1 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2 * numerous signal handling fixes to increase stability
3 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
4 * minor incompatible change: we now correctly canonize default
5 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
6 as per the MOP, rather than the historical (INITARG INITFUNCTION
7 INITFORM). (reported by Bruno Haible)
8 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
9 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
11 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
12 pathname is a directory pathname.
13 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
14 * dynamic space size on PPC has been increased to 768Mb. (thanks to
16 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
17 :METACLASS argument in addition to a class name. (reported by
18 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
19 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
20 (reported by Bruno Haible; patch from Gerd Moellmann)
21 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
23 * bug fix: more cleanups to the floating point exception handling on
24 x86-64 (thanks to James Knight)
25 * bug fix: the compiler does not try to constant fold calls of
26 COERCE and BIT-* functions when they return freshly constructed
27 objects. (reported by Nikodemus Siivola and Paul Dietz)
28 * optimization: improved the MIPS versions of generic (in the
29 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
30 * optimization: direct conversion of (unsigned-byte 32) to floats on
32 * optimization: structure instances with raw slots now use less
33 memory, and probably show better memory locality. (thanks to
35 * optimization: DYNAMIC-EXTENT declarations for lists and closures
36 are treated as requests for stack allocation on the x86-64,
37 Alpha-32, PPC and SPARC platforms.
38 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
39 the wrong file descriptor; implementation of SOCKET-OPEN-P.
40 (thanks to Tony Martinez)
42 ** gcing a dead thread can no longer lead to lockups
43 ** threads block signals until they are set up properly
44 ** errno is no longer shared by threads
45 ** interrupt-thread restores the eflags register on x86
46 ** fixed some lockups due to gc/thread interaction
47 * fixed some bugs revealed by Paul Dietz' test suite:
48 ** invalid dotted lists no longer raise a read error when
50 ** signal an error if a symbol that names a declaration is used
51 as the name of a type, or vice versa
52 ** allow using the (declare (typespec var*)) abbreviation for
53 (declare (type typespec var*)) with all type specifiers
54 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
55 ** always return NIL from PPRINT-POP when OBJECT is NIL
56 ** don't signal errors when pretty-printing malformed LABELS,
57 FLET or MACROLET forms
58 ** declarations in a DOLIST body are also in scope for the
60 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
62 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
65 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
66 * fixed cross-compiler leakages that prevented building a 32-bit
67 target with a 64-bit host compiler.
68 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
69 opened with :IF-EXISTS :OVERWRITE.
70 * fixed bug 281: error for an invalid qualifier in a short-form method
71 combination method is not signalled until the faulty method is called.
72 * bug fix: iteration variable type inferrer failed to deal with open
73 intervals. (reported by Alan Shields)
74 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
75 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
76 or more runtime options were provided to the sbcl binary.
77 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
78 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
79 * optimization: there's now a fast-path for fixnum arguments in the
80 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
81 * optimization: the code generated on x86-64 is more compact thanks
82 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
83 * bug fix: floating-point exception handling now partly works on
84 x86-64. (Thanks to James Knight)
85 * improvement to the MIPS backend from Thiemo Seufer: C-style
86 64-bit long long arguments and return values to alien functions
88 * fixed some bugs revealed by Paul Dietz' test suite:
89 ** the type-error signalled from WARN has a filled-in DATUM slot.
90 ** the type-error required when a stream is not associated with
91 a file has the stream as its datum.
92 ** type-errors on single-floats on x86-64 no longer have
93 :INVALID-OBJECT as the datum
94 ** the type-errors signalled for invalid function names now have
95 a correct expected type
96 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
97 for structures defined with a :TYPE. Documentation strings for
98 typed structures are no longer immediately discarded
99 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
100 broadcast streams and synonym streams. FILE-LENGTH now also works
101 on broadcast streams.
103 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
104 * incompatible change: the --noprogrammer option, deprecated since
105 version 0.7.5, has been removed. Please use the equivalent
106 --disable-debugger option instead.
107 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
109 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
110 inoperational for a while now) has been completely removed.
111 * null lexical environments are now printed as #<NULL-LEXENV>,
112 significantly reducing the amount of clutter in typical backtraces.
113 * documentation on weak pointers, finalization, and after GC hooks
114 has been added to the manual.
115 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
116 as other specialized array types, is much faster. SUBSEQ and
117 COPY-SEQ on such arrays have also been sped up.
118 * optimization: EQL is now more efficient when at least other argument
119 is known to be of type (OR FIXNUM (NOT NUMBER)).
120 * fixed bug: the runtime is now less vulnerable to changes in the
121 size of the SBCL object on OS X, and virtual memory is reserved for
122 all spaces that need to be at a fixed address.
123 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
124 * fixed bug: finalizers and after GC hooks that cause consing are now
126 * fixed bug: compiler error messages and summaries are now printed to
127 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
128 * fixed inference of the upper bound of an iteration variable.
129 (reported by Rajat Datta).
130 * fixed bug 211e: calling local functions with duplicated constant
131 keyword argument no longer causes a bogus style warning about an
133 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
134 ftype declaration to be disregarded. (reported by Dave Roberts)
135 * fixed bug 373: caused by erronous compilation of references to alien
136 variables in the runtime on ppc/darwin.
137 * fixed bug 376: CONJUGATE type deriver.
138 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
139 (reported by Baughn on #lisp)
140 * a cross-compiler bug on non-x86 platforms has been identified and
141 fixed. (thanks to Bruno Haible)
142 * improvements to the MIPS runtime code for increased stability.
143 (thanks to Thiemo Seufer)
144 * increased the maximimum compact environment size to allow
145 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
146 * partial workaround for bug 135: don't name the function we're
147 calling for hairy cases of EVAL, so as not to accumulate one environment
148 entry per call to EVAL. (reported by Kevin Reid)
149 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
150 * optimization: type testing for non-vector arrays should be faster.
151 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
153 * bug fix: setting 31st element of a bit vector to zero did not work
155 * bug fix: redefining a class definition which failed due to a
156 previous accessor / function clash now works (but see BUGS entry
157 #380 for more problems in this area). (thanks to Zach Beane)
158 * the long form of DEFINE-METHOD-COMBINATION disables method group
159 checking when given a single method group with pattern *. (thanks
161 * on x86 compiler supports stack allocation of results of simple
162 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
163 * support for building and running on Mac OS X 10.4 "Tiger" has been added
164 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
165 * fixed some bugs related to Unicode integration:
166 ** the restarts for recovering from input and output encoding
167 errors only appear when there is in fact such an error to
169 * fixed some bugs revealed by Paul Dietz' test suite:
170 ** MISC.549 and similar: late transformation of unsafe type
171 assertions into derived types caused unexpected code
173 ** SCALE-FLOAT type deriver is less wrong.
174 ** type derivers for EXP, LOG and similar functions compute result
175 types for complex arguments better.
176 ** (MISC.563) CONJUGATE type deriver works for very restricted
178 ** out-of-line type testers for character strings are available.
179 ** EQUAL compiler transform understands specialness of objects
181 ** accessing double-floats stored on the stack now works on x86-64.
182 ** debugger internals could sometimes create invalid lispobjs,
183 resulting in GC crashes.
184 ** MISC.548: type check weakening can convert required type into
186 ** initialization forms for bindings are not in scope of free special
189 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
190 * incompatible change: thread support for non-NPTL systems has
191 been removed - locking is buggy and unreliable. A threaded
192 SBCL build will now warn at startup and refuse to create
193 new threads, unless futex support is detected in the kernel
194 * incompatible change: the top level REPL now has only an ABORT
195 restart associated with it, not TOPLEVEL and ABORT as it used to.
196 TOP and TOPLEVEL are now available as debugger commands for
197 returning to the top level.
198 * incompatible change: forms evaluated in the REPL now use the
199 global optimization policy.
200 * incompatible change: user- and system-initialization files are
201 no longer processed with LOAD, but by READ and EVAL; hence the
202 global optimization policy, startup package, readtable, etc,
204 * The .fasl file format number has been incremented because of
205 various incompatible changes.
206 * internal entry point details and argument counts no longer appear
207 in backtraces unless explicitly requested by setting
208 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
209 * built-in and standard functions no longer have names like "top
210 level local call to FOO".
211 * fixed bug 32: functions defined in non-null lexical environments
212 now have more legible printed representation
213 * fixed bug 33: functions defined in non-null lexical environemnts
214 are now more amenable to inspection by INSPECT.
215 * workaround for bug 354: XEPs no longer appear in backtraces unless
216 explicitly requested.
217 * fixed bug: receiving the signal which results from attempting to
218 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
219 SIGBUS on other *BSDs) on architectures where the C stack is also
220 the Lisp stack (x86 and x86-64) from foreign code no longer leads
221 to debugger confusion or wild instability. (reported by Cheuksan
223 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
224 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
225 (reported by Lutz Euler)
226 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
227 compiled code. (reported by Paul Dietz)
228 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
229 specialized parameter to a method to something that is not TYPEP
230 the specializer is now possible.
231 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
232 face of package deletion.
233 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
234 pathname. (reported by David Tolpin on c.l.l)
235 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
236 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
237 STRING-STREAM and FILE-STREAM, respectively.
238 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
239 than previously; in addition, its test suite is now run on build.
240 (thanks to Luke Gorrie)
241 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
242 * optimization: added a immediate representation for single-floats
244 * fixed some bugs revealed by Paul Dietz' test suite:
245 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
246 ** attempting to create a package with a colliding nickname causes
247 correctable errors to be signalled.
248 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
249 ** several x86-64 backend bugs related to sign-extension of immediate
252 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
253 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
254 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
256 * fixed loading of multiply forward-referenced layouts.
257 (thanks to Cheuksan Wang)
258 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
259 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
260 (see COMPILE-FILE documentation for details of :PRINT option)
261 * fixed bugs 19 and 317: fixed-format floating point printing is
262 more accurate. This also fixes a bug reported by Adam Warner
263 related to the ~@F format directive.
264 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
266 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
267 dispatch table if the from-char is a dispatch macro character.
268 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
269 by the wordsize no longer ignore the last word. (reported by Lutz
271 * fixed bug in type checking of dynamic-extent variables. (reported
273 * optimization: sequence traversal functions use their freedom to
274 coerce function designators to functions.
275 * optimization: code with many calls to CLOS methods specialized on
276 CLOS classes has had redundant type checks removed.
277 * optimization: added declarations to speed up operations that access
278 the internal character database (for example STRING-UPCASE)
279 * optimization: comparison operations between floats and sufficiently small
280 fixnums no longer create extra rationals
281 * fixed some bugs related to Unicode integration:
282 ** portions of multibyte characters at the end of buffers for
283 character-based file input are correctly transferred to the
284 start of the buffer at the next read.
285 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
286 passing it through to OPEN.
287 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
288 argument given to internal calls to OPEN.
289 ** the built-in comment readers (introduced by character sequences
290 ";" and "#|") are more forgiving to encoding errors; they will
291 STYLE-WARN and then attempt to resync the stream at a character
292 boundary. (thanks to Teemu Kalvas)
293 * fixed some bugs in the x86-64 port:
294 ** Negative short int return values from c-calls are sign-extended
296 ** The stack is aligned to 16-bytes for c-calls, as required by
297 the ABI. (thanks to Cheuksan Wang)
298 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
299 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
300 for enums. (thanks to Vincent Arkesteijn)
301 ** Multiple small optimizations and bugfixes for floating point
303 * fixed some bugs revealed by Paul Dietz' test suite:
304 ** Space, Tab, Linefeed, Return and Page have the invalid
305 secondary constituent character trait.
306 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
308 ** WITH-INPUT-FROM-STRING should only update the index place on
310 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
312 ** Bit-array operations (BIT-AND and similar) worked incorrectly
313 with one-dimensional arrays with fill pointers.
314 ** TYPE-OF failed on a complex with an integer realpart and a
316 ** compiler failure during type inference for the code of form
317 (IF (EQL X (THE ...)) ...) (MISC.535).
319 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
320 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
321 platform. The port supports 61-bit fixnums, large memory spaces
322 and reloading shared object files.
323 * enhancement: saving cores with foreign code loaded is now
324 supported on ppc/Darwin in addition to the previously supported
326 * enhancement: the statistical profiler now walks deeper into the
327 call stack for more meaningful call-graphs and accrued time
328 reports (x86/x86-64 only). It also now reports time spent in
330 * enhancement: it is now possible to trace most individual methods
331 of a generic function in addition to tracing the generic function
333 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
335 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
336 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
337 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
338 produces an error. (thanks to Vincent Arkesteijn)
339 * bug fix: NAMESTRING on pathnames with :WILD components in their
340 directories works correctly. (thanks to Artem V. Andreev)
341 * fixed bug 125: compiler preserves identity of closures. (reported
343 * bug fixed: functions with &REST arguments sometimes failed with
344 "Undefined function" when compiled with (DEBUG 3). (reported by
346 * bug fix: overflow during compiling of setting element of a bit
347 vector with constant index and value. (reported by Timmy Douglas)
348 * build fix: fixed the dependence on *LOAD-PATHNAME* and
349 *COMPILE-FILE-PATHNAME* being absolute pathnames.
350 * on x86 compiler partially supports stack allocation of dynamic-extent
352 * GO and RETURN-FROM do not check the extent of their exit points
353 when compiled with SAFETY 0.
354 * fixed some bugs related to Unicode integration:
355 ** encoding and decoding errors are now much more robustly
356 handled; it should now be possible to recover even from invalid
357 input or output to the terminal. (thanks to Teemu Kalvas)
358 ** provided a first cut at implementing STRING-TO-OCTETS and
359 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
360 ** altered the SB-MD5 contributed module slightly, changing the
361 interface just enough for it to be supportable for builds where
362 lisp characters are not eight bits.
363 * fixed some bugs revealed by Paul Dietz' test suite:
364 ** the FORMATTER-generated functions for ~V[ conditionals require
365 the correct number of arguments.
366 ** READ-FROM-STRING returns the mandated second value when applied
367 to displaced strings.
368 ** the #\Rubout and #\Backspace characters are treated as invalid
369 constituent characters by the tokenizer.
371 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
372 * new feature: reloading changed shared object files with
373 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
374 * new feature: references to foreign variables and functions
375 can now be compiled and loaded before the corresponding shared
376 object file is loaded, as long as the foreign definitions are
377 available at runtime.
378 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
379 supported. (thanks to Dan Debertin)
380 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
381 just keywords. (thanks to Vincent Arkesteijn)
382 * fixed bug #331: structure-class instances corresponding to
383 DEFSTRUCT forms are now created eagerly.
384 * fixed bug #345: backtraces from calls to undefined functions work
385 on x86 as well. Related bug #61 is now also partially fixed on x86
386 (backtraces from throws to unknown catch tags.)
387 * bug fix: lambda-list parsing is now stricter vrt. order and number
388 of lambda-list keywords.
389 * bug fix: as specified by AMOP, an error is signalled if a
390 class-option appears multiple times in a DEFCLASS form. (reported
392 * bug fix: RANDOM can be compiled when the compiler derives the type
393 of its numeric argument as a disjoint set of small integers.
394 (reported by Paul Dietz)
395 * bug fix: starting a core saved with shared objects loaded when
396 those objects are not available no longer causes threaded SBCL to
397 hang. (reported by Sean Ross)
398 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
400 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
401 argument, which must always be NIL. (reported by Kalle Niemitalo)
402 * bug fix: printing 1.0d+23 no longer results in an error.
403 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
404 * bug fix: structure slot setters preserve evaluation order. (thanks
406 * fixed some bugs related to Unicode integration:
407 ** RUN-PROGRAM can allow its child to take input from a Lisp
408 stream. (reported by Stefan Scholl)
409 ** ASDF-INSTALL successfully downloads .tar.gz files in all
410 locales. (reported by Ken Causey)
411 * fixed some bugs revealed by Paul Dietz' test suite:
412 ** INCF, DECF and REMF evaluate their place form as specified in
414 ** FORMATTER expands ~{ ~} iteration directives with V or #
415 parameters correctly.
416 ** FORMATTER deals with the ~@[ ~] conditional directive where the
417 consequent uses no arguments correctly.
418 ** the system has a partial understanding of the (COMPLEX RATIO)
420 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
423 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
424 * new feature: a build-time option (controlled by the :SB-UNICODE
425 keyword feature, enabled by default) for building the system with
426 support for the entire 21-bit character space defined by the
428 * new feature: the system now has rudimentary external-format
429 support; the primary user-visible change at this time is that
430 characters with the high bit set (such as lower-case-e-acute) will
431 print correctly to a terminal in a UTF-8 environment.
432 * minor incompatible change: BASE-CHAR no longer names a class;
433 however, CHARACTER continues to do so, as required by ANSI.
434 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
435 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
436 has been moved to the SB-EXT package (temporarily re-exported from
438 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
439 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
441 * fixed bug #351: better error-handlind and reporting for malformed
443 * fixed bug #350: bignum-printing is now more memory-efficient,
444 allowing printing of very large bignums, eg. (expt 2 10000000).
445 (reported by Bruno Haible)
446 * fixed bug #302: better primitive-type selection for intersection
448 * fixed bug #308: non-graphic characters now all have names, as
449 required. (reported by Bruno Haible)
450 * bug fix: redefining a class with different superclasses now correctly
451 removes it from the direct-subclasses of its previous superclasses.
452 (reported by David Morse)
453 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
454 new value now works. (reported by Bruno Haible)
455 * bug fix: correct canonicalization of multiple non-standard slot
456 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
457 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
458 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
459 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
460 now exists, an signals an error.
461 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
462 during macroexpansion for non-iteration variables as well. (reported
463 by Bruno Haible for CMUCL)
464 * bug fix: Cyclic structures and unprintable objects in compiler
465 messages no longer cause errors. (reported by Bruno Haible)
466 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
467 and READ-FROM-STRING all now return a primary value of NIL if
468 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
469 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
470 (reported by Bruno Haible for CMUCL)
471 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
472 do not designate a condition. (reported by Bruno Haible for
474 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
475 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
476 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
477 permitting resolution in favour of any of the conflicting symbols.
478 (reported by Bruno Haible for CMUCL)
479 * bug fix: EQUAL compiler optimizations is less aggressive on
480 strings which can potentially compare true despite having distinct
481 specialized array element types.
482 * bug fix: unit enumerations can be defined without dividing by
483 zero. (thanks to Vincent Arkesteijn)
484 * FORMAT compile-time argument count checking has been enhanced.
485 (report from Bruno Haible for CMUCL)
486 * a partial workaround for the bug 262: the compiler does not try to
487 inline-expand a local function doing RETURN-FROM from a deleted
488 BLOCK. (thanks to Peter Denno for the bug report and to David
489 Wragg for the simple test case)
490 * fixed some bugs revealed by Paul Dietz' test suite:
491 ** RENAME-PACKAGE allows all package designators as new package
493 ** constraint propagation and lambda variable substitution are
494 more cautious in dealing with partially deleted code.
495 ** compiler sometimes lost reoptimization passes.
496 ** CERROR, given a condition as condition designator, passes its
497 remaining arguments to the continue format control without
499 ** Case-altering FORMAT directives work correctly on non-ASCII
501 ** The REAL type specifier handles bounds outside the floating
502 point ranges without signalling FLOATING-POINT-OVERFLOW.
503 ** Functions with IR1-transformations can create intercomponent
504 references to global functions.
505 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
507 ** FORMAT ~R treats a nil value for its first parameter correctly.
509 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
510 * enhancement: saving cores with foreign code loaded is now
511 supported on x86/NetBSD and sparc/Linux in addition to the previously
513 * bug fix: on some platforms repeated installations caused multiple
514 copies of HTML documentation to be installed -- should not happen
515 any more. (reported by Stefan Scholl)
516 * bug fix: parsing self-recursive alien record types multiple times
517 no longer causes infinite recursion. (reported by Thomas F. Burdick,
518 original patch by Helmut Eller for CMUCL)
519 * bug fix: stack-exhaustion detection works now on NetBSD as well.
520 (thanks to Richard Kreuter)
521 * bug fix: defining classes whose accessors are methods on existing
522 generic functions in other (locked) packages no longer signals
523 bogus package lock violations. (reported by François-René Rideau)
524 * bug fix: special variables as DEFMETHOD parameters no longer have
525 associated bogus type declarations. (reported by David Wragg and
527 * bug fix: read-write consistency on streams of element-type
528 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
529 * bug fix: redefiniton of the only method of a generic function with
530 no DEFGENERIC no longer emits a full WARNING. In addition,
531 redefinition of generic functions with no DEFGENERIC to an
532 incompatible lambda list now signals an error. (thanks to Zach
534 * bug fix: DEFGENERIC now works even when there's a function of the
535 same name in an enclosing lexical environment. (thanks to Zach
537 * fixed compiler failure, caused by instrumenting code during
538 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
539 * optimization: added loop analysis and improved register allocation
540 * fixed some bugs revealed by Paul Dietz' test suite:
541 ** POSITION on displaced vectors with non-zero displacement
542 returns the right answer.
543 ** (SIMPLE-STRING) is a valid type specifier for sequence
545 ** *PRINT-LEVEL* handling for slotless structures is pedantically
547 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
548 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
550 ** The justification version of the FORMAT ~< directive treats
551 non-zero minpad parameter correctly.
553 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
554 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
555 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
556 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
557 the supported interface.
558 * new feature: Single-stepping of code compiled with DEBUG 2 or
559 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
560 * new feature: saving cores with foreign code loaded is now
561 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
562 Timothy Moore's work for CMUCL)
563 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
564 parameters to * instead of NIL if no initform is supplied.
565 (reported by Johan Bockgaard)
566 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
567 correctly when FUNCALL appears as the car of the form. Note:
568 despite this FUNCALL forms are not currently subject to
569 compiler-macro expansion. (port of Raymond Toy's fix for the
570 same from CMUCL, reported by Johan Bockgaard)
571 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
572 (thanks for Teemu Kalvas)
573 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
574 case instead of terminating the iteration (thanks for Julian
575 Squires, Sean Champ and Raymond Toy)
576 * bug fix: incorrect expansion of defgeneric that caused a style
577 warning. (thanks for Zach Beane)
578 * bug fix: slot accessor effective method computation works properly
579 for classes with multiple non-standard applicable methods on
580 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
581 * on x86 compiler supports stack allocation of results of LIST and
582 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
583 CMUCL implementation by Gerd Moellmann)
584 * fixed some bugs revealed by Paul Dietz' test suite:
585 ** FORMAT strings with both the ~<~:;~> form of the justification
586 directive and pretty-printing directives cause an error.
588 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
589 * incompatible change: the internal functions
590 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
591 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
592 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
593 instead of the old functions.
594 * new feature: on platforms where "dladdr" is available foreign
595 function names now appear in backtraces. (based on Helmut Eller's
597 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
598 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
600 * documentation: ASDF manual is now distributed with SBCL.
601 * bug fix: SBCL can now load its contributed modules with REQUIRE
602 even if the system-provided entries have been removed from
603 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
605 * bug fix: backtraces involving undefined functions or assembly
606 routines are more informative. (thanks to Brian Downing)
607 * bug fix: mutually referent alien structures now work correctly.
608 (reported by Rick Taube)
609 * bug fix: structures defined by WITH-ALIEN can be referred to
610 within other definitions in the same WITH-ALIEN.
611 * bug fix: division operators (MOD, TRUNCATE and the like) with
612 constant zero divisors and integer dividends no longer generate
614 * bug fix: provide default methods for INPUT-STREAM-P and
615 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
616 * bug fix: improve the detection and resolution of MOP metacycles
617 (where effective method computation depends on the generic function
618 having its effective method computed). (reported by Bruno Haible)
619 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
620 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
621 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
622 * bug fix: pathnames with relative directory components can be
623 represented relative to default pathnames.
624 * optimization: in taking the GCD of bignums, reduce the two bignums
625 to approximately the same size (using Euclid's algorithm) before
626 applying the more sophisticated binary GCD. (thanks to Juho
628 * optimization: COUNT on bitvectors now operates word-at-a-time.
629 * optimization: ASH with a positive, but not necessarily constant,
630 (leftwards) shift, when in a modular context, is compiled to a
632 * fixed some bugs revealed by Paul Dietz' test suite:
633 ** FORMAT variable parameters ("~V<char>") are defaulted properly
634 if the corresponding argument is NIL.
635 ** FORMAT directives accepting a minpad parameter treat negative
637 ** literal commas as character parameters to FORMAT directives are
639 ** literal spaces directly after ~<Newline> directives within a
640 format-logical-block (~:< ~@:>) do not induce :FILL-style
641 conditional newlines.
642 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
643 as specified: it no longer includes conditional newlines).
644 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
646 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
647 :PER-LINE-PREFIX argument does not evaluate to a string.
649 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
650 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
651 the manual for details; for now, package locks can be disabled by
652 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
653 if no major problems are found then it is likely that they will be
654 compiled in unconditionally.
655 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
656 unsupported operators on all platforms. To load a shared library
657 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
658 object file, link it into a shared library outside of SBCL and
659 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
661 * bug workaround: ROOM T and the GC were not working together
662 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
663 reporting facility aren't preserved. That reporting has been
664 disabled, so now until and unless someone figures out how to make it
665 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
666 * minor incompatible change: as threatened around sbcl-0.8.0, the
667 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
668 The symbols are also exported from SB-PCL for backwards
669 compatibility, but more so than before SB-PCL should be treated as
670 an implementation-internal package.
671 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
673 * the SB-SPROF contrib now works on (most) non-x86 architectures.
674 It is known as of this release not to work on the Alpha, however.
675 * fixed bug #167: errors signalled due to illegal syntax in method
676 bodies are now more legible.
677 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
678 designators and can hence be used with TYPEP.
679 * fixed bug #333: CHECK-TYPE now ensures that the type error
680 signalled, if any, has the right object to be accessed by
681 TYPE-ERROR-DATUM. (reported by Tony Martinez)
682 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
683 5.1.2.3 for multiple-value place subforms. (reported by Kalle
685 * fixed bug #334: programmatic addition of slots using specialized
686 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
687 and :ALLOCATION :CLASS slots.
688 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
689 float exponents. (rereported by Peter Seibel)
690 * fixed a bug: #\Space (and other whitespace characters) are no
691 longer considered to be macro characters in standard syntax by
693 * fixed bug: initialization of condition class metaobjects no longer
694 causes an instance of the condition to be created. (reported by
696 * fixed bug: it is now possible to have slots such that
697 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
698 description is neither :INSTANCE nor :CLASS.
699 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
700 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
701 SLOT-VALUE-USING-CLASS), and no longer causes errors with
702 non-standard SLOT-DEFINITION-ALLOCATION values.
703 * fixed bugs: various race conditions handling exiting threads.
704 CL-PPCRE's thread torture test now passes.
705 * fixed arguably-a-bug: GC time proportional to number of threads in
706 system even when most of them are idle
707 * optimization: improved performance of BIT and SBIT on bit-vectors.
708 * .fasl file incompatibility: The fasl file version number has
709 been incremented because of changes associated with package locks.
711 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
712 * minor incompatible change: the system no longer provides
713 optimization hints (conditions which are TYPEP
714 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
715 mode; these hints are still emitted when higher SPEED optimization
717 * new contrib module: a sampling profiler (profiling by statistical
718 sampling, rather than by instrumenting functions) is available as
719 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
720 * the behaviour of the standard function ED is now customizeable by
721 third parties through a hook variable: see ED's documentation
722 string for information on the protocol.
723 * the compiler no longer emits efficiency notes for (FUNCALL X)
724 when the type of X is uncertain under default optimization
726 * fixed bug 276: mutating a binding of a specialized parameter to a
727 method to something that is not TYPEP the specializer is now
729 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
730 exhibit the required behaviour on the x86 platform. (thanks to
731 Peter van Eynde, Eric Marsden and Bruno Haible)
732 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
733 even for difficult arguments. (reported by Peter Graves)
734 * fixed bug 141a: the backquote printer now descends quoted
736 * fixed another bug in backquote printing: no more destructive
737 modification of the form's list structure. (reported by Brian
739 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
740 move between its address being taken and the call to
741 interrupt_thread, fixing a crashing race condition.
742 * the SB-POSIX contrib implementation has been adjusted so that it
743 no longer exhibits ridiculously poor performance when constructing
744 instances corresponding to C structs.
746 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
747 * minor incompatible change: the sb-grovel contrib now treats C
748 structures as alien (in the sense of SB-ALIEN) objects rather than
749 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
750 has implications for memory management of client code
751 (sb-grovel-returned objects must now be manually managed) and for
752 type safety (alien objects now have full types).
753 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
754 used to control emission of compiler diagnostics, rather than the
755 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
756 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
757 quality should be considered deprecated.
758 * install.sh now installs the user manual as well
759 * (not quite a new documentable feature, but worth considering in
760 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
761 beginnings of a semantically meaningful condition hierarchy is
762 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
764 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
765 designator as the defaults argument.
766 * fixed bug: Displaced arrays whose displaced-to array has become
767 too small now cause ARRAY-DIMENSION to signal an error, providing
768 for safer bounds-checking. (reported by Bruno Haible)
769 * fixed bug: DEFCLASS slot definitions with identical :READER and
770 :WRITER names now signal a reasonable error. (reported by Thomas
772 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
774 * fixed bug: Invalid slot specification errors now print correctly.
775 (thanks to Zach Beane)
776 * fixed bug 320: Shared to local slot value transfers in class
777 redefinitions now happen corrently from superclasses as
778 well. (reported by Bruno Haible)
779 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
781 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
782 improper lists correctly. (reported by Bruno Haible)
783 * fixed bug 313: source-transform for <fun-name> was erroneously
784 applied to a call of a value of a variable with name <fun-name>.
785 (reported by Antonio Menezes Leitao)
786 * fixed bug 307: The obsolete instance protocol ensures that
787 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
788 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
789 remove dead unknown-values globs from the middle of the stack.
790 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
791 easier to resume long computations after using *BREAK-ON-SIGNALS*
792 to diagnose and fix failures. (thanks to Nikodemus Siivola)
793 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
794 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
795 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
796 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
797 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
799 * fixed a (fairly theoretical) bug in string printing: if
800 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
801 does not have array-element-type equal to the most general string
803 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
804 function name. (thanks to Nikodemus Siivola)
805 * fixed bug: compile-time format string checker failed on
806 non-closed ~{. (reported by Thomas F Burdick)
807 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
808 don't warn on legal loop code involving more than one aggregate
809 boolean. (thanks to Nikodemus Siivola)
810 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
811 in keyword position to MAKE-PACKAGE was not being reported as
813 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
814 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
816 * fixed bugs 280 and 312: the checking for multiple definitions in a
817 file is less likely to become confused by uses of inline
819 * fixed bug: the #S reader macro performs the keyword coercion
820 specified for slot names. (reported by Kalle Niemitalo)
821 * fixed bug: lambda lists may contain symbols whose names start with
822 & but are not lambda-list-keywords; their occurrence triggers a
824 * fixed bug 321: define-method-combination argument lambda lists do
825 not cause invalid code to be generated when &optional or &aux
826 variables have default values. (reported by Bruno Haible)
827 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
828 have CLOS classes; this appears also to have fixed reported
829 instability in stack exhaustion detection.
830 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
831 MUFFLE-WARNING when no associated restart is present is now
833 * optimization: rearranged the expansion of various defining macros
834 so that each expands into only one top-level form in a
835 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
837 * optimization: used a previously little-used slot in symbols to
838 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
840 * fixed some bugs revealed by Paul Dietz' test suite:
841 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
843 ** Fixed an optimization bug related to inheritance of initforms
844 from local to shared slots.
845 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
846 ** CAST optimizer forgot to flush argument derived type.
847 ** print/read consistency on floats is now orders of magnitude
848 more likely. (thanks also to Bruno Haible for a similar report
850 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
851 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
852 for objects of type REAL. Make it so.
853 ** SXHASH is specified (infelicitously) to respect similarity,
854 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
855 it so. (thanks to Markus Ziegler)
856 ** on the Alpha, the compiler succeeds in compiling functions
857 returning a known number of arguments greater than 63.
858 ** fixed handling of invalid NIL arguments in keyword position
860 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
862 ** provided a readably-printable representation for RANDOM-STATE
864 ** ensured that pathnames, where they have a namestring, always
865 print using #P"..." syntax.
867 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
868 * Support for the forthcoming 2.0 version of the NetBSD kernel
869 running on x86 hardware has been added. (thanks to Perry
870 E. Metzger most immediately, and others for their past work)
871 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
872 current release version; SBCL's OpenBSD support had been broken
873 since about the time of OpenBSD's switch to ELF binary format.)
874 * [placeholder for DX summary]
875 ** user code with &REST lists declared dynamic-extent, under high
876 speed or space and low safety and debug optimization policy.
877 * The manual has been converted to Texinfo format and the debugger
878 chapter from the cmucl manual has been added.
879 * A facility has been added to extract documentation strings from
880 sbcl and store them as Texinfo-formatted snippets for inclusion in
881 the manual (via Texinfo's @include directive)
882 * bug fix: compiler emitted division in optimized DEREF. (thanks for
883 the test case to Dave Roberts)
884 * bug fix: multidimensional simple arrays loaded from FASLs had fill
885 pointers. (reported by Sean Ross)
886 * bug fix: PROFILE output is printed nicely even for large numerical
887 values. (thanks to Zach Beane)
888 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
889 greater than 32 handle EOF correctly.
890 * bug fix: on X86 an immediate argument of the IMUL instruction is
891 correctly printed in disassembly. (thanks to Lutz Euler)
892 * bug fix: class slots in redefined classes preserve their old
893 values. (thanks to Bruno Haible and Nikodemus Siivola)
894 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
895 succeeds. (reported by Marco Baringer)
896 * fixed some bugs revealed by Paul Dietz' test suite:
897 ** READ-SEQUENCE now works on ECHO-STREAMs.
898 ** RATIONALIZE works more according to its specification. (thanks
901 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
902 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
903 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
904 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
905 should matter to you only if you rebind the printer control
906 variables and then find you want different bindings in the
907 debugger than in the ordinary execution of your program.)
908 * The runtime build system has been tweaked to support building
909 (on SPARC/SunOS) using a C compiler which invokes Sun's own
910 assembler and linker. (thanks to Nikodemus Siivola)
911 * Unbound, undefined, undeclared variables now trigger full
912 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
913 more often programmer error than deliberate exploitation of undefined
915 * optimization: the hash algorithm for strings has changed to one
916 that is less vulnerable to spurious collisions. (thanks to Juho
918 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
919 needless bounds checking. (thanks to Juho Snellman)
920 * optimization: implemented multiplication as a modular
921 (UNSIGNED-BYTE 32) operation on the PPC backend.
922 * fixed some bugs revealed by Paul Dietz' test suite:
923 ** ADJUST-ARRAY now copies the datum in a zero rank array if
925 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
926 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
927 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
929 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
931 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
933 ** ECHO-STREAMs no longer attempt to echo the end of file value to
934 their output stream on EOF from read.
935 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
936 have been read to end-of-file.
937 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
939 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
940 description of determination of which consecutive characters
942 ** Printing the "Space" character with escaping on now yields "#\\ ",
943 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
944 ** Reading floating-point numbers with *READ-BASE* set to a number
945 less than 10 works correctly.
946 ** Reading floating-point numbers with *READ-BASE* set to a number
947 more than 10 works correctly.
948 ** Printing with *PRINT-READABLY* targets the standard readtable, not
949 the readtable currently in effect.
951 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
952 * minor incompatible change: parsing of namestrings on a physical
953 (Unix) host has changed; numbers after the final #\. in a
954 namestring are no longer interpreted as a version field. This is
955 intented to be largely invisible to the user, except that the
956 meaning of the namestring "*.*.*" has changed: it now refers to a
957 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
958 should usually be replaced by
959 (make-pathname :name :wild :type :wild :version :wild)
960 with the added benefit that this is more likely to be portable.
961 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
962 option to OPEN now signals an error if the file being opened
963 exists; this may have an impact on existing code.
964 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
965 signals received, on the PowerPC platforms (both Linux and
966 Darwin). (thanks to Pierre Mai for pointing out the location of
968 * several fixes on OS X: The system now builds and runs cleanly on
969 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
970 (thanks to Brian Mastenbrook)
971 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
972 second-resolution: integer multiples of 1/3600 between -24 and 24.
973 (thanks to Vincent Arkesteijn)
974 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
975 type when called with 1 argument; PEEK-CHAR checked type of
976 PEEK-TYPE only after having read first character from a
977 stream. (reported by Peter Graves)
978 * bug fix: the garbage collector now has much better locality
979 behaviour, and in particular no longer treats all memory as being
980 exhausted when many small objects point to each other in a deeply
982 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
983 recognized as being TYPEP their class.
984 * bug fix: the PUSHNEW documentation string has been corrected.
985 (thanks to Vincent Arkesteijn)
986 * bug fix: defaulting of the value for the last of an atypically
987 large number of multiple values being bound was not being
988 performed correctly on the Alpha or PPC platforms
989 * optimization: implemented multiplication as a modular
990 (UNSIGNED-BYTE 32) operation on the x86 backend.
991 * optimization: SEARCH on simple-base-strings can now be open-coded.
992 (see also contrib/compiler-extras.lisp for inspiration for
993 teaching the compiler about the Boyer-Moore algorithm).
994 * value, returned by MAX (and MIN) called with several EQUALP, but
995 not EQL, arguments now does not depend on compiler settings.
996 * fixed some bugs revealed by Paul Dietz' test suite:
997 ** in stack analysis liveness information is propagated from
998 non-local entry points.
999 ** pathwise CAST removing failed when the CAST node did not start
1001 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1002 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1004 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1005 host is already defined.
1006 ** RENAME-FILE works on streams instead of signalling an internal
1008 ** PEEK-CHAR uses the current readtable when determining whether
1009 or not a character is whitespace.
1010 ** MERGE-PATHNAMES handles the case when the pathname does not
1011 specify a name while the default-pathname specifies a version.
1012 ** Pathnames now stand a better chance of respecting print/read
1014 ** Attempting to use standardized file system operators with a
1015 pathname with invalid :DIRECTORY components signals a
1017 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1019 ** DIRECTORY on logical pathnames is more correct.
1020 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1021 signal a TYPE-ERROR if their argument is not a stream.
1022 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1023 designator argument does not designate a stream.
1024 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1025 examining the synonym.
1026 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1028 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1029 element-type larger than ([UN]SIGNED-BYTE 32).
1031 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1032 * When built with the :SB-FUTEX feature, threaded builds now take
1033 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1034 for faster/more reliable mutex and condition variable support.
1035 * Incompatible change (but one you probably shouldn't have been using
1036 anyway): the interface and code for arbitrating between multiple
1037 threads in the same user session has been redesigned.
1038 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1039 error when called without an explicit environment argument.
1040 (thanks to Nikodemus Siivola)
1041 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1042 confused about their position. (thanks to Adam Warner and Gerd
1044 * bug fix: Pretty printing backquoted forms with unquotations in the
1045 argument list position of various code constructs such as LAMBDA
1046 now works correctly. (reported by Paul Dietz)
1047 * bug fix: Pretty printing unquotations no longer loses all
1048 stream position information.
1049 * optimization: performance of string output streams is now less
1050 poor for multiple small sequence writes.
1051 * optimization: performance of CSUBTYPEP in the presence of complex
1052 expressions involving CONS and NOT many times has been improved.
1053 (reported by Paul Dietz)
1054 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1056 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1057 streams for the REPL, for the debugger, and so on. (thanks to
1059 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1060 * fixed some bugs revealed by Paul Dietz' test suite:
1061 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1062 is ignored if it is too small, rather than propagating through
1063 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1064 ** extremely complex negations of CONS types were not being
1065 sufficiently canonicalized, leading to inconsistencies in
1067 ** VALUES tranformer lost derived type.
1069 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1070 * fixed a bootstrapping bug: the build process no longer assumes
1071 that the various BOOLE-related constants have the same value in
1072 host and target lisps. (noted by Paul Dietz' test suite on an
1073 SBCL binary built from CLISP)
1074 * The system can now be dynamically linked on the MIPS platform,
1075 which enables dynamic loading of foreign code from Lisp. (thanks
1076 to Ralf Baechle for discussions on the MIPS ABI)
1077 * The system now records debugging information for its own source
1078 files in a filesystem-position-independent manner, relative to
1079 the "SYS" logical host.
1080 * fixed a compiler bug: MV-LET convertion did not check references
1081 to the "max args" entry point. (reported by Brian Downing)
1082 * tweaked disassembly notes to be less confident about proclaiming
1083 some instruction as an LRA. (thanks to Brian Downing)
1084 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1085 now each have their own history, command character, and other
1086 characteristics. (thanks to David Lichteblau)
1087 * fixed some bugs revealed by Paul Dietz' test suite:
1088 ** compiler failure in compiling LOGAND expressions including a
1090 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1091 shift greater than 32.
1092 ** FLUSH-DEST did not mark blocks for type check regeneration.
1093 ** HANDLER-CASE failed to accept declarations in handler clauses
1094 in some circumstances.
1096 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1097 * New code in contrib/sb-introspect (still probably not entirely
1098 stable yet) provides some support for smart Lisp development
1099 environments like SLIME.
1100 * The conditions signalled for errors occurring when loading .fasl
1101 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1102 in a way which should help ASDF recover gracefully.
1103 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1104 stuff has been cleaned up. If you code contrib/ stuff, this might
1105 affect you, and you can look at contrib/README, contrib/STANDARDS,
1106 and/or the 0.8.4.27 diff to check.
1107 * In full calls the compiler now does not generate checks for declared
1108 argument types for all arguments.
1109 * various threading fixes
1110 ** and some experimental patches which didn't make it into
1111 the main tree for this release, but which are shipped in
1112 contrib/experimental-thread.patch as a possible fix for some
1113 failures (deadlock, spinning...) in GC-intensive multithreaded
1115 * fixed PPC build problem (source code incompatibility of different
1116 library versions): added offsetof() hackery which attempts to divine
1117 where glibc maintainers put uc_mcontext today
1118 * fixed bug 282: compiler does not trust type assertions while passing
1119 arguments to a full call.
1120 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1121 &OPTIONAL VALUES type parameter.
1122 * fix bug 214: algorithm for noting rejected templates is now more
1123 similar to that of template seletion. (also reported by rydis on
1125 * fixed bug 141b: printing backquoted information readably and prettily
1126 inserts a space where necessary.
1127 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1128 (CHARACTER 10) are now reported as errors, rather than propagated
1129 as unknown types. (reported by piso on #lisp)
1130 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1131 with values NIL and :ERROR. (thanks to Milan Zamazal)
1132 * fixed bug 191c: CLOS now does proper keyword argument checking as
1133 described in CLHS 7.6.5 and 7.6.5.1.
1134 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1135 longer raise an error; further, using a list as a for-as-arithmetic
1136 counter now raises a meaningful error.
1137 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1138 now understood by sequence creation functions such as MAKE-SEQUENCE
1140 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1141 a STRING-INPUT-STREAM.
1142 * compiler enhancement: SIGNUM is now better able to derive the type
1144 * type declarations inside WITH-SLOTS are checked. (reported by
1146 * fixed some bugs revealed by Paul Dietz' test suite:
1147 ** incorrect optimization of TRUNCATE for a positive first
1148 argument and negative second.
1149 ** compiler failure in let-convertion during flushing dead code.
1150 ** compiler failure while deriving type of TRUNCATE on an
1151 interval, containing 0.
1152 ** ASH of a negative bignum by a negative bignum count now returns
1154 ** intersection of CONS types now canonicalizes properly, fixing
1155 inconsistencies in SUBTYPEP.
1157 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1158 * incompatible change: The --disable-debugger command line
1159 option now clobbers the debugger at a more fundamental
1160 level, by redefining #'INVOKE-DEBUGGER instead of by
1161 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1162 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1163 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1164 enter the debugger REPL and then suffer recursive errors
1165 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1166 while under the new system, BREAK in --disable-debugger mode
1167 terminates the system just as an unhandled error would.
1168 * fixed compiler performance when processing loops with a step >1;
1169 * bug fix: DOCUMENTATION now retrieves generic function
1170 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1171 support has been systematized, and now supports the methods
1172 specified by ANSI, along with a default method and a method for
1173 slot documentation. (reported by Nathan Froyd)
1174 * bug fix: effective methods associated with a generic function are
1175 no longer cached over a change of that generic function's method
1176 combination. (reported by Andreas Fuchs)
1177 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1178 names a non-existent directory. (thanks to Andreas Fuchs)
1179 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1180 the PPC platform, lead to stack corruption; this has been fixed.
1181 (reported by Rainer Joswig)
1182 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1183 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1184 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1185 arithmetic in general, is now much more reliable on the PPC
1187 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1188 platform now returns the right answer.
1189 * optimization: restored some effective method precomputation in
1190 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1191 precomputation is now tunable.
1192 * optimization: compiler-internal data structure use has been
1193 reviewed, and changes have been made that should improve the
1194 performance of the compiler by about 20%.
1195 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1196 simple-base-strings and simple-bit-vectors is improved.
1197 * optimization: the optimization of 32-bit logical and arithmetic
1198 functions introduced in version 0.8.3 on the x86 has been
1199 implemented on the mips, ppc and sparc platforms; an
1200 implementation of the same facility, but for 64-bit arithmetic,
1201 has been added for the alpha.
1202 * microoptimization: the compiler is better able to make use of the
1203 x86 LEA instruction for multiplication by constants.
1204 * bug fix: in some situations compiler did not report usage of
1205 generic arithmetic in (SPEED 3) policy.
1206 * bug 145b fix: compiler used wrong type specifier while converting
1207 MEMBER-types to numeric.
1208 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1210 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1211 index argument now works properly on the Alpha platform.
1212 * bug fix: floating point exception treatment on the Alpha platform
1214 * bug fix: FILE-POSITION works much better on string input and
1215 output streams. (thanks to Nikodemus Siivola)
1216 * bug fix: many threading/garbage collection symptoms sorted.
1217 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1218 might be pseudo-atomic.
1219 * internal change: Stopping for GC is now done with signals not
1220 ptrace. GC is now done in whichever thread wanted it, instead of
1222 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1224 * bug fix: result form in DO is not contained in the implicit
1226 * incompatible change: ICR structure is changed; the value part of
1227 CONTINUATION is now called LVAR; corresponding functions are
1228 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1229 * added type deriver for ISQRT (thanks to Robert E. Brown).
1230 * bug fix: better support for loading from the command line when an
1231 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1233 * fixed some bugs revealed by Paul Dietz' test suite:
1234 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1235 ** ROUND and FROUND now give the right answer when given very
1236 small float arguments.
1237 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1239 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1240 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1241 ** LCM should return a non-negative integer.
1242 ** PARSE-INTEGER returned the index of a terminator instead of the
1243 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1244 ** PARSE-INTEGER returned an incorrect index being applied to a
1246 ** LCM with two arguments of 0 returns 0 rather than signalling
1248 ** unsigned addition of a 32-bit constant with the high bit set no
1249 longer causes an internal compiler error on the x86.
1250 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1251 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1252 with negative last argument.
1253 ** byte specifiers with zero size and position no longer cause
1254 an error during type derivation.
1255 ** bignum multiplication on the Alpha platform now returns the
1257 * porting: The system now builds on SuSE AMD64, although it still
1258 generates a 32-bit binary.
1259 * .fasl file incompatibility: The fasl file version number has
1260 been incremented (because of the changes to internal compiler
1261 data structures referred to above).
1263 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1264 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1265 more accurately, on the Darwin kernel running on PowerPC hardware.
1266 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1267 * Compiler code deletion notes now signal a condition of type
1268 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1269 an associated MUFFLE-WARNING restart.
1270 * The compiler now performs limited argument count validation of
1271 constant format strings in FORMAT, and where appropriate in ERROR,
1272 CERROR and WARN. (thanks to Gerd Moellmann)
1273 * New ASDF-INSTALL contrib can be used for automatic download and
1274 installation of third-party Lisp code from CCLAN or other sites
1276 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1277 which forces another thread to execute a function supplied by the
1279 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1280 now accept and act upon their :ELEMENT-TYPE keyword argument.
1281 (reported by Martin Atzmueller, Edi Weitz)
1282 * bug fix: FILE-POSITION now accepts position designators up to
1283 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1284 is the greater. (thanks to Patrik Nordebo)
1285 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1286 Kalle Olavi Niemitalo)
1287 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1288 printable. (reported by Eric Marsden)
1289 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1290 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1291 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1292 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1293 (thanks to Stig Sandoe)
1294 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1295 UNREAD-CHAR now correctly includes the unread character in the
1296 target sequence. (thanks to Gerd Moellmann)
1297 * bug fix in threaded builds: the system can now be suspended and
1298 resumed by shell job control with minimal disruption.
1299 * bug fixes in times and timezones >2038AD
1300 * better handling of "where is GNU make?" problem in build scripts
1301 (thanks to Nikodemus Siivola)
1302 * new optimization: inside a named function any reference to a
1303 function with the same name is considered to be a self-reference;
1304 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1305 optimization quality.
1306 * new optimization on x86: logical functions and + now have
1307 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1308 used when the result is truncated to 32 bits.
1309 * VALUES declaration is partially enabled.
1310 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1311 * fixed some bugs revealed by Paul Dietz' test suite:
1312 ** The system now obeys the constraint imposed by
1313 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1314 types form a lattice under type intersection.
1315 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1316 ** ASSOC now ignores NIL elements in an alist.
1317 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1318 and (1+ MOST-POSITIVE-FIXNUM) answers.
1319 ** The addition of a method with invalid qualifiers to a generic
1320 function does not cause an error to be signalled immediately;
1321 a warning is signalled, and the error is generated only on
1322 calling the generic function.
1323 * changed .fasl file version number, in part to add type codes for
1324 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1325 obscure ANSI requirements
1327 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1328 * fixed bug 148: failure to inline-expand a local function left
1329 garbage, confusing the compiler.
1330 * fixed bugs 3cd: structure slot readers perform type check if the
1331 slot can have an invalid value (i.e. it is either not initialized
1332 or can be written with a less specific slot writer).
1333 * bug fix: the compiler now traps array references to elements off
1334 the end of an array; previously, the bounds checking in some
1335 circumstances could go off-by-one.
1336 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1338 * type declarations for array element types now obey the description
1339 on the CLHS page "Declaration TYPE", as per discussions on
1340 sbcl-help around 2003-05-08. This means that a declaration
1341 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1342 declaration, all references to BAR will be asserted or assumed
1343 (with THE, so dependent on compiler policy) to involve objects of
1344 type FOO. Note that no such declaration is implied in
1345 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1346 * declared types of functions from the "Conditions"
1347 chapter. (reported by Paul Dietz)
1348 * bug fix: CERROR accepts a function as its first argument.
1349 * bug fix: NTH an NTHCDR accept a bignum as index
1350 arguments. (reported by Adam Warner)
1351 * optimization: character compare routines now optimize comparing
1352 against a constant character. (reported by Gilbert Baumann)
1353 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1354 argument now works properly on the MIPS platform.
1355 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1356 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1357 declared to have a functional type, does not cause an error
1359 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1360 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1361 * fixed bug: if last continuation of a deleted block has a
1362 destination, this destination should be deleted too. (reported by
1364 * fixed a bug in the bootstrap process: the host compiler's values
1365 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1366 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1367 test case from Patrik Nordebo)
1368 * improved the ability of the disassembler on the PPC platform to
1369 provide helpful disassembly notes.
1370 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1371 the class in more cases than previously.
1372 * bug fix: FILE-POSITION now understands :START and :END for
1373 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1374 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1375 8) data. (thanks to David Lichteblau)
1376 * bug fix: it is possible to add a method to a generic function
1377 without lambda list.
1378 * bug fix: reader failed to signal END-OF-FILE inside an
1379 object representation. (reported by Nikodemus Siivola)
1380 * fixed some bugs revealed by Paul Dietz' test suite:
1381 ** LAST and [N]BUTLAST should accept a bignum.
1382 ** condition slot accessors are methods.
1383 ** (VECTOR NIL) is a subtype of STRING.
1385 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1386 * minor incompatible change: some nonsensical specialized lambda
1387 lists (used in DEFMETHOD) which were previously ignored now signal
1389 * minor incompatible change: the system is now aware of the types of
1390 variables in the COMMON-LISP package, and will signal errors for
1391 most violations of these type constraints (where previously they
1392 were silently accepted).
1393 * minor incompatible change: COMPILE-FILE now uses the freedom
1394 afforded (ANSI 3.2.2.3) to use derived function types for
1395 functions defined in the same file. This also permits the system
1396 to warn on static type mismatches and function
1397 redefinition. (Currently it does not work with high DEBUG level.)
1398 * minor incompatible change: VALUES declaration is disabled.
1399 * When issuing notes, the compiler now signals a condition of type
1400 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1401 restart for use in user handlers. It is expected that the
1402 COMPILER-NOTE condition will eventually become a condition
1403 supertype to a hierarchy of note types, which will then be
1404 handleable in a similar fashion. However, at the moment, no such
1405 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1406 but it's an implementation detail, not a classification for the
1408 * Changes in type checking closed the following bugs:
1409 ** type checking of unused values (192b, 194d, 203);
1410 ** template selection based on unsafe type assertions (192c, 236);
1411 ** type checking in branches (194bc).
1412 * A short form of VALUES type specifier has ANSI meaning (it has
1413 increased the number of situations when SBCL cannot perform type
1415 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1416 keywords or constants is permissible.
1417 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1418 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1419 classes. (thanks to Antonio Martinez)
1420 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1421 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1422 argument to a pretty-print stream. Instead, it leaves any such
1423 operations to DESCRIBE-OBJECT methods.
1424 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1426 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1427 lambda lists. (thanks to David Lichteblau)
1428 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1429 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1430 (thanks to Gerd Moellmann)
1431 * fixed bug 246: increased compilation speed of long
1432 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1434 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1436 * DEFCONSTANT now throws a condition of type
1437 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1438 constant to a non-EQL value; CONTINUE and ABORT restarts
1439 respectively change and preserve the value.
1440 * fixed bug 63: The code walker, part of the implementation of CLOS,
1441 is now better at handling symbol macros.
1442 * bug fix: There is no longer an internal implementation type named
1443 CL:LENGTH. (reported by Raymond Toy)
1444 * bug fix: In macro-like defining macros/special operators the
1445 implicit block does not enclose the lambda list.
1446 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1447 enclosed in parentheses) are not suitable as type specifiers, and
1448 their use properly signals an error now.
1449 * bug fix: An argument count mismatch for a type specifier in code
1450 being compiled no longer causes an unhandled error at compile
1451 time, but signals a compile-time warning.
1452 * fixed simple vector readable printing
1453 * bug fix: DESCRIBE takes more care over whether the class
1454 precedence list slot of a class is bound before accessing it.
1455 (reported by Markus Krummenacker)
1456 * bug fix: FORMATTER can successfully compile pretty-printer format
1457 strings which use variants of the ~* directive inside.
1458 * bug fix: SEARCH now applies its TEST predicate to the elements of
1459 the arguments in the correct order. (thanks to Wolfhard Buss)
1460 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1461 in type inference. (thanks to Robert E. Brown)
1462 * bug fix: Reading in symbols with an explicit package name of ""
1463 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1464 * fixed some bugs revealed by Paul Dietz' test suite:
1465 ** NIL is now allowed as a structure slot name.
1466 ** Arbitrary numbers, not just REALs, are allowed in certain
1467 circumstances in LOOP for-as-arithmetic clauses.
1468 ** Multiple class redefinitions before slot access no longer
1469 causes a type error.
1470 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1471 association between the name and a class.
1472 ** Generic functions with non-standard method-combination and over
1473 five methods all of which return constants no longer return NIL
1474 after the first few invocations. (thanks to Gerd Moellmann)
1475 ** CALL-NEXT-METHOD with no arguments now passes the original
1476 values of the arguments, even in the presence of assignment.
1477 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1479 ** DISASSEMBLE works with closures and funcallable instances.
1480 ** ADD-METHOD now returns the generic function, not the new method.
1481 ** FIND-METHOD signals an error if the lengths of the specializers
1482 is incompatible with the generic function, even if the ERRORP
1484 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1485 which its argument is a member.
1486 ** DEFCLASS only redefines the class named by its class-name
1487 argument if that name is the proper name of the class;
1488 otherwise, it creates a new class.
1489 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1490 of the UNBOUND-SLOT condition to the name of the slot.
1491 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1492 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1493 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1494 SLOT-MAKUNBOUND in the specified fashion.
1496 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1497 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1498 cross-compilation host. As a consequence, we can now bootstrap our
1499 way up to SBCL starting with a bare gcc toolchain and human-readable
1500 source code (first the source to CLISP, then the source to SBCL).
1501 * A contributed module containing a partial implementation of the
1502 simple-streams interface has been included. (thanks to Rudi
1504 * A contributed module implementing the RFC1321 Message Digest
1505 Algorithm, known as MD5, has been included.
1506 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1507 no longer has any effect, as the code controlled by this feature
1508 has been deleted. (As far as we know, no-one has ever built using
1509 this feature, and its semantics were confused in any case).
1510 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1511 work on conditions (as required by the ANSI specification),
1512 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1513 expected behaviour on conditions. Users should note, however,
1514 that such behaviour is not required by the ANSI specification,
1515 and so use of this behaviour may render their code unportable.
1516 * fixed some bugs revealed by Paul Dietz' test suite:
1517 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1519 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1520 ** SXHASH on condition objects no longer returns NIL.
1521 ** :ALLOCATION :CLASS slots are better treated; their values are
1522 updated on class redefinition, and initforms inherited from
1523 superclasses are applied.
1524 ** REMOVE-METHOD returns its generic function argument even when
1525 no method was removed.
1526 ** SHARED-INITIALIZE now initializes the values of the requested
1527 slots, including those with :ALLOCATION :CLASS.
1528 ** ALLOCATE-INSTANCE now works on structure classes defined via
1529 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1531 ** SLOT-EXISTS-P now works on conditions, as well as structures
1533 ** MAKE-LOAD-FORM now has the required methods on
1534 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1535 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1536 keyword, and now implements the SLOT-NAMES argument.
1537 ** methods with &OPTIONAL arguments no longer allow too many
1538 arguments to be passed in the call without error.
1539 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1540 option is consistent with the required arguments of the generic
1541 function lambda list.
1542 * bug fix: REQUIRE accepts a string designator. (Thanks to
1544 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1545 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1546 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1547 * bug fix: The compiler now checks for duplicated variables in macro
1549 * bug fix: SETQ on globals returns the correct value.
1550 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1551 not a COMPILER-ERROR (followed by some other strange error on
1552 choosing the CONTINUE restart).
1553 * bug fix: make.sh and friends are now more consistent in the way that
1554 they look for GNU "make".
1556 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1557 * experimental native threads support (on x86 Linux >=2.4 only).
1558 This is not compiled in by default: you need to add :SB-THREAD to
1559 the target features. See the "Beyond ANSI" chapter of the manual
1561 * fix for longstanding nonANSIism: The old distinction between
1562 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1563 The return value from CL:FIND-CLASS is now a CLOS class, and
1564 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1565 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1566 * An interface to the MetaObject Protocol, as described in Kiczales,
1567 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1568 MIT Press, 1991, is available from the SB-MOP package.
1569 * incompatible change: the SB-PCL package should now be considered
1570 a private implementation detail, and no longer a semi-private MOP
1572 * minor incompatible change: due to rearrangement for threads, the
1573 control stack and binding stack are now allocated at arbitrary
1574 addresses instead of being hardcoded per-port. Users affected by
1575 this probably have to be doing advanced things with shared
1576 libraries, and will know who they are.
1577 * minor incompatible change: Previously, all --eval forms used were
1578 processed with READ before any of them were processed with EVAL.
1579 Now each --eval form is processed with both READ and EVAL before
1580 the next --eval form is processed. (Thus package operations like
1581 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1582 work as the user might reasonably expect.)
1583 * minor incompatible change: *STANDARD-INPUT* is now only an
1584 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1586 * minor incompatible change: Y-OR-N-P is now character-oriented,
1587 not line oriented. Also, YES-OR-NO-P now works without errors.
1588 (thanks to Antonio Martinez)
1589 * sb-aclrepl module improvements: an integrated inspector, added
1590 repl features, and a bug fix to :trace command.
1591 * Known functions, which cannot be open coded by the backend, are
1592 considered to be able to check types of their arguments. (fixing
1593 a bug report by Nathan J. Froyd)
1594 * fixed a bug in computing method discriminating functions: It is
1595 now possible to define methods specialized on classes which have
1596 forward-referenced superclasses. (thanks to Gerd Moellmann)
1597 * fixed evaluation order in optional entries (reported by Gilbert
1599 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1600 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1601 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1602 required-by-AMOP NAME argument, as well as CLASS and
1603 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1604 * fixed bug 20: DEFMETHOD can define methods using names that are
1605 not the proper names of classes to designate class specializers.
1606 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1607 Unix file descriptors, instead of blowing up. (thanks to Antonio
1609 * Garbage collection refactoring: user-visible change is that a
1610 call to the GC function during WITHOUT-GCING will not do garbage
1611 collection until the end of the WITHOUT-GCING. If you were doing
1612 this you were probably losing anyway.
1613 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1614 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1615 TYPEP the latter but not the former.
1616 * The compiler issues a full WARNING for calls to undefined functions
1617 with names from the CL package.
1618 * MAP-INTO for a vector destination is open coded. (reported by
1619 Brian Downing on c.l.l)
1620 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1621 documentation string.
1622 * fixed some bugs revealed by Paul Dietz' test suite:
1623 ** COPY-ALIST now signals an error if its argument is a dotted
1625 ** Condition slots are now accessed more correctly in the presence
1626 of multiple initargs for a given slot.
1627 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1628 exclude restarts of the same name associated with a different
1630 ** DEFCLASS of forward-referenced classes with another
1631 forward-referenced class in the superclasses list no longer
1633 ** Condition slots are now initialized once each, not multiple
1634 times. (thanks to Gerd Moellmann)
1635 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1637 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1638 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1639 arguments contain duplicated elements.
1640 ** RESTART-CASE understands local macros.
1641 ** RESTART-CASE associates exactly its own restarts with a condition.
1642 ** ENDP in safe mode checks its argument to be of type LIST.
1643 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1644 in question is unbound.
1645 ** Optimized MAKE-INSTANCE functions no longer cause internal
1646 assertion failures in the presence of duplicate initargs.
1647 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1648 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1649 propagated even in the ctor optimized implementation of
1651 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1653 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1654 * a better implementation of SXHASH on (simple) bit vectors,
1655 measured both in execution speed and in distribution of results
1656 over the positive fixnums, has been installed. Likewise, a better
1657 implementation of EQUAL for simple bit vectors is now available.
1658 * fixed CEILING optimization for a divisor of form 2^k.
1659 * fixed bug 240 (emitting extra style warnings "using the lexical
1660 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1661 by Antonio Martinez)
1662 * fixed SXHASH, giving different results for NIL depending on type
1663 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1664 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1665 variable. (found by Rolf Wester)
1666 * SBCL does not ignore type declarations for special
1667 variables. (reported by rif on c.l.l 2003-03-05)
1668 * some bug fixes in contrib/sb-aclrepl/
1669 * fixed some bugs revealed by Paul Dietz' test suite:
1670 ** a bug in the CONS type specifier, whereby the CAR and CDR
1671 types got intertwined, has been fixed;
1672 ** the type system is now able to reason about the interaction
1673 between INTEGER and RATIO types more completely;
1674 ** APPEND, [N]REVERSE and NRECONC check that those their
1675 arguments, which must be proper lists, are really so;
1676 ** An array specialized to be unable to hold elements has been
1677 implemented, as required -- yes, really -- by ANSI;
1678 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1679 on malformed property lists;
1681 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1682 * incompatible packaging change: in line with Unix convention,
1683 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1684 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1685 * REQUIRE and PROVIDE are now optionally capable of doing something
1686 useful. See the documentation string for REQUIRE.
1687 * infrastructure for a managed SBCL contrib system: contributed
1688 modules in this release include:
1689 ** the ASDF system definition facility;
1690 ** an interface to the BSD Sockets API;
1691 ** an ACL-like convenience interface to the repl;
1692 (thanks to Kevin Rosenberg)
1693 ** an implementation of ROTATE-BYTE, with efficient implementation
1695 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1696 gives a closer approximation to the right answer than previously.
1697 (thanks to Raymond Toy)
1698 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1699 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1700 optional environment arguments, as required by ANSI.
1701 * fixed bugs in other functions taking environment objects, allowing
1702 calls with an explicit NIL environment argument to be compiled
1704 * fixed bug 228: primary return values from
1705 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1706 COMPILE or FUNCTION.
1707 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1708 :TYPE will no longer signal errors on innocuous objects.
1709 * fixed bug 231b: SETQ is better at respecting type declarations in
1710 the lexical environment.
1711 * fixed a bug in DEFCLASS: classes named by symbols with no or
1712 unprintable packages can now be defined.
1713 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1714 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1715 * fixed compiler failure related to checking types of functions.
1716 (reported by Robert E. Brown)
1717 * the compiler is now much more consistent in its error-checking
1718 treatment of bounding index arguments to sequence functions: in
1719 (SAFETY 3) code, errors will be signalled in almost all cases if
1720 invalid sequence bounding indices are passed to functions defined
1721 by ANSI to operate on sequences.
1722 * fixed a bug in the build procedure: documentation of SBCL-specific
1723 packages is now preserved and available in the final Lisp image.
1724 * lifted FDEFINITION lookup out of loops in the implementation of
1725 many list operations. (thanks to Robert E. Brown)
1726 * fixed a bug in the reader: the #n# reader macro now works for
1727 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1728 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1729 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1730 E. Brown and Tony Martinez respectively)
1731 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1732 SLOT-BOUNDP: in particular, it is now able to optimize them much
1733 better, and is now not vulnerable to having packages renamed.
1734 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1735 always called when a slot is not present in an instance. (thanks
1737 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1738 longer signal an error on unreading a character following EOF on
1739 the previous constituent stream. (thanks to Tony Martinez)
1740 * fixed some bugs revealed by Paul Dietz' test suite:
1741 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1742 not just nonnegative fixnums;
1743 ** the logical bit-array operators such as BIT-AND now accept an
1744 explicit NIL for their "opt-arg" argument (to indicate a
1745 freshly-consed result bit-array);
1746 ** ELT now signals an error on an invalid sequence index in safe
1748 ** the type system is now cleverer about negations of numeric
1749 types, and consequently understands the BIGNUM and RATIO types
1751 ** the type system is now cleverer about the interaction between
1752 INTEGER and RATIO types: while bugs still remain, many more
1753 cases are accurately computed;
1754 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1755 if it is in the last clause;
1756 ** CONSTANTLY now correctly returns a side-effect-free function in
1758 ** DECLARE is no longer treated as a special-operator; in
1759 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1760 * incremented fasl file version number due to the change in the
1761 DEFSTRUCT-SLOT-DESCRIPTION structure.
1763 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1764 * minor incompatible change: code processed by the "interpreter" or
1765 EVAL now has a compilation optimization policy of (DEBUG 2)
1766 (changed from (DEBUG 1)) to improve debuggability of interactive
1767 development, and to allow the use of the debug RETURN command in
1769 * an experimental implementation of the RETURN command for the
1770 debugger has been included. (thanks to Frederik Kuivinen)
1771 * fixed bug 62: constraints were not propagated into a loop.
1772 * fixed bug in embedded calls of SORT (reported and investigated by
1774 * fixed some bugs revealed by Paul Dietz' test suite:
1775 ** printing and reading of arrays with some dimensions having
1776 length 0 (thanks to Gerd Moellmann);
1777 ** BOA constructor with &AUX argument without a default value does
1778 not cause a type error;
1779 ** CONSTANTP now returns true for all self-evaluating objects.
1781 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1782 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1783 accessors that are related by inheritance, as specified in the
1784 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1785 Valtteri Vuorikoski)
1786 * The compiler is now able to inline functions that were defined in
1787 a complex lexical environment (e.g. inside a MACROLET).
1788 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1790 * Support for the upcoming FreeBSD-5.0 release has been included.
1791 (thanks to Dag-Erling Smorgrav)
1792 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1793 effect when it is not in a toplevel context.
1794 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1795 stand a better chance of being correct. (thanks to Gerd
1797 * fixed bug in COERCE, which now signals an error on coercing a
1798 rational to a bounded real type which excludes the expected
1800 * The compiler is now able to derive types more accurately from the
1801 COERCE and COMPILE functions.
1802 * fixed bug 223: functional binding is considered to be constant
1803 only for symbols in the CL package.
1804 * fixed bug 231: SETQ did not check the type of a variable being set
1805 (reported by Robert E. Brown)
1806 * A new optimization for MAKE-INSTANCE has been included, fixing
1807 various bugs (including relating to :ALLOCATION :CLASS slots and
1808 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1810 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1811 ** As required by ANSI, LOOP now disallows anonymous collection
1812 clauses such as COLLECT I in conjunction with aggregate boolean
1813 clauses such as THEREIS (= I 1);
1814 ** LOOP now signals an error when any variable is reused in the
1815 same loop (including the potentially useful construct analogous
1816 to WITH A = 1 WITH A = (1+ A);
1817 ** IT is only a special loop symbol within the first clause of a
1818 conditional loop clause;
1819 ** LOOP with a typed iteration variable over a hashtable now
1820 signals a type error iff it should.
1821 * fixed some other bugs revealed by Paul Dietz' test suite:
1822 ** FILE-STREAM now names the class previously known as FD-STREAM;
1823 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1824 argument) no longer signals an error;
1825 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1826 of the slot symbol, rather than using the current package
1827 ((:CONC-NAME "") continues to intern the slot's name in the
1829 * incremented fasl file version number, because of the incompatible
1830 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1831 of the new implementation of DEFINE-COMPILER-MACRO.
1833 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1834 * Support for building SBCL for MIPS platforms running in
1835 little-endian mode has now been checked in, and basic
1836 functionality on said platforms verified.
1837 * minor incompatible change: PCL now records the pathname of a file
1838 in which methods and the like are defined, rather than its
1840 * minor incompatible change: TRUENAME now considers the truename of
1841 a file naming a directory to be the pathname with :DIRECTORY
1842 component indicating that directory.
1843 * minor incompatible change: a NAMED clause in the extended form of
1844 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1845 reason for the previous behaviour is unclear.
1846 * more systematization and improvement of CLOS and MOP conformance
1847 in PCL (thanks to Gerd Moellman and Pierre Mai):
1848 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1850 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1851 primary methods with no specializers;
1852 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1854 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1855 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1856 FORWARD-REFERENCED-CLASSes; error reporting on
1857 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1859 ** SXHASH on CLOS instances now uses a slot internal to the
1860 instance to return different numbers on distinct instances,
1861 while preserving the same return value through invocations of
1863 ** DEFMETHOD signals errors when methods with longer incongruent
1864 lambda lists are added to generic functions;
1865 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1866 CLASS, as specified in AMOP;
1867 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1868 on the order returned by the primary method for classes of
1869 class STANDARD-CLASS;
1870 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1871 * fixed some bugs shown by Paul Dietz' test suite:
1872 ** DOLIST puts its body in TAGBODY;
1873 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1875 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1876 value producing form;
1877 ** if more variables are given to PROGV than values, extra
1878 variables are bound and made to have no value;
1879 ** NSUBSTITUTE on list arguments gets the right answer with
1881 ** ELT signals an error of type TYPE-ERROR when the index argument
1882 is not a valid sequence index;
1883 ** LOOP signals (at macroexpansion time) an error of type
1884 PROGRAM-ERROR when duplicate variable names are found;
1885 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1886 ** FUNCALL of special-operators now cause an error of type
1888 ** PSETQ now works as required in the presence of side-effecting
1889 symbol-macro places;
1890 ** NCONC accepts any object as its last argument;
1891 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1893 ** loop-for-as-package does not require a package to be explicitely
1895 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1897 * fixed bug 166: compiler preserves "there is a way to go"
1898 invariant when deleting code.
1899 * fixed bug 172: macro lambda lists with required arguments after
1900 &REST arguments now cause an error to be signalled. (thanks to
1902 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1903 bug. (thanks to Matthew Danish)
1904 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1906 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1907 arguments are lexically rebound. (thanks to Gerd Moellmann and
1909 * fixed bug 194: error messages are now more informative when there
1910 is no primary method applicable in a call to a generic
1911 function. (thanks to Gerd Moellmann)
1912 * fixed bug in command line argument checking (thanks to Julian
1914 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1915 (thanks to Matthew Danish)
1916 * incremented fasl file version number, because of the
1917 SXHASH-related changes in the layout of CLOS data structures
1919 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1920 * minor incompatible change: The runtime (the Unix executable named
1921 "sbcl") is now much pickier about the .core files it will load.
1922 Essentially it now requires .core files to descend from the same
1923 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1924 as the runtime does. (The intent is to prevent the crashes which
1925 can occur, and which can even be reported as mysterious failures,
1926 when people patch the sources or change the build parameters
1927 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1928 sbcl and .core files.)
1929 * fixed bug: VALUES-LIST is no longer optimized away.
1930 * fixed bug 142: The FFI conversion of C string values to Lisp
1931 string values no longer conses excessively. (thanks to Nathan
1932 Froyd porting Raymond Toy's fix to CMU CL)
1933 * began to systematize and improve MOP conformance in PCL (thanks to
1934 Nathan Froyd, Gerd Moellman and Pierre Mai):
1935 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1937 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1938 ** FINALIZE-INHERITANCE is now called on class finalization;
1939 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1940 argument precedence order.
1941 * fixed bug 202: The compiler no longer fails on functions whose
1942 derived types contradict their declared type.
1943 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1944 so it can be non-toplevel.
1945 * The fasl file version number has changed (because of the new
1946 implementation of DEFMACRO).
1947 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1948 safe code, that any length requirement by their type-specifier
1949 argument is valid. The exceptions to this are described in bug
1951 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1952 functions now check the types of their inputs as required by ANSI.
1953 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1954 introduced symbol is DECLAREd to be SPECIAL.
1955 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1956 * fixed bug: :COUNT argument to sequence functions may be negative
1957 * fixed bug: body of DO-SYMBOLS may contain declarations
1958 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1959 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1960 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1961 symbol macro only once
1962 * fixed printing of call frame when argument list is unavailable
1963 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1964 * compiler no longer signals WARNING on unknown keyword
1967 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1968 * A beta-quality port to the mips architecture running Linux,
1969 based on the old CMUCL backend, has been made. It has been tested
1970 on a big-endian kernel, and works sufficiently well to be able to
1971 rebuild itself; it has not been tested in little-endian mode.
1972 * fixed an inconsistency between gencgc.c and purify.c which made
1973 dumping/loading .core files unreliable
1974 * fixed bug 120a: The compiler now deals correctly with IFs where
1975 the consequent is the same as the alternative, instead of
1976 misderiving the return type. (thanks to Alexey Dejneka)
1977 * fixed bug 113: Logical pathnames are now dumpable (the logical
1978 host is resolved at load-time, throwing an error if it is not
1980 * fixed bug 174: FORMAT's error message is slightly clearer when a
1981 non-printing character is used in a format directive.
1982 * fixed several bugs in compiler checking of type declarations, i.e.
1983 violations of the Python "declarations are assertions" principle
1984 (thanks to Alexey Dejneka)
1985 * fixed several bugs in PCL's error checking (thanks to Gerd
1987 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1989 * fixed bug in compilation of functions as first class values
1990 (thanks to Antonio Martinez-Shotton)
1991 * The compiler's handling TYPE-ERRORs which it can prove will
1992 inevitably happen at runtime has been cleaned up and corrected
1993 in several ways. (thanks to Alexey Dejneka)
1994 * improved argument type checking for various basic arithmetic
1995 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1996 lavished on them in the past that they can be compiled in many
1997 ways in different special cases
1998 * fixed bug 181: compiler checks validity of user supplied type
2000 * cleaned up code flushing in optimization: Function calls which
2001 should signal errors for safety purposes (e.g. which ANSI says
2002 should signal errors when their arguments are of incorrect type)
2003 are no longer optimized away.
2004 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2005 * incremented fasl file version number, because changes in the
2006 implementation of sequence functions like COERCE caused
2007 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2008 in old inline expansions) to become undefined. (Actually these
2009 changes were later undone, so we might very well be binary
2010 compatible with 0.7.7 after all, but leaving the version number
2011 incremented seemed like the simplest and most conservative
2014 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2015 * An alpha-quality port to the parisc architecture running Linux,
2016 based on the old CMUCL backend, has been made. This, even more so
2017 than the other backends, should be considered still a work in
2018 progress; known problems include that the Linux kernel in 64-bit
2019 mode does not propagate the correct sigcontext structure to
2020 userspace, and consequently SBCL on a parisc64 kernel will not
2022 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2023 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2024 suppressed.) Also now that the compiler is looking at declarations
2025 in the environment, it checks optimization declarations as well,
2026 and suppresses inlining when (> DEBUG SPEED).
2027 * More fixes have been made to treatment of floating point exception
2028 treatment and other Unix signals. In particular, floating point
2029 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2030 * The detection and handling of control stack exhaustion (infinite
2031 or very deeply nested recursion) has changed. Stack exhaustion
2032 detection is now done by write-protecting pages at the OS level
2033 and applies at all optimization settings; when found, a
2034 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2035 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2036 be caught using IGNORE-ERRORS.
2037 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2038 correctly when transforming recursive LABELS functions to LETs.
2039 (thanks to Alexey Dejneka)
2040 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2041 that are names of constants or global variables.
2042 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2043 alien routines with docstrings.
2044 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2045 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2047 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2048 (thanks to Eric Marsden)
2049 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2050 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2051 * Bug fix: The SPARC backend can now compile functions involving
2052 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2053 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2054 to a functions expecting a PATHNAME-DESIGNATOR.
2055 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2056 lambda lists. (thanks to Alexey Dejneka)
2057 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2058 functions so that EXPORTing the name of the class doesn't cause
2059 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2060 to Antonio Martinez for reporting this)
2061 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2062 constant-folded without failing an assertion. (thanks to Einar
2063 Floystad Dorum for reporting this)
2064 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2065 types are now dealt with more correctly by the compiler.
2066 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2067 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2068 behaviour analogous to RENAME-FILE. This puts its behaviour more
2069 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2071 * The fasl file version number has changed again. (because of the
2072 bug fix involving the names of PCL MAKE-INSTANCE functions)
2074 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2075 * bug fix: Floating point exceptions are treated much more
2076 consistently on the x86/Linux and PPC/Linux platforms.
2077 * Array initialization with :INITIAL-ELEMENT is now much faster for
2078 cases when the compiler cannot open code the array creation, but
2079 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2080 array accesses have also seen a speed increase.
2081 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2082 specified. (This was at the root of some bad interactions between
2083 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2085 * bug fix: Internal error arguments for undefined functions are now
2086 computed correctly on the PPC/Linux platform.
2087 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2088 Raymond Toy's patch for CMU CL)
2089 * Support for the Solaris 9 operating environment has been included
2090 (thanks to Daniel Merritt)
2091 * A very ugly but hopefully complete draft of the missing FFI chapter
2092 of the manual has been created by reformatting the corresponding
2093 CMU CL manual chapter into (currently very ugly and incoherent)
2094 DocBook and bringing it up to date for SBCL behavior. Thus, the
2095 manual is now essentially complete, at least by my extreme
2096 once-and-only-once standards, whereby it's acceptable to refer to
2097 the doc strings of SB-EXT functions as the primary documentation.
2098 * The fasl file version number has changed again, due to cleanup of
2099 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2100 is no longer a static symbol.)
2102 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2103 * SBCL now builds with OpenMCL (version 0.12) as the
2104 cross-compilation host; also, more progress has been made toward
2105 bootstrapping under CLISP.
2106 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2108 * bug 158 fixed: The compiler can now deal with integer loop
2109 increments different from 1; fixing this turned out also to fix
2111 * bug 169 fixed: no more bogus warnings about using lexical bindings
2112 despite the presence of perfectly good SPECIAL declarations (thanks
2113 to David Lichteblau)
2114 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2115 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2116 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2118 * bug 180 fixed: Method combination specifications no longer ignore
2119 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2120 * bug fix: Structure type predicate functions now check their argument
2121 count as they should.
2122 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2123 correctly. (thanks to Pierre Mai)
2124 * minor incompatible change: The --noprogrammer option is deprecated
2125 in favor of the new --disable-debugger option, which is very similar.
2126 (The major difference is that it takes effect at a slightly different
2127 time at startup, causing handling of errors in --sysinit and
2128 --userinit files will be affected differently.) The
2129 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2130 been added to allow this functionality to be controlled from ordinary
2131 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2132 maintainers, who might want to run non-interactive scripts to
2133 build SBCL cores which will later be used interactively.)
2134 * minor incompatible change: The LOAD function no longer, when given
2135 a wild pathname to load, loads all files matching that pathname.
2136 Instead, an error of type FILE-ERROR is signalled.
2138 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2139 * bug 147 fixed: The compiler preserves its block link/count
2140 invariants more correctly now so that it doesn't crash. (thanks
2142 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2144 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2145 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2146 no-value-for-this-character as NIL (as specified by ANSI).
2147 * HOST-NAMESTRING on physical pathnames now returns a string that is
2148 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2149 (thanks to Christophe Rhodes)
2150 * The Alpha port handles icache flushing more correctly. (thanks to
2152 * More progress has been made toward bootstrapping under CLISP. (thanks
2153 to Christophe Rhodes)
2154 * The fasl file format has changed again, because dynamic loading
2155 on OpenBSD (which has non-ELF object files) motivated some cleanups
2156 in the way that foreign symbols are transformed and passed around.
2157 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2158 is no longer treated as whitespace by the reader, but instead as
2159 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2160 |A
\7fB|, instead of A as it used to.
2162 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2163 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2164 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2165 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2166 this as of 0.7.1.45, but was left out of the previous news section
2167 (thanks to Dan Barlow)
2168 * SBCL now runs on the Solaris operating system on SPARC architectures
2169 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2170 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2171 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2172 * SPARC backend cleanups, allowing builds of cores optimized for V8
2173 and V9 SPARCS, and also emission of code targeted to a particular
2174 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2176 * SBCL is closer to bootstrapping under CLISP, thanks to various
2177 fixes by Christophe Rhodes.
2178 * The fasl file format has changed again, to allow the compiler's
2179 INFO database to support symbol macros.
2180 * The user manual (in doc/) is formatted into HTML more nicely.
2181 (thanks to coreythomas)
2182 * The system is smarter about SUBTYPEP relationships, especially
2183 those involving NOT types (including types such as ATOM which are
2184 represented internally using NOT types). Thus SUBTYPEP is less
2185 likely to return (VALUES NIL NIL) in general, and in particular
2186 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2188 * The fasl file format has changed again, because the internal
2189 representation of types now includes a new slot to support the new
2190 SUBTYPEP-of-NOT-types logic.
2191 * (not a change in the main branch of SBCL, but a related prototype
2192 which can hopefully be merged into the main branch of SBCL in the
2193 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2194 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2195 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2196 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2197 dispatch macro character combinations. (thanks to Alexey Dejneka)
2198 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2199 that superficially look like logical namestrings correctly.
2200 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2202 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2203 default supplied-p) for &optional and &key arguments. (thanks to
2206 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2207 * incompatible change: The compiler is now less aggressive about
2208 tail call optimization, doing it only when (> SPACE DEBUG) or
2209 (> SPEED DEBUG). (This is an incompatible change because there are
2210 programs which relied on the old CMU-CL-style behavior to optimize
2211 away their unbounded recursion which will now die of stack overflow.)
2212 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2213 for non-GENCGC systems has been increased to 20M (since that
2214 seems much closer to the likely performance optimum for modern
2215 systems than the old 4M value was)
2216 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2217 and *DEBUG-PRINT-LEVEL*
2218 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2219 of CMU CL's support for SPARC, and various endianness and other
2220 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2221 * new syntactic sugar for the Unix command line: --load foo.bar is now
2222 an alternate notation for --eval '(load "foo.bar")'.
2224 ** The system now detects stack overflow and handles it gracefully,
2225 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2226 optimization settings. (This is a good thing in general, and
2227 its introduction in this version should be particularly timely
2228 for anyone whose code fails because of suppression of tail
2230 ** The system now hunts for the C variable "environ" in a more
2231 devious way, to avoid segfaults when the C library version differs
2232 between compile time and run time. (thanks to Christophe Rhodes)
2233 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2234 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2235 ** The compiler no longer issues bogus style warnings for undefined
2236 classes in the same source file as the DEFCLASSes which defined
2237 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2239 ** fixes in CONDITION class precedence list for undefined function
2240 errors (thanks to Alexei Dejneka)
2241 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2242 correctly. (thanks to Dan Barlow)
2243 ** portability fixes aiming at bootstrapping under CLISP (thanks
2244 to Dave McDonald and Christophe Rhodes)
2245 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2246 ** fixes in type translation and and type inference (thanks to
2248 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2249 ** various fixes in the new ports (thanks to Dan Barlow)
2250 * several changes related to debugging:
2251 ** suppression of tail recursion, as noted above
2252 ** stack overflow detection, as noted above
2253 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2254 is now the default. (For some time encapsulation has been more
2255 reliable than the breakpoint-based :ENCAPSULATE NIL
2256 implementation, at least on X86 systems; and I just noticed that
2257 encapsulation also seems closer to the spirit of the ANSI
2260 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2262 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2263 up properly again. (There was a packaging bug in 0.7.0 which
2264 left their definitions in SB-SYS::LOAD-FOREIGN and
2265 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2266 vital for most things which interface to C-level interfaces,
2267 like extensions working with sockets or databases or
2268 Perl-compatible regexes or whatever, and the need to fix
2269 this bug is the main reason that 0.7.1 was released so
2271 ** DEFGENERIC is now choosier about the methods it redefines, so that
2272 reLOADing a previously-LOADed file containing DEFGENERICs does
2273 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2274 cycle now works as it should. (thanks to Alexey Dejneka)
2275 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2276 fixed by Christophe Rhodes. (He actually submitted this patch
2277 months ago, and I delayed until after 0.7.0.)
2278 ** Bug 111 (internal compiler confusion about runtime checks on
2279 FUNCTION types) was fixed by Alexey Dejneka.
2280 * Some internal cleanups (getting rid of variables which aren't
2281 needed now that the byte interpreter is gone) caused the fasl
2282 file format number to change again.
2284 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2285 * major incompatible change: The default fasl file extension, i.e. the
2286 default extension for files produced by COMPILE-FILE, has changed
2287 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2289 ** There are many changes in the implementation of the compiler.
2290 SBCL is now essentially a compiler-only implementation of ANSI
2291 Common Lisp. EVAL still "interprets" a few special cases, but
2292 almost all the interesting cases are handled by creating
2293 a LAMBDA expression, calling COMPILE on it, then calling
2294 FUNCALL on the result.
2295 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2296 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2297 Since the code is newer, there might still be some new bugs
2298 (though not as many as before Martin Atzmueller's fixes:-). But
2299 the new code is substantially simpler and clearer, and hopefully
2300 any remaining bugs will be simpler, less fundamental, and more
2301 fixable then the bugs in the old code.
2302 ** The revised compiler is still a little unsteady on its feet.
2304 *** The debugging information it produces (particularly the names
2305 of FUNCTION objects) is sometimes much less useful than what
2306 the old compiler produced.
2307 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2308 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2309 within a MACROLET) has been temporarily weakened.
2310 ** There are new compiler optimizations for various functions:
2311 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2312 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2313 *** the math functions TRUNCATE, FLOOR, and CEILING
2314 *** the function-of-all-trades COERCE
2315 Mostly these should be transparent, but there's one
2316 potentially-annoying problem (bug 117): when the compiler
2317 inline-expands a function and does type analysis on the result,
2318 it can create control paths which have type mismatches, and
2319 when it can't prove that those control paths aren't taken,
2320 it will issue WARNINGs about the type mismatches. This is
2321 a particular problem in practice for the new sequence functions.
2322 It's not clear how this should be fixed, and for now, a
2323 workaround is given in the entry for 117 in the BUGS file.
2324 ** (Because of the interaction between the two previous items --
2325 occasional inlining problems and new inline expansions -- some
2326 of the new sequence function optimizations won't really kick in
2327 completely until debugging information, and then inlining, are
2328 straightened out in some future version.)
2329 * minor incompatible changes:
2330 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2331 DIRECTORY no longer implicitly promotes NIL slots of its
2332 pathname argument to :WILD. In particular, when you ask for the
2333 contents of a directory (which you used to be able to do without
2334 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2335 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2336 ** changes in behavior that ANSI explicitly defines to be
2337 implementation dependent:
2338 *** The new compiler-only implementation still conforms with ANSI,
2339 but acts a little different than before. Besides the obvious
2340 changes in performance tradeoffs (that the cost per form passed
2341 to EVAL has gone up, and the cost per form executed by EVAL
2342 has gone down), the behavior of the system changes a little
2343 because there are no longer any interpreted function objects.
2344 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2345 e.g. doing COMPILE on the output of interactive DEFUN is
2347 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2349 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2350 has changed from (:CL) to NIL.
2351 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2352 e.g. #\Newline and #\Tab, don't have names specified in the
2353 ANSI Common Lisp standard, is now based on their ASCII symbolic
2354 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2355 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2356 are no longer used for output.
2357 ** changes in internal implementation constants:
2358 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2359 4 million. (If your application spends a lot of time GCing and
2360 you have a lot of RAM, you might want to experiment with
2361 increasing it even more.)
2362 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2363 However, almost all old code should still continue to work without
2364 immediate update, as SB-C-CALL is now a (deprecated) nickname
2366 ** Old operator names in the style DEF-FOO are now deprecated in
2367 favor of new corresponding names DEFINE-FOO, for consistency with
2368 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2369 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2370 internal symbols, but a few supported extensions like
2371 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2372 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2373 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2374 etc. as you get deeper into recursive calls to the debugger
2375 command loop, instead of the old "5]", "5]]", "5]]]"
2376 sequence. (I was motivated to do this when squabbles between
2377 ILISP and SBCL left me very deeply nested in the debugger. In the
2378 short term, this change will probably provoke more ILISP/SBCL
2379 squabbles, but hopefully it will be an improvement in the long run.)
2380 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2381 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2382 function, since declaiming return types involving aliens is
2383 (1) annoyingly messy to do by hand and (2) vital to efficient
2384 compilation of code which calls such functions.
2385 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2386 longer reexported by the SB-EXT package. They're solely useful
2387 for alien code, so it seems more logical that you should get
2388 them from the SB-ALIEN package, not in SB-EXT.
2389 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2390 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2391 features. Instead, the code that they used to control is always
2392 built into the system.
2393 * many other bug fixes
2394 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2395 advantage of the new EVAL-WHEN stuff and to clean them up in
2396 general, and they are now more ANSI-compliant in a number of
2397 ways. Martin Atzmueller is responsible for a lot of this.
2398 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2400 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2401 *** correct ERROR type for various file operations
2402 *** some fixes for Lisp streams
2403 *** DEFMETHOD syntax checking
2404 *** changing old weird representation of debug information as
2405 strings (which, among their other deficiencies, don't transform
2406 correctly when you rename packages, and don't change their
2407 print representation when you change things like *PACKAGE*
2408 and *PRINT-LENGTH*) to symbols and lists of symbols
2409 He also made several improvements and fixed several bugs in DESCRIBE.
2410 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2412 *** misbehavior of WRITE-STRING/WRITE-LINE
2413 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2414 and several other LOOP problems as well
2415 *** DIRECTORY when similar filenames are present
2416 *** DEFGENERIC with :METHOD options
2417 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2418 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2419 *** argument ordering in FIND with :TEST option
2420 *** mishandled package designator argument in APROPOS-LIST
2421 *** various problems in the backquote readmacro
2422 *** a bug in APROPOS
2423 *** probably some others that I'm not describing very well here,
2424 since the CVS log documents them by reference to sbcl-devel
2425 messages, and the SourceForge archives aren't working well.:-(
2426 ** Dan Barlow improved the Alpha port (and is making progress on the
2427 PPC port, for those of you who think different).
2428 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2429 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2430 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2431 continues to work on the SPARC port (for those of you in a position
2432 to look down upon our little PC-compatible boxes from a great height).
2433 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2434 and uses that copy, rather than the current dynamic value, when
2435 it's trying to decide whether to truncate output. Thus e.g.
2436 (let ((*print-lines* 50))
2437 (pprint-logical-block (stream nil)
2439 (let ((*print-lines* 8))
2440 (print (aref possiblybigthings i) stream)))))
2441 should now truncate the logical block only at 50 lines, instead of
2442 often truncating it at 8 lines, as it did before.
2443 * The doc/cmucl/ directory, containing old CMU CL documentation
2444 from the time of the fork, is no longer part of the base system.
2445 SourceForge has shut down its anonymous FTP service, and with it
2446 my original plan for distributing the old CMU CL documentation
2447 there. For now, if you need these files you can download an old
2448 SBCL source release and extract them from it.
2449 * The fasl file version number changed again, for dozens of reasons,
2450 some of which are apparent above.
2452 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2453 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2454 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2455 patch, so that DISASSEMBLE output is much nicer.
2456 * The code in the SB-PROFILE package now seems reasonably stable.
2457 I still haven't decided what the final interface should look like
2458 (I'd like PROFILE to interact cleanly with TRACE, since both
2459 facilities use function encapsulation) but if you have a need
2460 for profiling now, you can probably use it successfully with
2461 the current CMU-CL-style interface.
2462 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2463 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2464 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2465 link now returns the dangling link itself, and for similar
2466 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2467 link itself. (In these cases the old code signalled an error and
2468 looped endlessly, respectively.) Thus, DIRECTORY now works even
2469 in the presence of dangling and cyclic symbolic links.
2470 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2471 is now a supported extension again, since the consensus on
2472 sbcl-devel was that it can be useful for ordinary development
2473 work, not just for debugging SBCL itself.
2474 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2475 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2476 that currently-defined functions might be redefined later with
2477 different return types.
2478 * Hash tables can be printed readably, as inspired by CMU CL code
2479 of Eric Marsden and SBCL code of Martin Atzmueller.
2480 * better error handling in CLOS method combination, thanks to
2481 Martin Atzmueller porting Pierre Mai's CMU CL patches
2482 * more overflow fixes for >16Mbyte I/O buffers
2483 * A bug in READ has been fixed, so that now a single Ctrl-D
2484 character suffices to cause end-of-file on character streams.
2485 In particular, now you only need one Ctrl-D at the command
2486 line (not two) to exit SBCL.
2487 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2489 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2490 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2491 does the right thing.
2492 * The implementation of some type tests, especially for CONDITION
2493 types, is now tidier and maybe faster, due to CMU CL code
2494 originally by Douglas Crosher, ported by Martin Atzmueller.
2495 * Some math functions have been fixed, and there are new
2496 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2497 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2498 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2499 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2500 likely to be built into sbcl-0.7.0.)
2501 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2502 CL, ported to SBCL by Martin Atzmueller)
2503 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2504 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2505 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2506 * There's a new slam.sh hack to shorten the edit/compile/debug
2507 cycle for low-level changes to SBCL itself, and a new
2508 :SB-AFTER-XC-CORE target feature to control the generation of
2509 the after-xc.core file needed by slam.sh.
2510 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2511 COMPILE-FILE is no longer supported, so that now every function
2512 gets an entry point, so that block compilation looks a little
2513 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2514 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2515 returns the number of bytes consed since the system started,
2516 rather than the number consed since the first time the function
2517 was called. (The new definition parallels ANSI functions like
2518 CL:GET-INTERNAL-RUN-TIME.)
2519 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2520 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2521 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2522 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2523 * Fasl file version numbers are now independent of the target CPU,
2524 since historically most system changes which required version
2525 number changes have affected all CPUs equally. Similarly,
2526 the byte fasl file version is now equal to the ordinary
2529 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2530 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2531 is no longer recognized. I apologize for this, because it was
2532 listed in SB-EXT as a supported extension, but I found that
2533 its existing behavior was poorly specified, as well as incorrectly
2534 specified, and it looked like too much of a mess to straighten it
2535 out. I have enough on my hands trying to get ANSI stuff to work..
2536 * many patches ported from CMU CL by Martin Atzmueller, with
2537 half a dozen bug fixes in pretty-printing and the debugger, and
2538 half a dozen others elsewhere
2539 * fixed bug 13: Floating point infinities are now supported again.
2540 They might still be a little bit flaky, but thanks to bug reports
2541 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2542 as flaky as they were.
2543 * The --noprogrammer command line option is now supported. (Its
2544 behavior is slightly different in detail from what the old man
2545 page claimed it would do, but it's still appropriate under the
2546 same circumstances that the man page talks about.)
2547 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2548 are now supported, and enabled by default. Thus, the compiler can
2549 handle many floating point and complex operations much less
2550 inefficiently. (Thus e.g. you can implement a complex FFT
2552 * The compiler now detects type mismatches between DECLAIM FTYPE
2553 and DEFUN better, and implements CHECK-TYPE more correctly, and
2554 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2555 thanks to patches from Martin Atzmueller.
2556 * various fixes to make the cross-compiler more portable to
2557 ANSI-conforming-but-different cross-compilation hosts (notably
2558 Lispworks for Windows, following bug reports from Arthur Lemmens)
2559 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2560 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2561 fixed thanks to Pierre Mai's CMU CL patches.
2562 * fixes to keep the system from overflowing internal counters when
2563 it tries to use i/o buffers larger than 16M bytes
2564 * fixed bug 45a: Various internal functions required to support
2565 complex special functions have been merged from CMU CL sources.
2566 (When I was first setting up SBCL, I misunderstood a compile-time
2567 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2568 * improved support for type intersection and union, fixing bug 12
2569 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2570 more obscure bugs as well
2571 * some steps toward byte-compiling non-performance-critical
2572 parts of the system, courtesy of patches from Martin Atzmueller
2573 * Christophe Rhodes has made some debian packages of sbcl at
2574 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2575 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2576 stable, but are nonetheless usable. When he's ready, I'd be happy
2577 to add them to the SourceForge "File Releases" section. (And if
2578 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2579 * new fasl file format version number (because of changes in
2580 internal representation of (OR ..) types to accommodate the new
2581 support for (AND ..) types, among other things)
2583 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2584 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2586 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2588 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2589 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2590 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2591 * DESCRIBE now works on structure objects again.
2592 * Most function call argument type mismatches are now handled as
2593 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2594 know whether the function will be redefined before the call is
2595 executed. (The compiler could flag local calls with full WARNINGs,
2596 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2597 it doesn't keep track of enough information to know whether calls
2598 are local in this sense.)
2599 * Compiler output is now more verbose, with messages truncated
2600 later than before. (There should be some supported way for users
2601 to override the default verbosity, but I haven't decided how to
2602 provide it yet, so this behavior is still controlled by the internal
2603 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2604 src/compiler/ir1util.lisp.)
2605 * Fasl file format version numbers have increased again, because
2606 support for the Gray streams extension changes the layout of the
2607 system's STREAM objects.
2608 * The Gray subclassable streams extension now works, thanks to a
2609 patch from Martin Atzmueller.
2610 * The full LOAD-FOREIGN extension (not just the primitive
2611 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2612 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2613 but like most other programs, it defaults to copying the Unix
2614 environment from the original process instead of starting the
2615 new process in an empty environment.
2616 * Extensions which manipulate the Unix environment now support
2617 an :ENVIRONMENT keyword option which doesn't smash case or
2618 do other bad things. The CMU-CL-style :ENV option is retained
2619 for porting convenience.
2620 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2621 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2623 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2625 * A patch from Martin Atzmueller seems to have solved the SIGINT
2626 problem, and as far as we know, signal-handling now works cleanly.
2627 (If you find any new bugs, please report them!)
2628 * The system no longer defaults Lisp source file names to types
2629 ".l", ".cl", or ".lsp", but only to ".lisp".
2630 * The compiler no longer uses special default file extensions for
2631 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2632 seems to expect a single default extension for all compiled code,
2633 and there's no compelling reason to try to stretch the standard
2634 to allow two different extensions.) Instead, byte-compiled files
2635 default to the same extension as native-compiled files.
2636 * Fasl file format version numbers have increased again, because
2637 a rearrangement of internal implementation packages made some
2638 dumped symbols in old fasl files unreadable in new cores.
2639 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2640 many fewer weird special cases.
2641 * Bug #17 (differing COMPILE-FILE behavior between logical and
2642 physical pathnames) has been fixed, and some related misbehavior too,
2643 thanks to a patch from Martin Atzmueller.
2644 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2645 by Tim Moore, ported to SBCL by Martin Atzmueller.
2646 * Martin Atzmueller fixed several filesystem-related problems,
2647 including bug #36, in part by porting CMU CL patches, which were
2648 written in part by Paul Werkowski.
2649 * More compiler warnings in src/runtime/ are gone, thanks to
2650 more patches from Martin Atzmueller.
2651 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2654 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2656 * DESCRIBE now works on CONDITION objects.
2657 * The debugger now handles errors which arise when trying to print
2658 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2660 * The build system now uses an additional file, customize-target-features.lisp,
2661 to allow local modifications to the target *FEATURES* list. (The point of
2662 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2663 debugging features enabled, without having to worry about propagating it
2664 into everyone's system when I do a "cvs update".) When no
2665 customize-target-features.lisp file exists, the target *FEATURES* list
2666 should be constructed the same way as before.
2667 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2668 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2669 warn about behavior which is undefined under the ANSI spec. Note: This
2670 is specified by ANSI, but it's not very popular with programmers.
2671 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2672 macro in the SBCL sources for an example of a workaround which you
2673 might use to make portable ANSI-standard code which does what you want.
2674 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2675 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2676 when it's used as a non-toplevel form. (This is required in order
2677 to implement the DEFCONSTANT-EQX macro.)
2678 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2679 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2680 funny things, due to bugs in the implementation of EVAL-WHEN
2681 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2682 non-toplevel DEFCONSTANTs are uncommon.)
2683 * The core file version number and fasl file version number have been
2684 incremented, because the old noncompliant DEFCONSTANT behavior involved
2685 calling functions which no longer exist, and because I also took the
2686 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2687 * fixed bug 1 (error handling before read-eval-print loop starts), and
2688 redid debugger restarts and related debugger commands somewhat while
2690 ** The QUIT debugger command is gone, since it did something
2691 rather different than the SB-EXT:QUIT command, and since it never
2692 worked properly outside the main toplevel read/eval/print loop.
2693 Invoking the new TOPLEVEL restart provides the same functionality.
2694 ** The GO debugger command is also gone, since you can just invoke
2695 the CONTINUE restart directly instead.
2696 ** The TOP debugger command is also gone, since it's redundant with the
2697 FRAME 0 command, and since it interfered with abbreviations for the
2699 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2701 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2702 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2703 some porting work by Martin Atzmueller)
2704 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2705 with a boatload of patches which he ported to SBCL. Now that those
2707 ** The system tries to make sure that its low-priority messages
2708 are prefixed by semicolons, to help people who like to use
2709 syntax highlighting in their ILISP buffer. (This patch
2710 was originally due to Raymond Toy.)
2711 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2712 patches originally written by Raymond Toy.
2713 ** The compiler understands coercion between single-value and
2714 multiple-VALUES type expressions better, getting rid of some very
2715 weird behavior, thanks to patches originally by Robert MacLachlan
2716 and Douglas Crosher.
2717 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2718 lambda lists, thanks to a patch originally by Pierre Mai.
2719 ** The system no longer bogusly warns about "abbreviated type
2721 ** The compiler gets less confused by inlining and RETURN-FROM,
2722 thanks to some patches originally by Tim Moore.
2723 ** The system no longer hangs when dumping circular lists to fasl
2724 files, thanks to a patch originally from Douglas Crosher.
2725 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2726 undefined function error.
2727 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2728 in structure slot initforms) for now, documented workaround instead:-|
2729 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2730 * fixed bug 5: added stubs for various Gray stream functions called
2731 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2732 installed, at least appropriate type errors are generated
2733 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2734 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2735 consistently in DEFMETHOD forms.
2736 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2737 it doesn't seem to affect SBCL after all
2738 * The C runtime system now builds with better optimization and many
2739 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2741 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2743 * The system is now under CVS at SourceForge (instead of the
2744 CVS repository on my home machine).
2745 * The new signal handling code has been tweaked to treat register
2746 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2747 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2748 downstream, e.g. in debug-int.lisp, has implicit dependencies
2749 on the unsignedness of integer representation of machine words,
2750 and that caused the system to bomb out with infinite regress
2751 when trying to recover from type errors involving signed values,
2752 e.g. (BUTLAST '(1 2 3) -1).)
2753 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2754 (This was one of the bugs Peter Van Eynde reported back in July.)
2755 * The system now uses code inspired by Colin Walters' O(N)
2756 implementation of MAP (from the cmucl-imp@cons.org mailing
2757 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2758 inline the MAP operation, and there is more than one
2759 sequence argument to the MAP call (so that it can't just
2760 do ETYPECASE once and for all based on the type of the
2761 single sequence argument). (The old non-inline implementation
2762 of the general M-argument sequence-of-length-N case required
2763 O(M*N*N) time when any of the sequence arguments were LISTs.)
2764 * The QUIT :UNIX-CODE keyword argument has been renamed to
2765 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2766 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2768 * Raymond Wiker's patches to port dynamic loading from Linux to
2769 FreeBSD have been added.
2770 * The BUGS file is now more nearly up to date, thanks in large part
2771 to Martin Atzmueller's review of it.
2772 * The debugger now flushes standard output streams before it begins
2773 its output ("debugger invoked" and so forth).
2774 * The core version number and fasl file version number have both
2775 been incremented, because of incompatible changes in the layout
2777 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2778 used to not be called for a saved Lisp image.)
2779 * Martin Atzmueller's version of a patch to fix a compiler crash,
2780 as posted on sbcl-devel 13 September 2000, has been installed.
2781 * Instead of installing Martin Atzmueller's patch for the
2782 compiler transform for SUBSEQ, I deleted the compiler transform,
2783 and transforms for some similar consing operations.
2784 * A bug in signal handling which kept TRACE from working on OpenBSD
2786 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2787 away by constant folding
2788 * The system now defines its address space constants in one place
2789 (in the Lisp sources), and propagates them automatically elsewhere
2790 (through GENESIS and the sbcl.h file). Therefore, patching the
2791 address map is less unnecessarily tedious and error-prone. The
2792 Lisp names of address space constants have also been systematized.
2793 * CVS tags like dollar-Header-dollar have been removed from
2794 the sources, because they have never saved me trouble and
2795 they've been source of trouble working with patches and other
2796 diff-related operations.
2797 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2798 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2800 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2802 * The system has been ported to OpenBSD.
2803 * The system now compiles with a simple "sh make.sh" on the systems
2804 that it's supported on. I.e., now you no longer need to tweak
2805 text in the target-features.lisp-expr and symlinks in src/runtime/
2806 by hand, the make.sh takes care of it for you.
2807 * The system is no longer so grossly inefficient when compiling code
2808 involving vectors implemented as general (not simple) vectors (VECTOR T),
2809 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2810 which dares to use the various sequence functions on non-simple
2811 vectors, takes less of a performance hit.
2812 * There is now a primitive type predicate VECTOR-T-P
2813 to test for the (VECTOR T) type, so that e.g.
2814 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2815 can now be compiled with some semblance of efficiency. (The old code
2816 turned the type declaration into a full call to %TYPEP at runtime!)
2817 * AREF on (VECTOR T) is still not fast, since it's still compiled
2818 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2819 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2820 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2821 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2822 without making full calls to SUBTYPEP at runtime.
2823 (Some analogous efficiency issues for non-simple vectors specialized to
2824 element types other than T, or for non-simple multidimensional arrays,
2825 have not been addressed. They could almost certainly be handled the
2826 same way if anyone is motivated to do so.)
2827 * The changes in array handling break binary compatibility, so
2828 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2829 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2830 instead of (VALUES T T).
2831 * By following the instructions that Dan Barlow posted to sbcl-devel
2832 on 2 July 2000, I was able to enable primitive dynamic object
2833 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2834 functionality is not implemented (since it calls ld to resolve
2835 library references automatically, requiring RUN-PROGRAM for its
2836 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2837 try to resolve library references) is now supported.
2838 * The system now flushes the standard output streams when it terminates,
2839 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2840 them at several other probably-convenient times, e.g. in each pass of
2841 the toplevel read-eval-print loop, and after evaluating a form given
2842 as an "--eval" command-line option. (These changes were motivated by a
2843 discussion of stream flushing issues on cmucl-imp in August 2000.)
2844 * The source transform for TYPEP of array types no longer assumes
2845 that an array whose element type is a not-yet-defined type
2846 is implemented as an array of T, but instead punts, so that the
2847 type will be interpreted at runtime.
2848 * There is now some support for cross-compiling in make.sh: each of
2849 the phases of make.sh has its own script. (This should be transparent
2850 to people doing ordinary, non-cross-compile builds.)
2851 * Since my laptop doesn't have hundreds of megabytes of memory like
2852 my desktop machine, I became more motivated to do some items on
2853 my to-do list in order to reduce the size of the system a little:
2854 ** Arrange for various needed-only-at-cold-init things to be
2855 uninterned after cold init. To support this, those things have
2856 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2857 symbols with such names are now uninterned after cold init).
2858 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2859 fasl files for cold load.
2860 ** Remove the old compiler structure pooling code (which used to
2861 be conditional on the target feature :SB-ALLOC) completely.
2862 ** Redo the representation of some data in cold init to be more compact.
2863 (I also looked into supporting byte compiled code at bootstrap time,
2864 which would probably reduce the size of the system a lot, but that
2865 looked too complicated, so I punted for now.)
2866 * The maximum signal nesting depth in the src/runtime/ support code has
2867 been reduced from 4096 to 256. (I don't know any reason for the very
2868 large old value. If the new smaller value turns out to break something,
2869 I'll probably just bump it back up.)
2870 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2872 * Many, many bugs reported by Peter Van Eynde have been added to
2873 the BUGS list; some have even been fixed.
2874 * While enabling dynamic object file loading, I tried to make the
2875 code easier to understand, renaming various functions and variables
2876 with less ambiguous names, and changing some function calling
2877 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2878 * While trying to figure out how to do the OpenBSD port, I tried to
2879 clean up some of the code in src/runtime/. In particular, I dropped
2880 support for non-POSIX signal handling, added various comments,
2881 tweaked the code to reduce the number of compilation warnings, and
2882 renamed some files to increase consistency.
2883 * To support the new automatic configuration functionality in make.sh,
2884 the source file target-features.lisp-expr has been replaced with the
2885 source file base-target-features.lisp-expr and the machine-generated
2886 file local-target-features.lisp-expr.
2887 * fixed a stupid quoting error in make.sh so that using CMU CL
2888 "lisp -batch" as cross-compilation host works again
2890 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2892 * DESCRIBE no longer tries to call itself recursively to describe
2893 bound/fbound values, so that it no longer fails on symbols which are
2894 bound to themselves (like keywords, T, and NIL).
2895 * DESCRIBE now works on generic functions.
2896 * The printer now prints less-screwed-up representations of closures
2897 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2898 * A private symbol is used instead of the :EMPTY keyword previously
2899 used to mark empty slots in hash tables. Thus
2900 (DEFVAR *HT* (MAKE-HASH-TABLE))
2901 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2902 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2903 now does what ANSI says that it should. (You can still get
2904 similar noncompliant behavior if bang on the hash table
2905 implementation with all the symbols you get back from
2906 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2907 This breaks binary compatibility, since tests for equality to
2908 :EMPTY are wired into things like the macroexpansion of
2909 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2911 * There's now a minimal placeholder implementation for CL:STEP,
2912 as required by ANSI.
2913 * An obscure bug in the interaction of the normal compiler, the byte
2914 compiler, inlining, and structure predicates has been patched
2915 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2916 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2918 * Missing ordinary arguments in a macro call are now detected even
2919 when the macro lambda list contains &KEY or &REST.
2920 * The debugger no longer complains about encountering the top of the
2921 stack when you type "FRAME 0" to explicitly instruct it to go to
2922 the top of the stack. And it now prints the frame you request even
2923 if it's the current frame (instead of saying "You are here.").
2924 * As specified by ANSI, the system now always prints keywords
2925 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2927 * The default initial SIZE of HASH-TABLEs is now smaller.
2928 * Type information from CLOS class dispatch is now propagated
2929 into DEFMETHOD bodies, so that e.g.
2930 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2932 is now basically equivalent to
2933 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2934 (DECLARE (TYPE SINGLE-FLOAT X))
2936 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2937 operation, without having to do run-time type dispatch.
2938 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2939 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2940 * Since I don't seem to be making big file reorganizations very often
2941 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2942 is overflowing my ability to conveniently back them up), I've finally
2943 checked the system into CVS. (The CVS repository is on my home system,
2944 not at SourceForge -- putting it on SourceForge might come later.)
2945 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2946 high-level GC-NOTIFY-FOO functions send their output. (There's
2947 still very little control of where low-level verbose GC functions
2948 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2949 less than it used to -- the GC-NOTIFY-FOO functions are now under
2950 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2951 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2952 in only one place in the source code, and propagates it automatically
2953 everywhere that it's needed. Thus e.g. when I bump the version from
2954 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2955 * The C source files now include boilerplate legalese and documentation
2956 at the head of each file (just as the Lisp source files already did).
2957 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2958 to his page will be replaced with a link to his new CLiki service.
2960 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2962 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2963 * The build process now looks for GNU make under the default name "gmake",
2964 instead of "make" as it used to. If GNU make is not available as "gmake"
2965 on your system, you can change this default behavior by setting the
2966 GNUMAKE environment variable.
2967 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2968 can build without error under CMU CL.
2970 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2972 * There is now a partial SBCL user manual (with some new text and some
2973 text cribbed from the CMU CL manual).
2974 * The beginnings of a profiler have been added (starting with the
2975 CMU CL profiler and simplifying and cleaning up). Eventually the
2976 main interface should be through the TRACE macro, but for now,
2977 it's still accessed through vaguely CMU-CL-style functions and macros
2978 exported from the package SB-PROFILE.
2979 * Some problems left over from porting CMU CL to the new
2980 cross-compilation bootstrap process have been cleaned up:
2981 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2982 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2983 part when getting rid of the compiler *BACKEND* stuff.)
2984 ** The value of *NULL-TYPE* was screwed up, because it was
2985 being initialized before the type system knew the final
2986 definition of the 'NULL type. This screwed up several key
2987 optimizations in the compiler, causing inefficiency in all sorts
2988 of places. (I found it because I wanted to understand why
2989 GET-INTERNAL-RUN-TIME was consing.)
2990 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2991 PROCLAIM FTYPE information. Unfortunately this broke binary
2992 compatibility again, since now the forms output by DEFGENERIC
2993 to refer to functions which didn't exist in 0.6.3.
2994 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2995 can use the new (as of 0.6.3) transform for SOME into MAP into
2997 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2998 values to (INTEGER 0 1000000), so that the time code will no longer
2999 occasionally get blown up by Linux returning 1000000 microseconds
3000 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3001 its output conform to the ANSI spec. (Alas, this makes its output
3002 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3003 * A full call to MAP NIL with a single sequence argument no longer conses.
3004 * fixes to problems pointed out by Martin Atzmueller:
3005 * The manual page no longer talks about multiprocessing as though
3006 it were currently supported.
3007 * The ILISP support patches have been removed from the distribution,
3008 because as of version 5.10.1, ILISP now supports SBCL without us
3009 having to maintain patches.
3010 * added a modified version of Raymond Toy's recent CMU CL patch for
3011 EQUALP comparison of HASH-TABLE
3013 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3015 * The system still can't cross-compile itself with
3016 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3017 entails), but at least it can compile more of itself that way
3018 than it used to be able to, and various buglets which were uncovered
3019 by trying to cross-compile itself that way have now been fixed.
3020 * This release breaks binary compatibility again. This time
3021 at least I've incremented the FASL file format version to 2, so that the
3022 problem can be detected reliably instead of just causing weird errors.
3023 * various new style warnings:
3024 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3025 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3026 ** using the lexical binding of a variable named in the *FOO* style
3027 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3028 as specified by ANSI.
3029 * *RANDOM-STATE* is no longer automatically initialized from
3030 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3031 default behavior of the system is to repeat its behavior every time
3032 it's run. If you'd like to change this behavior, you can always
3033 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3034 old convention there was no comparably easy way to get the system to
3035 repeat its behavior every time it was run.
3036 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3037 FTYPE declarations has been removed, in favor of their ANSI
3038 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3039 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3040 the types of their sequence arguments can be determined at compile time.
3041 This is done through a new open code expansion for MAP which eliminates
3042 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3043 argument types can be determined at compile time.
3044 * The optimizer now transforms COERCE into an identity operation when it
3045 can prove that the coerced object is already of the correct type. (This
3046 can be a win for machine generated code, including the output of other
3047 optimization transforms, such as the MAP transform above.)
3048 * Credit information has been moved from source file headers into CREDITS.
3049 * Source file headers have been made more standard.
3050 * The CASE macro now compiles without complaining even when it has
3053 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3055 * (Note that the way that the PCL macroexpansions were rewritten
3056 to accommodate the change in DEFGENERIC below breaks binary
3057 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3058 not run under sbcl-0.6.2. Once we get out of alpha releases,
3059 i.e. hit release 1.0.0, we'll probably try to maintain binary
3060 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3061 and sbcl-1.4.4. Until then, however, it might be fairly common
3062 for maintenance releases to break binary compatibility.)
3063 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3065 * The reporting of unhandled signals has been changed to print some
3066 explanatory text as well as the report form. (Previously only
3067 the report form was printed.)
3068 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3069 it defines, so that the compiler no longer issues undefined function
3070 warnings for compiled-but-not-yet-loaded generic functions.
3071 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3072 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3073 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3074 section of the ANSI spec are supported.
3075 * The "" nickname for the "KEYWORD" package has been removed.
3076 The reader still handles symbol tokens which begin with a package marker
3077 as keywords, but it doesn't expose its mechanism for doing so in the
3078 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3079 * The system now issues STYLE-WARNINGs for contradictory TYPE
3080 proclamations. (Warnings for contradictory FTYPE proclamations would
3081 be nice too, but those can't be done usefully unless the type system
3082 is made smarter about FUNCTION types.)
3083 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3084 systematized, so that "*target-*.lisp is supposed to exist only on the
3085 target and imply that there's a related file which exists on the
3086 host, and *host-*.lisp is supposed to exist only on the host and imply
3087 that there's a related file which exists on the target. This involves a
3088 lot of renaming. Hopefully the acute confusion caused by the renaming
3089 will be justified by the reduction in chronic confusion..
3090 ** runtime-type.lisp -> early-target-type.lisp
3091 ** target-type.lisp -> late-target-type.lisp
3092 ** early-host-format.lisp -> early-format.lisp
3093 ** late-host-format.lisp -> late-format.lisp
3094 ** host-error.lisp -> misc-error.lisp
3095 ** early-error.lisp -> early-target-error.lisp
3096 ** late-error.lisp -> late-target-error.lisp
3097 ** host-defboot.lisp -> early-defboot.lisp
3098 ** code/misc.lisp -> code/target-misc.lisp
3099 ** code/host-misc.lisp -> code/misc.lisp
3100 ** code/numbers.lisp -> code/target-numbers.lisp
3101 ** code/early-numbers.lisp -> numbers.lisp
3102 ** early-host-type.lisp -> early-type.lisp
3103 ** late-host-type.lisp -> late-type.lisp
3104 ** host-typep.lisp -> typep.lisp
3105 ** load.lisp -> target-load.lisp
3106 ** host-load.lisp -> load.lisp
3107 ** host-disassem.lisp -> disassem.lisp
3108 ** host-insts.lisp -> insts.lisp
3109 ** byte-comp.lisp -> target-byte-comp.lisp
3110 ** host-byte-comp.lisp -> byte-comp.lisp
3111 ** host-signal.lisp -> signal.lisp
3112 ** host-defstruct.lisp -> defstruct.lisp
3113 ** late-target-type.lisp -> deftypes-for-target.lisp
3114 Furthermore, several other previously target-only files foo.lisp (e.g.
3115 hash-table.lisp and random.lisp) have been split into a target-and-host
3116 foo.lisp file and a target-only target-foo.lisp file, with their key type
3117 definitions in the target-and-host part, so that the cross-compiler will
3118 know more about target types.
3119 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3120 gone away. In their place are various *BACKEND-FOO* variables
3121 corresponding to the slots of the old structure.
3122 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3123 into the target SBCL has been fixed.
3124 * The chill.lisp system for loading cold code into a running SBCL
3126 * Support for the CMU CL "scavenger hook" extension has been removed.
3127 (It was undocumented and unused in the CMU CL sources that SBCL was
3128 derived from, and stale in sbcl-0.6.1.)
3129 * Various errors in the cross-compiler type system were detected
3130 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3131 (enabling various consistency checks). Many of them were fixed,
3132 but some hard problems remain, so the compiler is back to
3133 running without *TYPE-SYSTEM-INITIALIZED* for now.
3134 * As part of the cross-compiler type system cleanup, I implemented
3135 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3136 * I have started adding UNCROSS calls throughout the type system
3137 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3138 on cross-compiler input files will be able to go away, and various
3140 * CONSTANTP now returns true for quoted forms (as explicitly required
3143 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3145 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3146 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3147 (MAKE-LIST -1). In the longer term, it ought to become true
3148 that declarations are assertions even at SAFETY 1. For now, it's not
3149 quite true even at SAFETY 3, but it's at least more nearly true..
3150 (Note that this change seems to increases the size of the system by
3151 O(5%) and to decrease the speed of the compiler by 20% or more.)
3152 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3153 for the problem of printing dozens of lines of distracting information
3154 about low-level system machinery as part of the top stack frame
3155 on entry to the debugger when an undefined function was called.
3156 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3157 is not reset to COMMON-LISP-USER.
3158 * Compilation of stuff related to dyncount.lisp has been made conditional
3159 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3160 smaller. The various dyncount-related symbols have been moved into
3161 a new "SB-DYNCOUNT" package.
3162 * tty-inspect.lisp has been renamed to inspect.lisp.
3163 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3164 feature has gone away. (When we eventually port to other flavors of
3165 libc and/or Unix, we'll try to make the differences between flavors
3166 invisible at the user level.)
3167 * Various other *FEATURES* tags, and/or their associated conditionals,
3168 have been removed if obsolescent, or given better documentation, or
3169 sometimes given more-mnemonic names.
3171 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3173 * tidied up "make.sh" script
3174 * tidied up system directory structure
3175 * better "clean.sh" behavior
3176 * added doc/FOR-CMUCL-DEVELOPERS
3177 * many many small tweaks to output format, e.g. removing possibly-confusing
3178 trailing #\. character in DESCRIBE-INSTANCE
3179 * (EQUALP #\A 'A) no longer signals an error.
3180 * new hashing code, including EQUALP hashing
3181 * tidied up Lisp initialization and toplevel
3182 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3183 * command line argument processing
3184 * added POSIX-GETENV function to deal with Unix-ish environment variables
3185 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3186 terminating SBCL on EOF
3187 * non-verbose GC by default
3188 * There is no more "sbcl" shell script; the sbcl file is now the C
3189 runtime executable (just like CMU CL).
3190 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3192 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3193 symbol and package information as Lisp native symbol and package objects
3194 instead of strings naming symbols and strings naming packages. This way,
3195 whenever packages are renamed (as in warm init), debug information is
3196 transformed along with everything else.
3197 * tweaked the optimization policy declarations which control the building
3198 of SBCL itself. Now, among other things, the system no longer saves
3199 source location debugging information. (This helps two problems at once
3200 by reducing SBCL size and by keeping SBCL from trying to look for its
3201 sources -- which may not exist -- when reporting errors.)
3202 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3203 debugging and testing purposes
3204 * cleaned up printing, making the printer call PRINT-OBJECT for
3205 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3206 methods, giving nearly-ANSI behavior
3207 * converted almost all special variables to use *FOO* naming convention
3208 * deleted PARSE-TIME functionality, since it can be done portably
3209 * moved some files out of cold init into warm init
3210 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3212 * regularized formatting of source files
3213 * added an install.sh script
3214 * fixed ridiculous memory usage of cross-compiler by making
3215 compiler/alloc.lisp not try to do pooling unless it can hook
3216 itself into the GC of the cross-compilation host. Now the system
3217 builds nicely on my old laptop.
3218 * added :SB-ALLOC in target-features.lisp-expr
3219 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3220 was not implemented)
3221 * re-did condition handling and note reporting in the compiler. Notes
3222 are no longer handled by signalling conditions. Style warnings
3223 and warnings are handled more correctly and reported in such a way
3224 that it's easy to find one or the other in your output (so that you
3225 can e.g. figure out which of many problems caused COMPILE-FILE to
3227 * changed the severity of several compiler warnings from full WARNING
3228 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3229 compiler note reporting so that it doesn't use the condition system
3230 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3231 COMPILE-FILE command)
3232 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3233 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3234 no longer supported, since their implementation was deeply intertwingled
3235 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3236 * removed bogus "support" for compiler macros named (SETF FOO), and
3237 removed the compiler macro for SETF INFO (but only after making a fool
3238 of myself on the cmucl-imp mailing list by posting a bogus patch for
3239 DEFINE-COMPILER-MACRO..)
3240 * Compiled files containing forms which have side effects on the Lisp
3241 reader (such as DEFPACKAGE forms) are now handled more correctly.
3242 (Compiler queuing of top level lambdas has been suppressed by setting
3243 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3244 * deleted various currently-unused source files, e.g. gengc.lisp. They
3245 may be added back at some point e.g. when porting to other architectures,
3246 but until they are it's distracting to distribute them and to try to
3248 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3249 there were so many of them they're just distractions, and UNCROSS is
3250 known to be able to handle the current sources
3251 * moved PROFILE functionality into TRACE, so that it will be clear
3252 how the wrapping and unwrapping of functions when you profile them
3253 interacts with the wrapping and unwrapping of functions when you
3254 trace them. (Actually, the functionality isn't there yet, but at least
3255 the interface specification is there. Hopefully, the functionality will
3256 arrive with some future maintenance release.)
3257 * removed host-oops.lisp
3258 * changed signature of QUIT function to allow UNIX-CODE argument
3259 * fixed READ-SEQUENCE bug
3260 * tweaked verbose GC output so that it looks more like the progress
3261 output that ANSI specifies for functions like LOAD
3262 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3263 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob