2151 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2153 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2154 up properly again. (There was a packaging bug in 0.7.0 which
2155 left their definitions in SB-SYS::LOAD-FOREIGN and
2156 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2157 vital for most things which interface to C-level interfaces,
2158 like extensions working with sockets or databases or
2159 Perl-compatible regexes or whatever, and the need to fix
2160 this bug is the main reason that 0.7.1 was released so
2162 ** DEFGENERIC is now choosier about the methods it redefines, so that
2163 reLOADing a previously-LOADed file containing DEFGENERICs does
2164 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2165 cycle now works as it should. (thanks to Alexey Dejneka)
2166 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2167 fixed by Christophe Rhodes. (He actually submitted this patch
2168 months ago, and I delayed until after 0.7.0.)
2169 ** Bug 111 (internal compiler confusion about runtime checks on
2170 FUNCTION types) was fixed by Alexey Dejneka.
2171 * Some internal cleanups (getting rid of variables which aren't
2172 needed now that the byte interpreter is gone) caused the fasl
2173 file format number to change again.
2175 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2176 * major incompatible change: The default fasl file extension, i.e. the
2177 default extension for files produced by COMPILE-FILE, has changed
2178 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2180 ** There are many changes in the implementation of the compiler.
2181 SBCL is now essentially a compiler-only implementation of ANSI
2182 Common Lisp. EVAL still "interprets" a few special cases, but
2183 almost all the interesting cases are handled by creating
2184 a LAMBDA expression, calling COMPILE on it, then calling
2185 FUNCALL on the result.
2186 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2187 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2188 Since the code is newer, there might still be some new bugs
2189 (though not as many as before Martin Atzmueller's fixes:-). But
2190 the new code is substantially simpler and clearer, and hopefully
2191 any remaining bugs will be simpler, less fundamental, and more
2192 fixable then the bugs in the old code.
2193 ** The revised compiler is still a little unsteady on its feet.
2195 *** The debugging information it produces (particularly the names
2196 of FUNCTION objects) is sometimes much less useful than what
2197 the old compiler produced.
2198 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2199 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2200 within a MACROLET) has been temporarily weakened.
2201 ** There are new compiler optimizations for various functions:
2202 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2203 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2204 *** the math functions TRUNCATE, FLOOR, and CEILING
2205 *** the function-of-all-trades COERCE
2206 Mostly these should be transparent, but there's one
2207 potentially-annoying problem (bug 117): when the compiler
2208 inline-expands a function and does type analysis on the result,
2209 it can create control paths which have type mismatches, and
2210 when it can't prove that those control paths aren't taken,
2211 it will issue WARNINGs about the type mismatches. This is
2212 a particular problem in practice for the new sequence functions.
2213 It's not clear how this should be fixed, and for now, a
2214 workaround is given in the entry for 117 in the BUGS file.
2215 ** (Because of the interaction between the two previous items --
2216 occasional inlining problems and new inline expansions -- some
2217 of the new sequence function optimizations won't really kick in
2218 completely until debugging information, and then inlining, are
2219 straightened out in some future version.)
2220 * minor incompatible changes:
2221 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2222 DIRECTORY no longer implicitly promotes NIL slots of its
2223 pathname argument to :WILD. In particular, when you ask for the
2224 contents of a directory (which you used to be able to do without
2225 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2226 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2227 ** changes in behavior that ANSI explicitly defines to be
2228 implementation dependent:
2229 *** The new compiler-only implementation still conforms with ANSI,
2230 but acts a little different than before. Besides the obvious
2231 changes in performance tradeoffs (that the cost per form passed
2232 to EVAL has gone up, and the cost per form executed by EVAL
2233 has gone down), the behavior of the system changes a little
2234 because there are no longer any interpreted function objects.
2235 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2236 e.g. doing COMPILE on the output of interactive DEFUN is
2238 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2240 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2241 has changed from (:CL) to NIL.
2242 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2243 e.g. #\Newline and #\Tab, don't have names specified in the
2244 ANSI Common Lisp standard, is now based on their ASCII symbolic
2245 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2246 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2247 are no longer used for output.
2248 ** changes in internal implementation constants:
2249 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2250 4 million. (If your application spends a lot of time GCing and
2251 you have a lot of RAM, you might want to experiment with
2252 increasing it even more.)
2253 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2254 However, almost all old code should still continue to work without
2255 immediate update, as SB-C-CALL is now a (deprecated) nickname
2257 ** Old operator names in the style DEF-FOO are now deprecated in
2258 favor of new corresponding names DEFINE-FOO, for consistency with
2259 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2260 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2261 internal symbols, but a few supported extensions like
2262 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2263 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2264 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2265 etc. as you get deeper into recursive calls to the debugger
2266 command loop, instead of the old "5]", "5]]", "5]]]"
2267 sequence. (I was motivated to do this when squabbles between
2268 ILISP and SBCL left me very deeply nested in the debugger. In the
2269 short term, this change will probably provoke more ILISP/SBCL
2270 squabbles, but hopefully it will be an improvement in the long run.)
2271 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2272 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2273 function, since declaiming return types involving aliens is
2274 (1) annoyingly messy to do by hand and (2) vital to efficient
2275 compilation of code which calls such functions.
2276 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2277 longer reexported by the SB-EXT package. They're solely useful
2278 for alien code, so it seems more logical that you should get
2279 them from the SB-ALIEN package, not in SB-EXT.
2280 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2281 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2282 features. Instead, the code that they used to control is always
2283 built into the system.
2284 * many other bug fixes
2285 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2286 advantage of the new EVAL-WHEN stuff and to clean them up in
2287 general, and they are now more ANSI-compliant in a number of
2288 ways. Martin Atzmueller is responsible for a lot of this.
2289 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2291 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2292 *** correct ERROR type for various file operations
2293 *** some fixes for Lisp streams
2294 *** DEFMETHOD syntax checking
2295 *** changing old weird representation of debug information as
2296 strings (which, among their other deficiencies, don't transform
2297 correctly when you rename packages, and don't change their
2298 print representation when you change things like *PACKAGE*
2299 and *PRINT-LENGTH*) to symbols and lists of symbols
2300 He also made several improvements and fixed several bugs in DESCRIBE.
2301 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2303 *** misbehavior of WRITE-STRING/WRITE-LINE
2304 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2305 and several other LOOP problems as well
2306 *** DIRECTORY when similar filenames are present
2307 *** DEFGENERIC with :METHOD options
2308 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2309 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2310 *** argument ordering in FIND with :TEST option
2311 *** mishandled package designator argument in APROPOS-LIST
2312 *** various problems in the backquote readmacro
2313 *** a bug in APROPOS
2314 *** probably some others that I'm not describing very well here,
2315 since the CVS log documents them by reference to sbcl-devel
2316 messages, and the SourceForge archives aren't working well.:-(
2317 ** Dan Barlow improved the Alpha port (and is making progress on the
2318 PPC port, for those of you who think different).
2319 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2320 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2321 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2322 continues to work on the SPARC port (for those of you in a position
2323 to look down upon our little PC-compatible boxes from a great height).
2324 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2325 and uses that copy, rather than the current dynamic value, when
2326 it's trying to decide whether to truncate output. Thus e.g.
2327 (let ((*print-lines* 50))
2328 (pprint-logical-block (stream nil)
2330 (let ((*print-lines* 8))
2331 (print (aref possiblybigthings i) stream)))))
2332 should now truncate the logical block only at 50 lines, instead of
2333 often truncating it at 8 lines, as it did before.
2334 * The doc/cmucl/ directory, containing old CMU CL documentation
2335 from the time of the fork, is no longer part of the base system.
2336 SourceForge has shut down its anonymous FTP service, and with it
2337 my original plan for distributing the old CMU CL documentation
2338 there. For now, if you need these files you can download an old
2339 SBCL source release and extract them from it.
2340 * The fasl file version number changed again, for dozens of reasons,
2341 some of which are apparent above.
2343 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2344 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2345 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2346 patch, so that DISASSEMBLE output is much nicer.
2347 * The code in the SB-PROFILE package now seems reasonably stable.
2348 I still haven't decided what the final interface should look like
2349 (I'd like PROFILE to interact cleanly with TRACE, since both
2350 facilities use function encapsulation) but if you have a need
2351 for profiling now, you can probably use it successfully with
2352 the current CMU-CL-style interface.
2353 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2354 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2355 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2356 link now returns the dangling link itself, and for similar
2357 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2358 link itself. (In these cases the old code signalled an error and
2359 looped endlessly, respectively.) Thus, DIRECTORY now works even
2360 in the presence of dangling and cyclic symbolic links.
2361 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2362 is now a supported extension again, since the consensus on
2363 sbcl-devel was that it can be useful for ordinary development
2364 work, not just for debugging SBCL itself.
2365 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2366 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2367 that currently-defined functions might be redefined later with
2368 different return types.
2369 * Hash tables can be printed readably, as inspired by CMU CL code
2370 of Eric Marsden and SBCL code of Martin Atzmueller.
2371 * better error handling in CLOS method combination, thanks to
2372 Martin Atzmueller porting Pierre Mai's CMU CL patches
2373 * more overflow fixes for >16Mbyte I/O buffers
2374 * A bug in READ has been fixed, so that now a single Ctrl-D
2375 character suffices to cause end-of-file on character streams.
2376 In particular, now you only need one Ctrl-D at the command
2377 line (not two) to exit SBCL.
2378 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2380 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2381 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2382 does the right thing.
2383 * The implementation of some type tests, especially for CONDITION
2384 types, is now tidier and maybe faster, due to CMU CL code
2385 originally by Douglas Crosher, ported by Martin Atzmueller.
2386 * Some math functions have been fixed, and there are new
2387 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2388 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2389 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2390 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2391 likely to be built into sbcl-0.7.0.)
2392 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2393 CL, ported to SBCL by Martin Atzmueller)
2394 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2395 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2396 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2397 * There's a new slam.sh hack to shorten the edit/compile/debug
2398 cycle for low-level changes to SBCL itself, and a new
2399 :SB-AFTER-XC-CORE target feature to control the generation of
2400 the after-xc.core file needed by slam.sh.
2401 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2402 COMPILE-FILE is no longer supported, so that now every function
2403 gets an entry point, so that block compilation looks a little
2404 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2405 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2406 returns the number of bytes consed since the system started,
2407 rather than the number consed since the first time the function
2408 was called. (The new definition parallels ANSI functions like
2409 CL:GET-INTERNAL-RUN-TIME.)
2410 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2411 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2412 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2413 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2414 * Fasl file version numbers are now independent of the target CPU,
2415 since historically most system changes which required version
2416 number changes have affected all CPUs equally. Similarly,
2417 the byte fasl file version is now equal to the ordinary
2420 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2421 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2422 is no longer recognized. I apologize for this, because it was
2423 listed in SB-EXT as a supported extension, but I found that
2424 its existing behavior was poorly specified, as well as incorrectly
2425 specified, and it looked like too much of a mess to straighten it
2426 out. I have enough on my hands trying to get ANSI stuff to work..
2427 * many patches ported from CMU CL by Martin Atzmueller, with
2428 half a dozen bug fixes in pretty-printing and the debugger, and
2429 half a dozen others elsewhere
2430 * fixed bug 13: Floating point infinities are now supported again.
2431 They might still be a little bit flaky, but thanks to bug reports
2432 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2433 as flaky as they were.
2434 * The --noprogrammer command line option is now supported. (Its
2435 behavior is slightly different in detail from what the old man
2436 page claimed it would do, but it's still appropriate under the
2437 same circumstances that the man page talks about.)
2438 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2439 are now supported, and enabled by default. Thus, the compiler can
2440 handle many floating point and complex operations much less
2441 inefficiently. (Thus e.g. you can implement a complex FFT
2443 * The compiler now detects type mismatches between DECLAIM FTYPE
2444 and DEFUN better, and implements CHECK-TYPE more correctly, and
2445 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2446 thanks to patches from Martin Atzmueller.
2447 * various fixes to make the cross-compiler more portable to
2448 ANSI-conforming-but-different cross-compilation hosts (notably
2449 Lispworks for Windows, following bug reports from Arthur Lemmens)
2450 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2451 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2452 fixed thanks to Pierre Mai's CMU CL patches.
2453 * fixes to keep the system from overflowing internal counters when
2454 it tries to use i/o buffers larger than 16M bytes
2455 * fixed bug 45a: Various internal functions required to support
2456 complex special functions have been merged from CMU CL sources.
2457 (When I was first setting up SBCL, I misunderstood a compile-time
2458 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2459 * improved support for type intersection and union, fixing bug 12
2460 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2461 more obscure bugs as well
2462 * some steps toward byte-compiling non-performance-critical
2463 parts of the system, courtesy of patches from Martin Atzmueller
2464 * Christophe Rhodes has made some debian packages of sbcl at
2465 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2466 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2467 stable, but are nonetheless usable. When he's ready, I'd be happy
2468 to add them to the SourceForge "File Releases" section. (And if
2469 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2470 * new fasl file format version number (because of changes in
2471 internal representation of (OR ..) types to accommodate the new
2472 support for (AND ..) types, among other things)
2474 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2475 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2477 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2479 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2480 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2481 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2482 * DESCRIBE now works on structure objects again.
2483 * Most function call argument type mismatches are now handled as
2484 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2485 know whether the function will be redefined before the call is
2486 executed. (The compiler could flag local calls with full WARNINGs,
2487 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2488 it doesn't keep track of enough information to know whether calls
2489 are local in this sense.)
2490 * Compiler output is now more verbose, with messages truncated
2491 later than before. (There should be some supported way for users
2492 to override the default verbosity, but I haven't decided how to
2493 provide it yet, so this behavior is still controlled by the internal
2494 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2495 src/compiler/ir1util.lisp.)
2496 * Fasl file format version numbers have increased again, because
2497 support for the Gray streams extension changes the layout of the
2498 system's STREAM objects.
2499 * The Gray subclassable streams extension now works, thanks to a
2500 patch from Martin Atzmueller.
2501 * The full LOAD-FOREIGN extension (not just the primitive
2502 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2503 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2504 but like most other programs, it defaults to copying the Unix
2505 environment from the original process instead of starting the
2506 new process in an empty environment.
2507 * Extensions which manipulate the Unix environment now support
2508 an :ENVIRONMENT keyword option which doesn't smash case or
2509 do other bad things. The CMU-CL-style :ENV option is retained
2510 for porting convenience.
2511 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2512 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2514 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2516 * A patch from Martin Atzmueller seems to have solved the SIGINT
2517 problem, and as far as we know, signal-handling now works cleanly.
2518 (If you find any new bugs, please report them!)
2519 * The system no longer defaults Lisp source file names to types
2520 ".l", ".cl", or ".lsp", but only to ".lisp".
2521 * The compiler no longer uses special default file extensions for
2522 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2523 seems to expect a single default extension for all compiled code,
2524 and there's no compelling reason to try to stretch the standard
2525 to allow two different extensions.) Instead, byte-compiled files
2526 default to the same extension as native-compiled files.
2527 * Fasl file format version numbers have increased again, because
2528 a rearrangement of internal implementation packages made some
2529 dumped symbols in old fasl files unreadable in new cores.
2530 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2531 many fewer weird special cases.
2532 * Bug #17 (differing COMPILE-FILE behavior between logical and
2533 physical pathnames) has been fixed, and some related misbehavior too,
2534 thanks to a patch from Martin Atzmueller.
2535 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2536 by Tim Moore, ported to SBCL by Martin Atzmueller.
2537 * Martin Atzmueller fixed several filesystem-related problems,
2538 including bug #36, in part by porting CMU CL patches, which were
2539 written in part by Paul Werkowski.
2540 * More compiler warnings in src/runtime/ are gone, thanks to
2541 more patches from Martin Atzmueller.
2542 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2545 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2547 * DESCRIBE now works on CONDITION objects.
2548 * The debugger now handles errors which arise when trying to print
2549 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2551 * The build system now uses an additional file, customize-target-features.lisp,
2552 to allow local modifications to the target *FEATURES* list. (The point of
2553 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2554 debugging features enabled, without having to worry about propagating it
2555 into everyone's system when I do a "cvs update".) When no
2556 customize-target-features.lisp file exists, the target *FEATURES* list
2557 should be constructed the same way as before.
2558 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2559 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2560 warn about behavior which is undefined under the ANSI spec. Note: This
2561 is specified by ANSI, but it's not very popular with programmers.
2562 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2563 macro in the SBCL sources for an example of a workaround which you
2564 might use to make portable ANSI-standard code which does what you want.
2565 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2566 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2567 when it's used as a non-toplevel form. (This is required in order
2568 to implement the DEFCONSTANT-EQX macro.)
2569 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2570 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2571 funny things, due to bugs in the implementation of EVAL-WHEN
2572 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2573 non-toplevel DEFCONSTANTs are uncommon.)
2574 * The core file version number and fasl file version number have been
2575 incremented, because the old noncompliant DEFCONSTANT behavior involved
2576 calling functions which no longer exist, and because I also took the
2577 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2578 * fixed bug 1 (error handling before read-eval-print loop starts), and
2579 redid debugger restarts and related debugger commands somewhat while
2581 ** The QUIT debugger command is gone, since it did something
2582 rather different than the SB-EXT:QUIT command, and since it never
2583 worked properly outside the main toplevel read/eval/print loop.
2584 Invoking the new TOPLEVEL restart provides the same functionality.
2585 ** The GO debugger command is also gone, since you can just invoke
2586 the CONTINUE restart directly instead.
2587 ** The TOP debugger command is also gone, since it's redundant with the
2588 FRAME 0 command, and since it interfered with abbreviations for the
2590 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2592 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2593 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2594 some porting work by Martin Atzmueller)
2595 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2596 with a boatload of patches which he ported to SBCL. Now that those
2598 ** The system tries to make sure that its low-priority messages
2599 are prefixed by semicolons, to help people who like to use
2600 syntax highlighting in their ILISP buffer. (This patch
2601 was originally due to Raymond Toy.)
2602 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2603 patches originally written by Raymond Toy.
2604 ** The compiler understands coercion between single-value and
2605 multiple-VALUES type expressions better, getting rid of some very
2606 weird behavior, thanks to patches originally by Robert MacLachlan
2607 and Douglas Crosher.
2608 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2609 lambda lists, thanks to a patch originally by Pierre Mai.
2610 ** The system no longer bogusly warns about "abbreviated type
2612 ** The compiler gets less confused by inlining and RETURN-FROM,
2613 thanks to some patches originally by Tim Moore.
2614 ** The system no longer hangs when dumping circular lists to fasl
2615 files, thanks to a patch originally from Douglas Crosher.
2616 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2617 undefined function error.
2618 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2619 in structure slot initforms) for now, documented workaround instead:-|
2620 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2621 * fixed bug 5: added stubs for various Gray stream functions called
2622 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2623 installed, at least appropriate type errors are generated
2624 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2625 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2626 consistently in DEFMETHOD forms.
2627 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2628 it doesn't seem to affect SBCL after all
2629 * The C runtime system now builds with better optimization and many
2630 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2632 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2634 * The system is now under CVS at SourceForge (instead of the
2635 CVS repository on my home machine).
2636 * The new signal handling code has been tweaked to treat register
2637 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2638 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2639 downstream, e.g. in debug-int.lisp, has implicit dependencies
2640 on the unsignedness of integer representation of machine words,
2641 and that caused the system to bomb out with infinite regress
2642 when trying to recover from type errors involving signed values,
2643 e.g. (BUTLAST '(1 2 3) -1).)
2644 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2645 (This was one of the bugs Peter Van Eynde reported back in July.)
2646 * The system now uses code inspired by Colin Walters' O(N)
2647 implementation of MAP (from the cmucl-imp@cons.org mailing
2648 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2649 inline the MAP operation, and there is more than one
2650 sequence argument to the MAP call (so that it can't just
2651 do ETYPECASE once and for all based on the type of the
2652 single sequence argument). (The old non-inline implementation
2653 of the general M-argument sequence-of-length-N case required
2654 O(M*N*N) time when any of the sequence arguments were LISTs.)
2655 * The QUIT :UNIX-CODE keyword argument has been renamed to
2656 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2657 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2659 * Raymond Wiker's patches to port dynamic loading from Linux to
2660 FreeBSD have been added.
2661 * The BUGS file is now more nearly up to date, thanks in large part
2662 to Martin Atzmueller's review of it.
2663 * The debugger now flushes standard output streams before it begins
2664 its output ("debugger invoked" and so forth).
2665 * The core version number and fasl file version number have both
2666 been incremented, because of incompatible changes in the layout
2668 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2669 used to not be called for a saved Lisp image.)
2670 * Martin Atzmueller's version of a patch to fix a compiler crash,
2671 as posted on sbcl-devel 13 September 2000, has been installed.
2672 * Instead of installing Martin Atzmueller's patch for the
2673 compiler transform for SUBSEQ, I deleted the compiler transform,
2674 and transforms for some similar consing operations.
2675 * A bug in signal handling which kept TRACE from working on OpenBSD
2677 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2678 away by constant folding
2679 * The system now defines its address space constants in one place
2680 (in the Lisp sources), and propagates them automatically elsewhere
2681 (through GENESIS and the sbcl.h file). Therefore, patching the
2682 address map is less unnecessarily tedious and error-prone. The
2683 Lisp names of address space constants have also been systematized.
2684 * CVS tags like dollar-Header-dollar have been removed from
2685 the sources, because they have never saved me trouble and
2686 they've been source of trouble working with patches and other
2687 diff-related operations.
2688 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2689 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2691 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2693 * The system has been ported to OpenBSD.
2694 * The system now compiles with a simple "sh make.sh" on the systems
2695 that it's supported on. I.e., now you no longer need to tweak
2696 text in the target-features.lisp-expr and symlinks in src/runtime/
2697 by hand, the make.sh takes care of it for you.
2698 * The system is no longer so grossly inefficient when compiling code
2699 involving vectors implemented as general (not simple) vectors (VECTOR T),
2700 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2701 which dares to use the various sequence functions on non-simple
2702 vectors, takes less of a performance hit.
2703 * There is now a primitive type predicate VECTOR-T-P
2704 to test for the (VECTOR T) type, so that e.g.
2705 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2706 can now be compiled with some semblance of efficiency. (The old code
2707 turned the type declaration into a full call to %TYPEP at runtime!)
2708 * AREF on (VECTOR T) is still not fast, since it's still compiled
2709 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2710 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2711 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2712 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2713 without making full calls to SUBTYPEP at runtime.
2714 (Some analogous efficiency issues for non-simple vectors specialized to
2715 element types other than T, or for non-simple multidimensional arrays,
2716 have not been addressed. They could almost certainly be handled the
2717 same way if anyone is motivated to do so.)
2718 * The changes in array handling break binary compatibility, so
2719 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2720 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2721 instead of (VALUES T T).
2722 * By following the instructions that Dan Barlow posted to sbcl-devel
2723 on 2 July 2000, I was able to enable primitive dynamic object
2724 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2725 functionality is not implemented (since it calls ld to resolve
2726 library references automatically, requiring RUN-PROGRAM for its
2727 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2728 try to resolve library references) is now supported.
2729 * The system now flushes the standard output streams when it terminates,
2730 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2731 them at several other probably-convenient times, e.g. in each pass of
2732 the toplevel read-eval-print loop, and after evaluating a form given
2733 as an "--eval" command-line option. (These changes were motivated by a
2734 discussion of stream flushing issues on cmucl-imp in August 2000.)
2735 * The source transform for TYPEP of array types no longer assumes
2736 that an array whose element type is a not-yet-defined type
2737 is implemented as an array of T, but instead punts, so that the
2738 type will be interpreted at runtime.
2739 * There is now some support for cross-compiling in make.sh: each of
2740 the phases of make.sh has its own script. (This should be transparent
2741 to people doing ordinary, non-cross-compile builds.)
2742 * Since my laptop doesn't have hundreds of megabytes of memory like
2743 my desktop machine, I became more motivated to do some items on
2744 my to-do list in order to reduce the size of the system a little:
2745 ** Arrange for various needed-only-at-cold-init things to be
2746 uninterned after cold init. To support this, those things have
2747 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2748 symbols with such names are now uninterned after cold init).
2749 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2750 fasl files for cold load.
2751 ** Remove the old compiler structure pooling code (which used to
2752 be conditional on the target feature :SB-ALLOC) completely.
2753 ** Redo the representation of some data in cold init to be more compact.
2754 (I also looked into supporting byte compiled code at bootstrap time,
2755 which would probably reduce the size of the system a lot, but that
2756 looked too complicated, so I punted for now.)
2757 * The maximum signal nesting depth in the src/runtime/ support code has
2758 been reduced from 4096 to 256. (I don't know any reason for the very
2759 large old value. If the new smaller value turns out to break something,
2760 I'll probably just bump it back up.)
2761 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2763 * Many, many bugs reported by Peter Van Eynde have been added to
2764 the BUGS list; some have even been fixed.
2765 * While enabling dynamic object file loading, I tried to make the
2766 code easier to understand, renaming various functions and variables
2767 with less ambiguous names, and changing some function calling
2768 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2769 * While trying to figure out how to do the OpenBSD port, I tried to
2770 clean up some of the code in src/runtime/. In particular, I dropped
2771 support for non-POSIX signal handling, added various comments,
2772 tweaked the code to reduce the number of compilation warnings, and
2773 renamed some files to increase consistency.
2774 * To support the new automatic configuration functionality in make.sh,
2775 the source file target-features.lisp-expr has been replaced with the
2776 source file base-target-features.lisp-expr and the machine-generated
2777 file local-target-features.lisp-expr.
2778 * fixed a stupid quoting error in make.sh so that using CMU CL
2779 "lisp -batch" as cross-compilation host works again
2781 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2783 * DESCRIBE no longer tries to call itself recursively to describe
2784 bound/fbound values, so that it no longer fails on symbols which are
2785 bound to themselves (like keywords, T, and NIL).
2786 * DESCRIBE now works on generic functions.
2787 * The printer now prints less-screwed-up representations of closures
2788 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2789 * A private symbol is used instead of the :EMPTY keyword previously
2790 used to mark empty slots in hash tables. Thus
2791 (DEFVAR *HT* (MAKE-HASH-TABLE))
2792 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2793 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2794 now does what ANSI says that it should. (You can still get
2795 similar noncompliant behavior if bang on the hash table
2796 implementation with all the symbols you get back from
2797 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2798 This breaks binary compatibility, since tests for equality to
2799 :EMPTY are wired into things like the macroexpansion of
2800 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2802 * There's now a minimal placeholder implementation for CL:STEP,
2803 as required by ANSI.
2804 * An obscure bug in the interaction of the normal compiler, the byte
2805 compiler, inlining, and structure predicates has been patched
2806 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2807 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2809 * Missing ordinary arguments in a macro call are now detected even
2810 when the macro lambda list contains &KEY or &REST.
2811 * The debugger no longer complains about encountering the top of the
2812 stack when you type "FRAME 0" to explicitly instruct it to go to
2813 the top of the stack. And it now prints the frame you request even
2814 if it's the current frame (instead of saying "You are here.").
2815 * As specified by ANSI, the system now always prints keywords
2816 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2818 * The default initial SIZE of HASH-TABLEs is now smaller.
2819 * Type information from CLOS class dispatch is now propagated
2820 into DEFMETHOD bodies, so that e.g.
2821 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2823 is now basically equivalent to
2824 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2825 (DECLARE (TYPE SINGLE-FLOAT X))
2827 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2828 operation, without having to do run-time type dispatch.
2829 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2830 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2831 * Since I don't seem to be making big file reorganizations very often
2832 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2833 is overflowing my ability to conveniently back them up), I've finally
2834 checked the system into CVS. (The CVS repository is on my home system,
2835 not at SourceForge -- putting it on SourceForge might come later.)
2836 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2837 high-level GC-NOTIFY-FOO functions send their output. (There's
2838 still very little control of where low-level verbose GC functions
2839 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2840 less than it used to -- the GC-NOTIFY-FOO functions are now under
2841 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2842 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2843 in only one place in the source code, and propagates it automatically
2844 everywhere that it's needed. Thus e.g. when I bump the version from
2845 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2846 * The C source files now include boilerplate legalese and documentation
2847 at the head of each file (just as the Lisp source files already did).
2848 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2849 to his page will be replaced with a link to his new CLiki service.
2851 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2853 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2854 * The build process now looks for GNU make under the default name "gmake",
2855 instead of "make" as it used to. If GNU make is not available as "gmake"
2856 on your system, you can change this default behavior by setting the
2857 GNUMAKE environment variable.
2858 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2859 can build without error under CMU CL.
2861 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2863 * There is now a partial SBCL user manual (with some new text and some
2864 text cribbed from the CMU CL manual).
2865 * The beginnings of a profiler have been added (starting with the
2866 CMU CL profiler and simplifying and cleaning up). Eventually the
2867 main interface should be through the TRACE macro, but for now,
2868 it's still accessed through vaguely CMU-CL-style functions and macros
2869 exported from the package SB-PROFILE.
2870 * Some problems left over from porting CMU CL to the new
2871 cross-compilation bootstrap process have been cleaned up:
2872 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2873 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2874 part when getting rid of the compiler *BACKEND* stuff.)
2875 ** The value of *NULL-TYPE* was screwed up, because it was
2876 being initialized before the type system knew the final
2877 definition of the 'NULL type. This screwed up several key
2878 optimizations in the compiler, causing inefficiency in all sorts
2879 of places. (I found it because I wanted to understand why
2880 GET-INTERNAL-RUN-TIME was consing.)
2881 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2882 PROCLAIM FTYPE information. Unfortunately this broke binary
2883 compatibility again, since now the forms output by DEFGENERIC
2884 to refer to functions which didn't exist in 0.6.3.
2885 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2886 can use the new (as of 0.6.3) transform for SOME into MAP into
2888 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2889 values to (INTEGER 0 1000000), so that the time code will no longer
2890 occasionally get blown up by Linux returning 1000000 microseconds
2891 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2892 its output conform to the ANSI spec. (Alas, this makes its output
2893 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2894 * A full call to MAP NIL with a single sequence argument no longer conses.
2895 * fixes to problems pointed out by Martin Atzmueller:
2896 * The manual page no longer talks about multiprocessing as though
2897 it were currently supported.
2898 * The ILISP support patches have been removed from the distribution,
2899 because as of version 5.10.1, ILISP now supports SBCL without us
2900 having to maintain patches.
2901 * added a modified version of Raymond Toy's recent CMU CL patch for
2902 EQUALP comparison of HASH-TABLE
2904 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2906 * The system still can't cross-compile itself with
2907 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2908 entails), but at least it can compile more of itself that way
2909 than it used to be able to, and various buglets which were uncovered
2910 by trying to cross-compile itself that way have now been fixed.
2911 * This release breaks binary compatibility again. This time
2912 at least I've incremented the FASL file format version to 2, so that the
2913 problem can be detected reliably instead of just causing weird errors.
2914 * various new style warnings:
2915 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2916 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2917 ** using the lexical binding of a variable named in the *FOO* style
2918 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2919 as specified by ANSI.
2920 * *RANDOM-STATE* is no longer automatically initialized from
2921 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2922 default behavior of the system is to repeat its behavior every time
2923 it's run. If you'd like to change this behavior, you can always
2924 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2925 old convention there was no comparably easy way to get the system to
2926 repeat its behavior every time it was run.
2927 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2928 FTYPE declarations has been removed, in favor of their ANSI
2929 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2930 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2931 the types of their sequence arguments can be determined at compile time.
2932 This is done through a new open code expansion for MAP which eliminates
2933 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2934 argument types can be determined at compile time.
2935 * The optimizer now transforms COERCE into an identity operation when it
2936 can prove that the coerced object is already of the correct type. (This
2937 can be a win for machine generated code, including the output of other
2938 optimization transforms, such as the MAP transform above.)
2939 * Credit information has been moved from source file headers into CREDITS.
2940 * Source file headers have been made more standard.
2941 * The CASE macro now compiles without complaining even when it has
2944 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2946 * (Note that the way that the PCL macroexpansions were rewritten
2947 to accommodate the change in DEFGENERIC below breaks binary
2948 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2949 not run under sbcl-0.6.2. Once we get out of alpha releases,
2950 i.e. hit release 1.0.0, we'll probably try to maintain binary
2951 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2952 and sbcl-1.4.4. Until then, however, it might be fairly common
2953 for maintenance releases to break binary compatibility.)
2954 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2956 * The reporting of unhandled signals has been changed to print some
2957 explanatory text as well as the report form. (Previously only
2958 the report form was printed.)
2959 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2960 it defines, so that the compiler no longer issues undefined function
2961 warnings for compiled-but-not-yet-loaded generic functions.
2962 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2963 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2964 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2965 section of the ANSI spec are supported.
2966 * The "" nickname for the "KEYWORD" package has been removed.
2967 The reader still handles symbol tokens which begin with a package marker
2968 as keywords, but it doesn't expose its mechanism for doing so in the
2969 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2970 * The system now issues STYLE-WARNINGs for contradictory TYPE
2971 proclamations. (Warnings for contradictory FTYPE proclamations would
2972 be nice too, but those can't be done usefully unless the type system
2973 is made smarter about FUNCTION types.)
2974 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2975 systematized, so that "*target-*.lisp is supposed to exist only on the
2976 target and imply that there's a related file which exists on the
2977 host, and *host-*.lisp is supposed to exist only on the host and imply
2978 that there's a related file which exists on the target. This involves a
2979 lot of renaming. Hopefully the acute confusion caused by the renaming
2980 will be justified by the reduction in chronic confusion..
2981 ** runtime-type.lisp -> early-target-type.lisp
2982 ** target-type.lisp -> late-target-type.lisp
2983 ** early-host-format.lisp -> early-format.lisp
2984 ** late-host-format.lisp -> late-format.lisp
2985 ** host-error.lisp -> misc-error.lisp
2986 ** early-error.lisp -> early-target-error.lisp
2987 ** late-error.lisp -> late-target-error.lisp
2988 ** host-defboot.lisp -> early-defboot.lisp
2989 ** code/misc.lisp -> code/target-misc.lisp
2990 ** code/host-misc.lisp -> code/misc.lisp
2991 ** code/numbers.lisp -> code/target-numbers.lisp
2992 ** code/early-numbers.lisp -> numbers.lisp
2993 ** early-host-type.lisp -> early-type.lisp
2994 ** late-host-type.lisp -> late-type.lisp
2995 ** host-typep.lisp -> typep.lisp
2996 ** load.lisp -> target-load.lisp
2997 ** host-load.lisp -> load.lisp
2998 ** host-disassem.lisp -> disassem.lisp
2999 ** host-insts.lisp -> insts.lisp
3000 ** byte-comp.lisp -> target-byte-comp.lisp
3001 ** host-byte-comp.lisp -> byte-comp.lisp
3002 ** host-signal.lisp -> signal.lisp
3003 ** host-defstruct.lisp -> defstruct.lisp
3004 ** late-target-type.lisp -> deftypes-for-target.lisp
3005 Furthermore, several other previously target-only files foo.lisp (e.g.
3006 hash-table.lisp and random.lisp) have been split into a target-and-host
3007 foo.lisp file and a target-only target-foo.lisp file, with their key type
3008 definitions in the target-and-host part, so that the cross-compiler will
3009 know more about target types.
3010 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3011 gone away. In their place are various *BACKEND-FOO* variables
3012 corresponding to the slots of the old structure.
3013 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3014 into the target SBCL has been fixed.
3015 * The chill.lisp system for loading cold code into a running SBCL
3017 * Support for the CMU CL "scavenger hook" extension has been removed.
3018 (It was undocumented and unused in the CMU CL sources that SBCL was
3019 derived from, and stale in sbcl-0.6.1.)
3020 * Various errors in the cross-compiler type system were detected
3021 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3022 (enabling various consistency checks). Many of them were fixed,
3023 but some hard problems remain, so the compiler is back to
3024 running without *TYPE-SYSTEM-INITIALIZED* for now.
3025 * As part of the cross-compiler type system cleanup, I implemented
3026 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3027 * I have started adding UNCROSS calls throughout the type system
3028 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3029 on cross-compiler input files will be able to go away, and various
3031 * CONSTANTP now returns true for quoted forms (as explicitly required
3034 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3036 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3037 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3038 (MAKE-LIST -1). In the longer term, it ought to become true
3039 that declarations are assertions even at SAFETY 1. For now, it's not
3040 quite true even at SAFETY 3, but it's at least more nearly true..
3041 (Note that this change seems to increases the size of the system by
3042 O(5%) and to decrease the speed of the compiler by 20% or more.)
3043 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3044 for the problem of printing dozens of lines of distracting information
3045 about low-level system machinery as part of the top stack frame
3046 on entry to the debugger when an undefined function was called.
3047 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3048 is not reset to COMMON-LISP-USER.
3049 * Compilation of stuff related to dyncount.lisp has been made conditional
3050 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3051 smaller. The various dyncount-related symbols have been moved into
3052 a new "SB-DYNCOUNT" package.
3053 * tty-inspect.lisp has been renamed to inspect.lisp.
3054 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3055 feature has gone away. (When we eventually port to other flavors of
3056 libc and/or Unix, we'll try to make the differences between flavors
3057 invisible at the user level.)
3058 * Various other *FEATURES* tags, and/or their associated conditionals,
3059 have been removed if obsolescent, or given better documentation, or
3060 sometimes given more-mnemonic names.
3062 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3064 * tidied up "make.sh" script
3065 * tidied up system directory structure
3066 * better "clean.sh" behavior
3067 * added doc/FOR-CMUCL-DEVELOPERS
3068 * many many small tweaks to output format, e.g. removing possibly-confusing
3069 trailing #\. character in DESCRIBE-INSTANCE
3070 * (EQUALP #\A 'A) no longer signals an error.
3071 * new hashing code, including EQUALP hashing
3072 * tidied up Lisp initialization and toplevel
3073 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3074 * command line argument processing
3075 * added POSIX-GETENV function to deal with Unix-ish environment variables
3076 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3077 terminating SBCL on EOF
3078 * non-verbose GC by default
3079 * There is no more "sbcl" shell script; the sbcl file is now the C
3080 runtime executable (just like CMU CL).
3081 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3083 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3084 symbol and package information as Lisp native symbol and package objects
3085 instead of strings naming symbols and strings naming packages. This way,
3086 whenever packages are renamed (as in warm init), debug information is
3087 transformed along with everything else.
3088 * tweaked the optimization policy declarations which control the building
3089 of SBCL itself. Now, among other things, the system no longer saves
3090 source location debugging information. (This helps two problems at once
3091 by reducing SBCL size and by keeping SBCL from trying to look for its
3092 sources -- which may not exist -- when reporting errors.)
3093 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3094 debugging and testing purposes
3095 * cleaned up printing, making the printer call PRINT-OBJECT for
3096 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3097 methods, giving nearly-ANSI behavior
3098 * converted almost all special variables to use *FOO* naming convention
3099 * deleted PARSE-TIME functionality, since it can be done portably
3100 * moved some files out of cold init into warm init
3101 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3103 * regularized formatting of source files
3104 * added an install.sh script
3105 * fixed ridiculous memory usage of cross-compiler by making
3106 compiler/alloc.lisp not try to do pooling unless it can hook
3107 itself into the GC of the cross-compilation host. Now the system
3108 builds nicely on my old laptop.
3109 * added :SB-ALLOC in target-features.lisp-expr
3110 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3111 was not implemented)
3112 * re-did condition handling and note reporting in the compiler. Notes
3113 are no longer handled by signalling conditions. Style warnings
3114 and warnings are handled more correctly and reported in such a way
3115 that it's easy to find one or the other in your output (so that you
3116 can e.g. figure out which of many problems caused COMPILE-FILE to
3118 * changed the severity of several compiler warnings from full WARNING
3119 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3120 compiler note reporting so that it doesn't use the condition system
3121 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3122 COMPILE-FILE command)
3123 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3124 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3125 no longer supported, since their implementation was deeply intertwingled
3126 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3127 * removed bogus "support" for compiler macros named (SETF FOO), and
3128 removed the compiler macro for SETF INFO (but only after making a fool
3129 of myself on the cmucl-imp mailing list by posting a bogus patch for
3130 DEFINE-COMPILER-MACRO..)
3131 * Compiled files containing forms which have side effects on the Lisp
3132 reader (such as DEFPACKAGE forms) are now handled more correctly.
3133 (Compiler queuing of top level lambdas has been suppressed by setting
3134 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3135 * deleted various currently-unused source files, e.g. gengc.lisp. They
3136 may be added back at some point e.g. when porting to other architectures,
3137 but until they are it's distracting to distribute them and to try to
3139 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3140 there were so many of them they're just distractions, and UNCROSS is
3141 known to be able to handle the current sources
3142 * moved PROFILE functionality into TRACE, so that it will be clear
3143 how the wrapping and unwrapping of functions when you profile them
3144 interacts with the wrapping and unwrapping of functions when you
3145 trace them. (Actually, the functionality isn't there yet, but at least
3146 the interface specification is there. Hopefully, the functionality will
3147 arrive with some future maintenance release.)
3148 * removed host-oops.lisp
3149 * changed signature of QUIT function to allow UNIX-CODE argument
3150 * fixed READ-SEQUENCE bug
3151 * tweaked verbose GC output so that it looks more like the progress
3152 output that ANSI specifies for functions like LOAD
3153 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3154 * added <http://sbcl.sourceforge.com> to the banner information printed by