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 * fixed some bugs related to Unicode integration:
31 ** the restarts for recovering from input and output encoding
32 errors only appear when there is in fact such an error to
34 * increased the maximimum compact environment size to allow
35 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
36 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
37 * optimization: type testing for non-vector arrays should be faster.
38 * fixed some bugs revealed by Paul Dietz' test suite:
39 ** MISC.549 and similar: late transformation of unsafe type
40 assertions into derived types caused unexpected code
42 ** SCALE-FLOAT type deriver is less wrong.
43 ** type derivers for EXP, LOG and similar functions compute result
44 types for complex arguments better.
45 ** (MISC.563) CONJUGATE type deriver works for very restricted
47 ** out-of-line type testers for character strings are available.
48 ** EQUAL compiler transform understands specialness of objects
50 ** accessing double-floats stored on the stack now works on x86-64.
51 ** debugger internals could sometimes create invalid lispobjs,
52 resulting in GC crashes.
54 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
55 * incompatible change: thread support for non-NPTL systems has
56 been removed - locking is buggy and unreliable. A threaded
57 SBCL build will now warn at startup and refuse to create
58 new threads, unless futex support is detected in the kernel
59 * incompatible change: the top level REPL now has only an ABORT
60 restart associated with it, not TOPLEVEL and ABORT as it used to.
61 TOP and TOPLEVEL are now available as debugger commands for
62 returning to the top level.
63 * incompatible change: forms evaluated in the REPL now use the
64 global optimization policy.
65 * incompatible change: user- and system-initialization files are
66 no longer processed with LOAD, but by READ and EVAL; hence the
67 global optimization policy, startup package, readtable, etc,
69 * The .fasl file format number has been incremented because of
70 various incompatible changes.
71 * internal entry point details and argument counts no longer appear
72 in backtraces unless explicitly requested by setting
73 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
74 * built-in and standard functions no longer have names like "top
75 level local call to FOO".
76 * fixed bug 32: functions defined in non-null lexical environments
77 now have more legible printed representation
78 * fixed bug 33: functions defined in non-null lexical environemnts
79 are now more amenable to inspection by INSPECT.
80 * workaround for bug 354: XEPs no longer appear in backtraces unless
82 * fixed bug: receiving the signal which results from attempting to
83 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
84 SIGBUS on other *BSDs) on architectures where the C stack is also
85 the Lisp stack (x86 and x86-64) from foreign code no longer leads
86 to debugger confusion or wild instability. (reported by Cheuksan
88 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
89 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
90 (reported by Lutz Euler)
91 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
92 compiled code. (reported by Paul Dietz)
93 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
94 specialized parameter to a method to something that is not TYPEP
95 the specializer is now possible.
96 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
97 face of package deletion.
98 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
99 pathname. (reported by David Tolpin on c.l.l)
100 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
101 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
102 STRING-STREAM and FILE-STREAM, respectively.
103 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
104 than previously; in addition, its test suite is now run on build.
105 (thanks to Luke Gorrie)
106 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
107 * optimization: added a immediate representation for single-floats
109 * fixed some bugs revealed by Paul Dietz' test suite:
110 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
111 ** attempting to create a package with a colliding nickname causes
112 correctable errors to be signalled.
113 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
114 ** several x86-64 backend bugs related to sign-extension of immediate
117 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
118 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
119 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
121 * fixed loading of multiply forward-referenced layouts.
122 (thanks to Cheuksan Wang)
123 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
124 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
125 (see COMPILE-FILE documentation for details of :PRINT option)
126 * fixed bugs 19 and 317: fixed-format floating point printing is
127 more accurate. This also fixes a bug reported by Adam Warner
128 related to the ~@F format directive.
129 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
131 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
132 dispatch table if the from-char is a dispatch macro character.
133 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
134 by the wordsize no longer ignore the last word. (reported by Lutz
136 * fixed bug in type checking of dynamic-extent variables. (reported
138 * optimization: sequence traversal functions use their freedom to
139 coerce function designators to functions.
140 * optimization: code with many calls to CLOS methods specialized on
141 CLOS classes has had redundant type checks removed.
142 * optimization: added declarations to speed up operations that access
143 the internal character database (for example STRING-UPCASE)
144 * optimization: comparison operations between floats and sufficiently small
145 fixnums no longer create extra rationals
146 * fixed some bugs related to Unicode integration:
147 ** portions of multibyte characters at the end of buffers for
148 character-based file input are correctly transferred to the
149 start of the buffer at the next read.
150 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
151 passing it through to OPEN.
152 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
153 argument given to internal calls to OPEN.
154 ** the built-in comment readers (introduced by character sequences
155 ";" and "#|") are more forgiving to encoding errors; they will
156 STYLE-WARN and then attempt to resync the stream at a character
157 boundary. (thanks to Teemu Kalvas)
158 * fixed some bugs in the x86-64 port:
159 ** Negative short int return values from c-calls are sign-extended
161 ** The stack is aligned to 16-bytes for c-calls, as required by
162 the ABI. (thanks to Cheuksan Wang)
163 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
164 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
165 for enums. (thanks to Vincent Arkesteijn)
166 ** Multiple small optimizations and bugfixes for floating point
168 * fixed some bugs revealed by Paul Dietz' test suite:
169 ** Space, Tab, Linefeed, Return and Page have the invalid
170 secondary constituent character trait.
171 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
173 ** WITH-INPUT-FROM-STRING should only update the index place on
175 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
177 ** Bit-array operations (BIT-AND and similar) worked incorrectly
178 with one-dimensional arrays with fill pointers.
179 ** TYPE-OF failed on a complex with an integer realpart and a
181 ** compiler failure during type inference for the code of form
182 (IF (EQL X (THE ...)) ...) (MISC.535).
184 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
185 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
186 platform. The port supports 61-bit fixnums, large memory spaces
187 and reloading shared object files.
188 * enhancement: saving cores with foreign code loaded is now
189 supported on ppc/Darwin in addition to the previously supported
191 * enhancement: the statistical profiler now walks deeper into the
192 call stack for more meaningful call-graphs and accrued time
193 reports (x86/x86-64 only). It also now reports time spent in
195 * enhancement: it is now possible to trace most individual methods
196 of a generic function in addition to tracing the generic function
198 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
200 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
201 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
202 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
203 produces an error. (thanks to Vincent Arkesteijn)
204 * bug fix: NAMESTRING on pathnames with :WILD components in their
205 directories works correctly. (thanks to Artem V. Andreev)
206 * fixed bug 125: compiler preserves identity of closures. (reported
208 * bug fixed: functions with &REST arguments sometimes failed with
209 "Undefined function" when compiled with (DEBUG 3). (reported by
211 * bug fix: overflow during compiling of setting element of a bit
212 vector with constant index and value. (reported by Timmy Douglas)
213 * build fix: fixed the dependence on *LOAD-PATHNAME* and
214 *COMPILE-FILE-PATHNAME* being absolute pathnames.
215 * on x86 compiler partially supports stack allocation of dynamic-extent
217 * GO and RETURN-FROM do not check the extent of their exit points
218 when compiled with SAFETY 0.
219 * fixed some bugs related to Unicode integration:
220 ** encoding and decoding errors are now much more robustly
221 handled; it should now be possible to recover even from invalid
222 input or output to the terminal. (thanks to Teemu Kalvas)
223 ** provided a first cut at implementing STRING-TO-OCTETS and
224 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
225 ** altered the SB-MD5 contributed module slightly, changing the
226 interface just enough for it to be supportable for builds where
227 lisp characters are not eight bits.
228 * fixed some bugs revealed by Paul Dietz' test suite:
229 ** the FORMATTER-generated functions for ~V[ conditionals require
230 the correct number of arguments.
231 ** READ-FROM-STRING returns the mandated second value when applied
232 to displaced strings.
233 ** the #\Rubout and #\Backspace characters are treated as invalid
234 constituent characters by the tokenizer.
236 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
237 * new feature: reloading changed shared object files with
238 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
239 * new feature: references to foreign variables and functions
240 can now be compiled and loaded before the corresponding shared
241 object file is loaded, as long as the foreign definitions are
242 available at runtime.
243 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
244 supported. (thanks to Dan Debertin)
245 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
246 just keywords. (thanks to Vincent Arkesteijn)
247 * fixed bug #331: structure-class instances corresponding to
248 DEFSTRUCT forms are now created eagerly.
249 * fixed bug #345: backtraces from calls to undefined functions work
250 on x86 as well. Related bug #61 is now also partially fixed on x86
251 (backtraces from throws to unknown catch tags.)
252 * bug fix: lambda-list parsing is now stricter vrt. order and number
253 of lambda-list keywords.
254 * bug fix: as specified by AMOP, an error is signalled if a
255 class-option appears multiple times in a DEFCLASS form. (reported
257 * bug fix: RANDOM can be compiled when the compiler derives the type
258 of its numeric argument as a disjoint set of small integers.
259 (reported by Paul Dietz)
260 * bug fix: starting a core saved with shared objects loaded when
261 those objects are not available no longer causes threaded SBCL to
262 hang. (reported by Sean Ross)
263 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
265 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
266 argument, which must always be NIL. (reported by Kalle Niemitalo)
267 * bug fix: printing 1.0d+23 no longer results in an error.
268 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
269 * bug fix: structure slot setters preserve evaluation order. (thanks
271 * fixed some bugs related to Unicode integration:
272 ** RUN-PROGRAM can allow its child to take input from a Lisp
273 stream. (reported by Stefan Scholl)
274 ** ASDF-INSTALL successfully downloads .tar.gz files in all
275 locales. (reported by Ken Causey)
276 * fixed some bugs revealed by Paul Dietz' test suite:
277 ** INCF, DECF and REMF evaluate their place form as specified in
279 ** FORMATTER expands ~{ ~} iteration directives with V or #
280 parameters correctly.
281 ** FORMATTER deals with the ~@[ ~] conditional directive where the
282 consequent uses no arguments correctly.
283 ** the system has a partial understanding of the (COMPLEX RATIO)
285 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
288 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
289 * new feature: a build-time option (controlled by the :SB-UNICODE
290 keyword feature, enabled by default) for building the system with
291 support for the entire 21-bit character space defined by the
293 * new feature: the system now has rudimentary external-format
294 support; the primary user-visible change at this time is that
295 characters with the high bit set (such as lower-case-e-acute) will
296 print correctly to a terminal in a UTF-8 environment.
297 * minor incompatible change: BASE-CHAR no longer names a class;
298 however, CHARACTER continues to do so, as required by ANSI.
299 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
300 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
301 has been moved to the SB-EXT package (temporarily re-exported from
303 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
304 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
306 * fixed bug #351: better error-handlind and reporting for malformed
308 * fixed bug #350: bignum-printing is now more memory-efficient,
309 allowing printing of very large bignums, eg. (expt 2 10000000).
310 (reported by Bruno Haible)
311 * fixed bug #302: better primitive-type selection for intersection
313 * fixed bug #308: non-graphic characters now all have names, as
314 required. (reported by Bruno Haible)
315 * bug fix: redefining a class with different superclasses now correctly
316 removes it from the direct-subclasses of its previous superclasses.
317 (reported by David Morse)
318 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
319 new value now works. (reported by Bruno Haible)
320 * bug fix: correct canonicalization of multiple non-standard slot
321 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
322 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
323 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
324 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
325 now exists, an signals an error.
326 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
327 during macroexpansion for non-iteration variables as well. (reported
328 by Bruno Haible for CMUCL)
329 * bug fix: Cyclic structures and unprintable objects in compiler
330 messages no longer cause errors. (reported by Bruno Haible)
331 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
332 and READ-FROM-STRING all now return a primary value of NIL if
333 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
334 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
335 (reported by Bruno Haible for CMUCL)
336 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
337 do not designate a condition. (reported by Bruno Haible for
339 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
340 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
341 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
342 permitting resolution in favour of any of the conflicting symbols.
343 (reported by Bruno Haible for CMUCL)
344 * bug fix: EQUAL compiler optimizations is less aggressive on
345 strings which can potentially compare true despite having distinct
346 specialized array element types.
347 * bug fix: unit enumerations can be defined without dividing by
348 zero. (thanks to Vincent Arkesteijn)
349 * FORMAT compile-time argument count checking has been enhanced.
350 (report from Bruno Haible for CMUCL)
351 * a partial workaround for the bug 262: the compiler does not try to
352 inline-expand a local function doing RETURN-FROM from a deleted
353 BLOCK. (thanks to Peter Denno for the bug report and to David
354 Wragg for the simple test case)
355 * fixed some bugs revealed by Paul Dietz' test suite:
356 ** RENAME-PACKAGE allows all package designators as new package
358 ** constraint propagation and lambda variable substitution are
359 more cautious in dealing with partially deleted code.
360 ** compiler sometimes lost reoptimization passes.
361 ** CERROR, given a condition as condition designator, passes its
362 remaining arguments to the continue format control without
364 ** Case-altering FORMAT directives work correctly on non-ASCII
366 ** The REAL type specifier handles bounds outside the floating
367 point ranges without signalling FLOATING-POINT-OVERFLOW.
368 ** Functions with IR1-transformations can create intercomponent
369 references to global functions.
370 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
372 ** FORMAT ~R treats a nil value for its first parameter correctly.
374 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
375 * enhancement: saving cores with foreign code loaded is now
376 supported on x86/NetBSD and sparc/Linux in addition to the previously
378 * bug fix: on some platforms repeated installations caused multiple
379 copies of HTML documentation to be installed -- should not happen
380 any more. (reported by Stefan Scholl)
381 * bug fix: parsing self-recursive alien record types multiple times
382 no longer causes infinite recursion. (reported by Thomas F. Burdick,
383 original patch by Helmut Eller for CMUCL)
384 * bug fix: stack-exhaustion detection works now on NetBSD as well.
385 (thanks to Richard Kreuter)
386 * bug fix: defining classes whose accessors are methods on existing
387 generic functions in other (locked) packages no longer signals
388 bogus package lock violations. (reported by François-René Rideau)
389 * bug fix: special variables as DEFMETHOD parameters no longer have
390 associated bogus type declarations. (reported by David Wragg and
392 * bug fix: read-write consistency on streams of element-type
393 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
394 * bug fix: redefiniton of the only method of a generic function with
395 no DEFGENERIC no longer emits a full WARNING. In addition,
396 redefinition of generic functions with no DEFGENERIC to an
397 incompatible lambda list now signals an error. (thanks to Zach
399 * bug fix: DEFGENERIC now works even when there's a function of the
400 same name in an enclosing lexical environment. (thanks to Zach
402 * fixed compiler failure, caused by instrumenting code during
403 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
404 * optimization: added loop analysis and improved register allocation
405 * fixed some bugs revealed by Paul Dietz' test suite:
406 ** POSITION on displaced vectors with non-zero displacement
407 returns the right answer.
408 ** (SIMPLE-STRING) is a valid type specifier for sequence
410 ** *PRINT-LEVEL* handling for slotless structures is pedantically
412 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
413 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
415 ** The justification version of the FORMAT ~< directive treats
416 non-zero minpad parameter correctly.
418 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
419 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
420 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
421 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
422 the supported interface.
423 * new feature: Single-stepping of code compiled with DEBUG 2 or
424 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
425 * new feature: saving cores with foreign code loaded is now
426 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
427 Timothy Moore's work for CMUCL)
428 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
429 parameters to * instead of NIL if no initform is supplied.
430 (reported by Johan Bockgaard)
431 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
432 correctly when FUNCALL appears as the car of the form. Note:
433 despite this FUNCALL forms are not currently subject to
434 compiler-macro expansion. (port of Raymond Toy's fix for the
435 same from CMUCL, reported by Johan Bockgaard)
436 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
437 (thanks for Teemu Kalvas)
438 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
439 case instead of terminating the iteration (thanks for Julian
440 Squires, Sean Champ and Raymond Toy)
441 * bug fix: incorrect expansion of defgeneric that caused a style
442 warning. (thanks for Zach Beane)
443 * bug fix: slot accessor effective method computation works properly
444 for classes with multiple non-standard applicable methods on
445 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
446 * on x86 compiler supports stack allocation of results of LIST and
447 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
448 CMUCL implementation by Gerd Moellmann)
449 * fixed some bugs revealed by Paul Dietz' test suite:
450 ** FORMAT strings with both the ~<~:;~> form of the justification
451 directive and pretty-printing directives cause an error.
453 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
454 * incompatible change: the internal functions
455 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
456 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
457 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
458 instead of the old functions.
459 * new feature: on platforms where "dladdr" is available foreign
460 function names now appear in backtraces. (based on Helmut Eller's
462 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
463 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
465 * documentation: ASDF manual is now distributed with SBCL.
466 * bug fix: SBCL can now load its contributed modules with REQUIRE
467 even if the system-provided entries have been removed from
468 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
470 * bug fix: backtraces involving undefined functions or assembly
471 routines are more informative. (thanks to Brian Downing)
472 * bug fix: mutually referent alien structures now work correctly.
473 (reported by Rick Taube)
474 * bug fix: structures defined by WITH-ALIEN can be referred to
475 within other definitions in the same WITH-ALIEN.
476 * bug fix: division operators (MOD, TRUNCATE and the like) with
477 constant zero divisors and integer dividends no longer generate
479 * bug fix: provide default methods for INPUT-STREAM-P and
480 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
481 * bug fix: improve the detection and resolution of MOP metacycles
482 (where effective method computation depends on the generic function
483 having its effective method computed). (reported by Bruno Haible)
484 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
485 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
486 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
487 * bug fix: pathnames with relative directory components can be
488 represented relative to default pathnames.
489 * optimization: in taking the GCD of bignums, reduce the two bignums
490 to approximately the same size (using Euclid's algorithm) before
491 applying the more sophisticated binary GCD. (thanks to Juho
493 * optimization: COUNT on bitvectors now operates word-at-a-time.
494 * optimization: ASH with a positive, but not necessarily constant,
495 (leftwards) shift, when in a modular context, is compiled to a
497 * fixed some bugs revealed by Paul Dietz' test suite:
498 ** FORMAT variable parameters ("~V<char>") are defaulted properly
499 if the corresponding argument is NIL.
500 ** FORMAT directives accepting a minpad parameter treat negative
502 ** literal commas as character parameters to FORMAT directives are
504 ** literal spaces directly after ~<Newline> directives within a
505 format-logical-block (~:< ~@:>) do not induce :FILL-style
506 conditional newlines.
507 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
508 as specified: it no longer includes conditional newlines).
509 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
511 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
512 :PER-LINE-PREFIX argument does not evaluate to a string.
514 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
515 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
516 the manual for details; for now, package locks can be disabled by
517 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
518 if no major problems are found then it is likely that they will be
519 compiled in unconditionally.
520 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
521 unsupported operators on all platforms. To load a shared library
522 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
523 object file, link it into a shared library outside of SBCL and
524 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
526 * bug workaround: ROOM T and the GC were not working together
527 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
528 reporting facility aren't preserved. That reporting has been
529 disabled, so now until and unless someone figures out how to make it
530 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
531 * minor incompatible change: as threatened around sbcl-0.8.0, the
532 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
533 The symbols are also exported from SB-PCL for backwards
534 compatibility, but more so than before SB-PCL should be treated as
535 an implementation-internal package.
536 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
538 * the SB-SPROF contrib now works on (most) non-x86 architectures.
539 It is known as of this release not to work on the Alpha, however.
540 * fixed bug #167: errors signalled due to illegal syntax in method
541 bodies are now more legible.
542 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
543 designators and can hence be used with TYPEP.
544 * fixed bug #333: CHECK-TYPE now ensures that the type error
545 signalled, if any, has the right object to be accessed by
546 TYPE-ERROR-DATUM. (reported by Tony Martinez)
547 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
548 5.1.2.3 for multiple-value place subforms. (reported by Kalle
550 * fixed bug #334: programmatic addition of slots using specialized
551 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
552 and :ALLOCATION :CLASS slots.
553 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
554 float exponents. (rereported by Peter Seibel)
555 * fixed a bug: #\Space (and other whitespace characters) are no
556 longer considered to be macro characters in standard syntax by
558 * fixed bug: initialization of condition class metaobjects no longer
559 causes an instance of the condition to be created. (reported by
561 * fixed bug: it is now possible to have slots such that
562 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
563 description is neither :INSTANCE nor :CLASS.
564 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
565 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
566 SLOT-VALUE-USING-CLASS), and no longer causes errors with
567 non-standard SLOT-DEFINITION-ALLOCATION values.
568 * fixed bugs: various race conditions handling exiting threads.
569 CL-PPCRE's thread torture test now passes.
570 * fixed arguably-a-bug: GC time proportional to number of threads in
571 system even when most of them are idle
572 * optimization: improved performance of BIT and SBIT on bit-vectors.
573 * .fasl file incompatibility: The fasl file version number has
574 been incremented because of changes associated with package locks.
576 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
577 * minor incompatible change: the system no longer provides
578 optimization hints (conditions which are TYPEP
579 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
580 mode; these hints are still emitted when higher SPEED optimization
582 * new contrib module: a sampling profiler (profiling by statistical
583 sampling, rather than by instrumenting functions) is available as
584 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
585 * the behaviour of the standard function ED is now customizeable by
586 third parties through a hook variable: see ED's documentation
587 string for information on the protocol.
588 * the compiler no longer emits efficiency notes for (FUNCALL X)
589 when the type of X is uncertain under default optimization
591 * fixed bug 276: mutating a binding of a specialized parameter to a
592 method to something that is not TYPEP the specializer is now
594 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
595 exhibit the required behaviour on the x86 platform. (thanks to
596 Peter van Eynde, Eric Marsden and Bruno Haible)
597 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
598 even for difficult arguments. (reported by Peter Graves)
599 * fixed bug 141a: the backquote printer now descends quoted
601 * fixed another bug in backquote printing: no more destructive
602 modification of the form's list structure. (reported by Brian
604 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
605 move between its address being taken and the call to
606 interrupt_thread, fixing a crashing race condition.
607 * the SB-POSIX contrib implementation has been adjusted so that it
608 no longer exhibits ridiculously poor performance when constructing
609 instances corresponding to C structs.
611 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
612 * minor incompatible change: the sb-grovel contrib now treats C
613 structures as alien (in the sense of SB-ALIEN) objects rather than
614 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
615 has implications for memory management of client code
616 (sb-grovel-returned objects must now be manually managed) and for
617 type safety (alien objects now have full types).
618 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
619 used to control emission of compiler diagnostics, rather than the
620 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
621 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
622 quality should be considered deprecated.
623 * install.sh now installs the user manual as well
624 * (not quite a new documentable feature, but worth considering in
625 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
626 beginnings of a semantically meaningful condition hierarchy is
627 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
629 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
630 designator as the defaults argument.
631 * fixed bug: Displaced arrays whose displaced-to array has become
632 too small now cause ARRAY-DIMENSION to signal an error, providing
633 for safer bounds-checking. (reported by Bruno Haible)
634 * fixed bug: DEFCLASS slot definitions with identical :READER and
635 :WRITER names now signal a reasonable error. (reported by Thomas
637 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
639 * fixed bug: Invalid slot specification errors now print correctly.
640 (thanks to Zach Beane)
641 * fixed bug 320: Shared to local slot value transfers in class
642 redefinitions now happen corrently from superclasses as
643 well. (reported by Bruno Haible)
644 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
646 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
647 improper lists correctly. (reported by Bruno Haible)
648 * fixed bug 313: source-transform for <fun-name> was erroneously
649 applied to a call of a value of a variable with name <fun-name>.
650 (reported by Antonio Menezes Leitao)
651 * fixed bug 307: The obsolete instance protocol ensures that
652 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
653 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
654 remove dead unknown-values globs from the middle of the stack.
655 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
656 easier to resume long computations after using *BREAK-ON-SIGNALS*
657 to diagnose and fix failures. (thanks to Nikodemus Siivola)
658 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
659 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
660 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
661 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
662 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
664 * fixed a (fairly theoretical) bug in string printing: if
665 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
666 does not have array-element-type equal to the most general string
668 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
669 function name. (thanks to Nikodemus Siivola)
670 * fixed bug: compile-time format string checker failed on
671 non-closed ~{. (reported by Thomas F Burdick)
672 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
673 don't warn on legal loop code involving more than one aggregate
674 boolean. (thanks to Nikodemus Siivola)
675 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
676 in keyword position to MAKE-PACKAGE was not being reported as
678 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
679 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
681 * fixed bugs 280 and 312: the checking for multiple definitions in a
682 file is less likely to become confused by uses of inline
684 * fixed bug: the #S reader macro performs the keyword coercion
685 specified for slot names. (reported by Kalle Niemitalo)
686 * fixed bug: lambda lists may contain symbols whose names start with
687 & but are not lambda-list-keywords; their occurrence triggers a
689 * fixed bug 321: define-method-combination argument lambda lists do
690 not cause invalid code to be generated when &optional or &aux
691 variables have default values. (reported by Bruno Haible)
692 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
693 have CLOS classes; this appears also to have fixed reported
694 instability in stack exhaustion detection.
695 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
696 MUFFLE-WARNING when no associated restart is present is now
698 * optimization: rearranged the expansion of various defining macros
699 so that each expands into only one top-level form in a
700 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
702 * optimization: used a previously little-used slot in symbols to
703 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
705 * fixed some bugs revealed by Paul Dietz' test suite:
706 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
708 ** Fixed an optimization bug related to inheritance of initforms
709 from local to shared slots.
710 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
711 ** CAST optimizer forgot to flush argument derived type.
712 ** print/read consistency on floats is now orders of magnitude
713 more likely. (thanks also to Bruno Haible for a similar report
715 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
716 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
717 for objects of type REAL. Make it so.
718 ** SXHASH is specified (infelicitously) to respect similarity,
719 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
720 it so. (thanks to Markus Ziegler)
721 ** on the Alpha, the compiler succeeds in compiling functions
722 returning a known number of arguments greater than 63.
723 ** fixed handling of invalid NIL arguments in keyword position
725 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
727 ** provided a readably-printable representation for RANDOM-STATE
729 ** ensured that pathnames, where they have a namestring, always
730 print using #P"..." syntax.
732 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
733 * Support for the forthcoming 2.0 version of the NetBSD kernel
734 running on x86 hardware has been added. (thanks to Perry
735 E. Metzger most immediately, and others for their past work)
736 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
737 current release version; SBCL's OpenBSD support had been broken
738 since about the time of OpenBSD's switch to ELF binary format.)
739 * [placeholder for DX summary]
740 ** user code with &REST lists declared dynamic-extent, under high
741 speed or space and low safety and debug optimization policy.
742 * The manual has been converted to Texinfo format and the debugger
743 chapter from the cmucl manual has been added.
744 * A facility has been added to extract documentation strings from
745 sbcl and store them as Texinfo-formatted snippets for inclusion in
746 the manual (via Texinfo's @include directive)
747 * bug fix: compiler emitted division in optimized DEREF. (thanks for
748 the test case to Dave Roberts)
749 * bug fix: multidimensional simple arrays loaded from FASLs had fill
750 pointers. (reported by Sean Ross)
751 * bug fix: PROFILE output is printed nicely even for large numerical
752 values. (thanks to Zach Beane)
753 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
754 greater than 32 handle EOF correctly.
755 * bug fix: on X86 an immediate argument of the IMUL instruction is
756 correctly printed in disassembly. (thanks to Lutz Euler)
757 * bug fix: class slots in redefined classes preserve their old
758 values. (thanks to Bruno Haible and Nikodemus Siivola)
759 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
760 succeeds. (reported by Marco Baringer)
761 * fixed some bugs revealed by Paul Dietz' test suite:
762 ** READ-SEQUENCE now works on ECHO-STREAMs.
763 ** RATIONALIZE works more according to its specification. (thanks
766 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
767 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
768 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
769 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
770 should matter to you only if you rebind the printer control
771 variables and then find you want different bindings in the
772 debugger than in the ordinary execution of your program.)
773 * The runtime build system has been tweaked to support building
774 (on SPARC/SunOS) using a C compiler which invokes Sun's own
775 assembler and linker. (thanks to Nikodemus Siivola)
776 * Unbound, undefined, undeclared variables now trigger full
777 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
778 more often programmer error than deliberate exploitation of undefined
780 * optimization: the hash algorithm for strings has changed to one
781 that is less vulnerable to spurious collisions. (thanks to Juho
783 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
784 needless bounds checking. (thanks to Juho Snellman)
785 * optimization: implemented multiplication as a modular
786 (UNSIGNED-BYTE 32) operation on the PPC backend.
787 * fixed some bugs revealed by Paul Dietz' test suite:
788 ** ADJUST-ARRAY now copies the datum in a zero rank array if
790 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
791 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
792 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
794 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
796 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
798 ** ECHO-STREAMs no longer attempt to echo the end of file value to
799 their output stream on EOF from read.
800 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
801 have been read to end-of-file.
802 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
804 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
805 description of determination of which consecutive characters
807 ** Printing the "Space" character with escaping on now yields "#\\ ",
808 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
809 ** Reading floating-point numbers with *READ-BASE* set to a number
810 less than 10 works correctly.
811 ** Reading floating-point numbers with *READ-BASE* set to a number
812 more than 10 works correctly.
813 ** Printing with *PRINT-READABLY* targets the standard readtable, not
814 the readtable currently in effect.
816 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
817 * minor incompatible change: parsing of namestrings on a physical
818 (Unix) host has changed; numbers after the final #\. in a
819 namestring are no longer interpreted as a version field. This is
820 intented to be largely invisible to the user, except that the
821 meaning of the namestring "*.*.*" has changed: it now refers to a
822 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
823 should usually be replaced by
824 (make-pathname :name :wild :type :wild :version :wild)
825 with the added benefit that this is more likely to be portable.
826 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
827 option to OPEN now signals an error if the file being opened
828 exists; this may have an impact on existing code.
829 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
830 signals received, on the PowerPC platforms (both Linux and
831 Darwin). (thanks to Pierre Mai for pointing out the location of
833 * several fixes on OS X: The system now builds and runs cleanly on
834 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
835 (thanks to Brian Mastenbrook)
836 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
837 second-resolution: integer multiples of 1/3600 between -24 and 24.
838 (thanks to Vincent Arkesteijn)
839 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
840 type when called with 1 argument; PEEK-CHAR checked type of
841 PEEK-TYPE only after having read first character from a
842 stream. (reported by Peter Graves)
843 * bug fix: the garbage collector now has much better locality
844 behaviour, and in particular no longer treats all memory as being
845 exhausted when many small objects point to each other in a deeply
847 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
848 recognized as being TYPEP their class.
849 * bug fix: the PUSHNEW documentation string has been corrected.
850 (thanks to Vincent Arkesteijn)
851 * bug fix: defaulting of the value for the last of an atypically
852 large number of multiple values being bound was not being
853 performed correctly on the Alpha or PPC platforms
854 * optimization: implemented multiplication as a modular
855 (UNSIGNED-BYTE 32) operation on the x86 backend.
856 * optimization: SEARCH on simple-base-strings can now be open-coded.
857 (see also contrib/compiler-extras.lisp for inspiration for
858 teaching the compiler about the Boyer-Moore algorithm).
859 * value, returned by MAX (and MIN) called with several EQUALP, but
860 not EQL, arguments now does not depend on compiler settings.
861 * fixed some bugs revealed by Paul Dietz' test suite:
862 ** in stack analysis liveness information is propagated from
863 non-local entry points.
864 ** pathwise CAST removing failed when the CAST node did not start
866 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
867 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
869 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
870 host is already defined.
871 ** RENAME-FILE works on streams instead of signalling an internal
873 ** PEEK-CHAR uses the current readtable when determining whether
874 or not a character is whitespace.
875 ** MERGE-PATHNAMES handles the case when the pathname does not
876 specify a name while the default-pathname specifies a version.
877 ** Pathnames now stand a better chance of respecting print/read
879 ** Attempting to use standardized file system operators with a
880 pathname with invalid :DIRECTORY components signals a
882 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
884 ** DIRECTORY on logical pathnames is more correct.
885 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
886 signal a TYPE-ERROR if their argument is not a stream.
887 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
888 designator argument does not designate a stream.
889 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
890 examining the synonym.
891 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
893 ** OPEN and WITH-OPEN-STREAM allow opening streams with
894 element-type larger than ([UN]SIGNED-BYTE 32).
896 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
897 * When built with the :SB-FUTEX feature, threaded builds now take
898 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
899 for faster/more reliable mutex and condition variable support.
900 * Incompatible change (but one you probably shouldn't have been using
901 anyway): the interface and code for arbitrating between multiple
902 threads in the same user session has been redesigned.
903 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
904 error when called without an explicit environment argument.
905 (thanks to Nikodemus Siivola)
906 * bug fix: buffered :DIRECTION :IO streams are less likely to become
907 confused about their position. (thanks to Adam Warner and Gerd
909 * bug fix: Pretty printing backquoted forms with unquotations in the
910 argument list position of various code constructs such as LAMBDA
911 now works correctly. (reported by Paul Dietz)
912 * bug fix: Pretty printing unquotations no longer loses all
913 stream position information.
914 * optimization: performance of string output streams is now less
915 poor for multiple small sequence writes.
916 * optimization: performance of CSUBTYPEP in the presence of complex
917 expressions involving CONS and NOT many times has been improved.
918 (reported by Paul Dietz)
919 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
921 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
922 streams for the REPL, for the debugger, and so on. (thanks to
924 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
925 * fixed some bugs revealed by Paul Dietz' test suite:
926 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
927 is ignored if it is too small, rather than propagating through
928 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
929 ** extremely complex negations of CONS types were not being
930 sufficiently canonicalized, leading to inconsistencies in
932 ** VALUES tranformer lost derived type.
934 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
935 * fixed a bootstrapping bug: the build process no longer assumes
936 that the various BOOLE-related constants have the same value in
937 host and target lisps. (noted by Paul Dietz' test suite on an
938 SBCL binary built from CLISP)
939 * The system can now be dynamically linked on the MIPS platform,
940 which enables dynamic loading of foreign code from Lisp. (thanks
941 to Ralf Baechle for discussions on the MIPS ABI)
942 * The system now records debugging information for its own source
943 files in a filesystem-position-independent manner, relative to
944 the "SYS" logical host.
945 * fixed a compiler bug: MV-LET convertion did not check references
946 to the "max args" entry point. (reported by Brian Downing)
947 * tweaked disassembly notes to be less confident about proclaiming
948 some instruction as an LRA. (thanks to Brian Downing)
949 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
950 now each have their own history, command character, and other
951 characteristics. (thanks to David Lichteblau)
952 * fixed some bugs revealed by Paul Dietz' test suite:
953 ** compiler failure in compiling LOGAND expressions including a
955 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
956 shift greater than 32.
957 ** FLUSH-DEST did not mark blocks for type check regeneration.
958 ** HANDLER-CASE failed to accept declarations in handler clauses
959 in some circumstances.
961 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
962 * New code in contrib/sb-introspect (still probably not entirely
963 stable yet) provides some support for smart Lisp development
964 environments like SLIME.
965 * The conditions signalled for errors occurring when loading .fasl
966 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
967 in a way which should help ASDF recover gracefully.
968 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
969 stuff has been cleaned up. If you code contrib/ stuff, this might
970 affect you, and you can look at contrib/README, contrib/STANDARDS,
971 and/or the 0.8.4.27 diff to check.
972 * In full calls the compiler now does not generate checks for declared
973 argument types for all arguments.
974 * various threading fixes
975 ** and some experimental patches which didn't make it into
976 the main tree for this release, but which are shipped in
977 contrib/experimental-thread.patch as a possible fix for some
978 failures (deadlock, spinning...) in GC-intensive multithreaded
980 * fixed PPC build problem (source code incompatibility of different
981 library versions): added offsetof() hackery which attempts to divine
982 where glibc maintainers put uc_mcontext today
983 * fixed bug 282: compiler does not trust type assertions while passing
984 arguments to a full call.
985 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
986 &OPTIONAL VALUES type parameter.
987 * fix bug 214: algorithm for noting rejected templates is now more
988 similar to that of template seletion. (also reported by rydis on
990 * fixed bug 141b: printing backquoted information readably and prettily
991 inserts a space where necessary.
992 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
993 (CHARACTER 10) are now reported as errors, rather than propagated
994 as unknown types. (reported by piso on #lisp)
995 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
996 with values NIL and :ERROR. (thanks to Milan Zamazal)
997 * fixed bug 191c: CLOS now does proper keyword argument checking as
998 described in CLHS 7.6.5 and 7.6.5.1.
999 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1000 longer raise an error; further, using a list as a for-as-arithmetic
1001 counter now raises a meaningful error.
1002 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1003 now understood by sequence creation functions such as MAKE-SEQUENCE
1005 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1006 a STRING-INPUT-STREAM.
1007 * compiler enhancement: SIGNUM is now better able to derive the type
1009 * type declarations inside WITH-SLOTS are checked. (reported by
1011 * fixed some bugs revealed by Paul Dietz' test suite:
1012 ** incorrect optimization of TRUNCATE for a positive first
1013 argument and negative second.
1014 ** compiler failure in let-convertion during flushing dead code.
1015 ** compiler failure while deriving type of TRUNCATE on an
1016 interval, containing 0.
1017 ** ASH of a negative bignum by a negative bignum count now returns
1019 ** intersection of CONS types now canonicalizes properly, fixing
1020 inconsistencies in SUBTYPEP.
1022 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1023 * incompatible change: The --disable-debugger command line
1024 option now clobbers the debugger at a more fundamental
1025 level, by redefining #'INVOKE-DEBUGGER instead of by
1026 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1027 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1028 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1029 enter the debugger REPL and then suffer recursive errors
1030 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1031 while under the new system, BREAK in --disable-debugger mode
1032 terminates the system just as an unhandled error would.
1033 * fixed compiler performance when processing loops with a step >1;
1034 * bug fix: DOCUMENTATION now retrieves generic function
1035 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1036 support has been systematized, and now supports the methods
1037 specified by ANSI, along with a default method and a method for
1038 slot documentation. (reported by Nathan Froyd)
1039 * bug fix: effective methods associated with a generic function are
1040 no longer cached over a change of that generic function's method
1041 combination. (reported by Andreas Fuchs)
1042 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1043 names a non-existent directory. (thanks to Andreas Fuchs)
1044 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1045 the PPC platform, lead to stack corruption; this has been fixed.
1046 (reported by Rainer Joswig)
1047 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1048 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1049 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1050 arithmetic in general, is now much more reliable on the PPC
1052 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1053 platform now returns the right answer.
1054 * optimization: restored some effective method precomputation in
1055 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1056 precomputation is now tunable.
1057 * optimization: compiler-internal data structure use has been
1058 reviewed, and changes have been made that should improve the
1059 performance of the compiler by about 20%.
1060 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1061 simple-base-strings and simple-bit-vectors is improved.
1062 * optimization: the optimization of 32-bit logical and arithmetic
1063 functions introduced in version 0.8.3 on the x86 has been
1064 implemented on the mips, ppc and sparc platforms; an
1065 implementation of the same facility, but for 64-bit arithmetic,
1066 has been added for the alpha.
1067 * microoptimization: the compiler is better able to make use of the
1068 x86 LEA instruction for multiplication by constants.
1069 * bug fix: in some situations compiler did not report usage of
1070 generic arithmetic in (SPEED 3) policy.
1071 * bug 145b fix: compiler used wrong type specifier while converting
1072 MEMBER-types to numeric.
1073 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1075 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1076 index argument now works properly on the Alpha platform.
1077 * bug fix: floating point exception treatment on the Alpha platform
1079 * bug fix: FILE-POSITION works much better on string input and
1080 output streams. (thanks to Nikodemus Siivola)
1081 * bug fix: many threading/garbage collection symptoms sorted.
1082 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1083 might be pseudo-atomic.
1084 * internal change: Stopping for GC is now done with signals not
1085 ptrace. GC is now done in whichever thread wanted it, instead of
1087 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1089 * bug fix: result form in DO is not contained in the implicit
1091 * incompatible change: ICR structure is changed; the value part of
1092 CONTINUATION is now called LVAR; corresponding functions are
1093 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1094 * added type deriver for ISQRT (thanks to Robert E. Brown).
1095 * bug fix: better support for loading from the command line when an
1096 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1098 * fixed some bugs revealed by Paul Dietz' test suite:
1099 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1100 ** ROUND and FROUND now give the right answer when given very
1101 small float arguments.
1102 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1104 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1105 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1106 ** LCM should return a non-negative integer.
1107 ** PARSE-INTEGER returned the index of a terminator instead of the
1108 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1109 ** PARSE-INTEGER returned an incorrect index being applied to a
1111 ** LCM with two arguments of 0 returns 0 rather than signalling
1113 ** unsigned addition of a 32-bit constant with the high bit set no
1114 longer causes an internal compiler error on the x86.
1115 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1116 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1117 with negative last argument.
1118 ** byte specifiers with zero size and position no longer cause
1119 an error during type derivation.
1120 ** bignum multiplication on the Alpha platform now returns the
1122 * porting: The system now builds on SuSE AMD64, although it still
1123 generates a 32-bit binary.
1124 * .fasl file incompatibility: The fasl file version number has
1125 been incremented (because of the changes to internal compiler
1126 data structures referred to above).
1128 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1129 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1130 more accurately, on the Darwin kernel running on PowerPC hardware.
1131 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1132 * Compiler code deletion notes now signal a condition of type
1133 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1134 an associated MUFFLE-WARNING restart.
1135 * The compiler now performs limited argument count validation of
1136 constant format strings in FORMAT, and where appropriate in ERROR,
1137 CERROR and WARN. (thanks to Gerd Moellmann)
1138 * New ASDF-INSTALL contrib can be used for automatic download and
1139 installation of third-party Lisp code from CCLAN or other sites
1141 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1142 which forces another thread to execute a function supplied by the
1144 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1145 now accept and act upon their :ELEMENT-TYPE keyword argument.
1146 (reported by Martin Atzmueller, Edi Weitz)
1147 * bug fix: FILE-POSITION now accepts position designators up to
1148 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1149 is the greater. (thanks to Patrik Nordebo)
1150 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1151 Kalle Olavi Niemitalo)
1152 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1153 printable. (reported by Eric Marsden)
1154 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1155 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1156 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1157 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1158 (thanks to Stig Sandoe)
1159 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1160 UNREAD-CHAR now correctly includes the unread character in the
1161 target sequence. (thanks to Gerd Moellmann)
1162 * bug fix in threaded builds: the system can now be suspended and
1163 resumed by shell job control with minimal disruption.
1164 * bug fixes in times and timezones >2038AD
1165 * better handling of "where is GNU make?" problem in build scripts
1166 (thanks to Nikodemus Siivola)
1167 * new optimization: inside a named function any reference to a
1168 function with the same name is considered to be a self-reference;
1169 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1170 optimization quality.
1171 * new optimization on x86: logical functions and + now have
1172 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1173 used when the result is truncated to 32 bits.
1174 * VALUES declaration is partially enabled.
1175 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1176 * fixed some bugs revealed by Paul Dietz' test suite:
1177 ** The system now obeys the constraint imposed by
1178 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1179 types form a lattice under type intersection.
1180 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1181 ** ASSOC now ignores NIL elements in an alist.
1182 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1183 and (1+ MOST-POSITIVE-FIXNUM) answers.
1184 ** The addition of a method with invalid qualifiers to a generic
1185 function does not cause an error to be signalled immediately;
1186 a warning is signalled, and the error is generated only on
1187 calling the generic function.
1188 * changed .fasl file version number, in part to add type codes for
1189 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1190 obscure ANSI requirements
1192 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1193 * fixed bug 148: failure to inline-expand a local function left
1194 garbage, confusing the compiler.
1195 * fixed bugs 3cd: structure slot readers perform type check if the
1196 slot can have an invalid value (i.e. it is either not initialized
1197 or can be written with a less specific slot writer).
1198 * bug fix: the compiler now traps array references to elements off
1199 the end of an array; previously, the bounds checking in some
1200 circumstances could go off-by-one.
1201 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1203 * type declarations for array element types now obey the description
1204 on the CLHS page "Declaration TYPE", as per discussions on
1205 sbcl-help around 2003-05-08. This means that a declaration
1206 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1207 declaration, all references to BAR will be asserted or assumed
1208 (with THE, so dependent on compiler policy) to involve objects of
1209 type FOO. Note that no such declaration is implied in
1210 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1211 * declared types of functions from the "Conditions"
1212 chapter. (reported by Paul Dietz)
1213 * bug fix: CERROR accepts a function as its first argument.
1214 * bug fix: NTH an NTHCDR accept a bignum as index
1215 arguments. (reported by Adam Warner)
1216 * optimization: character compare routines now optimize comparing
1217 against a constant character. (reported by Gilbert Baumann)
1218 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1219 argument now works properly on the MIPS platform.
1220 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1221 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1222 declared to have a functional type, does not cause an error
1224 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1225 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1226 * fixed bug: if last continuation of a deleted block has a
1227 destination, this destination should be deleted too. (reported by
1229 * fixed a bug in the bootstrap process: the host compiler's values
1230 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1231 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1232 test case from Patrik Nordebo)
1233 * improved the ability of the disassembler on the PPC platform to
1234 provide helpful disassembly notes.
1235 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1236 the class in more cases than previously.
1237 * bug fix: FILE-POSITION now understands :START and :END for
1238 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1239 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1240 8) data. (thanks to David Lichteblau)
1241 * bug fix: it is possible to add a method to a generic function
1242 without lambda list.
1243 * bug fix: reader failed to signal END-OF-FILE inside an
1244 object representation. (reported by Nikodemus Siivola)
1245 * fixed some bugs revealed by Paul Dietz' test suite:
1246 ** LAST and [N]BUTLAST should accept a bignum.
1247 ** condition slot accessors are methods.
1248 ** (VECTOR NIL) is a subtype of STRING.
1250 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1251 * minor incompatible change: some nonsensical specialized lambda
1252 lists (used in DEFMETHOD) which were previously ignored now signal
1254 * minor incompatible change: the system is now aware of the types of
1255 variables in the COMMON-LISP package, and will signal errors for
1256 most violations of these type constraints (where previously they
1257 were silently accepted).
1258 * minor incompatible change: COMPILE-FILE now uses the freedom
1259 afforded (ANSI 3.2.2.3) to use derived function types for
1260 functions defined in the same file. This also permits the system
1261 to warn on static type mismatches and function
1262 redefinition. (Currently it does not work with high DEBUG level.)
1263 * minor incompatible change: VALUES declaration is disabled.
1264 * When issuing notes, the compiler now signals a condition of type
1265 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1266 restart for use in user handlers. It is expected that the
1267 COMPILER-NOTE condition will eventually become a condition
1268 supertype to a hierarchy of note types, which will then be
1269 handleable in a similar fashion. However, at the moment, no such
1270 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1271 but it's an implementation detail, not a classification for the
1273 * Changes in type checking closed the following bugs:
1274 ** type checking of unused values (192b, 194d, 203);
1275 ** template selection based on unsafe type assertions (192c, 236);
1276 ** type checking in branches (194bc).
1277 * A short form of VALUES type specifier has ANSI meaning (it has
1278 increased the number of situations when SBCL cannot perform type
1280 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1281 keywords or constants is permissible.
1282 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1283 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1284 classes. (thanks to Antonio Martinez)
1285 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1286 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1287 argument to a pretty-print stream. Instead, it leaves any such
1288 operations to DESCRIBE-OBJECT methods.
1289 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1291 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1292 lambda lists. (thanks to David Lichteblau)
1293 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1294 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1295 (thanks to Gerd Moellmann)
1296 * fixed bug 246: increased compilation speed of long
1297 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1299 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1301 * DEFCONSTANT now throws a condition of type
1302 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1303 constant to a non-EQL value; CONTINUE and ABORT restarts
1304 respectively change and preserve the value.
1305 * fixed bug 63: The code walker, part of the implementation of CLOS,
1306 is now better at handling symbol macros.
1307 * bug fix: There is no longer an internal implementation type named
1308 CL:LENGTH. (reported by Raymond Toy)
1309 * bug fix: In macro-like defining macros/special operators the
1310 implicit block does not enclose the lambda list.
1311 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1312 enclosed in parentheses) are not suitable as type specifiers, and
1313 their use properly signals an error now.
1314 * bug fix: An argument count mismatch for a type specifier in code
1315 being compiled no longer causes an unhandled error at compile
1316 time, but signals a compile-time warning.
1317 * fixed simple vector readable printing
1318 * bug fix: DESCRIBE takes more care over whether the class
1319 precedence list slot of a class is bound before accessing it.
1320 (reported by Markus Krummenacker)
1321 * bug fix: FORMATTER can successfully compile pretty-printer format
1322 strings which use variants of the ~* directive inside.
1323 * bug fix: SEARCH now applies its TEST predicate to the elements of
1324 the arguments in the correct order. (thanks to Wolfhard Buss)
1325 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1326 in type inference. (thanks to Robert E. Brown)
1327 * bug fix: Reading in symbols with an explicit package name of ""
1328 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1329 * fixed some bugs revealed by Paul Dietz' test suite:
1330 ** NIL is now allowed as a structure slot name.
1331 ** Arbitrary numbers, not just REALs, are allowed in certain
1332 circumstances in LOOP for-as-arithmetic clauses.
1333 ** Multiple class redefinitions before slot access no longer
1334 causes a type error.
1335 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1336 association between the name and a class.
1337 ** Generic functions with non-standard method-combination and over
1338 five methods all of which return constants no longer return NIL
1339 after the first few invocations. (thanks to Gerd Moellmann)
1340 ** CALL-NEXT-METHOD with no arguments now passes the original
1341 values of the arguments, even in the presence of assignment.
1342 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1344 ** DISASSEMBLE works with closures and funcallable instances.
1345 ** ADD-METHOD now returns the generic function, not the new method.
1346 ** FIND-METHOD signals an error if the lengths of the specializers
1347 is incompatible with the generic function, even if the ERRORP
1349 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1350 which its argument is a member.
1351 ** DEFCLASS only redefines the class named by its class-name
1352 argument if that name is the proper name of the class;
1353 otherwise, it creates a new class.
1354 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1355 of the UNBOUND-SLOT condition to the name of the slot.
1356 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1357 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1358 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1359 SLOT-MAKUNBOUND in the specified fashion.
1361 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1362 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1363 cross-compilation host. As a consequence, we can now bootstrap our
1364 way up to SBCL starting with a bare gcc toolchain and human-readable
1365 source code (first the source to CLISP, then the source to SBCL).
1366 * A contributed module containing a partial implementation of the
1367 simple-streams interface has been included. (thanks to Rudi
1369 * A contributed module implementing the RFC1321 Message Digest
1370 Algorithm, known as MD5, has been included.
1371 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1372 no longer has any effect, as the code controlled by this feature
1373 has been deleted. (As far as we know, no-one has ever built using
1374 this feature, and its semantics were confused in any case).
1375 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1376 work on conditions (as required by the ANSI specification),
1377 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1378 expected behaviour on conditions. Users should note, however,
1379 that such behaviour is not required by the ANSI specification,
1380 and so use of this behaviour may render their code unportable.
1381 * fixed some bugs revealed by Paul Dietz' test suite:
1382 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1384 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1385 ** SXHASH on condition objects no longer returns NIL.
1386 ** :ALLOCATION :CLASS slots are better treated; their values are
1387 updated on class redefinition, and initforms inherited from
1388 superclasses are applied.
1389 ** REMOVE-METHOD returns its generic function argument even when
1390 no method was removed.
1391 ** SHARED-INITIALIZE now initializes the values of the requested
1392 slots, including those with :ALLOCATION :CLASS.
1393 ** ALLOCATE-INSTANCE now works on structure classes defined via
1394 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1396 ** SLOT-EXISTS-P now works on conditions, as well as structures
1398 ** MAKE-LOAD-FORM now has the required methods on
1399 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1400 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1401 keyword, and now implements the SLOT-NAMES argument.
1402 ** methods with &OPTIONAL arguments no longer allow too many
1403 arguments to be passed in the call without error.
1404 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1405 option is consistent with the required arguments of the generic
1406 function lambda list.
1407 * bug fix: REQUIRE accepts a string designator. (Thanks to
1409 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1410 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1411 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1412 * bug fix: The compiler now checks for duplicated variables in macro
1414 * bug fix: SETQ on globals returns the correct value.
1415 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1416 not a COMPILER-ERROR (followed by some other strange error on
1417 choosing the CONTINUE restart).
1418 * bug fix: make.sh and friends are now more consistent in the way that
1419 they look for GNU "make".
1421 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1422 * experimental native threads support (on x86 Linux >=2.4 only).
1423 This is not compiled in by default: you need to add :SB-THREAD to
1424 the target features. See the "Beyond ANSI" chapter of the manual
1426 * fix for longstanding nonANSIism: The old distinction between
1427 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1428 The return value from CL:FIND-CLASS is now a CLOS class, and
1429 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1430 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1431 * An interface to the MetaObject Protocol, as described in Kiczales,
1432 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1433 MIT Press, 1991, is available from the SB-MOP package.
1434 * incompatible change: the SB-PCL package should now be considered
1435 a private implementation detail, and no longer a semi-private MOP
1437 * minor incompatible change: due to rearrangement for threads, the
1438 control stack and binding stack are now allocated at arbitrary
1439 addresses instead of being hardcoded per-port. Users affected by
1440 this probably have to be doing advanced things with shared
1441 libraries, and will know who they are.
1442 * minor incompatible change: Previously, all --eval forms used were
1443 processed with READ before any of them were processed with EVAL.
1444 Now each --eval form is processed with both READ and EVAL before
1445 the next --eval form is processed. (Thus package operations like
1446 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1447 work as the user might reasonably expect.)
1448 * minor incompatible change: *STANDARD-INPUT* is now only an
1449 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1451 * minor incompatible change: Y-OR-N-P is now character-oriented,
1452 not line oriented. Also, YES-OR-NO-P now works without errors.
1453 (thanks to Antonio Martinez)
1454 * sb-aclrepl module improvements: an integrated inspector, added
1455 repl features, and a bug fix to :trace command.
1456 * Known functions, which cannot be open coded by the backend, are
1457 considered to be able to check types of their arguments. (fixing
1458 a bug report by Nathan J. Froyd)
1459 * fixed a bug in computing method discriminating functions: It is
1460 now possible to define methods specialized on classes which have
1461 forward-referenced superclasses. (thanks to Gerd Moellmann)
1462 * fixed evaluation order in optional entries (reported by Gilbert
1464 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1465 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1466 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1467 required-by-AMOP NAME argument, as well as CLASS and
1468 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1469 * fixed bug 20: DEFMETHOD can define methods using names that are
1470 not the proper names of classes to designate class specializers.
1471 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1472 Unix file descriptors, instead of blowing up. (thanks to Antonio
1474 * Garbage collection refactoring: user-visible change is that a
1475 call to the GC function during WITHOUT-GCING will not do garbage
1476 collection until the end of the WITHOUT-GCING. If you were doing
1477 this you were probably losing anyway.
1478 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1479 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1480 TYPEP the latter but not the former.
1481 * The compiler issues a full WARNING for calls to undefined functions
1482 with names from the CL package.
1483 * MAP-INTO for a vector destination is open coded. (reported by
1484 Brian Downing on c.l.l)
1485 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1486 documentation string.
1487 * fixed some bugs revealed by Paul Dietz' test suite:
1488 ** COPY-ALIST now signals an error if its argument is a dotted
1490 ** Condition slots are now accessed more correctly in the presence
1491 of multiple initargs for a given slot.
1492 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1493 exclude restarts of the same name associated with a different
1495 ** DEFCLASS of forward-referenced classes with another
1496 forward-referenced class in the superclasses list no longer
1498 ** Condition slots are now initialized once each, not multiple
1499 times. (thanks to Gerd Moellmann)
1500 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1502 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1503 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1504 arguments contain duplicated elements.
1505 ** RESTART-CASE understands local macros.
1506 ** RESTART-CASE associates exactly its own restarts with a condition.
1507 ** ENDP in safe mode checks its argument to be of type LIST.
1508 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1509 in question is unbound.
1510 ** Optimized MAKE-INSTANCE functions no longer cause internal
1511 assertion failures in the presence of duplicate initargs.
1512 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1513 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1514 propagated even in the ctor optimized implementation of
1516 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1518 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1519 * a better implementation of SXHASH on (simple) bit vectors,
1520 measured both in execution speed and in distribution of results
1521 over the positive fixnums, has been installed. Likewise, a better
1522 implementation of EQUAL for simple bit vectors is now available.
1523 * fixed CEILING optimization for a divisor of form 2^k.
1524 * fixed bug 240 (emitting extra style warnings "using the lexical
1525 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1526 by Antonio Martinez)
1527 * fixed SXHASH, giving different results for NIL depending on type
1528 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1529 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1530 variable. (found by Rolf Wester)
1531 * SBCL does not ignore type declarations for special
1532 variables. (reported by rif on c.l.l 2003-03-05)
1533 * some bug fixes in contrib/sb-aclrepl/
1534 * fixed some bugs revealed by Paul Dietz' test suite:
1535 ** a bug in the CONS type specifier, whereby the CAR and CDR
1536 types got intertwined, has been fixed;
1537 ** the type system is now able to reason about the interaction
1538 between INTEGER and RATIO types more completely;
1539 ** APPEND, [N]REVERSE and NRECONC check that those their
1540 arguments, which must be proper lists, are really so;
1541 ** An array specialized to be unable to hold elements has been
1542 implemented, as required -- yes, really -- by ANSI;
1543 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1544 on malformed property lists;
1546 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1547 * incompatible packaging change: in line with Unix convention,
1548 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1549 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1550 * REQUIRE and PROVIDE are now optionally capable of doing something
1551 useful. See the documentation string for REQUIRE.
1552 * infrastructure for a managed SBCL contrib system: contributed
1553 modules in this release include:
1554 ** the ASDF system definition facility;
1555 ** an interface to the BSD Sockets API;
1556 ** an ACL-like convenience interface to the repl;
1557 (thanks to Kevin Rosenberg)
1558 ** an implementation of ROTATE-BYTE, with efficient implementation
1560 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1561 gives a closer approximation to the right answer than previously.
1562 (thanks to Raymond Toy)
1563 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1564 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1565 optional environment arguments, as required by ANSI.
1566 * fixed bugs in other functions taking environment objects, allowing
1567 calls with an explicit NIL environment argument to be compiled
1569 * fixed bug 228: primary return values from
1570 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1571 COMPILE or FUNCTION.
1572 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1573 :TYPE will no longer signal errors on innocuous objects.
1574 * fixed bug 231b: SETQ is better at respecting type declarations in
1575 the lexical environment.
1576 * fixed a bug in DEFCLASS: classes named by symbols with no or
1577 unprintable packages can now be defined.
1578 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1579 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1580 * fixed compiler failure related to checking types of functions.
1581 (reported by Robert E. Brown)
1582 * the compiler is now much more consistent in its error-checking
1583 treatment of bounding index arguments to sequence functions: in
1584 (SAFETY 3) code, errors will be signalled in almost all cases if
1585 invalid sequence bounding indices are passed to functions defined
1586 by ANSI to operate on sequences.
1587 * fixed a bug in the build procedure: documentation of SBCL-specific
1588 packages is now preserved and available in the final Lisp image.
1589 * lifted FDEFINITION lookup out of loops in the implementation of
1590 many list operations. (thanks to Robert E. Brown)
1591 * fixed a bug in the reader: the #n# reader macro now works for
1592 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1593 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1594 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1595 E. Brown and Tony Martinez respectively)
1596 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1597 SLOT-BOUNDP: in particular, it is now able to optimize them much
1598 better, and is now not vulnerable to having packages renamed.
1599 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1600 always called when a slot is not present in an instance. (thanks
1602 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1603 longer signal an error on unreading a character following EOF on
1604 the previous constituent stream. (thanks to Tony Martinez)
1605 * fixed some bugs revealed by Paul Dietz' test suite:
1606 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1607 not just nonnegative fixnums;
1608 ** the logical bit-array operators such as BIT-AND now accept an
1609 explicit NIL for their "opt-arg" argument (to indicate a
1610 freshly-consed result bit-array);
1611 ** ELT now signals an error on an invalid sequence index in safe
1613 ** the type system is now cleverer about negations of numeric
1614 types, and consequently understands the BIGNUM and RATIO types
1616 ** the type system is now cleverer about the interaction between
1617 INTEGER and RATIO types: while bugs still remain, many more
1618 cases are accurately computed;
1619 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1620 if it is in the last clause;
1621 ** CONSTANTLY now correctly returns a side-effect-free function in
1623 ** DECLARE is no longer treated as a special-operator; in
1624 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1625 * incremented fasl file version number due to the change in the
1626 DEFSTRUCT-SLOT-DESCRIPTION structure.
1628 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1629 * minor incompatible change: code processed by the "interpreter" or
1630 EVAL now has a compilation optimization policy of (DEBUG 2)
1631 (changed from (DEBUG 1)) to improve debuggability of interactive
1632 development, and to allow the use of the debug RETURN command in
1634 * an experimental implementation of the RETURN command for the
1635 debugger has been included. (thanks to Frederik Kuivinen)
1636 * fixed bug 62: constraints were not propagated into a loop.
1637 * fixed bug in embedded calls of SORT (reported and investigated by
1639 * fixed some bugs revealed by Paul Dietz' test suite:
1640 ** printing and reading of arrays with some dimensions having
1641 length 0 (thanks to Gerd Moellmann);
1642 ** BOA constructor with &AUX argument without a default value does
1643 not cause a type error;
1644 ** CONSTANTP now returns true for all self-evaluating objects.
1646 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1647 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1648 accessors that are related by inheritance, as specified in the
1649 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1650 Valtteri Vuorikoski)
1651 * The compiler is now able to inline functions that were defined in
1652 a complex lexical environment (e.g. inside a MACROLET).
1653 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1655 * Support for the upcoming FreeBSD-5.0 release has been included.
1656 (thanks to Dag-Erling Smorgrav)
1657 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1658 effect when it is not in a toplevel context.
1659 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1660 stand a better chance of being correct. (thanks to Gerd
1662 * fixed bug in COERCE, which now signals an error on coercing a
1663 rational to a bounded real type which excludes the expected
1665 * The compiler is now able to derive types more accurately from the
1666 COERCE and COMPILE functions.
1667 * fixed bug 223: functional binding is considered to be constant
1668 only for symbols in the CL package.
1669 * fixed bug 231: SETQ did not check the type of a variable being set
1670 (reported by Robert E. Brown)
1671 * A new optimization for MAKE-INSTANCE has been included, fixing
1672 various bugs (including relating to :ALLOCATION :CLASS slots and
1673 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1675 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1676 ** As required by ANSI, LOOP now disallows anonymous collection
1677 clauses such as COLLECT I in conjunction with aggregate boolean
1678 clauses such as THEREIS (= I 1);
1679 ** LOOP now signals an error when any variable is reused in the
1680 same loop (including the potentially useful construct analogous
1681 to WITH A = 1 WITH A = (1+ A);
1682 ** IT is only a special loop symbol within the first clause of a
1683 conditional loop clause;
1684 ** LOOP with a typed iteration variable over a hashtable now
1685 signals a type error iff it should.
1686 * fixed some other bugs revealed by Paul Dietz' test suite:
1687 ** FILE-STREAM now names the class previously known as FD-STREAM;
1688 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1689 argument) no longer signals an error;
1690 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1691 of the slot symbol, rather than using the current package
1692 ((:CONC-NAME "") continues to intern the slot's name in the
1694 * incremented fasl file version number, because of the incompatible
1695 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1696 of the new implementation of DEFINE-COMPILER-MACRO.
1698 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1699 * Support for building SBCL for MIPS platforms running in
1700 little-endian mode has now been checked in, and basic
1701 functionality on said platforms verified.
1702 * minor incompatible change: PCL now records the pathname of a file
1703 in which methods and the like are defined, rather than its
1705 * minor incompatible change: TRUENAME now considers the truename of
1706 a file naming a directory to be the pathname with :DIRECTORY
1707 component indicating that directory.
1708 * minor incompatible change: a NAMED clause in the extended form of
1709 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1710 reason for the previous behaviour is unclear.
1711 * more systematization and improvement of CLOS and MOP conformance
1712 in PCL (thanks to Gerd Moellman and Pierre Mai):
1713 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1715 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1716 primary methods with no specializers;
1717 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1719 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1720 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1721 FORWARD-REFERENCED-CLASSes; error reporting on
1722 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1724 ** SXHASH on CLOS instances now uses a slot internal to the
1725 instance to return different numbers on distinct instances,
1726 while preserving the same return value through invocations of
1728 ** DEFMETHOD signals errors when methods with longer incongruent
1729 lambda lists are added to generic functions;
1730 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1731 CLASS, as specified in AMOP;
1732 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1733 on the order returned by the primary method for classes of
1734 class STANDARD-CLASS;
1735 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1736 * fixed some bugs shown by Paul Dietz' test suite:
1737 ** DOLIST puts its body in TAGBODY;
1738 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1740 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1741 value producing form;
1742 ** if more variables are given to PROGV than values, extra
1743 variables are bound and made to have no value;
1744 ** NSUBSTITUTE on list arguments gets the right answer with
1746 ** ELT signals an error of type TYPE-ERROR when the index argument
1747 is not a valid sequence index;
1748 ** LOOP signals (at macroexpansion time) an error of type
1749 PROGRAM-ERROR when duplicate variable names are found;
1750 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1751 ** FUNCALL of special-operators now cause an error of type
1753 ** PSETQ now works as required in the presence of side-effecting
1754 symbol-macro places;
1755 ** NCONC accepts any object as its last argument;
1756 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1758 ** loop-for-as-package does not require a package to be explicitely
1760 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1762 * fixed bug 166: compiler preserves "there is a way to go"
1763 invariant when deleting code.
1764 * fixed bug 172: macro lambda lists with required arguments after
1765 &REST arguments now cause an error to be signalled. (thanks to
1767 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1768 bug. (thanks to Matthew Danish)
1769 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1771 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1772 arguments are lexically rebound. (thanks to Gerd Moellmann and
1774 * fixed bug 194: error messages are now more informative when there
1775 is no primary method applicable in a call to a generic
1776 function. (thanks to Gerd Moellmann)
1777 * fixed bug in command line argument checking (thanks to Julian
1779 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1780 (thanks to Matthew Danish)
1781 * incremented fasl file version number, because of the
1782 SXHASH-related changes in the layout of CLOS data structures
1784 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1785 * minor incompatible change: The runtime (the Unix executable named
1786 "sbcl") is now much pickier about the .core files it will load.
1787 Essentially it now requires .core files to descend from the same
1788 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1789 as the runtime does. (The intent is to prevent the crashes which
1790 can occur, and which can even be reported as mysterious failures,
1791 when people patch the sources or change the build parameters
1792 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1793 sbcl and .core files.)
1794 * fixed bug: VALUES-LIST is no longer optimized away.
1795 * fixed bug 142: The FFI conversion of C string values to Lisp
1796 string values no longer conses excessively. (thanks to Nathan
1797 Froyd porting Raymond Toy's fix to CMU CL)
1798 * began to systematize and improve MOP conformance in PCL (thanks to
1799 Nathan Froyd, Gerd Moellman and Pierre Mai):
1800 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1802 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1803 ** FINALIZE-INHERITANCE is now called on class finalization;
1804 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1805 argument precedence order.
1806 * fixed bug 202: The compiler no longer fails on functions whose
1807 derived types contradict their declared type.
1808 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1809 so it can be non-toplevel.
1810 * The fasl file version number has changed (because of the new
1811 implementation of DEFMACRO).
1812 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1813 safe code, that any length requirement by their type-specifier
1814 argument is valid. The exceptions to this are described in bug
1816 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1817 functions now check the types of their inputs as required by ANSI.
1818 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1819 introduced symbol is DECLAREd to be SPECIAL.
1820 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1821 * fixed bug: :COUNT argument to sequence functions may be negative
1822 * fixed bug: body of DO-SYMBOLS may contain declarations
1823 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1824 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1825 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1826 symbol macro only once
1827 * fixed printing of call frame when argument list is unavailable
1828 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1829 * compiler no longer signals WARNING on unknown keyword
1832 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1833 * A beta-quality port to the mips architecture running Linux,
1834 based on the old CMUCL backend, has been made. It has been tested
1835 on a big-endian kernel, and works sufficiently well to be able to
1836 rebuild itself; it has not been tested in little-endian mode.
1837 * fixed an inconsistency between gencgc.c and purify.c which made
1838 dumping/loading .core files unreliable
1839 * fixed bug 120a: The compiler now deals correctly with IFs where
1840 the consequent is the same as the alternative, instead of
1841 misderiving the return type. (thanks to Alexey Dejneka)
1842 * fixed bug 113: Logical pathnames are now dumpable (the logical
1843 host is resolved at load-time, throwing an error if it is not
1845 * fixed bug 174: FORMAT's error message is slightly clearer when a
1846 non-printing character is used in a format directive.
1847 * fixed several bugs in compiler checking of type declarations, i.e.
1848 violations of the Python "declarations are assertions" principle
1849 (thanks to Alexey Dejneka)
1850 * fixed several bugs in PCL's error checking (thanks to Gerd
1852 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1854 * fixed bug in compilation of functions as first class values
1855 (thanks to Antonio Martinez-Shotton)
1856 * The compiler's handling TYPE-ERRORs which it can prove will
1857 inevitably happen at runtime has been cleaned up and corrected
1858 in several ways. (thanks to Alexey Dejneka)
1859 * improved argument type checking for various basic arithmetic
1860 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1861 lavished on them in the past that they can be compiled in many
1862 ways in different special cases
1863 * fixed bug 181: compiler checks validity of user supplied type
1865 * cleaned up code flushing in optimization: Function calls which
1866 should signal errors for safety purposes (e.g. which ANSI says
1867 should signal errors when their arguments are of incorrect type)
1868 are no longer optimized away.
1869 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1870 * incremented fasl file version number, because changes in the
1871 implementation of sequence functions like COERCE caused
1872 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1873 in old inline expansions) to become undefined. (Actually these
1874 changes were later undone, so we might very well be binary
1875 compatible with 0.7.7 after all, but leaving the version number
1876 incremented seemed like the simplest and most conservative
1879 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1880 * An alpha-quality port to the parisc architecture running Linux,
1881 based on the old CMUCL backend, has been made. This, even more so
1882 than the other backends, should be considered still a work in
1883 progress; known problems include that the Linux kernel in 64-bit
1884 mode does not propagate the correct sigcontext structure to
1885 userspace, and consequently SBCL on a parisc64 kernel will not
1887 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1888 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1889 suppressed.) Also now that the compiler is looking at declarations
1890 in the environment, it checks optimization declarations as well,
1891 and suppresses inlining when (> DEBUG SPEED).
1892 * More fixes have been made to treatment of floating point exception
1893 treatment and other Unix signals. In particular, floating point
1894 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1895 * The detection and handling of control stack exhaustion (infinite
1896 or very deeply nested recursion) has changed. Stack exhaustion
1897 detection is now done by write-protecting pages at the OS level
1898 and applies at all optimization settings; when found, a
1899 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1900 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1901 be caught using IGNORE-ERRORS.
1902 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1903 correctly when transforming recursive LABELS functions to LETs.
1904 (thanks to Alexey Dejneka)
1905 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1906 that are names of constants or global variables.
1907 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1908 alien routines with docstrings.
1909 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1910 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1912 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1913 (thanks to Eric Marsden)
1914 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1915 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1916 * Bug fix: The SPARC backend can now compile functions involving
1917 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1918 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1919 to a functions expecting a PATHNAME-DESIGNATOR.
1920 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1921 lambda lists. (thanks to Alexey Dejneka)
1922 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1923 functions so that EXPORTing the name of the class doesn't cause
1924 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1925 to Antonio Martinez for reporting this)
1926 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1927 constant-folded without failing an assertion. (thanks to Einar
1928 Floystad Dorum for reporting this)
1929 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1930 types are now dealt with more correctly by the compiler.
1931 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1932 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1933 behaviour analogous to RENAME-FILE. This puts its behaviour more
1934 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1936 * The fasl file version number has changed again. (because of the
1937 bug fix involving the names of PCL MAKE-INSTANCE functions)
1939 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1940 * bug fix: Floating point exceptions are treated much more
1941 consistently on the x86/Linux and PPC/Linux platforms.
1942 * Array initialization with :INITIAL-ELEMENT is now much faster for
1943 cases when the compiler cannot open code the array creation, but
1944 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1945 array accesses have also seen a speed increase.
1946 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1947 specified. (This was at the root of some bad interactions between
1948 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1950 * bug fix: Internal error arguments for undefined functions are now
1951 computed correctly on the PPC/Linux platform.
1952 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1953 Raymond Toy's patch for CMU CL)
1954 * Support for the Solaris 9 operating environment has been included
1955 (thanks to Daniel Merritt)
1956 * A very ugly but hopefully complete draft of the missing FFI chapter
1957 of the manual has been created by reformatting the corresponding
1958 CMU CL manual chapter into (currently very ugly and incoherent)
1959 DocBook and bringing it up to date for SBCL behavior. Thus, the
1960 manual is now essentially complete, at least by my extreme
1961 once-and-only-once standards, whereby it's acceptable to refer to
1962 the doc strings of SB-EXT functions as the primary documentation.
1963 * The fasl file version number has changed again, due to cleanup of
1964 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1965 is no longer a static symbol.)
1967 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1968 * SBCL now builds with OpenMCL (version 0.12) as the
1969 cross-compilation host; also, more progress has been made toward
1970 bootstrapping under CLISP.
1971 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1973 * bug 158 fixed: The compiler can now deal with integer loop
1974 increments different from 1; fixing this turned out also to fix
1976 * bug 169 fixed: no more bogus warnings about using lexical bindings
1977 despite the presence of perfectly good SPECIAL declarations (thanks
1978 to David Lichteblau)
1979 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1980 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1981 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1983 * bug 180 fixed: Method combination specifications no longer ignore
1984 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1985 * bug fix: Structure type predicate functions now check their argument
1986 count as they should.
1987 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1988 correctly. (thanks to Pierre Mai)
1989 * minor incompatible change: The --noprogrammer option is deprecated
1990 in favor of the new --disable-debugger option, which is very similar.
1991 (The major difference is that it takes effect at a slightly different
1992 time at startup, causing handling of errors in --sysinit and
1993 --userinit files will be affected differently.) The
1994 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1995 been added to allow this functionality to be controlled from ordinary
1996 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1997 maintainers, who might want to run non-interactive scripts to
1998 build SBCL cores which will later be used interactively.)
1999 * minor incompatible change: The LOAD function no longer, when given
2000 a wild pathname to load, loads all files matching that pathname.
2001 Instead, an error of type FILE-ERROR is signalled.
2003 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2004 * bug 147 fixed: The compiler preserves its block link/count
2005 invariants more correctly now so that it doesn't crash. (thanks
2007 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2009 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2010 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2011 no-value-for-this-character as NIL (as specified by ANSI).
2012 * HOST-NAMESTRING on physical pathnames now returns a string that is
2013 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2014 (thanks to Christophe Rhodes)
2015 * The Alpha port handles icache flushing more correctly. (thanks to
2017 * More progress has been made toward bootstrapping under CLISP. (thanks
2018 to Christophe Rhodes)
2019 * The fasl file format has changed again, because dynamic loading
2020 on OpenBSD (which has non-ELF object files) motivated some cleanups
2021 in the way that foreign symbols are transformed and passed around.
2022 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2023 is no longer treated as whitespace by the reader, but instead as
2024 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2025 |A
\7fB|, instead of A as it used to.
2027 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2028 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2029 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2030 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2031 this as of 0.7.1.45, but was left out of the previous news section
2032 (thanks to Dan Barlow)
2033 * SBCL now runs on the Solaris operating system on SPARC architectures
2034 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2035 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2036 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2037 * SPARC backend cleanups, allowing builds of cores optimized for V8
2038 and V9 SPARCS, and also emission of code targeted to a particular
2039 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2041 * SBCL is closer to bootstrapping under CLISP, thanks to various
2042 fixes by Christophe Rhodes.
2043 * The fasl file format has changed again, to allow the compiler's
2044 INFO database to support symbol macros.
2045 * The user manual (in doc/) is formatted into HTML more nicely.
2046 (thanks to coreythomas)
2047 * The system is smarter about SUBTYPEP relationships, especially
2048 those involving NOT types (including types such as ATOM which are
2049 represented internally using NOT types). Thus SUBTYPEP is less
2050 likely to return (VALUES NIL NIL) in general, and in particular
2051 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2053 * The fasl file format has changed again, because the internal
2054 representation of types now includes a new slot to support the new
2055 SUBTYPEP-of-NOT-types logic.
2056 * (not a change in the main branch of SBCL, but a related prototype
2057 which can hopefully be merged into the main branch of SBCL in the
2058 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2059 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2060 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2061 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2062 dispatch macro character combinations. (thanks to Alexey Dejneka)
2063 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2064 that superficially look like logical namestrings correctly.
2065 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2067 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2068 default supplied-p) for &optional and &key arguments. (thanks to
2071 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2072 * incompatible change: The compiler is now less aggressive about
2073 tail call optimization, doing it only when (> SPACE DEBUG) or
2074 (> SPEED DEBUG). (This is an incompatible change because there are
2075 programs which relied on the old CMU-CL-style behavior to optimize
2076 away their unbounded recursion which will now die of stack overflow.)
2077 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2078 for non-GENCGC systems has been increased to 20M (since that
2079 seems much closer to the likely performance optimum for modern
2080 systems than the old 4M value was)
2081 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2082 and *DEBUG-PRINT-LEVEL*
2083 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2084 of CMU CL's support for SPARC, and various endianness and other
2085 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2086 * new syntactic sugar for the Unix command line: --load foo.bar is now
2087 an alternate notation for --eval '(load "foo.bar")'.
2089 ** The system now detects stack overflow and handles it gracefully,
2090 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2091 optimization settings. (This is a good thing in general, and
2092 its introduction in this version should be particularly timely
2093 for anyone whose code fails because of suppression of tail
2095 ** The system now hunts for the C variable "environ" in a more
2096 devious way, to avoid segfaults when the C library version differs
2097 between compile time and run time. (thanks to Christophe Rhodes)
2098 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2099 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2100 ** The compiler no longer issues bogus style warnings for undefined
2101 classes in the same source file as the DEFCLASSes which defined
2102 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2104 ** fixes in CONDITION class precedence list for undefined function
2105 errors (thanks to Alexei Dejneka)
2106 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2107 correctly. (thanks to Dan Barlow)
2108 ** portability fixes aiming at bootstrapping under CLISP (thanks
2109 to Dave McDonald and Christophe Rhodes)
2110 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2111 ** fixes in type translation and and type inference (thanks to
2113 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2114 ** various fixes in the new ports (thanks to Dan Barlow)
2115 * several changes related to debugging:
2116 ** suppression of tail recursion, as noted above
2117 ** stack overflow detection, as noted above
2118 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2119 is now the default. (For some time encapsulation has been more
2120 reliable than the breakpoint-based :ENCAPSULATE NIL
2121 implementation, at least on X86 systems; and I just noticed that
2122 encapsulation also seems closer to the spirit of the ANSI
2125 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2127 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2128 up properly again. (There was a packaging bug in 0.7.0 which
2129 left their definitions in SB-SYS::LOAD-FOREIGN and
2130 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2131 vital for most things which interface to C-level interfaces,
2132 like extensions working with sockets or databases or
2133 Perl-compatible regexes or whatever, and the need to fix
2134 this bug is the main reason that 0.7.1 was released so
2136 ** DEFGENERIC is now choosier about the methods it redefines, so that
2137 reLOADing a previously-LOADed file containing DEFGENERICs does
2138 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2139 cycle now works as it should. (thanks to Alexey Dejneka)
2140 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2141 fixed by Christophe Rhodes. (He actually submitted this patch
2142 months ago, and I delayed until after 0.7.0.)
2143 ** Bug 111 (internal compiler confusion about runtime checks on
2144 FUNCTION types) was fixed by Alexey Dejneka.
2145 * Some internal cleanups (getting rid of variables which aren't
2146 needed now that the byte interpreter is gone) caused the fasl
2147 file format number to change again.
2149 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2150 * major incompatible change: The default fasl file extension, i.e. the
2151 default extension for files produced by COMPILE-FILE, has changed
2152 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2154 ** There are many changes in the implementation of the compiler.
2155 SBCL is now essentially a compiler-only implementation of ANSI
2156 Common Lisp. EVAL still "interprets" a few special cases, but
2157 almost all the interesting cases are handled by creating
2158 a LAMBDA expression, calling COMPILE on it, then calling
2159 FUNCALL on the result.
2160 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2161 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2162 Since the code is newer, there might still be some new bugs
2163 (though not as many as before Martin Atzmueller's fixes:-). But
2164 the new code is substantially simpler and clearer, and hopefully
2165 any remaining bugs will be simpler, less fundamental, and more
2166 fixable then the bugs in the old code.
2167 ** The revised compiler is still a little unsteady on its feet.
2169 *** The debugging information it produces (particularly the names
2170 of FUNCTION objects) is sometimes much less useful than what
2171 the old compiler produced.
2172 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2173 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2174 within a MACROLET) has been temporarily weakened.
2175 ** There are new compiler optimizations for various functions:
2176 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2177 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2178 *** the math functions TRUNCATE, FLOOR, and CEILING
2179 *** the function-of-all-trades COERCE
2180 Mostly these should be transparent, but there's one
2181 potentially-annoying problem (bug 117): when the compiler
2182 inline-expands a function and does type analysis on the result,
2183 it can create control paths which have type mismatches, and
2184 when it can't prove that those control paths aren't taken,
2185 it will issue WARNINGs about the type mismatches. This is
2186 a particular problem in practice for the new sequence functions.
2187 It's not clear how this should be fixed, and for now, a
2188 workaround is given in the entry for 117 in the BUGS file.
2189 ** (Because of the interaction between the two previous items --
2190 occasional inlining problems and new inline expansions -- some
2191 of the new sequence function optimizations won't really kick in
2192 completely until debugging information, and then inlining, are
2193 straightened out in some future version.)
2194 * minor incompatible changes:
2195 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2196 DIRECTORY no longer implicitly promotes NIL slots of its
2197 pathname argument to :WILD. In particular, when you ask for the
2198 contents of a directory (which you used to be able to do without
2199 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2200 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2201 ** changes in behavior that ANSI explicitly defines to be
2202 implementation dependent:
2203 *** The new compiler-only implementation still conforms with ANSI,
2204 but acts a little different than before. Besides the obvious
2205 changes in performance tradeoffs (that the cost per form passed
2206 to EVAL has gone up, and the cost per form executed by EVAL
2207 has gone down), the behavior of the system changes a little
2208 because there are no longer any interpreted function objects.
2209 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2210 e.g. doing COMPILE on the output of interactive DEFUN is
2212 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2214 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2215 has changed from (:CL) to NIL.
2216 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2217 e.g. #\Newline and #\Tab, don't have names specified in the
2218 ANSI Common Lisp standard, is now based on their ASCII symbolic
2219 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2220 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2221 are no longer used for output.
2222 ** changes in internal implementation constants:
2223 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2224 4 million. (If your application spends a lot of time GCing and
2225 you have a lot of RAM, you might want to experiment with
2226 increasing it even more.)
2227 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2228 However, almost all old code should still continue to work without
2229 immediate update, as SB-C-CALL is now a (deprecated) nickname
2231 ** Old operator names in the style DEF-FOO are now deprecated in
2232 favor of new corresponding names DEFINE-FOO, for consistency with
2233 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2234 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2235 internal symbols, but a few supported extensions like
2236 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2237 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2238 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2239 etc. as you get deeper into recursive calls to the debugger
2240 command loop, instead of the old "5]", "5]]", "5]]]"
2241 sequence. (I was motivated to do this when squabbles between
2242 ILISP and SBCL left me very deeply nested in the debugger. In the
2243 short term, this change will probably provoke more ILISP/SBCL
2244 squabbles, but hopefully it will be an improvement in the long run.)
2245 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2246 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2247 function, since declaiming return types involving aliens is
2248 (1) annoyingly messy to do by hand and (2) vital to efficient
2249 compilation of code which calls such functions.
2250 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2251 longer reexported by the SB-EXT package. They're solely useful
2252 for alien code, so it seems more logical that you should get
2253 them from the SB-ALIEN package, not in SB-EXT.
2254 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2255 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2256 features. Instead, the code that they used to control is always
2257 built into the system.
2258 * many other bug fixes
2259 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2260 advantage of the new EVAL-WHEN stuff and to clean them up in
2261 general, and they are now more ANSI-compliant in a number of
2262 ways. Martin Atzmueller is responsible for a lot of this.
2263 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2265 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2266 *** correct ERROR type for various file operations
2267 *** some fixes for Lisp streams
2268 *** DEFMETHOD syntax checking
2269 *** changing old weird representation of debug information as
2270 strings (which, among their other deficiencies, don't transform
2271 correctly when you rename packages, and don't change their
2272 print representation when you change things like *PACKAGE*
2273 and *PRINT-LENGTH*) to symbols and lists of symbols
2274 He also made several improvements and fixed several bugs in DESCRIBE.
2275 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2277 *** misbehavior of WRITE-STRING/WRITE-LINE
2278 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2279 and several other LOOP problems as well
2280 *** DIRECTORY when similar filenames are present
2281 *** DEFGENERIC with :METHOD options
2282 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2283 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2284 *** argument ordering in FIND with :TEST option
2285 *** mishandled package designator argument in APROPOS-LIST
2286 *** various problems in the backquote readmacro
2287 *** a bug in APROPOS
2288 *** probably some others that I'm not describing very well here,
2289 since the CVS log documents them by reference to sbcl-devel
2290 messages, and the SourceForge archives aren't working well.:-(
2291 ** Dan Barlow improved the Alpha port (and is making progress on the
2292 PPC port, for those of you who think different).
2293 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2294 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2295 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2296 continues to work on the SPARC port (for those of you in a position
2297 to look down upon our little PC-compatible boxes from a great height).
2298 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2299 and uses that copy, rather than the current dynamic value, when
2300 it's trying to decide whether to truncate output. Thus e.g.
2301 (let ((*print-lines* 50))
2302 (pprint-logical-block (stream nil)
2304 (let ((*print-lines* 8))
2305 (print (aref possiblybigthings i) stream)))))
2306 should now truncate the logical block only at 50 lines, instead of
2307 often truncating it at 8 lines, as it did before.
2308 * The doc/cmucl/ directory, containing old CMU CL documentation
2309 from the time of the fork, is no longer part of the base system.
2310 SourceForge has shut down its anonymous FTP service, and with it
2311 my original plan for distributing the old CMU CL documentation
2312 there. For now, if you need these files you can download an old
2313 SBCL source release and extract them from it.
2314 * The fasl file version number changed again, for dozens of reasons,
2315 some of which are apparent above.
2317 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2318 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2319 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2320 patch, so that DISASSEMBLE output is much nicer.
2321 * The code in the SB-PROFILE package now seems reasonably stable.
2322 I still haven't decided what the final interface should look like
2323 (I'd like PROFILE to interact cleanly with TRACE, since both
2324 facilities use function encapsulation) but if you have a need
2325 for profiling now, you can probably use it successfully with
2326 the current CMU-CL-style interface.
2327 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2328 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2329 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2330 link now returns the dangling link itself, and for similar
2331 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2332 link itself. (In these cases the old code signalled an error and
2333 looped endlessly, respectively.) Thus, DIRECTORY now works even
2334 in the presence of dangling and cyclic symbolic links.
2335 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2336 is now a supported extension again, since the consensus on
2337 sbcl-devel was that it can be useful for ordinary development
2338 work, not just for debugging SBCL itself.
2339 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2340 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2341 that currently-defined functions might be redefined later with
2342 different return types.
2343 * Hash tables can be printed readably, as inspired by CMU CL code
2344 of Eric Marsden and SBCL code of Martin Atzmueller.
2345 * better error handling in CLOS method combination, thanks to
2346 Martin Atzmueller porting Pierre Mai's CMU CL patches
2347 * more overflow fixes for >16Mbyte I/O buffers
2348 * A bug in READ has been fixed, so that now a single Ctrl-D
2349 character suffices to cause end-of-file on character streams.
2350 In particular, now you only need one Ctrl-D at the command
2351 line (not two) to exit SBCL.
2352 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2354 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2355 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2356 does the right thing.
2357 * The implementation of some type tests, especially for CONDITION
2358 types, is now tidier and maybe faster, due to CMU CL code
2359 originally by Douglas Crosher, ported by Martin Atzmueller.
2360 * Some math functions have been fixed, and there are new
2361 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2362 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2363 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2364 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2365 likely to be built into sbcl-0.7.0.)
2366 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2367 CL, ported to SBCL by Martin Atzmueller)
2368 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2369 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2370 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2371 * There's a new slam.sh hack to shorten the edit/compile/debug
2372 cycle for low-level changes to SBCL itself, and a new
2373 :SB-AFTER-XC-CORE target feature to control the generation of
2374 the after-xc.core file needed by slam.sh.
2375 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2376 COMPILE-FILE is no longer supported, so that now every function
2377 gets an entry point, so that block compilation looks a little
2378 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2379 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2380 returns the number of bytes consed since the system started,
2381 rather than the number consed since the first time the function
2382 was called. (The new definition parallels ANSI functions like
2383 CL:GET-INTERNAL-RUN-TIME.)
2384 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2385 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2386 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2387 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2388 * Fasl file version numbers are now independent of the target CPU,
2389 since historically most system changes which required version
2390 number changes have affected all CPUs equally. Similarly,
2391 the byte fasl file version is now equal to the ordinary
2394 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2395 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2396 is no longer recognized. I apologize for this, because it was
2397 listed in SB-EXT as a supported extension, but I found that
2398 its existing behavior was poorly specified, as well as incorrectly
2399 specified, and it looked like too much of a mess to straighten it
2400 out. I have enough on my hands trying to get ANSI stuff to work..
2401 * many patches ported from CMU CL by Martin Atzmueller, with
2402 half a dozen bug fixes in pretty-printing and the debugger, and
2403 half a dozen others elsewhere
2404 * fixed bug 13: Floating point infinities are now supported again.
2405 They might still be a little bit flaky, but thanks to bug reports
2406 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2407 as flaky as they were.
2408 * The --noprogrammer command line option is now supported. (Its
2409 behavior is slightly different in detail from what the old man
2410 page claimed it would do, but it's still appropriate under the
2411 same circumstances that the man page talks about.)
2412 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2413 are now supported, and enabled by default. Thus, the compiler can
2414 handle many floating point and complex operations much less
2415 inefficiently. (Thus e.g. you can implement a complex FFT
2417 * The compiler now detects type mismatches between DECLAIM FTYPE
2418 and DEFUN better, and implements CHECK-TYPE more correctly, and
2419 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2420 thanks to patches from Martin Atzmueller.
2421 * various fixes to make the cross-compiler more portable to
2422 ANSI-conforming-but-different cross-compilation hosts (notably
2423 Lispworks for Windows, following bug reports from Arthur Lemmens)
2424 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2425 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2426 fixed thanks to Pierre Mai's CMU CL patches.
2427 * fixes to keep the system from overflowing internal counters when
2428 it tries to use i/o buffers larger than 16M bytes
2429 * fixed bug 45a: Various internal functions required to support
2430 complex special functions have been merged from CMU CL sources.
2431 (When I was first setting up SBCL, I misunderstood a compile-time
2432 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2433 * improved support for type intersection and union, fixing bug 12
2434 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2435 more obscure bugs as well
2436 * some steps toward byte-compiling non-performance-critical
2437 parts of the system, courtesy of patches from Martin Atzmueller
2438 * Christophe Rhodes has made some debian packages of sbcl at
2439 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2440 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2441 stable, but are nonetheless usable. When he's ready, I'd be happy
2442 to add them to the SourceForge "File Releases" section. (And if
2443 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2444 * new fasl file format version number (because of changes in
2445 internal representation of (OR ..) types to accommodate the new
2446 support for (AND ..) types, among other things)
2448 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2449 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2451 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2453 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2454 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2455 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2456 * DESCRIBE now works on structure objects again.
2457 * Most function call argument type mismatches are now handled as
2458 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2459 know whether the function will be redefined before the call is
2460 executed. (The compiler could flag local calls with full WARNINGs,
2461 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2462 it doesn't keep track of enough information to know whether calls
2463 are local in this sense.)
2464 * Compiler output is now more verbose, with messages truncated
2465 later than before. (There should be some supported way for users
2466 to override the default verbosity, but I haven't decided how to
2467 provide it yet, so this behavior is still controlled by the internal
2468 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2469 src/compiler/ir1util.lisp.)
2470 * Fasl file format version numbers have increased again, because
2471 support for the Gray streams extension changes the layout of the
2472 system's STREAM objects.
2473 * The Gray subclassable streams extension now works, thanks to a
2474 patch from Martin Atzmueller.
2475 * The full LOAD-FOREIGN extension (not just the primitive
2476 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2477 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2478 but like most other programs, it defaults to copying the Unix
2479 environment from the original process instead of starting the
2480 new process in an empty environment.
2481 * Extensions which manipulate the Unix environment now support
2482 an :ENVIRONMENT keyword option which doesn't smash case or
2483 do other bad things. The CMU-CL-style :ENV option is retained
2484 for porting convenience.
2485 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2486 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2488 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2490 * A patch from Martin Atzmueller seems to have solved the SIGINT
2491 problem, and as far as we know, signal-handling now works cleanly.
2492 (If you find any new bugs, please report them!)
2493 * The system no longer defaults Lisp source file names to types
2494 ".l", ".cl", or ".lsp", but only to ".lisp".
2495 * The compiler no longer uses special default file extensions for
2496 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2497 seems to expect a single default extension for all compiled code,
2498 and there's no compelling reason to try to stretch the standard
2499 to allow two different extensions.) Instead, byte-compiled files
2500 default to the same extension as native-compiled files.
2501 * Fasl file format version numbers have increased again, because
2502 a rearrangement of internal implementation packages made some
2503 dumped symbols in old fasl files unreadable in new cores.
2504 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2505 many fewer weird special cases.
2506 * Bug #17 (differing COMPILE-FILE behavior between logical and
2507 physical pathnames) has been fixed, and some related misbehavior too,
2508 thanks to a patch from Martin Atzmueller.
2509 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2510 by Tim Moore, ported to SBCL by Martin Atzmueller.
2511 * Martin Atzmueller fixed several filesystem-related problems,
2512 including bug #36, in part by porting CMU CL patches, which were
2513 written in part by Paul Werkowski.
2514 * More compiler warnings in src/runtime/ are gone, thanks to
2515 more patches from Martin Atzmueller.
2516 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2519 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2521 * DESCRIBE now works on CONDITION objects.
2522 * The debugger now handles errors which arise when trying to print
2523 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2525 * The build system now uses an additional file, customize-target-features.lisp,
2526 to allow local modifications to the target *FEATURES* list. (The point of
2527 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2528 debugging features enabled, without having to worry about propagating it
2529 into everyone's system when I do a "cvs update".) When no
2530 customize-target-features.lisp file exists, the target *FEATURES* list
2531 should be constructed the same way as before.
2532 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2533 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2534 warn about behavior which is undefined under the ANSI spec. Note: This
2535 is specified by ANSI, but it's not very popular with programmers.
2536 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2537 macro in the SBCL sources for an example of a workaround which you
2538 might use to make portable ANSI-standard code which does what you want.
2539 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2540 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2541 when it's used as a non-toplevel form. (This is required in order
2542 to implement the DEFCONSTANT-EQX macro.)
2543 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2544 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2545 funny things, due to bugs in the implementation of EVAL-WHEN
2546 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2547 non-toplevel DEFCONSTANTs are uncommon.)
2548 * The core file version number and fasl file version number have been
2549 incremented, because the old noncompliant DEFCONSTANT behavior involved
2550 calling functions which no longer exist, and because I also took the
2551 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2552 * fixed bug 1 (error handling before read-eval-print loop starts), and
2553 redid debugger restarts and related debugger commands somewhat while
2555 ** The QUIT debugger command is gone, since it did something
2556 rather different than the SB-EXT:QUIT command, and since it never
2557 worked properly outside the main toplevel read/eval/print loop.
2558 Invoking the new TOPLEVEL restart provides the same functionality.
2559 ** The GO debugger command is also gone, since you can just invoke
2560 the CONTINUE restart directly instead.
2561 ** The TOP debugger command is also gone, since it's redundant with the
2562 FRAME 0 command, and since it interfered with abbreviations for the
2564 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2566 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2567 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2568 some porting work by Martin Atzmueller)
2569 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2570 with a boatload of patches which he ported to SBCL. Now that those
2572 ** The system tries to make sure that its low-priority messages
2573 are prefixed by semicolons, to help people who like to use
2574 syntax highlighting in their ILISP buffer. (This patch
2575 was originally due to Raymond Toy.)
2576 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2577 patches originally written by Raymond Toy.
2578 ** The compiler understands coercion between single-value and
2579 multiple-VALUES type expressions better, getting rid of some very
2580 weird behavior, thanks to patches originally by Robert MacLachlan
2581 and Douglas Crosher.
2582 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2583 lambda lists, thanks to a patch originally by Pierre Mai.
2584 ** The system no longer bogusly warns about "abbreviated type
2586 ** The compiler gets less confused by inlining and RETURN-FROM,
2587 thanks to some patches originally by Tim Moore.
2588 ** The system no longer hangs when dumping circular lists to fasl
2589 files, thanks to a patch originally from Douglas Crosher.
2590 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2591 undefined function error.
2592 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2593 in structure slot initforms) for now, documented workaround instead:-|
2594 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2595 * fixed bug 5: added stubs for various Gray stream functions called
2596 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2597 installed, at least appropriate type errors are generated
2598 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2599 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2600 consistently in DEFMETHOD forms.
2601 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2602 it doesn't seem to affect SBCL after all
2603 * The C runtime system now builds with better optimization and many
2604 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2606 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2608 * The system is now under CVS at SourceForge (instead of the
2609 CVS repository on my home machine).
2610 * The new signal handling code has been tweaked to treat register
2611 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2612 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2613 downstream, e.g. in debug-int.lisp, has implicit dependencies
2614 on the unsignedness of integer representation of machine words,
2615 and that caused the system to bomb out with infinite regress
2616 when trying to recover from type errors involving signed values,
2617 e.g. (BUTLAST '(1 2 3) -1).)
2618 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2619 (This was one of the bugs Peter Van Eynde reported back in July.)
2620 * The system now uses code inspired by Colin Walters' O(N)
2621 implementation of MAP (from the cmucl-imp@cons.org mailing
2622 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2623 inline the MAP operation, and there is more than one
2624 sequence argument to the MAP call (so that it can't just
2625 do ETYPECASE once and for all based on the type of the
2626 single sequence argument). (The old non-inline implementation
2627 of the general M-argument sequence-of-length-N case required
2628 O(M*N*N) time when any of the sequence arguments were LISTs.)
2629 * The QUIT :UNIX-CODE keyword argument has been renamed to
2630 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2631 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2633 * Raymond Wiker's patches to port dynamic loading from Linux to
2634 FreeBSD have been added.
2635 * The BUGS file is now more nearly up to date, thanks in large part
2636 to Martin Atzmueller's review of it.
2637 * The debugger now flushes standard output streams before it begins
2638 its output ("debugger invoked" and so forth).
2639 * The core version number and fasl file version number have both
2640 been incremented, because of incompatible changes in the layout
2642 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2643 used to not be called for a saved Lisp image.)
2644 * Martin Atzmueller's version of a patch to fix a compiler crash,
2645 as posted on sbcl-devel 13 September 2000, has been installed.
2646 * Instead of installing Martin Atzmueller's patch for the
2647 compiler transform for SUBSEQ, I deleted the compiler transform,
2648 and transforms for some similar consing operations.
2649 * A bug in signal handling which kept TRACE from working on OpenBSD
2651 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2652 away by constant folding
2653 * The system now defines its address space constants in one place
2654 (in the Lisp sources), and propagates them automatically elsewhere
2655 (through GENESIS and the sbcl.h file). Therefore, patching the
2656 address map is less unnecessarily tedious and error-prone. The
2657 Lisp names of address space constants have also been systematized.
2658 * CVS tags like dollar-Header-dollar have been removed from
2659 the sources, because they have never saved me trouble and
2660 they've been source of trouble working with patches and other
2661 diff-related operations.
2662 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2663 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2665 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2667 * The system has been ported to OpenBSD.
2668 * The system now compiles with a simple "sh make.sh" on the systems
2669 that it's supported on. I.e., now you no longer need to tweak
2670 text in the target-features.lisp-expr and symlinks in src/runtime/
2671 by hand, the make.sh takes care of it for you.
2672 * The system is no longer so grossly inefficient when compiling code
2673 involving vectors implemented as general (not simple) vectors (VECTOR T),
2674 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2675 which dares to use the various sequence functions on non-simple
2676 vectors, takes less of a performance hit.
2677 * There is now a primitive type predicate VECTOR-T-P
2678 to test for the (VECTOR T) type, so that e.g.
2679 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2680 can now be compiled with some semblance of efficiency. (The old code
2681 turned the type declaration into a full call to %TYPEP at runtime!)
2682 * AREF on (VECTOR T) is still not fast, since it's still compiled
2683 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2684 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2685 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2686 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2687 without making full calls to SUBTYPEP at runtime.
2688 (Some analogous efficiency issues for non-simple vectors specialized to
2689 element types other than T, or for non-simple multidimensional arrays,
2690 have not been addressed. They could almost certainly be handled the
2691 same way if anyone is motivated to do so.)
2692 * The changes in array handling break binary compatibility, so
2693 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2694 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2695 instead of (VALUES T T).
2696 * By following the instructions that Dan Barlow posted to sbcl-devel
2697 on 2 July 2000, I was able to enable primitive dynamic object
2698 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2699 functionality is not implemented (since it calls ld to resolve
2700 library references automatically, requiring RUN-PROGRAM for its
2701 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2702 try to resolve library references) is now supported.
2703 * The system now flushes the standard output streams when it terminates,
2704 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2705 them at several other probably-convenient times, e.g. in each pass of
2706 the toplevel read-eval-print loop, and after evaluating a form given
2707 as an "--eval" command-line option. (These changes were motivated by a
2708 discussion of stream flushing issues on cmucl-imp in August 2000.)
2709 * The source transform for TYPEP of array types no longer assumes
2710 that an array whose element type is a not-yet-defined type
2711 is implemented as an array of T, but instead punts, so that the
2712 type will be interpreted at runtime.
2713 * There is now some support for cross-compiling in make.sh: each of
2714 the phases of make.sh has its own script. (This should be transparent
2715 to people doing ordinary, non-cross-compile builds.)
2716 * Since my laptop doesn't have hundreds of megabytes of memory like
2717 my desktop machine, I became more motivated to do some items on
2718 my to-do list in order to reduce the size of the system a little:
2719 ** Arrange for various needed-only-at-cold-init things to be
2720 uninterned after cold init. To support this, those things have
2721 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2722 symbols with such names are now uninterned after cold init).
2723 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2724 fasl files for cold load.
2725 ** Remove the old compiler structure pooling code (which used to
2726 be conditional on the target feature :SB-ALLOC) completely.
2727 ** Redo the representation of some data in cold init to be more compact.
2728 (I also looked into supporting byte compiled code at bootstrap time,
2729 which would probably reduce the size of the system a lot, but that
2730 looked too complicated, so I punted for now.)
2731 * The maximum signal nesting depth in the src/runtime/ support code has
2732 been reduced from 4096 to 256. (I don't know any reason for the very
2733 large old value. If the new smaller value turns out to break something,
2734 I'll probably just bump it back up.)
2735 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2737 * Many, many bugs reported by Peter Van Eynde have been added to
2738 the BUGS list; some have even been fixed.
2739 * While enabling dynamic object file loading, I tried to make the
2740 code easier to understand, renaming various functions and variables
2741 with less ambiguous names, and changing some function calling
2742 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2743 * While trying to figure out how to do the OpenBSD port, I tried to
2744 clean up some of the code in src/runtime/. In particular, I dropped
2745 support for non-POSIX signal handling, added various comments,
2746 tweaked the code to reduce the number of compilation warnings, and
2747 renamed some files to increase consistency.
2748 * To support the new automatic configuration functionality in make.sh,
2749 the source file target-features.lisp-expr has been replaced with the
2750 source file base-target-features.lisp-expr and the machine-generated
2751 file local-target-features.lisp-expr.
2752 * fixed a stupid quoting error in make.sh so that using CMU CL
2753 "lisp -batch" as cross-compilation host works again
2755 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2757 * DESCRIBE no longer tries to call itself recursively to describe
2758 bound/fbound values, so that it no longer fails on symbols which are
2759 bound to themselves (like keywords, T, and NIL).
2760 * DESCRIBE now works on generic functions.
2761 * The printer now prints less-screwed-up representations of closures
2762 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2763 * A private symbol is used instead of the :EMPTY keyword previously
2764 used to mark empty slots in hash tables. Thus
2765 (DEFVAR *HT* (MAKE-HASH-TABLE))
2766 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2767 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2768 now does what ANSI says that it should. (You can still get
2769 similar noncompliant behavior if bang on the hash table
2770 implementation with all the symbols you get back from
2771 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2772 This breaks binary compatibility, since tests for equality to
2773 :EMPTY are wired into things like the macroexpansion of
2774 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2776 * There's now a minimal placeholder implementation for CL:STEP,
2777 as required by ANSI.
2778 * An obscure bug in the interaction of the normal compiler, the byte
2779 compiler, inlining, and structure predicates has been patched
2780 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2781 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2783 * Missing ordinary arguments in a macro call are now detected even
2784 when the macro lambda list contains &KEY or &REST.
2785 * The debugger no longer complains about encountering the top of the
2786 stack when you type "FRAME 0" to explicitly instruct it to go to
2787 the top of the stack. And it now prints the frame you request even
2788 if it's the current frame (instead of saying "You are here.").
2789 * As specified by ANSI, the system now always prints keywords
2790 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2792 * The default initial SIZE of HASH-TABLEs is now smaller.
2793 * Type information from CLOS class dispatch is now propagated
2794 into DEFMETHOD bodies, so that e.g.
2795 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2797 is now basically equivalent to
2798 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2799 (DECLARE (TYPE SINGLE-FLOAT X))
2801 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2802 operation, without having to do run-time type dispatch.
2803 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2804 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2805 * Since I don't seem to be making big file reorganizations very often
2806 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2807 is overflowing my ability to conveniently back them up), I've finally
2808 checked the system into CVS. (The CVS repository is on my home system,
2809 not at SourceForge -- putting it on SourceForge might come later.)
2810 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2811 high-level GC-NOTIFY-FOO functions send their output. (There's
2812 still very little control of where low-level verbose GC functions
2813 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2814 less than it used to -- the GC-NOTIFY-FOO functions are now under
2815 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2816 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2817 in only one place in the source code, and propagates it automatically
2818 everywhere that it's needed. Thus e.g. when I bump the version from
2819 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2820 * The C source files now include boilerplate legalese and documentation
2821 at the head of each file (just as the Lisp source files already did).
2822 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2823 to his page will be replaced with a link to his new CLiki service.
2825 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2827 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2828 * The build process now looks for GNU make under the default name "gmake",
2829 instead of "make" as it used to. If GNU make is not available as "gmake"
2830 on your system, you can change this default behavior by setting the
2831 GNUMAKE environment variable.
2832 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2833 can build without error under CMU CL.
2835 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2837 * There is now a partial SBCL user manual (with some new text and some
2838 text cribbed from the CMU CL manual).
2839 * The beginnings of a profiler have been added (starting with the
2840 CMU CL profiler and simplifying and cleaning up). Eventually the
2841 main interface should be through the TRACE macro, but for now,
2842 it's still accessed through vaguely CMU-CL-style functions and macros
2843 exported from the package SB-PROFILE.
2844 * Some problems left over from porting CMU CL to the new
2845 cross-compilation bootstrap process have been cleaned up:
2846 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2847 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2848 part when getting rid of the compiler *BACKEND* stuff.)
2849 ** The value of *NULL-TYPE* was screwed up, because it was
2850 being initialized before the type system knew the final
2851 definition of the 'NULL type. This screwed up several key
2852 optimizations in the compiler, causing inefficiency in all sorts
2853 of places. (I found it because I wanted to understand why
2854 GET-INTERNAL-RUN-TIME was consing.)
2855 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2856 PROCLAIM FTYPE information. Unfortunately this broke binary
2857 compatibility again, since now the forms output by DEFGENERIC
2858 to refer to functions which didn't exist in 0.6.3.
2859 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2860 can use the new (as of 0.6.3) transform for SOME into MAP into
2862 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2863 values to (INTEGER 0 1000000), so that the time code will no longer
2864 occasionally get blown up by Linux returning 1000000 microseconds
2865 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2866 its output conform to the ANSI spec. (Alas, this makes its output
2867 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2868 * A full call to MAP NIL with a single sequence argument no longer conses.
2869 * fixes to problems pointed out by Martin Atzmueller:
2870 * The manual page no longer talks about multiprocessing as though
2871 it were currently supported.
2872 * The ILISP support patches have been removed from the distribution,
2873 because as of version 5.10.1, ILISP now supports SBCL without us
2874 having to maintain patches.
2875 * added a modified version of Raymond Toy's recent CMU CL patch for
2876 EQUALP comparison of HASH-TABLE
2878 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2880 * The system still can't cross-compile itself with
2881 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2882 entails), but at least it can compile more of itself that way
2883 than it used to be able to, and various buglets which were uncovered
2884 by trying to cross-compile itself that way have now been fixed.
2885 * This release breaks binary compatibility again. This time
2886 at least I've incremented the FASL file format version to 2, so that the
2887 problem can be detected reliably instead of just causing weird errors.
2888 * various new style warnings:
2889 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2890 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2891 ** using the lexical binding of a variable named in the *FOO* style
2892 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2893 as specified by ANSI.
2894 * *RANDOM-STATE* is no longer automatically initialized from
2895 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2896 default behavior of the system is to repeat its behavior every time
2897 it's run. If you'd like to change this behavior, you can always
2898 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2899 old convention there was no comparably easy way to get the system to
2900 repeat its behavior every time it was run.
2901 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2902 FTYPE declarations has been removed, in favor of their ANSI
2903 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2904 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2905 the types of their sequence arguments can be determined at compile time.
2906 This is done through a new open code expansion for MAP which eliminates
2907 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2908 argument types can be determined at compile time.
2909 * The optimizer now transforms COERCE into an identity operation when it
2910 can prove that the coerced object is already of the correct type. (This
2911 can be a win for machine generated code, including the output of other
2912 optimization transforms, such as the MAP transform above.)
2913 * Credit information has been moved from source file headers into CREDITS.
2914 * Source file headers have been made more standard.
2915 * The CASE macro now compiles without complaining even when it has
2918 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2920 * (Note that the way that the PCL macroexpansions were rewritten
2921 to accommodate the change in DEFGENERIC below breaks binary
2922 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2923 not run under sbcl-0.6.2. Once we get out of alpha releases,
2924 i.e. hit release 1.0.0, we'll probably try to maintain binary
2925 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2926 and sbcl-1.4.4. Until then, however, it might be fairly common
2927 for maintenance releases to break binary compatibility.)
2928 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2930 * The reporting of unhandled signals has been changed to print some
2931 explanatory text as well as the report form. (Previously only
2932 the report form was printed.)
2933 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2934 it defines, so that the compiler no longer issues undefined function
2935 warnings for compiled-but-not-yet-loaded generic functions.
2936 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2937 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2938 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2939 section of the ANSI spec are supported.
2940 * The "" nickname for the "KEYWORD" package has been removed.
2941 The reader still handles symbol tokens which begin with a package marker
2942 as keywords, but it doesn't expose its mechanism for doing so in the
2943 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2944 * The system now issues STYLE-WARNINGs for contradictory TYPE
2945 proclamations. (Warnings for contradictory FTYPE proclamations would
2946 be nice too, but those can't be done usefully unless the type system
2947 is made smarter about FUNCTION types.)
2948 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2949 systematized, so that "*target-*.lisp is supposed to exist only on the
2950 target and imply that there's a related file which exists on the
2951 host, and *host-*.lisp is supposed to exist only on the host and imply
2952 that there's a related file which exists on the target. This involves a
2953 lot of renaming. Hopefully the acute confusion caused by the renaming
2954 will be justified by the reduction in chronic confusion..
2955 ** runtime-type.lisp -> early-target-type.lisp
2956 ** target-type.lisp -> late-target-type.lisp
2957 ** early-host-format.lisp -> early-format.lisp
2958 ** late-host-format.lisp -> late-format.lisp
2959 ** host-error.lisp -> misc-error.lisp
2960 ** early-error.lisp -> early-target-error.lisp
2961 ** late-error.lisp -> late-target-error.lisp
2962 ** host-defboot.lisp -> early-defboot.lisp
2963 ** code/misc.lisp -> code/target-misc.lisp
2964 ** code/host-misc.lisp -> code/misc.lisp
2965 ** code/numbers.lisp -> code/target-numbers.lisp
2966 ** code/early-numbers.lisp -> numbers.lisp
2967 ** early-host-type.lisp -> early-type.lisp
2968 ** late-host-type.lisp -> late-type.lisp
2969 ** host-typep.lisp -> typep.lisp
2970 ** load.lisp -> target-load.lisp
2971 ** host-load.lisp -> load.lisp
2972 ** host-disassem.lisp -> disassem.lisp
2973 ** host-insts.lisp -> insts.lisp
2974 ** byte-comp.lisp -> target-byte-comp.lisp
2975 ** host-byte-comp.lisp -> byte-comp.lisp
2976 ** host-signal.lisp -> signal.lisp
2977 ** host-defstruct.lisp -> defstruct.lisp
2978 ** late-target-type.lisp -> deftypes-for-target.lisp
2979 Furthermore, several other previously target-only files foo.lisp (e.g.
2980 hash-table.lisp and random.lisp) have been split into a target-and-host
2981 foo.lisp file and a target-only target-foo.lisp file, with their key type
2982 definitions in the target-and-host part, so that the cross-compiler will
2983 know more about target types.
2984 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2985 gone away. In their place are various *BACKEND-FOO* variables
2986 corresponding to the slots of the old structure.
2987 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2988 into the target SBCL has been fixed.
2989 * The chill.lisp system for loading cold code into a running SBCL
2991 * Support for the CMU CL "scavenger hook" extension has been removed.
2992 (It was undocumented and unused in the CMU CL sources that SBCL was
2993 derived from, and stale in sbcl-0.6.1.)
2994 * Various errors in the cross-compiler type system were detected
2995 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2996 (enabling various consistency checks). Many of them were fixed,
2997 but some hard problems remain, so the compiler is back to
2998 running without *TYPE-SYSTEM-INITIALIZED* for now.
2999 * As part of the cross-compiler type system cleanup, I implemented
3000 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3001 * I have started adding UNCROSS calls throughout the type system
3002 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3003 on cross-compiler input files will be able to go away, and various
3005 * CONSTANTP now returns true for quoted forms (as explicitly required
3008 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3010 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3011 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3012 (MAKE-LIST -1). In the longer term, it ought to become true
3013 that declarations are assertions even at SAFETY 1. For now, it's not
3014 quite true even at SAFETY 3, but it's at least more nearly true..
3015 (Note that this change seems to increases the size of the system by
3016 O(5%) and to decrease the speed of the compiler by 20% or more.)
3017 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3018 for the problem of printing dozens of lines of distracting information
3019 about low-level system machinery as part of the top stack frame
3020 on entry to the debugger when an undefined function was called.
3021 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3022 is not reset to COMMON-LISP-USER.
3023 * Compilation of stuff related to dyncount.lisp has been made conditional
3024 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3025 smaller. The various dyncount-related symbols have been moved into
3026 a new "SB-DYNCOUNT" package.
3027 * tty-inspect.lisp has been renamed to inspect.lisp.
3028 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3029 feature has gone away. (When we eventually port to other flavors of
3030 libc and/or Unix, we'll try to make the differences between flavors
3031 invisible at the user level.)
3032 * Various other *FEATURES* tags, and/or their associated conditionals,
3033 have been removed if obsolescent, or given better documentation, or
3034 sometimes given more-mnemonic names.
3036 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3038 * tidied up "make.sh" script
3039 * tidied up system directory structure
3040 * better "clean.sh" behavior
3041 * added doc/FOR-CMUCL-DEVELOPERS
3042 * many many small tweaks to output format, e.g. removing possibly-confusing
3043 trailing #\. character in DESCRIBE-INSTANCE
3044 * (EQUALP #\A 'A) no longer signals an error.
3045 * new hashing code, including EQUALP hashing
3046 * tidied up Lisp initialization and toplevel
3047 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3048 * command line argument processing
3049 * added POSIX-GETENV function to deal with Unix-ish environment variables
3050 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3051 terminating SBCL on EOF
3052 * non-verbose GC by default
3053 * There is no more "sbcl" shell script; the sbcl file is now the C
3054 runtime executable (just like CMU CL).
3055 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3057 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3058 symbol and package information as Lisp native symbol and package objects
3059 instead of strings naming symbols and strings naming packages. This way,
3060 whenever packages are renamed (as in warm init), debug information is
3061 transformed along with everything else.
3062 * tweaked the optimization policy declarations which control the building
3063 of SBCL itself. Now, among other things, the system no longer saves
3064 source location debugging information. (This helps two problems at once
3065 by reducing SBCL size and by keeping SBCL from trying to look for its
3066 sources -- which may not exist -- when reporting errors.)
3067 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3068 debugging and testing purposes
3069 * cleaned up printing, making the printer call PRINT-OBJECT for
3070 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3071 methods, giving nearly-ANSI behavior
3072 * converted almost all special variables to use *FOO* naming convention
3073 * deleted PARSE-TIME functionality, since it can be done portably
3074 * moved some files out of cold init into warm init
3075 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3077 * regularized formatting of source files
3078 * added an install.sh script
3079 * fixed ridiculous memory usage of cross-compiler by making
3080 compiler/alloc.lisp not try to do pooling unless it can hook
3081 itself into the GC of the cross-compilation host. Now the system
3082 builds nicely on my old laptop.
3083 * added :SB-ALLOC in target-features.lisp-expr
3084 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3085 was not implemented)
3086 * re-did condition handling and note reporting in the compiler. Notes
3087 are no longer handled by signalling conditions. Style warnings
3088 and warnings are handled more correctly and reported in such a way
3089 that it's easy to find one or the other in your output (so that you
3090 can e.g. figure out which of many problems caused COMPILE-FILE to
3092 * changed the severity of several compiler warnings from full WARNING
3093 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3094 compiler note reporting so that it doesn't use the condition system
3095 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3096 COMPILE-FILE command)
3097 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3098 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3099 no longer supported, since their implementation was deeply intertwingled
3100 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3101 * removed bogus "support" for compiler macros named (SETF FOO), and
3102 removed the compiler macro for SETF INFO (but only after making a fool
3103 of myself on the cmucl-imp mailing list by posting a bogus patch for
3104 DEFINE-COMPILER-MACRO..)
3105 * Compiled files containing forms which have side effects on the Lisp
3106 reader (such as DEFPACKAGE forms) are now handled more correctly.
3107 (Compiler queuing of top level lambdas has been suppressed by setting
3108 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3109 * deleted various currently-unused source files, e.g. gengc.lisp. They
3110 may be added back at some point e.g. when porting to other architectures,
3111 but until they are it's distracting to distribute them and to try to
3113 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3114 there were so many of them they're just distractions, and UNCROSS is
3115 known to be able to handle the current sources
3116 * moved PROFILE functionality into TRACE, so that it will be clear
3117 how the wrapping and unwrapping of functions when you profile them
3118 interacts with the wrapping and unwrapping of functions when you
3119 trace them. (Actually, the functionality isn't there yet, but at least
3120 the interface specification is there. Hopefully, the functionality will
3121 arrive with some future maintenance release.)
3122 * removed host-oops.lisp
3123 * changed signature of QUIT function to allow UNIX-CODE argument
3124 * fixed READ-SEQUENCE bug
3125 * tweaked verbose GC output so that it looks more like the progress
3126 output that ANSI specifies for functions like LOAD
3127 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3128 * added <http://sbcl.sourceforge.com> to the banner information printed by