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 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
12 as other specialized array types, is much faster. SUBSEQ and
13 COPY-SEQ on such arrays have also been sped up.
14 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
15 * fixed bug: finalizers and after GC hooks that cause consing are now
17 * fixed bug: compiler error messages and summaries are now printed to
18 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
19 * fixed inference of the upper bound of an iteration variable.
20 (reported by Rajat Datta).
21 * fixed bug 373: caused by erronous compilation of references to alien
22 variables in the runtime on ppc/darwin.
23 * fixed bug 376: CONJUGATE type deriver.
24 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
25 (reported by Baughn on #lisp)
26 * a cross-compiler bug on non-x86 platforms has been identified and
27 fixed. (thanks to Bruno Haible)
28 * improvements to the MIPS runtime code for increased stability.
29 (thanks to Thiemo Seufer)
30 * increased the maximimum compact environment size to allow
31 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
32 * partial workaround for bug 135: don't name the function we're
33 calling for hairy cases of EVAL, so as not to accumulate one environment
34 entry per call to EVAL. (reported by Kevin Reid)
35 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
36 * optimization: type testing for non-vector arrays should be faster.
37 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
39 * fixed some bugs related to Unicode integration:
40 ** the restarts for recovering from input and output encoding
41 errors only appear when there is in fact such an error to
43 * fixed some bugs revealed by Paul Dietz' test suite:
44 ** MISC.549 and similar: late transformation of unsafe type
45 assertions into derived types caused unexpected code
47 ** SCALE-FLOAT type deriver is less wrong.
48 ** type derivers for EXP, LOG and similar functions compute result
49 types for complex arguments better.
50 ** (MISC.563) CONJUGATE type deriver works for very restricted
52 ** out-of-line type testers for character strings are available.
53 ** EQUAL compiler transform understands specialness of objects
55 ** accessing double-floats stored on the stack now works on x86-64.
56 ** debugger internals could sometimes create invalid lispobjs,
57 resulting in GC crashes.
59 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
60 * incompatible change: thread support for non-NPTL systems has
61 been removed - locking is buggy and unreliable. A threaded
62 SBCL build will now warn at startup and refuse to create
63 new threads, unless futex support is detected in the kernel
64 * incompatible change: the top level REPL now has only an ABORT
65 restart associated with it, not TOPLEVEL and ABORT as it used to.
66 TOP and TOPLEVEL are now available as debugger commands for
67 returning to the top level.
68 * incompatible change: forms evaluated in the REPL now use the
69 global optimization policy.
70 * incompatible change: user- and system-initialization files are
71 no longer processed with LOAD, but by READ and EVAL; hence the
72 global optimization policy, startup package, readtable, etc,
74 * The .fasl file format number has been incremented because of
75 various incompatible changes.
76 * internal entry point details and argument counts no longer appear
77 in backtraces unless explicitly requested by setting
78 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
79 * built-in and standard functions no longer have names like "top
80 level local call to FOO".
81 * fixed bug 32: functions defined in non-null lexical environments
82 now have more legible printed representation
83 * fixed bug 33: functions defined in non-null lexical environemnts
84 are now more amenable to inspection by INSPECT.
85 * workaround for bug 354: XEPs no longer appear in backtraces unless
87 * fixed bug: receiving the signal which results from attempting to
88 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
89 SIGBUS on other *BSDs) on architectures where the C stack is also
90 the Lisp stack (x86 and x86-64) from foreign code no longer leads
91 to debugger confusion or wild instability. (reported by Cheuksan
93 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
94 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
95 (reported by Lutz Euler)
96 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
97 compiled code. (reported by Paul Dietz)
98 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
99 specialized parameter to a method to something that is not TYPEP
100 the specializer is now possible.
101 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
102 face of package deletion.
103 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
104 pathname. (reported by David Tolpin on c.l.l)
105 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
106 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
107 STRING-STREAM and FILE-STREAM, respectively.
108 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
109 than previously; in addition, its test suite is now run on build.
110 (thanks to Luke Gorrie)
111 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
112 * optimization: added a immediate representation for single-floats
114 * fixed some bugs revealed by Paul Dietz' test suite:
115 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
116 ** attempting to create a package with a colliding nickname causes
117 correctable errors to be signalled.
118 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
119 ** several x86-64 backend bugs related to sign-extension of immediate
122 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
123 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
124 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
126 * fixed loading of multiply forward-referenced layouts.
127 (thanks to Cheuksan Wang)
128 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
129 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
130 (see COMPILE-FILE documentation for details of :PRINT option)
131 * fixed bugs 19 and 317: fixed-format floating point printing is
132 more accurate. This also fixes a bug reported by Adam Warner
133 related to the ~@F format directive.
134 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
136 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
137 dispatch table if the from-char is a dispatch macro character.
138 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
139 by the wordsize no longer ignore the last word. (reported by Lutz
141 * fixed bug in type checking of dynamic-extent variables. (reported
143 * optimization: sequence traversal functions use their freedom to
144 coerce function designators to functions.
145 * optimization: code with many calls to CLOS methods specialized on
146 CLOS classes has had redundant type checks removed.
147 * optimization: added declarations to speed up operations that access
148 the internal character database (for example STRING-UPCASE)
149 * optimization: comparison operations between floats and sufficiently small
150 fixnums no longer create extra rationals
151 * fixed some bugs related to Unicode integration:
152 ** portions of multibyte characters at the end of buffers for
153 character-based file input are correctly transferred to the
154 start of the buffer at the next read.
155 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
156 passing it through to OPEN.
157 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
158 argument given to internal calls to OPEN.
159 ** the built-in comment readers (introduced by character sequences
160 ";" and "#|") are more forgiving to encoding errors; they will
161 STYLE-WARN and then attempt to resync the stream at a character
162 boundary. (thanks to Teemu Kalvas)
163 * fixed some bugs in the x86-64 port:
164 ** Negative short int return values from c-calls are sign-extended
166 ** The stack is aligned to 16-bytes for c-calls, as required by
167 the ABI. (thanks to Cheuksan Wang)
168 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
169 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
170 for enums. (thanks to Vincent Arkesteijn)
171 ** Multiple small optimizations and bugfixes for floating point
173 * fixed some bugs revealed by Paul Dietz' test suite:
174 ** Space, Tab, Linefeed, Return and Page have the invalid
175 secondary constituent character trait.
176 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
178 ** WITH-INPUT-FROM-STRING should only update the index place on
180 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
182 ** Bit-array operations (BIT-AND and similar) worked incorrectly
183 with one-dimensional arrays with fill pointers.
184 ** TYPE-OF failed on a complex with an integer realpart and a
186 ** compiler failure during type inference for the code of form
187 (IF (EQL X (THE ...)) ...) (MISC.535).
189 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
190 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
191 platform. The port supports 61-bit fixnums, large memory spaces
192 and reloading shared object files.
193 * enhancement: saving cores with foreign code loaded is now
194 supported on ppc/Darwin in addition to the previously supported
196 * enhancement: the statistical profiler now walks deeper into the
197 call stack for more meaningful call-graphs and accrued time
198 reports (x86/x86-64 only). It also now reports time spent in
200 * enhancement: it is now possible to trace most individual methods
201 of a generic function in addition to tracing the generic function
203 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
205 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
206 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
207 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
208 produces an error. (thanks to Vincent Arkesteijn)
209 * bug fix: NAMESTRING on pathnames with :WILD components in their
210 directories works correctly. (thanks to Artem V. Andreev)
211 * fixed bug 125: compiler preserves identity of closures. (reported
213 * bug fixed: functions with &REST arguments sometimes failed with
214 "Undefined function" when compiled with (DEBUG 3). (reported by
216 * bug fix: overflow during compiling of setting element of a bit
217 vector with constant index and value. (reported by Timmy Douglas)
218 * build fix: fixed the dependence on *LOAD-PATHNAME* and
219 *COMPILE-FILE-PATHNAME* being absolute pathnames.
220 * on x86 compiler partially supports stack allocation of dynamic-extent
222 * GO and RETURN-FROM do not check the extent of their exit points
223 when compiled with SAFETY 0.
224 * fixed some bugs related to Unicode integration:
225 ** encoding and decoding errors are now much more robustly
226 handled; it should now be possible to recover even from invalid
227 input or output to the terminal. (thanks to Teemu Kalvas)
228 ** provided a first cut at implementing STRING-TO-OCTETS and
229 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
230 ** altered the SB-MD5 contributed module slightly, changing the
231 interface just enough for it to be supportable for builds where
232 lisp characters are not eight bits.
233 * fixed some bugs revealed by Paul Dietz' test suite:
234 ** the FORMATTER-generated functions for ~V[ conditionals require
235 the correct number of arguments.
236 ** READ-FROM-STRING returns the mandated second value when applied
237 to displaced strings.
238 ** the #\Rubout and #\Backspace characters are treated as invalid
239 constituent characters by the tokenizer.
241 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
242 * new feature: reloading changed shared object files with
243 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
244 * new feature: references to foreign variables and functions
245 can now be compiled and loaded before the corresponding shared
246 object file is loaded, as long as the foreign definitions are
247 available at runtime.
248 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
249 supported. (thanks to Dan Debertin)
250 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
251 just keywords. (thanks to Vincent Arkesteijn)
252 * fixed bug #331: structure-class instances corresponding to
253 DEFSTRUCT forms are now created eagerly.
254 * fixed bug #345: backtraces from calls to undefined functions work
255 on x86 as well. Related bug #61 is now also partially fixed on x86
256 (backtraces from throws to unknown catch tags.)
257 * bug fix: lambda-list parsing is now stricter vrt. order and number
258 of lambda-list keywords.
259 * bug fix: as specified by AMOP, an error is signalled if a
260 class-option appears multiple times in a DEFCLASS form. (reported
262 * bug fix: RANDOM can be compiled when the compiler derives the type
263 of its numeric argument as a disjoint set of small integers.
264 (reported by Paul Dietz)
265 * bug fix: starting a core saved with shared objects loaded when
266 those objects are not available no longer causes threaded SBCL to
267 hang. (reported by Sean Ross)
268 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
270 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
271 argument, which must always be NIL. (reported by Kalle Niemitalo)
272 * bug fix: printing 1.0d+23 no longer results in an error.
273 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
274 * bug fix: structure slot setters preserve evaluation order. (thanks
276 * fixed some bugs related to Unicode integration:
277 ** RUN-PROGRAM can allow its child to take input from a Lisp
278 stream. (reported by Stefan Scholl)
279 ** ASDF-INSTALL successfully downloads .tar.gz files in all
280 locales. (reported by Ken Causey)
281 * fixed some bugs revealed by Paul Dietz' test suite:
282 ** INCF, DECF and REMF evaluate their place form as specified in
284 ** FORMATTER expands ~{ ~} iteration directives with V or #
285 parameters correctly.
286 ** FORMATTER deals with the ~@[ ~] conditional directive where the
287 consequent uses no arguments correctly.
288 ** the system has a partial understanding of the (COMPLEX RATIO)
290 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
293 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
294 * new feature: a build-time option (controlled by the :SB-UNICODE
295 keyword feature, enabled by default) for building the system with
296 support for the entire 21-bit character space defined by the
298 * new feature: the system now has rudimentary external-format
299 support; the primary user-visible change at this time is that
300 characters with the high bit set (such as lower-case-e-acute) will
301 print correctly to a terminal in a UTF-8 environment.
302 * minor incompatible change: BASE-CHAR no longer names a class;
303 however, CHARACTER continues to do so, as required by ANSI.
304 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
305 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
306 has been moved to the SB-EXT package (temporarily re-exported from
308 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
309 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
311 * fixed bug #351: better error-handlind and reporting for malformed
313 * fixed bug #350: bignum-printing is now more memory-efficient,
314 allowing printing of very large bignums, eg. (expt 2 10000000).
315 (reported by Bruno Haible)
316 * fixed bug #302: better primitive-type selection for intersection
318 * fixed bug #308: non-graphic characters now all have names, as
319 required. (reported by Bruno Haible)
320 * bug fix: redefining a class with different superclasses now correctly
321 removes it from the direct-subclasses of its previous superclasses.
322 (reported by David Morse)
323 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
324 new value now works. (reported by Bruno Haible)
325 * bug fix: correct canonicalization of multiple non-standard slot
326 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
327 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
328 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
329 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
330 now exists, an signals an error.
331 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
332 during macroexpansion for non-iteration variables as well. (reported
333 by Bruno Haible for CMUCL)
334 * bug fix: Cyclic structures and unprintable objects in compiler
335 messages no longer cause errors. (reported by Bruno Haible)
336 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
337 and READ-FROM-STRING all now return a primary value of NIL if
338 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
339 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
340 (reported by Bruno Haible for CMUCL)
341 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
342 do not designate a condition. (reported by Bruno Haible for
344 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
345 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
346 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
347 permitting resolution in favour of any of the conflicting symbols.
348 (reported by Bruno Haible for CMUCL)
349 * bug fix: EQUAL compiler optimizations is less aggressive on
350 strings which can potentially compare true despite having distinct
351 specialized array element types.
352 * bug fix: unit enumerations can be defined without dividing by
353 zero. (thanks to Vincent Arkesteijn)
354 * FORMAT compile-time argument count checking has been enhanced.
355 (report from Bruno Haible for CMUCL)
356 * a partial workaround for the bug 262: the compiler does not try to
357 inline-expand a local function doing RETURN-FROM from a deleted
358 BLOCK. (thanks to Peter Denno for the bug report and to David
359 Wragg for the simple test case)
360 * fixed some bugs revealed by Paul Dietz' test suite:
361 ** RENAME-PACKAGE allows all package designators as new package
363 ** constraint propagation and lambda variable substitution are
364 more cautious in dealing with partially deleted code.
365 ** compiler sometimes lost reoptimization passes.
366 ** CERROR, given a condition as condition designator, passes its
367 remaining arguments to the continue format control without
369 ** Case-altering FORMAT directives work correctly on non-ASCII
371 ** The REAL type specifier handles bounds outside the floating
372 point ranges without signalling FLOATING-POINT-OVERFLOW.
373 ** Functions with IR1-transformations can create intercomponent
374 references to global functions.
375 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
377 ** FORMAT ~R treats a nil value for its first parameter correctly.
379 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
380 * enhancement: saving cores with foreign code loaded is now
381 supported on x86/NetBSD and sparc/Linux in addition to the previously
383 * bug fix: on some platforms repeated installations caused multiple
384 copies of HTML documentation to be installed -- should not happen
385 any more. (reported by Stefan Scholl)
386 * bug fix: parsing self-recursive alien record types multiple times
387 no longer causes infinite recursion. (reported by Thomas F. Burdick,
388 original patch by Helmut Eller for CMUCL)
389 * bug fix: stack-exhaustion detection works now on NetBSD as well.
390 (thanks to Richard Kreuter)
391 * bug fix: defining classes whose accessors are methods on existing
392 generic functions in other (locked) packages no longer signals
393 bogus package lock violations. (reported by François-René Rideau)
394 * bug fix: special variables as DEFMETHOD parameters no longer have
395 associated bogus type declarations. (reported by David Wragg and
397 * bug fix: read-write consistency on streams of element-type
398 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
399 * bug fix: redefiniton of the only method of a generic function with
400 no DEFGENERIC no longer emits a full WARNING. In addition,
401 redefinition of generic functions with no DEFGENERIC to an
402 incompatible lambda list now signals an error. (thanks to Zach
404 * bug fix: DEFGENERIC now works even when there's a function of the
405 same name in an enclosing lexical environment. (thanks to Zach
407 * fixed compiler failure, caused by instrumenting code during
408 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
409 * optimization: added loop analysis and improved register allocation
410 * fixed some bugs revealed by Paul Dietz' test suite:
411 ** POSITION on displaced vectors with non-zero displacement
412 returns the right answer.
413 ** (SIMPLE-STRING) is a valid type specifier for sequence
415 ** *PRINT-LEVEL* handling for slotless structures is pedantically
417 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
418 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
420 ** The justification version of the FORMAT ~< directive treats
421 non-zero minpad parameter correctly.
423 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
424 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
425 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
426 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
427 the supported interface.
428 * new feature: Single-stepping of code compiled with DEBUG 2 or
429 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
430 * new feature: saving cores with foreign code loaded is now
431 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
432 Timothy Moore's work for CMUCL)
433 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
434 parameters to * instead of NIL if no initform is supplied.
435 (reported by Johan Bockgaard)
436 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
437 correctly when FUNCALL appears as the car of the form. Note:
438 despite this FUNCALL forms are not currently subject to
439 compiler-macro expansion. (port of Raymond Toy's fix for the
440 same from CMUCL, reported by Johan Bockgaard)
441 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
442 (thanks for Teemu Kalvas)
443 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
444 case instead of terminating the iteration (thanks for Julian
445 Squires, Sean Champ and Raymond Toy)
446 * bug fix: incorrect expansion of defgeneric that caused a style
447 warning. (thanks for Zach Beane)
448 * bug fix: slot accessor effective method computation works properly
449 for classes with multiple non-standard applicable methods on
450 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
451 * on x86 compiler supports stack allocation of results of LIST and
452 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
453 CMUCL implementation by Gerd Moellmann)
454 * fixed some bugs revealed by Paul Dietz' test suite:
455 ** FORMAT strings with both the ~<~:;~> form of the justification
456 directive and pretty-printing directives cause an error.
458 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
459 * incompatible change: the internal functions
460 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
461 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
462 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
463 instead of the old functions.
464 * new feature: on platforms where "dladdr" is available foreign
465 function names now appear in backtraces. (based on Helmut Eller's
467 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
468 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
470 * documentation: ASDF manual is now distributed with SBCL.
471 * bug fix: SBCL can now load its contributed modules with REQUIRE
472 even if the system-provided entries have been removed from
473 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
475 * bug fix: backtraces involving undefined functions or assembly
476 routines are more informative. (thanks to Brian Downing)
477 * bug fix: mutually referent alien structures now work correctly.
478 (reported by Rick Taube)
479 * bug fix: structures defined by WITH-ALIEN can be referred to
480 within other definitions in the same WITH-ALIEN.
481 * bug fix: division operators (MOD, TRUNCATE and the like) with
482 constant zero divisors and integer dividends no longer generate
484 * bug fix: provide default methods for INPUT-STREAM-P and
485 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
486 * bug fix: improve the detection and resolution of MOP metacycles
487 (where effective method computation depends on the generic function
488 having its effective method computed). (reported by Bruno Haible)
489 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
490 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
491 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
492 * bug fix: pathnames with relative directory components can be
493 represented relative to default pathnames.
494 * optimization: in taking the GCD of bignums, reduce the two bignums
495 to approximately the same size (using Euclid's algorithm) before
496 applying the more sophisticated binary GCD. (thanks to Juho
498 * optimization: COUNT on bitvectors now operates word-at-a-time.
499 * optimization: ASH with a positive, but not necessarily constant,
500 (leftwards) shift, when in a modular context, is compiled to a
502 * fixed some bugs revealed by Paul Dietz' test suite:
503 ** FORMAT variable parameters ("~V<char>") are defaulted properly
504 if the corresponding argument is NIL.
505 ** FORMAT directives accepting a minpad parameter treat negative
507 ** literal commas as character parameters to FORMAT directives are
509 ** literal spaces directly after ~<Newline> directives within a
510 format-logical-block (~:< ~@:>) do not induce :FILL-style
511 conditional newlines.
512 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
513 as specified: it no longer includes conditional newlines).
514 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
516 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
517 :PER-LINE-PREFIX argument does not evaluate to a string.
519 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
520 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
521 the manual for details; for now, package locks can be disabled by
522 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
523 if no major problems are found then it is likely that they will be
524 compiled in unconditionally.
525 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
526 unsupported operators on all platforms. To load a shared library
527 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
528 object file, link it into a shared library outside of SBCL and
529 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
531 * bug workaround: ROOM T and the GC were not working together
532 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
533 reporting facility aren't preserved. That reporting has been
534 disabled, so now until and unless someone figures out how to make it
535 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
536 * minor incompatible change: as threatened around sbcl-0.8.0, the
537 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
538 The symbols are also exported from SB-PCL for backwards
539 compatibility, but more so than before SB-PCL should be treated as
540 an implementation-internal package.
541 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
543 * the SB-SPROF contrib now works on (most) non-x86 architectures.
544 It is known as of this release not to work on the Alpha, however.
545 * fixed bug #167: errors signalled due to illegal syntax in method
546 bodies are now more legible.
547 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
548 designators and can hence be used with TYPEP.
549 * fixed bug #333: CHECK-TYPE now ensures that the type error
550 signalled, if any, has the right object to be accessed by
551 TYPE-ERROR-DATUM. (reported by Tony Martinez)
552 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
553 5.1.2.3 for multiple-value place subforms. (reported by Kalle
555 * fixed bug #334: programmatic addition of slots using specialized
556 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
557 and :ALLOCATION :CLASS slots.
558 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
559 float exponents. (rereported by Peter Seibel)
560 * fixed a bug: #\Space (and other whitespace characters) are no
561 longer considered to be macro characters in standard syntax by
563 * fixed bug: initialization of condition class metaobjects no longer
564 causes an instance of the condition to be created. (reported by
566 * fixed bug: it is now possible to have slots such that
567 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
568 description is neither :INSTANCE nor :CLASS.
569 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
570 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
571 SLOT-VALUE-USING-CLASS), and no longer causes errors with
572 non-standard SLOT-DEFINITION-ALLOCATION values.
573 * fixed bugs: various race conditions handling exiting threads.
574 CL-PPCRE's thread torture test now passes.
575 * fixed arguably-a-bug: GC time proportional to number of threads in
576 system even when most of them are idle
577 * optimization: improved performance of BIT and SBIT on bit-vectors.
578 * .fasl file incompatibility: The fasl file version number has
579 been incremented because of changes associated with package locks.
581 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
582 * minor incompatible change: the system no longer provides
583 optimization hints (conditions which are TYPEP
584 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
585 mode; these hints are still emitted when higher SPEED optimization
587 * new contrib module: a sampling profiler (profiling by statistical
588 sampling, rather than by instrumenting functions) is available as
589 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
590 * the behaviour of the standard function ED is now customizeable by
591 third parties through a hook variable: see ED's documentation
592 string for information on the protocol.
593 * the compiler no longer emits efficiency notes for (FUNCALL X)
594 when the type of X is uncertain under default optimization
596 * fixed bug 276: mutating a binding of a specialized parameter to a
597 method to something that is not TYPEP the specializer is now
599 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
600 exhibit the required behaviour on the x86 platform. (thanks to
601 Peter van Eynde, Eric Marsden and Bruno Haible)
602 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
603 even for difficult arguments. (reported by Peter Graves)
604 * fixed bug 141a: the backquote printer now descends quoted
606 * fixed another bug in backquote printing: no more destructive
607 modification of the form's list structure. (reported by Brian
609 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
610 move between its address being taken and the call to
611 interrupt_thread, fixing a crashing race condition.
612 * the SB-POSIX contrib implementation has been adjusted so that it
613 no longer exhibits ridiculously poor performance when constructing
614 instances corresponding to C structs.
616 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
617 * minor incompatible change: the sb-grovel contrib now treats C
618 structures as alien (in the sense of SB-ALIEN) objects rather than
619 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
620 has implications for memory management of client code
621 (sb-grovel-returned objects must now be manually managed) and for
622 type safety (alien objects now have full types).
623 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
624 used to control emission of compiler diagnostics, rather than the
625 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
626 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
627 quality should be considered deprecated.
628 * install.sh now installs the user manual as well
629 * (not quite a new documentable feature, but worth considering in
630 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
631 beginnings of a semantically meaningful condition hierarchy is
632 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
634 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
635 designator as the defaults argument.
636 * fixed bug: Displaced arrays whose displaced-to array has become
637 too small now cause ARRAY-DIMENSION to signal an error, providing
638 for safer bounds-checking. (reported by Bruno Haible)
639 * fixed bug: DEFCLASS slot definitions with identical :READER and
640 :WRITER names now signal a reasonable error. (reported by Thomas
642 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
644 * fixed bug: Invalid slot specification errors now print correctly.
645 (thanks to Zach Beane)
646 * fixed bug 320: Shared to local slot value transfers in class
647 redefinitions now happen corrently from superclasses as
648 well. (reported by Bruno Haible)
649 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
651 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
652 improper lists correctly. (reported by Bruno Haible)
653 * fixed bug 313: source-transform for <fun-name> was erroneously
654 applied to a call of a value of a variable with name <fun-name>.
655 (reported by Antonio Menezes Leitao)
656 * fixed bug 307: The obsolete instance protocol ensures that
657 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
658 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
659 remove dead unknown-values globs from the middle of the stack.
660 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
661 easier to resume long computations after using *BREAK-ON-SIGNALS*
662 to diagnose and fix failures. (thanks to Nikodemus Siivola)
663 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
664 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
665 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
666 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
667 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
669 * fixed a (fairly theoretical) bug in string printing: if
670 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
671 does not have array-element-type equal to the most general string
673 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
674 function name. (thanks to Nikodemus Siivola)
675 * fixed bug: compile-time format string checker failed on
676 non-closed ~{. (reported by Thomas F Burdick)
677 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
678 don't warn on legal loop code involving more than one aggregate
679 boolean. (thanks to Nikodemus Siivola)
680 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
681 in keyword position to MAKE-PACKAGE was not being reported as
683 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
684 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
686 * fixed bugs 280 and 312: the checking for multiple definitions in a
687 file is less likely to become confused by uses of inline
689 * fixed bug: the #S reader macro performs the keyword coercion
690 specified for slot names. (reported by Kalle Niemitalo)
691 * fixed bug: lambda lists may contain symbols whose names start with
692 & but are not lambda-list-keywords; their occurrence triggers a
694 * fixed bug 321: define-method-combination argument lambda lists do
695 not cause invalid code to be generated when &optional or &aux
696 variables have default values. (reported by Bruno Haible)
697 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
698 have CLOS classes; this appears also to have fixed reported
699 instability in stack exhaustion detection.
700 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
701 MUFFLE-WARNING when no associated restart is present is now
703 * optimization: rearranged the expansion of various defining macros
704 so that each expands into only one top-level form in a
705 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
707 * optimization: used a previously little-used slot in symbols to
708 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
710 * fixed some bugs revealed by Paul Dietz' test suite:
711 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
713 ** Fixed an optimization bug related to inheritance of initforms
714 from local to shared slots.
715 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
716 ** CAST optimizer forgot to flush argument derived type.
717 ** print/read consistency on floats is now orders of magnitude
718 more likely. (thanks also to Bruno Haible for a similar report
720 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
721 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
722 for objects of type REAL. Make it so.
723 ** SXHASH is specified (infelicitously) to respect similarity,
724 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
725 it so. (thanks to Markus Ziegler)
726 ** on the Alpha, the compiler succeeds in compiling functions
727 returning a known number of arguments greater than 63.
728 ** fixed handling of invalid NIL arguments in keyword position
730 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
732 ** provided a readably-printable representation for RANDOM-STATE
734 ** ensured that pathnames, where they have a namestring, always
735 print using #P"..." syntax.
737 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
738 * Support for the forthcoming 2.0 version of the NetBSD kernel
739 running on x86 hardware has been added. (thanks to Perry
740 E. Metzger most immediately, and others for their past work)
741 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
742 current release version; SBCL's OpenBSD support had been broken
743 since about the time of OpenBSD's switch to ELF binary format.)
744 * [placeholder for DX summary]
745 ** user code with &REST lists declared dynamic-extent, under high
746 speed or space and low safety and debug optimization policy.
747 * The manual has been converted to Texinfo format and the debugger
748 chapter from the cmucl manual has been added.
749 * A facility has been added to extract documentation strings from
750 sbcl and store them as Texinfo-formatted snippets for inclusion in
751 the manual (via Texinfo's @include directive)
752 * bug fix: compiler emitted division in optimized DEREF. (thanks for
753 the test case to Dave Roberts)
754 * bug fix: multidimensional simple arrays loaded from FASLs had fill
755 pointers. (reported by Sean Ross)
756 * bug fix: PROFILE output is printed nicely even for large numerical
757 values. (thanks to Zach Beane)
758 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
759 greater than 32 handle EOF correctly.
760 * bug fix: on X86 an immediate argument of the IMUL instruction is
761 correctly printed in disassembly. (thanks to Lutz Euler)
762 * bug fix: class slots in redefined classes preserve their old
763 values. (thanks to Bruno Haible and Nikodemus Siivola)
764 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
765 succeeds. (reported by Marco Baringer)
766 * fixed some bugs revealed by Paul Dietz' test suite:
767 ** READ-SEQUENCE now works on ECHO-STREAMs.
768 ** RATIONALIZE works more according to its specification. (thanks
771 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
772 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
773 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
774 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
775 should matter to you only if you rebind the printer control
776 variables and then find you want different bindings in the
777 debugger than in the ordinary execution of your program.)
778 * The runtime build system has been tweaked to support building
779 (on SPARC/SunOS) using a C compiler which invokes Sun's own
780 assembler and linker. (thanks to Nikodemus Siivola)
781 * Unbound, undefined, undeclared variables now trigger full
782 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
783 more often programmer error than deliberate exploitation of undefined
785 * optimization: the hash algorithm for strings has changed to one
786 that is less vulnerable to spurious collisions. (thanks to Juho
788 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
789 needless bounds checking. (thanks to Juho Snellman)
790 * optimization: implemented multiplication as a modular
791 (UNSIGNED-BYTE 32) operation on the PPC backend.
792 * fixed some bugs revealed by Paul Dietz' test suite:
793 ** ADJUST-ARRAY now copies the datum in a zero rank array if
795 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
796 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
797 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
799 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
801 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
803 ** ECHO-STREAMs no longer attempt to echo the end of file value to
804 their output stream on EOF from read.
805 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
806 have been read to end-of-file.
807 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
809 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
810 description of determination of which consecutive characters
812 ** Printing the "Space" character with escaping on now yields "#\\ ",
813 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
814 ** Reading floating-point numbers with *READ-BASE* set to a number
815 less than 10 works correctly.
816 ** Reading floating-point numbers with *READ-BASE* set to a number
817 more than 10 works correctly.
818 ** Printing with *PRINT-READABLY* targets the standard readtable, not
819 the readtable currently in effect.
821 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
822 * minor incompatible change: parsing of namestrings on a physical
823 (Unix) host has changed; numbers after the final #\. in a
824 namestring are no longer interpreted as a version field. This is
825 intented to be largely invisible to the user, except that the
826 meaning of the namestring "*.*.*" has changed: it now refers to a
827 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
828 should usually be replaced by
829 (make-pathname :name :wild :type :wild :version :wild)
830 with the added benefit that this is more likely to be portable.
831 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
832 option to OPEN now signals an error if the file being opened
833 exists; this may have an impact on existing code.
834 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
835 signals received, on the PowerPC platforms (both Linux and
836 Darwin). (thanks to Pierre Mai for pointing out the location of
838 * several fixes on OS X: The system now builds and runs cleanly on
839 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
840 (thanks to Brian Mastenbrook)
841 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
842 second-resolution: integer multiples of 1/3600 between -24 and 24.
843 (thanks to Vincent Arkesteijn)
844 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
845 type when called with 1 argument; PEEK-CHAR checked type of
846 PEEK-TYPE only after having read first character from a
847 stream. (reported by Peter Graves)
848 * bug fix: the garbage collector now has much better locality
849 behaviour, and in particular no longer treats all memory as being
850 exhausted when many small objects point to each other in a deeply
852 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
853 recognized as being TYPEP their class.
854 * bug fix: the PUSHNEW documentation string has been corrected.
855 (thanks to Vincent Arkesteijn)
856 * bug fix: defaulting of the value for the last of an atypically
857 large number of multiple values being bound was not being
858 performed correctly on the Alpha or PPC platforms
859 * optimization: implemented multiplication as a modular
860 (UNSIGNED-BYTE 32) operation on the x86 backend.
861 * optimization: SEARCH on simple-base-strings can now be open-coded.
862 (see also contrib/compiler-extras.lisp for inspiration for
863 teaching the compiler about the Boyer-Moore algorithm).
864 * value, returned by MAX (and MIN) called with several EQUALP, but
865 not EQL, arguments now does not depend on compiler settings.
866 * fixed some bugs revealed by Paul Dietz' test suite:
867 ** in stack analysis liveness information is propagated from
868 non-local entry points.
869 ** pathwise CAST removing failed when the CAST node did not start
871 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
872 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
874 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
875 host is already defined.
876 ** RENAME-FILE works on streams instead of signalling an internal
878 ** PEEK-CHAR uses the current readtable when determining whether
879 or not a character is whitespace.
880 ** MERGE-PATHNAMES handles the case when the pathname does not
881 specify a name while the default-pathname specifies a version.
882 ** Pathnames now stand a better chance of respecting print/read
884 ** Attempting to use standardized file system operators with a
885 pathname with invalid :DIRECTORY components signals a
887 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
889 ** DIRECTORY on logical pathnames is more correct.
890 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
891 signal a TYPE-ERROR if their argument is not a stream.
892 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
893 designator argument does not designate a stream.
894 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
895 examining the synonym.
896 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
898 ** OPEN and WITH-OPEN-STREAM allow opening streams with
899 element-type larger than ([UN]SIGNED-BYTE 32).
901 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
902 * When built with the :SB-FUTEX feature, threaded builds now take
903 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
904 for faster/more reliable mutex and condition variable support.
905 * Incompatible change (but one you probably shouldn't have been using
906 anyway): the interface and code for arbitrating between multiple
907 threads in the same user session has been redesigned.
908 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
909 error when called without an explicit environment argument.
910 (thanks to Nikodemus Siivola)
911 * bug fix: buffered :DIRECTION :IO streams are less likely to become
912 confused about their position. (thanks to Adam Warner and Gerd
914 * bug fix: Pretty printing backquoted forms with unquotations in the
915 argument list position of various code constructs such as LAMBDA
916 now works correctly. (reported by Paul Dietz)
917 * bug fix: Pretty printing unquotations no longer loses all
918 stream position information.
919 * optimization: performance of string output streams is now less
920 poor for multiple small sequence writes.
921 * optimization: performance of CSUBTYPEP in the presence of complex
922 expressions involving CONS and NOT many times has been improved.
923 (reported by Paul Dietz)
924 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
926 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
927 streams for the REPL, for the debugger, and so on. (thanks to
929 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
930 * fixed some bugs revealed by Paul Dietz' test suite:
931 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
932 is ignored if it is too small, rather than propagating through
933 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
934 ** extremely complex negations of CONS types were not being
935 sufficiently canonicalized, leading to inconsistencies in
937 ** VALUES tranformer lost derived type.
939 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
940 * fixed a bootstrapping bug: the build process no longer assumes
941 that the various BOOLE-related constants have the same value in
942 host and target lisps. (noted by Paul Dietz' test suite on an
943 SBCL binary built from CLISP)
944 * The system can now be dynamically linked on the MIPS platform,
945 which enables dynamic loading of foreign code from Lisp. (thanks
946 to Ralf Baechle for discussions on the MIPS ABI)
947 * The system now records debugging information for its own source
948 files in a filesystem-position-independent manner, relative to
949 the "SYS" logical host.
950 * fixed a compiler bug: MV-LET convertion did not check references
951 to the "max args" entry point. (reported by Brian Downing)
952 * tweaked disassembly notes to be less confident about proclaiming
953 some instruction as an LRA. (thanks to Brian Downing)
954 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
955 now each have their own history, command character, and other
956 characteristics. (thanks to David Lichteblau)
957 * fixed some bugs revealed by Paul Dietz' test suite:
958 ** compiler failure in compiling LOGAND expressions including a
960 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
961 shift greater than 32.
962 ** FLUSH-DEST did not mark blocks for type check regeneration.
963 ** HANDLER-CASE failed to accept declarations in handler clauses
964 in some circumstances.
966 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
967 * New code in contrib/sb-introspect (still probably not entirely
968 stable yet) provides some support for smart Lisp development
969 environments like SLIME.
970 * The conditions signalled for errors occurring when loading .fasl
971 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
972 in a way which should help ASDF recover gracefully.
973 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
974 stuff has been cleaned up. If you code contrib/ stuff, this might
975 affect you, and you can look at contrib/README, contrib/STANDARDS,
976 and/or the 0.8.4.27 diff to check.
977 * In full calls the compiler now does not generate checks for declared
978 argument types for all arguments.
979 * various threading fixes
980 ** and some experimental patches which didn't make it into
981 the main tree for this release, but which are shipped in
982 contrib/experimental-thread.patch as a possible fix for some
983 failures (deadlock, spinning...) in GC-intensive multithreaded
985 * fixed PPC build problem (source code incompatibility of different
986 library versions): added offsetof() hackery which attempts to divine
987 where glibc maintainers put uc_mcontext today
988 * fixed bug 282: compiler does not trust type assertions while passing
989 arguments to a full call.
990 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
991 &OPTIONAL VALUES type parameter.
992 * fix bug 214: algorithm for noting rejected templates is now more
993 similar to that of template seletion. (also reported by rydis on
995 * fixed bug 141b: printing backquoted information readably and prettily
996 inserts a space where necessary.
997 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
998 (CHARACTER 10) are now reported as errors, rather than propagated
999 as unknown types. (reported by piso on #lisp)
1000 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1001 with values NIL and :ERROR. (thanks to Milan Zamazal)
1002 * fixed bug 191c: CLOS now does proper keyword argument checking as
1003 described in CLHS 7.6.5 and 7.6.5.1.
1004 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1005 longer raise an error; further, using a list as a for-as-arithmetic
1006 counter now raises a meaningful error.
1007 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1008 now understood by sequence creation functions such as MAKE-SEQUENCE
1010 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1011 a STRING-INPUT-STREAM.
1012 * compiler enhancement: SIGNUM is now better able to derive the type
1014 * type declarations inside WITH-SLOTS are checked. (reported by
1016 * fixed some bugs revealed by Paul Dietz' test suite:
1017 ** incorrect optimization of TRUNCATE for a positive first
1018 argument and negative second.
1019 ** compiler failure in let-convertion during flushing dead code.
1020 ** compiler failure while deriving type of TRUNCATE on an
1021 interval, containing 0.
1022 ** ASH of a negative bignum by a negative bignum count now returns
1024 ** intersection of CONS types now canonicalizes properly, fixing
1025 inconsistencies in SUBTYPEP.
1027 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1028 * incompatible change: The --disable-debugger command line
1029 option now clobbers the debugger at a more fundamental
1030 level, by redefining #'INVOKE-DEBUGGER instead of by
1031 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1032 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1033 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1034 enter the debugger REPL and then suffer recursive errors
1035 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1036 while under the new system, BREAK in --disable-debugger mode
1037 terminates the system just as an unhandled error would.
1038 * fixed compiler performance when processing loops with a step >1;
1039 * bug fix: DOCUMENTATION now retrieves generic function
1040 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1041 support has been systematized, and now supports the methods
1042 specified by ANSI, along with a default method and a method for
1043 slot documentation. (reported by Nathan Froyd)
1044 * bug fix: effective methods associated with a generic function are
1045 no longer cached over a change of that generic function's method
1046 combination. (reported by Andreas Fuchs)
1047 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1048 names a non-existent directory. (thanks to Andreas Fuchs)
1049 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1050 the PPC platform, lead to stack corruption; this has been fixed.
1051 (reported by Rainer Joswig)
1052 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1053 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1054 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1055 arithmetic in general, is now much more reliable on the PPC
1057 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1058 platform now returns the right answer.
1059 * optimization: restored some effective method precomputation in
1060 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1061 precomputation is now tunable.
1062 * optimization: compiler-internal data structure use has been
1063 reviewed, and changes have been made that should improve the
1064 performance of the compiler by about 20%.
1065 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1066 simple-base-strings and simple-bit-vectors is improved.
1067 * optimization: the optimization of 32-bit logical and arithmetic
1068 functions introduced in version 0.8.3 on the x86 has been
1069 implemented on the mips, ppc and sparc platforms; an
1070 implementation of the same facility, but for 64-bit arithmetic,
1071 has been added for the alpha.
1072 * microoptimization: the compiler is better able to make use of the
1073 x86 LEA instruction for multiplication by constants.
1074 * bug fix: in some situations compiler did not report usage of
1075 generic arithmetic in (SPEED 3) policy.
1076 * bug 145b fix: compiler used wrong type specifier while converting
1077 MEMBER-types to numeric.
1078 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1080 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1081 index argument now works properly on the Alpha platform.
1082 * bug fix: floating point exception treatment on the Alpha platform
1084 * bug fix: FILE-POSITION works much better on string input and
1085 output streams. (thanks to Nikodemus Siivola)
1086 * bug fix: many threading/garbage collection symptoms sorted.
1087 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1088 might be pseudo-atomic.
1089 * internal change: Stopping for GC is now done with signals not
1090 ptrace. GC is now done in whichever thread wanted it, instead of
1092 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1094 * bug fix: result form in DO is not contained in the implicit
1096 * incompatible change: ICR structure is changed; the value part of
1097 CONTINUATION is now called LVAR; corresponding functions are
1098 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1099 * added type deriver for ISQRT (thanks to Robert E. Brown).
1100 * bug fix: better support for loading from the command line when an
1101 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1103 * fixed some bugs revealed by Paul Dietz' test suite:
1104 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1105 ** ROUND and FROUND now give the right answer when given very
1106 small float arguments.
1107 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1109 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1110 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1111 ** LCM should return a non-negative integer.
1112 ** PARSE-INTEGER returned the index of a terminator instead of the
1113 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1114 ** PARSE-INTEGER returned an incorrect index being applied to a
1116 ** LCM with two arguments of 0 returns 0 rather than signalling
1118 ** unsigned addition of a 32-bit constant with the high bit set no
1119 longer causes an internal compiler error on the x86.
1120 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1121 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1122 with negative last argument.
1123 ** byte specifiers with zero size and position no longer cause
1124 an error during type derivation.
1125 ** bignum multiplication on the Alpha platform now returns the
1127 * porting: The system now builds on SuSE AMD64, although it still
1128 generates a 32-bit binary.
1129 * .fasl file incompatibility: The fasl file version number has
1130 been incremented (because of the changes to internal compiler
1131 data structures referred to above).
1133 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1134 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1135 more accurately, on the Darwin kernel running on PowerPC hardware.
1136 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1137 * Compiler code deletion notes now signal a condition of type
1138 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1139 an associated MUFFLE-WARNING restart.
1140 * The compiler now performs limited argument count validation of
1141 constant format strings in FORMAT, and where appropriate in ERROR,
1142 CERROR and WARN. (thanks to Gerd Moellmann)
1143 * New ASDF-INSTALL contrib can be used for automatic download and
1144 installation of third-party Lisp code from CCLAN or other sites
1146 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1147 which forces another thread to execute a function supplied by the
1149 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1150 now accept and act upon their :ELEMENT-TYPE keyword argument.
1151 (reported by Martin Atzmueller, Edi Weitz)
1152 * bug fix: FILE-POSITION now accepts position designators up to
1153 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1154 is the greater. (thanks to Patrik Nordebo)
1155 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1156 Kalle Olavi Niemitalo)
1157 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1158 printable. (reported by Eric Marsden)
1159 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1160 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1161 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1162 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1163 (thanks to Stig Sandoe)
1164 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1165 UNREAD-CHAR now correctly includes the unread character in the
1166 target sequence. (thanks to Gerd Moellmann)
1167 * bug fix in threaded builds: the system can now be suspended and
1168 resumed by shell job control with minimal disruption.
1169 * bug fixes in times and timezones >2038AD
1170 * better handling of "where is GNU make?" problem in build scripts
1171 (thanks to Nikodemus Siivola)
1172 * new optimization: inside a named function any reference to a
1173 function with the same name is considered to be a self-reference;
1174 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1175 optimization quality.
1176 * new optimization on x86: logical functions and + now have
1177 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1178 used when the result is truncated to 32 bits.
1179 * VALUES declaration is partially enabled.
1180 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1181 * fixed some bugs revealed by Paul Dietz' test suite:
1182 ** The system now obeys the constraint imposed by
1183 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1184 types form a lattice under type intersection.
1185 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1186 ** ASSOC now ignores NIL elements in an alist.
1187 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1188 and (1+ MOST-POSITIVE-FIXNUM) answers.
1189 ** The addition of a method with invalid qualifiers to a generic
1190 function does not cause an error to be signalled immediately;
1191 a warning is signalled, and the error is generated only on
1192 calling the generic function.
1193 * changed .fasl file version number, in part to add type codes for
1194 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1195 obscure ANSI requirements
1197 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1198 * fixed bug 148: failure to inline-expand a local function left
1199 garbage, confusing the compiler.
1200 * fixed bugs 3cd: structure slot readers perform type check if the
1201 slot can have an invalid value (i.e. it is either not initialized
1202 or can be written with a less specific slot writer).
1203 * bug fix: the compiler now traps array references to elements off
1204 the end of an array; previously, the bounds checking in some
1205 circumstances could go off-by-one.
1206 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1208 * type declarations for array element types now obey the description
1209 on the CLHS page "Declaration TYPE", as per discussions on
1210 sbcl-help around 2003-05-08. This means that a declaration
1211 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1212 declaration, all references to BAR will be asserted or assumed
1213 (with THE, so dependent on compiler policy) to involve objects of
1214 type FOO. Note that no such declaration is implied in
1215 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1216 * declared types of functions from the "Conditions"
1217 chapter. (reported by Paul Dietz)
1218 * bug fix: CERROR accepts a function as its first argument.
1219 * bug fix: NTH an NTHCDR accept a bignum as index
1220 arguments. (reported by Adam Warner)
1221 * optimization: character compare routines now optimize comparing
1222 against a constant character. (reported by Gilbert Baumann)
1223 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1224 argument now works properly on the MIPS platform.
1225 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1226 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1227 declared to have a functional type, does not cause an error
1229 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1230 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1231 * fixed bug: if last continuation of a deleted block has a
1232 destination, this destination should be deleted too. (reported by
1234 * fixed a bug in the bootstrap process: the host compiler's values
1235 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1236 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1237 test case from Patrik Nordebo)
1238 * improved the ability of the disassembler on the PPC platform to
1239 provide helpful disassembly notes.
1240 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1241 the class in more cases than previously.
1242 * bug fix: FILE-POSITION now understands :START and :END for
1243 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1244 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1245 8) data. (thanks to David Lichteblau)
1246 * bug fix: it is possible to add a method to a generic function
1247 without lambda list.
1248 * bug fix: reader failed to signal END-OF-FILE inside an
1249 object representation. (reported by Nikodemus Siivola)
1250 * fixed some bugs revealed by Paul Dietz' test suite:
1251 ** LAST and [N]BUTLAST should accept a bignum.
1252 ** condition slot accessors are methods.
1253 ** (VECTOR NIL) is a subtype of STRING.
1255 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1256 * minor incompatible change: some nonsensical specialized lambda
1257 lists (used in DEFMETHOD) which were previously ignored now signal
1259 * minor incompatible change: the system is now aware of the types of
1260 variables in the COMMON-LISP package, and will signal errors for
1261 most violations of these type constraints (where previously they
1262 were silently accepted).
1263 * minor incompatible change: COMPILE-FILE now uses the freedom
1264 afforded (ANSI 3.2.2.3) to use derived function types for
1265 functions defined in the same file. This also permits the system
1266 to warn on static type mismatches and function
1267 redefinition. (Currently it does not work with high DEBUG level.)
1268 * minor incompatible change: VALUES declaration is disabled.
1269 * When issuing notes, the compiler now signals a condition of type
1270 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1271 restart for use in user handlers. It is expected that the
1272 COMPILER-NOTE condition will eventually become a condition
1273 supertype to a hierarchy of note types, which will then be
1274 handleable in a similar fashion. However, at the moment, no such
1275 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1276 but it's an implementation detail, not a classification for the
1278 * Changes in type checking closed the following bugs:
1279 ** type checking of unused values (192b, 194d, 203);
1280 ** template selection based on unsafe type assertions (192c, 236);
1281 ** type checking in branches (194bc).
1282 * A short form of VALUES type specifier has ANSI meaning (it has
1283 increased the number of situations when SBCL cannot perform type
1285 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1286 keywords or constants is permissible.
1287 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1288 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1289 classes. (thanks to Antonio Martinez)
1290 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1291 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1292 argument to a pretty-print stream. Instead, it leaves any such
1293 operations to DESCRIBE-OBJECT methods.
1294 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1296 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1297 lambda lists. (thanks to David Lichteblau)
1298 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1299 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1300 (thanks to Gerd Moellmann)
1301 * fixed bug 246: increased compilation speed of long
1302 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1304 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1306 * DEFCONSTANT now throws a condition of type
1307 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1308 constant to a non-EQL value; CONTINUE and ABORT restarts
1309 respectively change and preserve the value.
1310 * fixed bug 63: The code walker, part of the implementation of CLOS,
1311 is now better at handling symbol macros.
1312 * bug fix: There is no longer an internal implementation type named
1313 CL:LENGTH. (reported by Raymond Toy)
1314 * bug fix: In macro-like defining macros/special operators the
1315 implicit block does not enclose the lambda list.
1316 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1317 enclosed in parentheses) are not suitable as type specifiers, and
1318 their use properly signals an error now.
1319 * bug fix: An argument count mismatch for a type specifier in code
1320 being compiled no longer causes an unhandled error at compile
1321 time, but signals a compile-time warning.
1322 * fixed simple vector readable printing
1323 * bug fix: DESCRIBE takes more care over whether the class
1324 precedence list slot of a class is bound before accessing it.
1325 (reported by Markus Krummenacker)
1326 * bug fix: FORMATTER can successfully compile pretty-printer format
1327 strings which use variants of the ~* directive inside.
1328 * bug fix: SEARCH now applies its TEST predicate to the elements of
1329 the arguments in the correct order. (thanks to Wolfhard Buss)
1330 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1331 in type inference. (thanks to Robert E. Brown)
1332 * bug fix: Reading in symbols with an explicit package name of ""
1333 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1334 * fixed some bugs revealed by Paul Dietz' test suite:
1335 ** NIL is now allowed as a structure slot name.
1336 ** Arbitrary numbers, not just REALs, are allowed in certain
1337 circumstances in LOOP for-as-arithmetic clauses.
1338 ** Multiple class redefinitions before slot access no longer
1339 causes a type error.
1340 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1341 association between the name and a class.
1342 ** Generic functions with non-standard method-combination and over
1343 five methods all of which return constants no longer return NIL
1344 after the first few invocations. (thanks to Gerd Moellmann)
1345 ** CALL-NEXT-METHOD with no arguments now passes the original
1346 values of the arguments, even in the presence of assignment.
1347 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1349 ** DISASSEMBLE works with closures and funcallable instances.
1350 ** ADD-METHOD now returns the generic function, not the new method.
1351 ** FIND-METHOD signals an error if the lengths of the specializers
1352 is incompatible with the generic function, even if the ERRORP
1354 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1355 which its argument is a member.
1356 ** DEFCLASS only redefines the class named by its class-name
1357 argument if that name is the proper name of the class;
1358 otherwise, it creates a new class.
1359 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1360 of the UNBOUND-SLOT condition to the name of the slot.
1361 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1362 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1363 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1364 SLOT-MAKUNBOUND in the specified fashion.
1366 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1367 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1368 cross-compilation host. As a consequence, we can now bootstrap our
1369 way up to SBCL starting with a bare gcc toolchain and human-readable
1370 source code (first the source to CLISP, then the source to SBCL).
1371 * A contributed module containing a partial implementation of the
1372 simple-streams interface has been included. (thanks to Rudi
1374 * A contributed module implementing the RFC1321 Message Digest
1375 Algorithm, known as MD5, has been included.
1376 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1377 no longer has any effect, as the code controlled by this feature
1378 has been deleted. (As far as we know, no-one has ever built using
1379 this feature, and its semantics were confused in any case).
1380 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1381 work on conditions (as required by the ANSI specification),
1382 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1383 expected behaviour on conditions. Users should note, however,
1384 that such behaviour is not required by the ANSI specification,
1385 and so use of this behaviour may render their code unportable.
1386 * fixed some bugs revealed by Paul Dietz' test suite:
1387 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1389 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1390 ** SXHASH on condition objects no longer returns NIL.
1391 ** :ALLOCATION :CLASS slots are better treated; their values are
1392 updated on class redefinition, and initforms inherited from
1393 superclasses are applied.
1394 ** REMOVE-METHOD returns its generic function argument even when
1395 no method was removed.
1396 ** SHARED-INITIALIZE now initializes the values of the requested
1397 slots, including those with :ALLOCATION :CLASS.
1398 ** ALLOCATE-INSTANCE now works on structure classes defined via
1399 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1401 ** SLOT-EXISTS-P now works on conditions, as well as structures
1403 ** MAKE-LOAD-FORM now has the required methods on
1404 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1405 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1406 keyword, and now implements the SLOT-NAMES argument.
1407 ** methods with &OPTIONAL arguments no longer allow too many
1408 arguments to be passed in the call without error.
1409 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1410 option is consistent with the required arguments of the generic
1411 function lambda list.
1412 * bug fix: REQUIRE accepts a string designator. (Thanks to
1414 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1415 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1416 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1417 * bug fix: The compiler now checks for duplicated variables in macro
1419 * bug fix: SETQ on globals returns the correct value.
1420 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1421 not a COMPILER-ERROR (followed by some other strange error on
1422 choosing the CONTINUE restart).
1423 * bug fix: make.sh and friends are now more consistent in the way that
1424 they look for GNU "make".
1426 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1427 * experimental native threads support (on x86 Linux >=2.4 only).
1428 This is not compiled in by default: you need to add :SB-THREAD to
1429 the target features. See the "Beyond ANSI" chapter of the manual
1431 * fix for longstanding nonANSIism: The old distinction between
1432 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1433 The return value from CL:FIND-CLASS is now a CLOS class, and
1434 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1435 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1436 * An interface to the MetaObject Protocol, as described in Kiczales,
1437 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1438 MIT Press, 1991, is available from the SB-MOP package.
1439 * incompatible change: the SB-PCL package should now be considered
1440 a private implementation detail, and no longer a semi-private MOP
1442 * minor incompatible change: due to rearrangement for threads, the
1443 control stack and binding stack are now allocated at arbitrary
1444 addresses instead of being hardcoded per-port. Users affected by
1445 this probably have to be doing advanced things with shared
1446 libraries, and will know who they are.
1447 * minor incompatible change: Previously, all --eval forms used were
1448 processed with READ before any of them were processed with EVAL.
1449 Now each --eval form is processed with both READ and EVAL before
1450 the next --eval form is processed. (Thus package operations like
1451 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1452 work as the user might reasonably expect.)
1453 * minor incompatible change: *STANDARD-INPUT* is now only an
1454 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1456 * minor incompatible change: Y-OR-N-P is now character-oriented,
1457 not line oriented. Also, YES-OR-NO-P now works without errors.
1458 (thanks to Antonio Martinez)
1459 * sb-aclrepl module improvements: an integrated inspector, added
1460 repl features, and a bug fix to :trace command.
1461 * Known functions, which cannot be open coded by the backend, are
1462 considered to be able to check types of their arguments. (fixing
1463 a bug report by Nathan J. Froyd)
1464 * fixed a bug in computing method discriminating functions: It is
1465 now possible to define methods specialized on classes which have
1466 forward-referenced superclasses. (thanks to Gerd Moellmann)
1467 * fixed evaluation order in optional entries (reported by Gilbert
1469 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1470 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1471 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1472 required-by-AMOP NAME argument, as well as CLASS and
1473 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1474 * fixed bug 20: DEFMETHOD can define methods using names that are
1475 not the proper names of classes to designate class specializers.
1476 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1477 Unix file descriptors, instead of blowing up. (thanks to Antonio
1479 * Garbage collection refactoring: user-visible change is that a
1480 call to the GC function during WITHOUT-GCING will not do garbage
1481 collection until the end of the WITHOUT-GCING. If you were doing
1482 this you were probably losing anyway.
1483 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1484 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1485 TYPEP the latter but not the former.
1486 * The compiler issues a full WARNING for calls to undefined functions
1487 with names from the CL package.
1488 * MAP-INTO for a vector destination is open coded. (reported by
1489 Brian Downing on c.l.l)
1490 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1491 documentation string.
1492 * fixed some bugs revealed by Paul Dietz' test suite:
1493 ** COPY-ALIST now signals an error if its argument is a dotted
1495 ** Condition slots are now accessed more correctly in the presence
1496 of multiple initargs for a given slot.
1497 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1498 exclude restarts of the same name associated with a different
1500 ** DEFCLASS of forward-referenced classes with another
1501 forward-referenced class in the superclasses list no longer
1503 ** Condition slots are now initialized once each, not multiple
1504 times. (thanks to Gerd Moellmann)
1505 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1507 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1508 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1509 arguments contain duplicated elements.
1510 ** RESTART-CASE understands local macros.
1511 ** RESTART-CASE associates exactly its own restarts with a condition.
1512 ** ENDP in safe mode checks its argument to be of type LIST.
1513 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1514 in question is unbound.
1515 ** Optimized MAKE-INSTANCE functions no longer cause internal
1516 assertion failures in the presence of duplicate initargs.
1517 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1518 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1519 propagated even in the ctor optimized implementation of
1521 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1523 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1524 * a better implementation of SXHASH on (simple) bit vectors,
1525 measured both in execution speed and in distribution of results
1526 over the positive fixnums, has been installed. Likewise, a better
1527 implementation of EQUAL for simple bit vectors is now available.
1528 * fixed CEILING optimization for a divisor of form 2^k.
1529 * fixed bug 240 (emitting extra style warnings "using the lexical
1530 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1531 by Antonio Martinez)
1532 * fixed SXHASH, giving different results for NIL depending on type
1533 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1534 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1535 variable. (found by Rolf Wester)
1536 * SBCL does not ignore type declarations for special
1537 variables. (reported by rif on c.l.l 2003-03-05)
1538 * some bug fixes in contrib/sb-aclrepl/
1539 * fixed some bugs revealed by Paul Dietz' test suite:
1540 ** a bug in the CONS type specifier, whereby the CAR and CDR
1541 types got intertwined, has been fixed;
1542 ** the type system is now able to reason about the interaction
1543 between INTEGER and RATIO types more completely;
1544 ** APPEND, [N]REVERSE and NRECONC check that those their
1545 arguments, which must be proper lists, are really so;
1546 ** An array specialized to be unable to hold elements has been
1547 implemented, as required -- yes, really -- by ANSI;
1548 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1549 on malformed property lists;
1551 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1552 * incompatible packaging change: in line with Unix convention,
1553 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1554 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1555 * REQUIRE and PROVIDE are now optionally capable of doing something
1556 useful. See the documentation string for REQUIRE.
1557 * infrastructure for a managed SBCL contrib system: contributed
1558 modules in this release include:
1559 ** the ASDF system definition facility;
1560 ** an interface to the BSD Sockets API;
1561 ** an ACL-like convenience interface to the repl;
1562 (thanks to Kevin Rosenberg)
1563 ** an implementation of ROTATE-BYTE, with efficient implementation
1565 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1566 gives a closer approximation to the right answer than previously.
1567 (thanks to Raymond Toy)
1568 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1569 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1570 optional environment arguments, as required by ANSI.
1571 * fixed bugs in other functions taking environment objects, allowing
1572 calls with an explicit NIL environment argument to be compiled
1574 * fixed bug 228: primary return values from
1575 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1576 COMPILE or FUNCTION.
1577 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1578 :TYPE will no longer signal errors on innocuous objects.
1579 * fixed bug 231b: SETQ is better at respecting type declarations in
1580 the lexical environment.
1581 * fixed a bug in DEFCLASS: classes named by symbols with no or
1582 unprintable packages can now be defined.
1583 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1584 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1585 * fixed compiler failure related to checking types of functions.
1586 (reported by Robert E. Brown)
1587 * the compiler is now much more consistent in its error-checking
1588 treatment of bounding index arguments to sequence functions: in
1589 (SAFETY 3) code, errors will be signalled in almost all cases if
1590 invalid sequence bounding indices are passed to functions defined
1591 by ANSI to operate on sequences.
1592 * fixed a bug in the build procedure: documentation of SBCL-specific
1593 packages is now preserved and available in the final Lisp image.
1594 * lifted FDEFINITION lookup out of loops in the implementation of
1595 many list operations. (thanks to Robert E. Brown)
1596 * fixed a bug in the reader: the #n# reader macro now works for
1597 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1598 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1599 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1600 E. Brown and Tony Martinez respectively)
1601 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1602 SLOT-BOUNDP: in particular, it is now able to optimize them much
1603 better, and is now not vulnerable to having packages renamed.
1604 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1605 always called when a slot is not present in an instance. (thanks
1607 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1608 longer signal an error on unreading a character following EOF on
1609 the previous constituent stream. (thanks to Tony Martinez)
1610 * fixed some bugs revealed by Paul Dietz' test suite:
1611 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1612 not just nonnegative fixnums;
1613 ** the logical bit-array operators such as BIT-AND now accept an
1614 explicit NIL for their "opt-arg" argument (to indicate a
1615 freshly-consed result bit-array);
1616 ** ELT now signals an error on an invalid sequence index in safe
1618 ** the type system is now cleverer about negations of numeric
1619 types, and consequently understands the BIGNUM and RATIO types
1621 ** the type system is now cleverer about the interaction between
1622 INTEGER and RATIO types: while bugs still remain, many more
1623 cases are accurately computed;
1624 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1625 if it is in the last clause;
1626 ** CONSTANTLY now correctly returns a side-effect-free function in
1628 ** DECLARE is no longer treated as a special-operator; in
1629 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1630 * incremented fasl file version number due to the change in the
1631 DEFSTRUCT-SLOT-DESCRIPTION structure.
1633 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1634 * minor incompatible change: code processed by the "interpreter" or
1635 EVAL now has a compilation optimization policy of (DEBUG 2)
1636 (changed from (DEBUG 1)) to improve debuggability of interactive
1637 development, and to allow the use of the debug RETURN command in
1639 * an experimental implementation of the RETURN command for the
1640 debugger has been included. (thanks to Frederik Kuivinen)
1641 * fixed bug 62: constraints were not propagated into a loop.
1642 * fixed bug in embedded calls of SORT (reported and investigated by
1644 * fixed some bugs revealed by Paul Dietz' test suite:
1645 ** printing and reading of arrays with some dimensions having
1646 length 0 (thanks to Gerd Moellmann);
1647 ** BOA constructor with &AUX argument without a default value does
1648 not cause a type error;
1649 ** CONSTANTP now returns true for all self-evaluating objects.
1651 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1652 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1653 accessors that are related by inheritance, as specified in the
1654 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1655 Valtteri Vuorikoski)
1656 * The compiler is now able to inline functions that were defined in
1657 a complex lexical environment (e.g. inside a MACROLET).
1658 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1660 * Support for the upcoming FreeBSD-5.0 release has been included.
1661 (thanks to Dag-Erling Smorgrav)
1662 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1663 effect when it is not in a toplevel context.
1664 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1665 stand a better chance of being correct. (thanks to Gerd
1667 * fixed bug in COERCE, which now signals an error on coercing a
1668 rational to a bounded real type which excludes the expected
1670 * The compiler is now able to derive types more accurately from the
1671 COERCE and COMPILE functions.
1672 * fixed bug 223: functional binding is considered to be constant
1673 only for symbols in the CL package.
1674 * fixed bug 231: SETQ did not check the type of a variable being set
1675 (reported by Robert E. Brown)
1676 * A new optimization for MAKE-INSTANCE has been included, fixing
1677 various bugs (including relating to :ALLOCATION :CLASS slots and
1678 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1680 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1681 ** As required by ANSI, LOOP now disallows anonymous collection
1682 clauses such as COLLECT I in conjunction with aggregate boolean
1683 clauses such as THEREIS (= I 1);
1684 ** LOOP now signals an error when any variable is reused in the
1685 same loop (including the potentially useful construct analogous
1686 to WITH A = 1 WITH A = (1+ A);
1687 ** IT is only a special loop symbol within the first clause of a
1688 conditional loop clause;
1689 ** LOOP with a typed iteration variable over a hashtable now
1690 signals a type error iff it should.
1691 * fixed some other bugs revealed by Paul Dietz' test suite:
1692 ** FILE-STREAM now names the class previously known as FD-STREAM;
1693 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1694 argument) no longer signals an error;
1695 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1696 of the slot symbol, rather than using the current package
1697 ((:CONC-NAME "") continues to intern the slot's name in the
1699 * incremented fasl file version number, because of the incompatible
1700 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1701 of the new implementation of DEFINE-COMPILER-MACRO.
1703 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1704 * Support for building SBCL for MIPS platforms running in
1705 little-endian mode has now been checked in, and basic
1706 functionality on said platforms verified.
1707 * minor incompatible change: PCL now records the pathname of a file
1708 in which methods and the like are defined, rather than its
1710 * minor incompatible change: TRUENAME now considers the truename of
1711 a file naming a directory to be the pathname with :DIRECTORY
1712 component indicating that directory.
1713 * minor incompatible change: a NAMED clause in the extended form of
1714 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1715 reason for the previous behaviour is unclear.
1716 * more systematization and improvement of CLOS and MOP conformance
1717 in PCL (thanks to Gerd Moellman and Pierre Mai):
1718 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1720 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1721 primary methods with no specializers;
1722 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1724 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1725 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1726 FORWARD-REFERENCED-CLASSes; error reporting on
1727 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1729 ** SXHASH on CLOS instances now uses a slot internal to the
1730 instance to return different numbers on distinct instances,
1731 while preserving the same return value through invocations of
1733 ** DEFMETHOD signals errors when methods with longer incongruent
1734 lambda lists are added to generic functions;
1735 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1736 CLASS, as specified in AMOP;
1737 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1738 on the order returned by the primary method for classes of
1739 class STANDARD-CLASS;
1740 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1741 * fixed some bugs shown by Paul Dietz' test suite:
1742 ** DOLIST puts its body in TAGBODY;
1743 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1745 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1746 value producing form;
1747 ** if more variables are given to PROGV than values, extra
1748 variables are bound and made to have no value;
1749 ** NSUBSTITUTE on list arguments gets the right answer with
1751 ** ELT signals an error of type TYPE-ERROR when the index argument
1752 is not a valid sequence index;
1753 ** LOOP signals (at macroexpansion time) an error of type
1754 PROGRAM-ERROR when duplicate variable names are found;
1755 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1756 ** FUNCALL of special-operators now cause an error of type
1758 ** PSETQ now works as required in the presence of side-effecting
1759 symbol-macro places;
1760 ** NCONC accepts any object as its last argument;
1761 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1763 ** loop-for-as-package does not require a package to be explicitely
1765 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1767 * fixed bug 166: compiler preserves "there is a way to go"
1768 invariant when deleting code.
1769 * fixed bug 172: macro lambda lists with required arguments after
1770 &REST arguments now cause an error to be signalled. (thanks to
1772 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1773 bug. (thanks to Matthew Danish)
1774 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1776 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1777 arguments are lexically rebound. (thanks to Gerd Moellmann and
1779 * fixed bug 194: error messages are now more informative when there
1780 is no primary method applicable in a call to a generic
1781 function. (thanks to Gerd Moellmann)
1782 * fixed bug in command line argument checking (thanks to Julian
1784 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1785 (thanks to Matthew Danish)
1786 * incremented fasl file version number, because of the
1787 SXHASH-related changes in the layout of CLOS data structures
1789 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1790 * minor incompatible change: The runtime (the Unix executable named
1791 "sbcl") is now much pickier about the .core files it will load.
1792 Essentially it now requires .core files to descend from the same
1793 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1794 as the runtime does. (The intent is to prevent the crashes which
1795 can occur, and which can even be reported as mysterious failures,
1796 when people patch the sources or change the build parameters
1797 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1798 sbcl and .core files.)
1799 * fixed bug: VALUES-LIST is no longer optimized away.
1800 * fixed bug 142: The FFI conversion of C string values to Lisp
1801 string values no longer conses excessively. (thanks to Nathan
1802 Froyd porting Raymond Toy's fix to CMU CL)
1803 * began to systematize and improve MOP conformance in PCL (thanks to
1804 Nathan Froyd, Gerd Moellman and Pierre Mai):
1805 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1807 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1808 ** FINALIZE-INHERITANCE is now called on class finalization;
1809 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1810 argument precedence order.
1811 * fixed bug 202: The compiler no longer fails on functions whose
1812 derived types contradict their declared type.
1813 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1814 so it can be non-toplevel.
1815 * The fasl file version number has changed (because of the new
1816 implementation of DEFMACRO).
1817 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1818 safe code, that any length requirement by their type-specifier
1819 argument is valid. The exceptions to this are described in bug
1821 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1822 functions now check the types of their inputs as required by ANSI.
1823 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1824 introduced symbol is DECLAREd to be SPECIAL.
1825 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1826 * fixed bug: :COUNT argument to sequence functions may be negative
1827 * fixed bug: body of DO-SYMBOLS may contain declarations
1828 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1829 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1830 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1831 symbol macro only once
1832 * fixed printing of call frame when argument list is unavailable
1833 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1834 * compiler no longer signals WARNING on unknown keyword
1837 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1838 * A beta-quality port to the mips architecture running Linux,
1839 based on the old CMUCL backend, has been made. It has been tested
1840 on a big-endian kernel, and works sufficiently well to be able to
1841 rebuild itself; it has not been tested in little-endian mode.
1842 * fixed an inconsistency between gencgc.c and purify.c which made
1843 dumping/loading .core files unreliable
1844 * fixed bug 120a: The compiler now deals correctly with IFs where
1845 the consequent is the same as the alternative, instead of
1846 misderiving the return type. (thanks to Alexey Dejneka)
1847 * fixed bug 113: Logical pathnames are now dumpable (the logical
1848 host is resolved at load-time, throwing an error if it is not
1850 * fixed bug 174: FORMAT's error message is slightly clearer when a
1851 non-printing character is used in a format directive.
1852 * fixed several bugs in compiler checking of type declarations, i.e.
1853 violations of the Python "declarations are assertions" principle
1854 (thanks to Alexey Dejneka)
1855 * fixed several bugs in PCL's error checking (thanks to Gerd
1857 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1859 * fixed bug in compilation of functions as first class values
1860 (thanks to Antonio Martinez-Shotton)
1861 * The compiler's handling TYPE-ERRORs which it can prove will
1862 inevitably happen at runtime has been cleaned up and corrected
1863 in several ways. (thanks to Alexey Dejneka)
1864 * improved argument type checking for various basic arithmetic
1865 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1866 lavished on them in the past that they can be compiled in many
1867 ways in different special cases
1868 * fixed bug 181: compiler checks validity of user supplied type
1870 * cleaned up code flushing in optimization: Function calls which
1871 should signal errors for safety purposes (e.g. which ANSI says
1872 should signal errors when their arguments are of incorrect type)
1873 are no longer optimized away.
1874 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1875 * incremented fasl file version number, because changes in the
1876 implementation of sequence functions like COERCE caused
1877 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1878 in old inline expansions) to become undefined. (Actually these
1879 changes were later undone, so we might very well be binary
1880 compatible with 0.7.7 after all, but leaving the version number
1881 incremented seemed like the simplest and most conservative
1884 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1885 * An alpha-quality port to the parisc architecture running Linux,
1886 based on the old CMUCL backend, has been made. This, even more so
1887 than the other backends, should be considered still a work in
1888 progress; known problems include that the Linux kernel in 64-bit
1889 mode does not propagate the correct sigcontext structure to
1890 userspace, and consequently SBCL on a parisc64 kernel will not
1892 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1893 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1894 suppressed.) Also now that the compiler is looking at declarations
1895 in the environment, it checks optimization declarations as well,
1896 and suppresses inlining when (> DEBUG SPEED).
1897 * More fixes have been made to treatment of floating point exception
1898 treatment and other Unix signals. In particular, floating point
1899 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1900 * The detection and handling of control stack exhaustion (infinite
1901 or very deeply nested recursion) has changed. Stack exhaustion
1902 detection is now done by write-protecting pages at the OS level
1903 and applies at all optimization settings; when found, a
1904 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1905 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1906 be caught using IGNORE-ERRORS.
1907 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1908 correctly when transforming recursive LABELS functions to LETs.
1909 (thanks to Alexey Dejneka)
1910 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1911 that are names of constants or global variables.
1912 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1913 alien routines with docstrings.
1914 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1915 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1917 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1918 (thanks to Eric Marsden)
1919 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1920 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1921 * Bug fix: The SPARC backend can now compile functions involving
1922 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1923 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1924 to a functions expecting a PATHNAME-DESIGNATOR.
1925 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1926 lambda lists. (thanks to Alexey Dejneka)
1927 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1928 functions so that EXPORTing the name of the class doesn't cause
1929 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1930 to Antonio Martinez for reporting this)
1931 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1932 constant-folded without failing an assertion. (thanks to Einar
1933 Floystad Dorum for reporting this)
1934 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1935 types are now dealt with more correctly by the compiler.
1936 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1937 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1938 behaviour analogous to RENAME-FILE. This puts its behaviour more
1939 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1941 * The fasl file version number has changed again. (because of the
1942 bug fix involving the names of PCL MAKE-INSTANCE functions)
1944 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1945 * bug fix: Floating point exceptions are treated much more
1946 consistently on the x86/Linux and PPC/Linux platforms.
1947 * Array initialization with :INITIAL-ELEMENT is now much faster for
1948 cases when the compiler cannot open code the array creation, but
1949 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1950 array accesses have also seen a speed increase.
1951 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1952 specified. (This was at the root of some bad interactions between
1953 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1955 * bug fix: Internal error arguments for undefined functions are now
1956 computed correctly on the PPC/Linux platform.
1957 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1958 Raymond Toy's patch for CMU CL)
1959 * Support for the Solaris 9 operating environment has been included
1960 (thanks to Daniel Merritt)
1961 * A very ugly but hopefully complete draft of the missing FFI chapter
1962 of the manual has been created by reformatting the corresponding
1963 CMU CL manual chapter into (currently very ugly and incoherent)
1964 DocBook and bringing it up to date for SBCL behavior. Thus, the
1965 manual is now essentially complete, at least by my extreme
1966 once-and-only-once standards, whereby it's acceptable to refer to
1967 the doc strings of SB-EXT functions as the primary documentation.
1968 * The fasl file version number has changed again, due to cleanup of
1969 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1970 is no longer a static symbol.)
1972 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1973 * SBCL now builds with OpenMCL (version 0.12) as the
1974 cross-compilation host; also, more progress has been made toward
1975 bootstrapping under CLISP.
1976 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1978 * bug 158 fixed: The compiler can now deal with integer loop
1979 increments different from 1; fixing this turned out also to fix
1981 * bug 169 fixed: no more bogus warnings about using lexical bindings
1982 despite the presence of perfectly good SPECIAL declarations (thanks
1983 to David Lichteblau)
1984 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1985 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1986 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1988 * bug 180 fixed: Method combination specifications no longer ignore
1989 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1990 * bug fix: Structure type predicate functions now check their argument
1991 count as they should.
1992 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1993 correctly. (thanks to Pierre Mai)
1994 * minor incompatible change: The --noprogrammer option is deprecated
1995 in favor of the new --disable-debugger option, which is very similar.
1996 (The major difference is that it takes effect at a slightly different
1997 time at startup, causing handling of errors in --sysinit and
1998 --userinit files will be affected differently.) The
1999 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2000 been added to allow this functionality to be controlled from ordinary
2001 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2002 maintainers, who might want to run non-interactive scripts to
2003 build SBCL cores which will later be used interactively.)
2004 * minor incompatible change: The LOAD function no longer, when given
2005 a wild pathname to load, loads all files matching that pathname.
2006 Instead, an error of type FILE-ERROR is signalled.
2008 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2009 * bug 147 fixed: The compiler preserves its block link/count
2010 invariants more correctly now so that it doesn't crash. (thanks
2012 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2014 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2015 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2016 no-value-for-this-character as NIL (as specified by ANSI).
2017 * HOST-NAMESTRING on physical pathnames now returns a string that is
2018 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2019 (thanks to Christophe Rhodes)
2020 * The Alpha port handles icache flushing more correctly. (thanks to
2022 * More progress has been made toward bootstrapping under CLISP. (thanks
2023 to Christophe Rhodes)
2024 * The fasl file format has changed again, because dynamic loading
2025 on OpenBSD (which has non-ELF object files) motivated some cleanups
2026 in the way that foreign symbols are transformed and passed around.
2027 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2028 is no longer treated as whitespace by the reader, but instead as
2029 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2030 |A
\7fB|, instead of A as it used to.
2032 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2033 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2034 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2035 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2036 this as of 0.7.1.45, but was left out of the previous news section
2037 (thanks to Dan Barlow)
2038 * SBCL now runs on the Solaris operating system on SPARC architectures
2039 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2040 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2041 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2042 * SPARC backend cleanups, allowing builds of cores optimized for V8
2043 and V9 SPARCS, and also emission of code targeted to a particular
2044 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2046 * SBCL is closer to bootstrapping under CLISP, thanks to various
2047 fixes by Christophe Rhodes.
2048 * The fasl file format has changed again, to allow the compiler's
2049 INFO database to support symbol macros.
2050 * The user manual (in doc/) is formatted into HTML more nicely.
2051 (thanks to coreythomas)
2052 * The system is smarter about SUBTYPEP relationships, especially
2053 those involving NOT types (including types such as ATOM which are
2054 represented internally using NOT types). Thus SUBTYPEP is less
2055 likely to return (VALUES NIL NIL) in general, and in particular
2056 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2058 * The fasl file format has changed again, because the internal
2059 representation of types now includes a new slot to support the new
2060 SUBTYPEP-of-NOT-types logic.
2061 * (not a change in the main branch of SBCL, but a related prototype
2062 which can hopefully be merged into the main branch of SBCL in the
2063 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2064 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2065 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2066 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2067 dispatch macro character combinations. (thanks to Alexey Dejneka)
2068 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2069 that superficially look like logical namestrings correctly.
2070 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2072 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2073 default supplied-p) for &optional and &key arguments. (thanks to
2076 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2077 * incompatible change: The compiler is now less aggressive about
2078 tail call optimization, doing it only when (> SPACE DEBUG) or
2079 (> SPEED DEBUG). (This is an incompatible change because there are
2080 programs which relied on the old CMU-CL-style behavior to optimize
2081 away their unbounded recursion which will now die of stack overflow.)
2082 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2083 for non-GENCGC systems has been increased to 20M (since that
2084 seems much closer to the likely performance optimum for modern
2085 systems than the old 4M value was)
2086 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2087 and *DEBUG-PRINT-LEVEL*
2088 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2089 of CMU CL's support for SPARC, and various endianness and other
2090 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2091 * new syntactic sugar for the Unix command line: --load foo.bar is now
2092 an alternate notation for --eval '(load "foo.bar")'.
2094 ** The system now detects stack overflow and handles it gracefully,
2095 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2096 optimization settings. (This is a good thing in general, and
2097 its introduction in this version should be particularly timely
2098 for anyone whose code fails because of suppression of tail
2100 ** The system now hunts for the C variable "environ" in a more
2101 devious way, to avoid segfaults when the C library version differs
2102 between compile time and run time. (thanks to Christophe Rhodes)
2103 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2104 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2105 ** The compiler no longer issues bogus style warnings for undefined
2106 classes in the same source file as the DEFCLASSes which defined
2107 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2109 ** fixes in CONDITION class precedence list for undefined function
2110 errors (thanks to Alexei Dejneka)
2111 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2112 correctly. (thanks to Dan Barlow)
2113 ** portability fixes aiming at bootstrapping under CLISP (thanks
2114 to Dave McDonald and Christophe Rhodes)
2115 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2116 ** fixes in type translation and and type inference (thanks to
2118 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2119 ** various fixes in the new ports (thanks to Dan Barlow)
2120 * several changes related to debugging:
2121 ** suppression of tail recursion, as noted above
2122 ** stack overflow detection, as noted above
2123 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2124 is now the default. (For some time encapsulation has been more
2125 reliable than the breakpoint-based :ENCAPSULATE NIL
2126 implementation, at least on X86 systems; and I just noticed that
2127 encapsulation also seems closer to the spirit of the ANSI
2130 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2132 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2133 up properly again. (There was a packaging bug in 0.7.0 which
2134 left their definitions in SB-SYS::LOAD-FOREIGN and
2135 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2136 vital for most things which interface to C-level interfaces,
2137 like extensions working with sockets or databases or
2138 Perl-compatible regexes or whatever, and the need to fix
2139 this bug is the main reason that 0.7.1 was released so
2141 ** DEFGENERIC is now choosier about the methods it redefines, so that
2142 reLOADing a previously-LOADed file containing DEFGENERICs does
2143 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2144 cycle now works as it should. (thanks to Alexey Dejneka)
2145 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2146 fixed by Christophe Rhodes. (He actually submitted this patch
2147 months ago, and I delayed until after 0.7.0.)
2148 ** Bug 111 (internal compiler confusion about runtime checks on
2149 FUNCTION types) was fixed by Alexey Dejneka.
2150 * Some internal cleanups (getting rid of variables which aren't
2151 needed now that the byte interpreter is gone) caused the fasl
2152 file format number to change again.
2154 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2155 * major incompatible change: The default fasl file extension, i.e. the
2156 default extension for files produced by COMPILE-FILE, has changed
2157 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2159 ** There are many changes in the implementation of the compiler.
2160 SBCL is now essentially a compiler-only implementation of ANSI
2161 Common Lisp. EVAL still "interprets" a few special cases, but
2162 almost all the interesting cases are handled by creating
2163 a LAMBDA expression, calling COMPILE on it, then calling
2164 FUNCALL on the result.
2165 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2166 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2167 Since the code is newer, there might still be some new bugs
2168 (though not as many as before Martin Atzmueller's fixes:-). But
2169 the new code is substantially simpler and clearer, and hopefully
2170 any remaining bugs will be simpler, less fundamental, and more
2171 fixable then the bugs in the old code.
2172 ** The revised compiler is still a little unsteady on its feet.
2174 *** The debugging information it produces (particularly the names
2175 of FUNCTION objects) is sometimes much less useful than what
2176 the old compiler produced.
2177 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2178 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2179 within a MACROLET) has been temporarily weakened.
2180 ** There are new compiler optimizations for various functions:
2181 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2182 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2183 *** the math functions TRUNCATE, FLOOR, and CEILING
2184 *** the function-of-all-trades COERCE
2185 Mostly these should be transparent, but there's one
2186 potentially-annoying problem (bug 117): when the compiler
2187 inline-expands a function and does type analysis on the result,
2188 it can create control paths which have type mismatches, and
2189 when it can't prove that those control paths aren't taken,
2190 it will issue WARNINGs about the type mismatches. This is
2191 a particular problem in practice for the new sequence functions.
2192 It's not clear how this should be fixed, and for now, a
2193 workaround is given in the entry for 117 in the BUGS file.
2194 ** (Because of the interaction between the two previous items --
2195 occasional inlining problems and new inline expansions -- some
2196 of the new sequence function optimizations won't really kick in
2197 completely until debugging information, and then inlining, are
2198 straightened out in some future version.)
2199 * minor incompatible changes:
2200 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2201 DIRECTORY no longer implicitly promotes NIL slots of its
2202 pathname argument to :WILD. In particular, when you ask for the
2203 contents of a directory (which you used to be able to do without
2204 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2205 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2206 ** changes in behavior that ANSI explicitly defines to be
2207 implementation dependent:
2208 *** The new compiler-only implementation still conforms with ANSI,
2209 but acts a little different than before. Besides the obvious
2210 changes in performance tradeoffs (that the cost per form passed
2211 to EVAL has gone up, and the cost per form executed by EVAL
2212 has gone down), the behavior of the system changes a little
2213 because there are no longer any interpreted function objects.
2214 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2215 e.g. doing COMPILE on the output of interactive DEFUN is
2217 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2219 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2220 has changed from (:CL) to NIL.
2221 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2222 e.g. #\Newline and #\Tab, don't have names specified in the
2223 ANSI Common Lisp standard, is now based on their ASCII symbolic
2224 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2225 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2226 are no longer used for output.
2227 ** changes in internal implementation constants:
2228 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2229 4 million. (If your application spends a lot of time GCing and
2230 you have a lot of RAM, you might want to experiment with
2231 increasing it even more.)
2232 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2233 However, almost all old code should still continue to work without
2234 immediate update, as SB-C-CALL is now a (deprecated) nickname
2236 ** Old operator names in the style DEF-FOO are now deprecated in
2237 favor of new corresponding names DEFINE-FOO, for consistency with
2238 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2239 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2240 internal symbols, but a few supported extensions like
2241 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2242 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2243 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2244 etc. as you get deeper into recursive calls to the debugger
2245 command loop, instead of the old "5]", "5]]", "5]]]"
2246 sequence. (I was motivated to do this when squabbles between
2247 ILISP and SBCL left me very deeply nested in the debugger. In the
2248 short term, this change will probably provoke more ILISP/SBCL
2249 squabbles, but hopefully it will be an improvement in the long run.)
2250 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2251 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2252 function, since declaiming return types involving aliens is
2253 (1) annoyingly messy to do by hand and (2) vital to efficient
2254 compilation of code which calls such functions.
2255 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2256 longer reexported by the SB-EXT package. They're solely useful
2257 for alien code, so it seems more logical that you should get
2258 them from the SB-ALIEN package, not in SB-EXT.
2259 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2260 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2261 features. Instead, the code that they used to control is always
2262 built into the system.
2263 * many other bug fixes
2264 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2265 advantage of the new EVAL-WHEN stuff and to clean them up in
2266 general, and they are now more ANSI-compliant in a number of
2267 ways. Martin Atzmueller is responsible for a lot of this.
2268 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2270 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2271 *** correct ERROR type for various file operations
2272 *** some fixes for Lisp streams
2273 *** DEFMETHOD syntax checking
2274 *** changing old weird representation of debug information as
2275 strings (which, among their other deficiencies, don't transform
2276 correctly when you rename packages, and don't change their
2277 print representation when you change things like *PACKAGE*
2278 and *PRINT-LENGTH*) to symbols and lists of symbols
2279 He also made several improvements and fixed several bugs in DESCRIBE.
2280 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2282 *** misbehavior of WRITE-STRING/WRITE-LINE
2283 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2284 and several other LOOP problems as well
2285 *** DIRECTORY when similar filenames are present
2286 *** DEFGENERIC with :METHOD options
2287 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2288 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2289 *** argument ordering in FIND with :TEST option
2290 *** mishandled package designator argument in APROPOS-LIST
2291 *** various problems in the backquote readmacro
2292 *** a bug in APROPOS
2293 *** probably some others that I'm not describing very well here,
2294 since the CVS log documents them by reference to sbcl-devel
2295 messages, and the SourceForge archives aren't working well.:-(
2296 ** Dan Barlow improved the Alpha port (and is making progress on the
2297 PPC port, for those of you who think different).
2298 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2299 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2300 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2301 continues to work on the SPARC port (for those of you in a position
2302 to look down upon our little PC-compatible boxes from a great height).
2303 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2304 and uses that copy, rather than the current dynamic value, when
2305 it's trying to decide whether to truncate output. Thus e.g.
2306 (let ((*print-lines* 50))
2307 (pprint-logical-block (stream nil)
2309 (let ((*print-lines* 8))
2310 (print (aref possiblybigthings i) stream)))))
2311 should now truncate the logical block only at 50 lines, instead of
2312 often truncating it at 8 lines, as it did before.
2313 * The doc/cmucl/ directory, containing old CMU CL documentation
2314 from the time of the fork, is no longer part of the base system.
2315 SourceForge has shut down its anonymous FTP service, and with it
2316 my original plan for distributing the old CMU CL documentation
2317 there. For now, if you need these files you can download an old
2318 SBCL source release and extract them from it.
2319 * The fasl file version number changed again, for dozens of reasons,
2320 some of which are apparent above.
2322 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2323 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2324 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2325 patch, so that DISASSEMBLE output is much nicer.
2326 * The code in the SB-PROFILE package now seems reasonably stable.
2327 I still haven't decided what the final interface should look like
2328 (I'd like PROFILE to interact cleanly with TRACE, since both
2329 facilities use function encapsulation) but if you have a need
2330 for profiling now, you can probably use it successfully with
2331 the current CMU-CL-style interface.
2332 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2333 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2334 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2335 link now returns the dangling link itself, and for similar
2336 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2337 link itself. (In these cases the old code signalled an error and
2338 looped endlessly, respectively.) Thus, DIRECTORY now works even
2339 in the presence of dangling and cyclic symbolic links.
2340 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2341 is now a supported extension again, since the consensus on
2342 sbcl-devel was that it can be useful for ordinary development
2343 work, not just for debugging SBCL itself.
2344 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2345 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2346 that currently-defined functions might be redefined later with
2347 different return types.
2348 * Hash tables can be printed readably, as inspired by CMU CL code
2349 of Eric Marsden and SBCL code of Martin Atzmueller.
2350 * better error handling in CLOS method combination, thanks to
2351 Martin Atzmueller porting Pierre Mai's CMU CL patches
2352 * more overflow fixes for >16Mbyte I/O buffers
2353 * A bug in READ has been fixed, so that now a single Ctrl-D
2354 character suffices to cause end-of-file on character streams.
2355 In particular, now you only need one Ctrl-D at the command
2356 line (not two) to exit SBCL.
2357 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2359 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2360 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2361 does the right thing.
2362 * The implementation of some type tests, especially for CONDITION
2363 types, is now tidier and maybe faster, due to CMU CL code
2364 originally by Douglas Crosher, ported by Martin Atzmueller.
2365 * Some math functions have been fixed, and there are new
2366 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2367 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2368 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2369 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2370 likely to be built into sbcl-0.7.0.)
2371 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2372 CL, ported to SBCL by Martin Atzmueller)
2373 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2374 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2375 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2376 * There's a new slam.sh hack to shorten the edit/compile/debug
2377 cycle for low-level changes to SBCL itself, and a new
2378 :SB-AFTER-XC-CORE target feature to control the generation of
2379 the after-xc.core file needed by slam.sh.
2380 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2381 COMPILE-FILE is no longer supported, so that now every function
2382 gets an entry point, so that block compilation looks a little
2383 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2384 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2385 returns the number of bytes consed since the system started,
2386 rather than the number consed since the first time the function
2387 was called. (The new definition parallels ANSI functions like
2388 CL:GET-INTERNAL-RUN-TIME.)
2389 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2390 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2391 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2392 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2393 * Fasl file version numbers are now independent of the target CPU,
2394 since historically most system changes which required version
2395 number changes have affected all CPUs equally. Similarly,
2396 the byte fasl file version is now equal to the ordinary
2399 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2400 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2401 is no longer recognized. I apologize for this, because it was
2402 listed in SB-EXT as a supported extension, but I found that
2403 its existing behavior was poorly specified, as well as incorrectly
2404 specified, and it looked like too much of a mess to straighten it
2405 out. I have enough on my hands trying to get ANSI stuff to work..
2406 * many patches ported from CMU CL by Martin Atzmueller, with
2407 half a dozen bug fixes in pretty-printing and the debugger, and
2408 half a dozen others elsewhere
2409 * fixed bug 13: Floating point infinities are now supported again.
2410 They might still be a little bit flaky, but thanks to bug reports
2411 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2412 as flaky as they were.
2413 * The --noprogrammer command line option is now supported. (Its
2414 behavior is slightly different in detail from what the old man
2415 page claimed it would do, but it's still appropriate under the
2416 same circumstances that the man page talks about.)
2417 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2418 are now supported, and enabled by default. Thus, the compiler can
2419 handle many floating point and complex operations much less
2420 inefficiently. (Thus e.g. you can implement a complex FFT
2422 * The compiler now detects type mismatches between DECLAIM FTYPE
2423 and DEFUN better, and implements CHECK-TYPE more correctly, and
2424 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2425 thanks to patches from Martin Atzmueller.
2426 * various fixes to make the cross-compiler more portable to
2427 ANSI-conforming-but-different cross-compilation hosts (notably
2428 Lispworks for Windows, following bug reports from Arthur Lemmens)
2429 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2430 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2431 fixed thanks to Pierre Mai's CMU CL patches.
2432 * fixes to keep the system from overflowing internal counters when
2433 it tries to use i/o buffers larger than 16M bytes
2434 * fixed bug 45a: Various internal functions required to support
2435 complex special functions have been merged from CMU CL sources.
2436 (When I was first setting up SBCL, I misunderstood a compile-time
2437 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2438 * improved support for type intersection and union, fixing bug 12
2439 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2440 more obscure bugs as well
2441 * some steps toward byte-compiling non-performance-critical
2442 parts of the system, courtesy of patches from Martin Atzmueller
2443 * Christophe Rhodes has made some debian packages of sbcl at
2444 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2445 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2446 stable, but are nonetheless usable. When he's ready, I'd be happy
2447 to add them to the SourceForge "File Releases" section. (And if
2448 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2449 * new fasl file format version number (because of changes in
2450 internal representation of (OR ..) types to accommodate the new
2451 support for (AND ..) types, among other things)
2453 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2454 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2456 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2458 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2459 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2460 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2461 * DESCRIBE now works on structure objects again.
2462 * Most function call argument type mismatches are now handled as
2463 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2464 know whether the function will be redefined before the call is
2465 executed. (The compiler could flag local calls with full WARNINGs,
2466 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2467 it doesn't keep track of enough information to know whether calls
2468 are local in this sense.)
2469 * Compiler output is now more verbose, with messages truncated
2470 later than before. (There should be some supported way for users
2471 to override the default verbosity, but I haven't decided how to
2472 provide it yet, so this behavior is still controlled by the internal
2473 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2474 src/compiler/ir1util.lisp.)
2475 * Fasl file format version numbers have increased again, because
2476 support for the Gray streams extension changes the layout of the
2477 system's STREAM objects.
2478 * The Gray subclassable streams extension now works, thanks to a
2479 patch from Martin Atzmueller.
2480 * The full LOAD-FOREIGN extension (not just the primitive
2481 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2482 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2483 but like most other programs, it defaults to copying the Unix
2484 environment from the original process instead of starting the
2485 new process in an empty environment.
2486 * Extensions which manipulate the Unix environment now support
2487 an :ENVIRONMENT keyword option which doesn't smash case or
2488 do other bad things. The CMU-CL-style :ENV option is retained
2489 for porting convenience.
2490 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2491 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2493 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2495 * A patch from Martin Atzmueller seems to have solved the SIGINT
2496 problem, and as far as we know, signal-handling now works cleanly.
2497 (If you find any new bugs, please report them!)
2498 * The system no longer defaults Lisp source file names to types
2499 ".l", ".cl", or ".lsp", but only to ".lisp".
2500 * The compiler no longer uses special default file extensions for
2501 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2502 seems to expect a single default extension for all compiled code,
2503 and there's no compelling reason to try to stretch the standard
2504 to allow two different extensions.) Instead, byte-compiled files
2505 default to the same extension as native-compiled files.
2506 * Fasl file format version numbers have increased again, because
2507 a rearrangement of internal implementation packages made some
2508 dumped symbols in old fasl files unreadable in new cores.
2509 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2510 many fewer weird special cases.
2511 * Bug #17 (differing COMPILE-FILE behavior between logical and
2512 physical pathnames) has been fixed, and some related misbehavior too,
2513 thanks to a patch from Martin Atzmueller.
2514 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2515 by Tim Moore, ported to SBCL by Martin Atzmueller.
2516 * Martin Atzmueller fixed several filesystem-related problems,
2517 including bug #36, in part by porting CMU CL patches, which were
2518 written in part by Paul Werkowski.
2519 * More compiler warnings in src/runtime/ are gone, thanks to
2520 more patches from Martin Atzmueller.
2521 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2524 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2526 * DESCRIBE now works on CONDITION objects.
2527 * The debugger now handles errors which arise when trying to print
2528 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2530 * The build system now uses an additional file, customize-target-features.lisp,
2531 to allow local modifications to the target *FEATURES* list. (The point of
2532 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2533 debugging features enabled, without having to worry about propagating it
2534 into everyone's system when I do a "cvs update".) When no
2535 customize-target-features.lisp file exists, the target *FEATURES* list
2536 should be constructed the same way as before.
2537 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2538 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2539 warn about behavior which is undefined under the ANSI spec. Note: This
2540 is specified by ANSI, but it's not very popular with programmers.
2541 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2542 macro in the SBCL sources for an example of a workaround which you
2543 might use to make portable ANSI-standard code which does what you want.
2544 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2545 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2546 when it's used as a non-toplevel form. (This is required in order
2547 to implement the DEFCONSTANT-EQX macro.)
2548 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2549 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2550 funny things, due to bugs in the implementation of EVAL-WHEN
2551 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2552 non-toplevel DEFCONSTANTs are uncommon.)
2553 * The core file version number and fasl file version number have been
2554 incremented, because the old noncompliant DEFCONSTANT behavior involved
2555 calling functions which no longer exist, and because I also took the
2556 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2557 * fixed bug 1 (error handling before read-eval-print loop starts), and
2558 redid debugger restarts and related debugger commands somewhat while
2560 ** The QUIT debugger command is gone, since it did something
2561 rather different than the SB-EXT:QUIT command, and since it never
2562 worked properly outside the main toplevel read/eval/print loop.
2563 Invoking the new TOPLEVEL restart provides the same functionality.
2564 ** The GO debugger command is also gone, since you can just invoke
2565 the CONTINUE restart directly instead.
2566 ** The TOP debugger command is also gone, since it's redundant with the
2567 FRAME 0 command, and since it interfered with abbreviations for the
2569 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2571 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2572 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2573 some porting work by Martin Atzmueller)
2574 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2575 with a boatload of patches which he ported to SBCL. Now that those
2577 ** The system tries to make sure that its low-priority messages
2578 are prefixed by semicolons, to help people who like to use
2579 syntax highlighting in their ILISP buffer. (This patch
2580 was originally due to Raymond Toy.)
2581 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2582 patches originally written by Raymond Toy.
2583 ** The compiler understands coercion between single-value and
2584 multiple-VALUES type expressions better, getting rid of some very
2585 weird behavior, thanks to patches originally by Robert MacLachlan
2586 and Douglas Crosher.
2587 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2588 lambda lists, thanks to a patch originally by Pierre Mai.
2589 ** The system no longer bogusly warns about "abbreviated type
2591 ** The compiler gets less confused by inlining and RETURN-FROM,
2592 thanks to some patches originally by Tim Moore.
2593 ** The system no longer hangs when dumping circular lists to fasl
2594 files, thanks to a patch originally from Douglas Crosher.
2595 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2596 undefined function error.
2597 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2598 in structure slot initforms) for now, documented workaround instead:-|
2599 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2600 * fixed bug 5: added stubs for various Gray stream functions called
2601 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2602 installed, at least appropriate type errors are generated
2603 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2604 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2605 consistently in DEFMETHOD forms.
2606 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2607 it doesn't seem to affect SBCL after all
2608 * The C runtime system now builds with better optimization and many
2609 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2611 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2613 * The system is now under CVS at SourceForge (instead of the
2614 CVS repository on my home machine).
2615 * The new signal handling code has been tweaked to treat register
2616 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2617 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2618 downstream, e.g. in debug-int.lisp, has implicit dependencies
2619 on the unsignedness of integer representation of machine words,
2620 and that caused the system to bomb out with infinite regress
2621 when trying to recover from type errors involving signed values,
2622 e.g. (BUTLAST '(1 2 3) -1).)
2623 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2624 (This was one of the bugs Peter Van Eynde reported back in July.)
2625 * The system now uses code inspired by Colin Walters' O(N)
2626 implementation of MAP (from the cmucl-imp@cons.org mailing
2627 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2628 inline the MAP operation, and there is more than one
2629 sequence argument to the MAP call (so that it can't just
2630 do ETYPECASE once and for all based on the type of the
2631 single sequence argument). (The old non-inline implementation
2632 of the general M-argument sequence-of-length-N case required
2633 O(M*N*N) time when any of the sequence arguments were LISTs.)
2634 * The QUIT :UNIX-CODE keyword argument has been renamed to
2635 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2636 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2638 * Raymond Wiker's patches to port dynamic loading from Linux to
2639 FreeBSD have been added.
2640 * The BUGS file is now more nearly up to date, thanks in large part
2641 to Martin Atzmueller's review of it.
2642 * The debugger now flushes standard output streams before it begins
2643 its output ("debugger invoked" and so forth).
2644 * The core version number and fasl file version number have both
2645 been incremented, because of incompatible changes in the layout
2647 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2648 used to not be called for a saved Lisp image.)
2649 * Martin Atzmueller's version of a patch to fix a compiler crash,
2650 as posted on sbcl-devel 13 September 2000, has been installed.
2651 * Instead of installing Martin Atzmueller's patch for the
2652 compiler transform for SUBSEQ, I deleted the compiler transform,
2653 and transforms for some similar consing operations.
2654 * A bug in signal handling which kept TRACE from working on OpenBSD
2656 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2657 away by constant folding
2658 * The system now defines its address space constants in one place
2659 (in the Lisp sources), and propagates them automatically elsewhere
2660 (through GENESIS and the sbcl.h file). Therefore, patching the
2661 address map is less unnecessarily tedious and error-prone. The
2662 Lisp names of address space constants have also been systematized.
2663 * CVS tags like dollar-Header-dollar have been removed from
2664 the sources, because they have never saved me trouble and
2665 they've been source of trouble working with patches and other
2666 diff-related operations.
2667 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2668 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2670 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2672 * The system has been ported to OpenBSD.
2673 * The system now compiles with a simple "sh make.sh" on the systems
2674 that it's supported on. I.e., now you no longer need to tweak
2675 text in the target-features.lisp-expr and symlinks in src/runtime/
2676 by hand, the make.sh takes care of it for you.
2677 * The system is no longer so grossly inefficient when compiling code
2678 involving vectors implemented as general (not simple) vectors (VECTOR T),
2679 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2680 which dares to use the various sequence functions on non-simple
2681 vectors, takes less of a performance hit.
2682 * There is now a primitive type predicate VECTOR-T-P
2683 to test for the (VECTOR T) type, so that e.g.
2684 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2685 can now be compiled with some semblance of efficiency. (The old code
2686 turned the type declaration into a full call to %TYPEP at runtime!)
2687 * AREF on (VECTOR T) is still not fast, since it's still compiled
2688 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2689 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2690 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2691 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2692 without making full calls to SUBTYPEP at runtime.
2693 (Some analogous efficiency issues for non-simple vectors specialized to
2694 element types other than T, or for non-simple multidimensional arrays,
2695 have not been addressed. They could almost certainly be handled the
2696 same way if anyone is motivated to do so.)
2697 * The changes in array handling break binary compatibility, so
2698 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2699 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2700 instead of (VALUES T T).
2701 * By following the instructions that Dan Barlow posted to sbcl-devel
2702 on 2 July 2000, I was able to enable primitive dynamic object
2703 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2704 functionality is not implemented (since it calls ld to resolve
2705 library references automatically, requiring RUN-PROGRAM for its
2706 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2707 try to resolve library references) is now supported.
2708 * The system now flushes the standard output streams when it terminates,
2709 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2710 them at several other probably-convenient times, e.g. in each pass of
2711 the toplevel read-eval-print loop, and after evaluating a form given
2712 as an "--eval" command-line option. (These changes were motivated by a
2713 discussion of stream flushing issues on cmucl-imp in August 2000.)
2714 * The source transform for TYPEP of array types no longer assumes
2715 that an array whose element type is a not-yet-defined type
2716 is implemented as an array of T, but instead punts, so that the
2717 type will be interpreted at runtime.
2718 * There is now some support for cross-compiling in make.sh: each of
2719 the phases of make.sh has its own script. (This should be transparent
2720 to people doing ordinary, non-cross-compile builds.)
2721 * Since my laptop doesn't have hundreds of megabytes of memory like
2722 my desktop machine, I became more motivated to do some items on
2723 my to-do list in order to reduce the size of the system a little:
2724 ** Arrange for various needed-only-at-cold-init things to be
2725 uninterned after cold init. To support this, those things have
2726 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2727 symbols with such names are now uninterned after cold init).
2728 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2729 fasl files for cold load.
2730 ** Remove the old compiler structure pooling code (which used to
2731 be conditional on the target feature :SB-ALLOC) completely.
2732 ** Redo the representation of some data in cold init to be more compact.
2733 (I also looked into supporting byte compiled code at bootstrap time,
2734 which would probably reduce the size of the system a lot, but that
2735 looked too complicated, so I punted for now.)
2736 * The maximum signal nesting depth in the src/runtime/ support code has
2737 been reduced from 4096 to 256. (I don't know any reason for the very
2738 large old value. If the new smaller value turns out to break something,
2739 I'll probably just bump it back up.)
2740 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2742 * Many, many bugs reported by Peter Van Eynde have been added to
2743 the BUGS list; some have even been fixed.
2744 * While enabling dynamic object file loading, I tried to make the
2745 code easier to understand, renaming various functions and variables
2746 with less ambiguous names, and changing some function calling
2747 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2748 * While trying to figure out how to do the OpenBSD port, I tried to
2749 clean up some of the code in src/runtime/. In particular, I dropped
2750 support for non-POSIX signal handling, added various comments,
2751 tweaked the code to reduce the number of compilation warnings, and
2752 renamed some files to increase consistency.
2753 * To support the new automatic configuration functionality in make.sh,
2754 the source file target-features.lisp-expr has been replaced with the
2755 source file base-target-features.lisp-expr and the machine-generated
2756 file local-target-features.lisp-expr.
2757 * fixed a stupid quoting error in make.sh so that using CMU CL
2758 "lisp -batch" as cross-compilation host works again
2760 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2762 * DESCRIBE no longer tries to call itself recursively to describe
2763 bound/fbound values, so that it no longer fails on symbols which are
2764 bound to themselves (like keywords, T, and NIL).
2765 * DESCRIBE now works on generic functions.
2766 * The printer now prints less-screwed-up representations of closures
2767 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2768 * A private symbol is used instead of the :EMPTY keyword previously
2769 used to mark empty slots in hash tables. Thus
2770 (DEFVAR *HT* (MAKE-HASH-TABLE))
2771 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2772 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2773 now does what ANSI says that it should. (You can still get
2774 similar noncompliant behavior if bang on the hash table
2775 implementation with all the symbols you get back from
2776 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2777 This breaks binary compatibility, since tests for equality to
2778 :EMPTY are wired into things like the macroexpansion of
2779 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2781 * There's now a minimal placeholder implementation for CL:STEP,
2782 as required by ANSI.
2783 * An obscure bug in the interaction of the normal compiler, the byte
2784 compiler, inlining, and structure predicates has been patched
2785 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2786 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2788 * Missing ordinary arguments in a macro call are now detected even
2789 when the macro lambda list contains &KEY or &REST.
2790 * The debugger no longer complains about encountering the top of the
2791 stack when you type "FRAME 0" to explicitly instruct it to go to
2792 the top of the stack. And it now prints the frame you request even
2793 if it's the current frame (instead of saying "You are here.").
2794 * As specified by ANSI, the system now always prints keywords
2795 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2797 * The default initial SIZE of HASH-TABLEs is now smaller.
2798 * Type information from CLOS class dispatch is now propagated
2799 into DEFMETHOD bodies, so that e.g.
2800 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2802 is now basically equivalent to
2803 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2804 (DECLARE (TYPE SINGLE-FLOAT X))
2806 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2807 operation, without having to do run-time type dispatch.
2808 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2809 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2810 * Since I don't seem to be making big file reorganizations very often
2811 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2812 is overflowing my ability to conveniently back them up), I've finally
2813 checked the system into CVS. (The CVS repository is on my home system,
2814 not at SourceForge -- putting it on SourceForge might come later.)
2815 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2816 high-level GC-NOTIFY-FOO functions send their output. (There's
2817 still very little control of where low-level verbose GC functions
2818 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2819 less than it used to -- the GC-NOTIFY-FOO functions are now under
2820 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2821 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2822 in only one place in the source code, and propagates it automatically
2823 everywhere that it's needed. Thus e.g. when I bump the version from
2824 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2825 * The C source files now include boilerplate legalese and documentation
2826 at the head of each file (just as the Lisp source files already did).
2827 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2828 to his page will be replaced with a link to his new CLiki service.
2830 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2832 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2833 * The build process now looks for GNU make under the default name "gmake",
2834 instead of "make" as it used to. If GNU make is not available as "gmake"
2835 on your system, you can change this default behavior by setting the
2836 GNUMAKE environment variable.
2837 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2838 can build without error under CMU CL.
2840 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2842 * There is now a partial SBCL user manual (with some new text and some
2843 text cribbed from the CMU CL manual).
2844 * The beginnings of a profiler have been added (starting with the
2845 CMU CL profiler and simplifying and cleaning up). Eventually the
2846 main interface should be through the TRACE macro, but for now,
2847 it's still accessed through vaguely CMU-CL-style functions and macros
2848 exported from the package SB-PROFILE.
2849 * Some problems left over from porting CMU CL to the new
2850 cross-compilation bootstrap process have been cleaned up:
2851 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2852 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2853 part when getting rid of the compiler *BACKEND* stuff.)
2854 ** The value of *NULL-TYPE* was screwed up, because it was
2855 being initialized before the type system knew the final
2856 definition of the 'NULL type. This screwed up several key
2857 optimizations in the compiler, causing inefficiency in all sorts
2858 of places. (I found it because I wanted to understand why
2859 GET-INTERNAL-RUN-TIME was consing.)
2860 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2861 PROCLAIM FTYPE information. Unfortunately this broke binary
2862 compatibility again, since now the forms output by DEFGENERIC
2863 to refer to functions which didn't exist in 0.6.3.
2864 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2865 can use the new (as of 0.6.3) transform for SOME into MAP into
2867 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2868 values to (INTEGER 0 1000000), so that the time code will no longer
2869 occasionally get blown up by Linux returning 1000000 microseconds
2870 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2871 its output conform to the ANSI spec. (Alas, this makes its output
2872 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2873 * A full call to MAP NIL with a single sequence argument no longer conses.
2874 * fixes to problems pointed out by Martin Atzmueller:
2875 * The manual page no longer talks about multiprocessing as though
2876 it were currently supported.
2877 * The ILISP support patches have been removed from the distribution,
2878 because as of version 5.10.1, ILISP now supports SBCL without us
2879 having to maintain patches.
2880 * added a modified version of Raymond Toy's recent CMU CL patch for
2881 EQUALP comparison of HASH-TABLE
2883 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2885 * The system still can't cross-compile itself with
2886 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2887 entails), but at least it can compile more of itself that way
2888 than it used to be able to, and various buglets which were uncovered
2889 by trying to cross-compile itself that way have now been fixed.
2890 * This release breaks binary compatibility again. This time
2891 at least I've incremented the FASL file format version to 2, so that the
2892 problem can be detected reliably instead of just causing weird errors.
2893 * various new style warnings:
2894 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2895 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2896 ** using the lexical binding of a variable named in the *FOO* style
2897 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2898 as specified by ANSI.
2899 * *RANDOM-STATE* is no longer automatically initialized from
2900 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2901 default behavior of the system is to repeat its behavior every time
2902 it's run. If you'd like to change this behavior, you can always
2903 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2904 old convention there was no comparably easy way to get the system to
2905 repeat its behavior every time it was run.
2906 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2907 FTYPE declarations has been removed, in favor of their ANSI
2908 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2909 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2910 the types of their sequence arguments can be determined at compile time.
2911 This is done through a new open code expansion for MAP which eliminates
2912 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2913 argument types can be determined at compile time.
2914 * The optimizer now transforms COERCE into an identity operation when it
2915 can prove that the coerced object is already of the correct type. (This
2916 can be a win for machine generated code, including the output of other
2917 optimization transforms, such as the MAP transform above.)
2918 * Credit information has been moved from source file headers into CREDITS.
2919 * Source file headers have been made more standard.
2920 * The CASE macro now compiles without complaining even when it has
2923 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2925 * (Note that the way that the PCL macroexpansions were rewritten
2926 to accommodate the change in DEFGENERIC below breaks binary
2927 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2928 not run under sbcl-0.6.2. Once we get out of alpha releases,
2929 i.e. hit release 1.0.0, we'll probably try to maintain binary
2930 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2931 and sbcl-1.4.4. Until then, however, it might be fairly common
2932 for maintenance releases to break binary compatibility.)
2933 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2935 * The reporting of unhandled signals has been changed to print some
2936 explanatory text as well as the report form. (Previously only
2937 the report form was printed.)
2938 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2939 it defines, so that the compiler no longer issues undefined function
2940 warnings for compiled-but-not-yet-loaded generic functions.
2941 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2942 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2943 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2944 section of the ANSI spec are supported.
2945 * The "" nickname for the "KEYWORD" package has been removed.
2946 The reader still handles symbol tokens which begin with a package marker
2947 as keywords, but it doesn't expose its mechanism for doing so in the
2948 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2949 * The system now issues STYLE-WARNINGs for contradictory TYPE
2950 proclamations. (Warnings for contradictory FTYPE proclamations would
2951 be nice too, but those can't be done usefully unless the type system
2952 is made smarter about FUNCTION types.)
2953 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2954 systematized, so that "*target-*.lisp is supposed to exist only on the
2955 target and imply that there's a related file which exists on the
2956 host, and *host-*.lisp is supposed to exist only on the host and imply
2957 that there's a related file which exists on the target. This involves a
2958 lot of renaming. Hopefully the acute confusion caused by the renaming
2959 will be justified by the reduction in chronic confusion..
2960 ** runtime-type.lisp -> early-target-type.lisp
2961 ** target-type.lisp -> late-target-type.lisp
2962 ** early-host-format.lisp -> early-format.lisp
2963 ** late-host-format.lisp -> late-format.lisp
2964 ** host-error.lisp -> misc-error.lisp
2965 ** early-error.lisp -> early-target-error.lisp
2966 ** late-error.lisp -> late-target-error.lisp
2967 ** host-defboot.lisp -> early-defboot.lisp
2968 ** code/misc.lisp -> code/target-misc.lisp
2969 ** code/host-misc.lisp -> code/misc.lisp
2970 ** code/numbers.lisp -> code/target-numbers.lisp
2971 ** code/early-numbers.lisp -> numbers.lisp
2972 ** early-host-type.lisp -> early-type.lisp
2973 ** late-host-type.lisp -> late-type.lisp
2974 ** host-typep.lisp -> typep.lisp
2975 ** load.lisp -> target-load.lisp
2976 ** host-load.lisp -> load.lisp
2977 ** host-disassem.lisp -> disassem.lisp
2978 ** host-insts.lisp -> insts.lisp
2979 ** byte-comp.lisp -> target-byte-comp.lisp
2980 ** host-byte-comp.lisp -> byte-comp.lisp
2981 ** host-signal.lisp -> signal.lisp
2982 ** host-defstruct.lisp -> defstruct.lisp
2983 ** late-target-type.lisp -> deftypes-for-target.lisp
2984 Furthermore, several other previously target-only files foo.lisp (e.g.
2985 hash-table.lisp and random.lisp) have been split into a target-and-host
2986 foo.lisp file and a target-only target-foo.lisp file, with their key type
2987 definitions in the target-and-host part, so that the cross-compiler will
2988 know more about target types.
2989 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2990 gone away. In their place are various *BACKEND-FOO* variables
2991 corresponding to the slots of the old structure.
2992 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2993 into the target SBCL has been fixed.
2994 * The chill.lisp system for loading cold code into a running SBCL
2996 * Support for the CMU CL "scavenger hook" extension has been removed.
2997 (It was undocumented and unused in the CMU CL sources that SBCL was
2998 derived from, and stale in sbcl-0.6.1.)
2999 * Various errors in the cross-compiler type system were detected
3000 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3001 (enabling various consistency checks). Many of them were fixed,
3002 but some hard problems remain, so the compiler is back to
3003 running without *TYPE-SYSTEM-INITIALIZED* for now.
3004 * As part of the cross-compiler type system cleanup, I implemented
3005 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3006 * I have started adding UNCROSS calls throughout the type system
3007 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3008 on cross-compiler input files will be able to go away, and various
3010 * CONSTANTP now returns true for quoted forms (as explicitly required
3013 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3015 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3016 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3017 (MAKE-LIST -1). In the longer term, it ought to become true
3018 that declarations are assertions even at SAFETY 1. For now, it's not
3019 quite true even at SAFETY 3, but it's at least more nearly true..
3020 (Note that this change seems to increases the size of the system by
3021 O(5%) and to decrease the speed of the compiler by 20% or more.)
3022 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3023 for the problem of printing dozens of lines of distracting information
3024 about low-level system machinery as part of the top stack frame
3025 on entry to the debugger when an undefined function was called.
3026 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3027 is not reset to COMMON-LISP-USER.
3028 * Compilation of stuff related to dyncount.lisp has been made conditional
3029 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3030 smaller. The various dyncount-related symbols have been moved into
3031 a new "SB-DYNCOUNT" package.
3032 * tty-inspect.lisp has been renamed to inspect.lisp.
3033 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3034 feature has gone away. (When we eventually port to other flavors of
3035 libc and/or Unix, we'll try to make the differences between flavors
3036 invisible at the user level.)
3037 * Various other *FEATURES* tags, and/or their associated conditionals,
3038 have been removed if obsolescent, or given better documentation, or
3039 sometimes given more-mnemonic names.
3041 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3043 * tidied up "make.sh" script
3044 * tidied up system directory structure
3045 * better "clean.sh" behavior
3046 * added doc/FOR-CMUCL-DEVELOPERS
3047 * many many small tweaks to output format, e.g. removing possibly-confusing
3048 trailing #\. character in DESCRIBE-INSTANCE
3049 * (EQUALP #\A 'A) no longer signals an error.
3050 * new hashing code, including EQUALP hashing
3051 * tidied up Lisp initialization and toplevel
3052 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3053 * command line argument processing
3054 * added POSIX-GETENV function to deal with Unix-ish environment variables
3055 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3056 terminating SBCL on EOF
3057 * non-verbose GC by default
3058 * There is no more "sbcl" shell script; the sbcl file is now the C
3059 runtime executable (just like CMU CL).
3060 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3062 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3063 symbol and package information as Lisp native symbol and package objects
3064 instead of strings naming symbols and strings naming packages. This way,
3065 whenever packages are renamed (as in warm init), debug information is
3066 transformed along with everything else.
3067 * tweaked the optimization policy declarations which control the building
3068 of SBCL itself. Now, among other things, the system no longer saves
3069 source location debugging information. (This helps two problems at once
3070 by reducing SBCL size and by keeping SBCL from trying to look for its
3071 sources -- which may not exist -- when reporting errors.)
3072 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3073 debugging and testing purposes
3074 * cleaned up printing, making the printer call PRINT-OBJECT for
3075 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3076 methods, giving nearly-ANSI behavior
3077 * converted almost all special variables to use *FOO* naming convention
3078 * deleted PARSE-TIME functionality, since it can be done portably
3079 * moved some files out of cold init into warm init
3080 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3082 * regularized formatting of source files
3083 * added an install.sh script
3084 * fixed ridiculous memory usage of cross-compiler by making
3085 compiler/alloc.lisp not try to do pooling unless it can hook
3086 itself into the GC of the cross-compilation host. Now the system
3087 builds nicely on my old laptop.
3088 * added :SB-ALLOC in target-features.lisp-expr
3089 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3090 was not implemented)
3091 * re-did condition handling and note reporting in the compiler. Notes
3092 are no longer handled by signalling conditions. Style warnings
3093 and warnings are handled more correctly and reported in such a way
3094 that it's easy to find one or the other in your output (so that you
3095 can e.g. figure out which of many problems caused COMPILE-FILE to
3097 * changed the severity of several compiler warnings from full WARNING
3098 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3099 compiler note reporting so that it doesn't use the condition system
3100 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3101 COMPILE-FILE command)
3102 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3103 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3104 no longer supported, since their implementation was deeply intertwingled
3105 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3106 * removed bogus "support" for compiler macros named (SETF FOO), and
3107 removed the compiler macro for SETF INFO (but only after making a fool
3108 of myself on the cmucl-imp mailing list by posting a bogus patch for
3109 DEFINE-COMPILER-MACRO..)
3110 * Compiled files containing forms which have side effects on the Lisp
3111 reader (such as DEFPACKAGE forms) are now handled more correctly.
3112 (Compiler queuing of top level lambdas has been suppressed by setting
3113 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3114 * deleted various currently-unused source files, e.g. gengc.lisp. They
3115 may be added back at some point e.g. when porting to other architectures,
3116 but until they are it's distracting to distribute them and to try to
3118 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3119 there were so many of them they're just distractions, and UNCROSS is
3120 known to be able to handle the current sources
3121 * moved PROFILE functionality into TRACE, so that it will be clear
3122 how the wrapping and unwrapping of functions when you profile them
3123 interacts with the wrapping and unwrapping of functions when you
3124 trace them. (Actually, the functionality isn't there yet, but at least
3125 the interface specification is there. Hopefully, the functionality will
3126 arrive with some future maintenance release.)
3127 * removed host-oops.lisp
3128 * changed signature of QUIT function to allow UNIX-CODE argument
3129 * fixed READ-SEQUENCE bug
3130 * tweaked verbose GC output so that it looks more like the progress
3131 output that ANSI specifies for functions like LOAD
3132 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3133 * added <http://sbcl.sourceforge.com> to the banner information printed by