1 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
4 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
5 * numerous signal handling fixes to increase stability
6 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
7 * minor incompatible change: we now correctly canonize default
8 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
9 as per the MOP, rather than the historical (INITARG INITFUNCTION
10 INITFORM). (reported by Bruno Haible)
11 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
12 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
14 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
15 pathname is a directory pathname.
16 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
17 * dynamic space size on PPC has been increased to 768Mb. (thanks to
19 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
20 :METACLASS argument in addition to a class name. (reported by
21 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
22 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
23 (reported by Bruno Haible; patch from Gerd Moellmann)
24 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
26 * bug fix: more cleanups to the floating point exception handling on
27 x86-64 (thanks to James Knight)
28 * bug fix: the compiler does not try to constant fold calls of
29 COERCE and BIT-* functions when they return freshly constructed
30 objects. (reported by Nikodemus Siivola and Paul Dietz)
31 * optimization: improved the MIPS versions of generic (in the
32 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
33 * optimization: direct conversion of (unsigned-byte 32) to floats on
35 * optimization: structure instances with raw slots now use less
36 memory, and probably show better memory locality. (thanks to
38 * optimization: DYNAMIC-EXTENT declarations for lists and closures
39 are treated as requests for stack allocation on the x86-64,
40 Alpha-32, PPC and SPARC platforms.
41 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
42 the wrong file descriptor; implementation of SOCKET-OPEN-P.
43 (thanks to Tony Martinez)
45 ** gcing a dead thread can no longer lead to lockups
46 ** threads block signals until they are set up properly
47 ** errno is no longer shared by threads
48 ** interrupt-thread restores the eflags register on x86
49 ** fixed some lockups due to gc/thread interaction
50 * fixed some bugs revealed by Paul Dietz' test suite:
51 ** invalid dotted lists no longer raise a read error when
53 ** signal an error if a symbol that names a declaration is used
54 as the name of a type, or vice versa
55 ** allow using the (declare (typespec var*)) abbreviation for
56 (declare (type typespec var*)) with all type specifiers
57 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
58 ** always return NIL from PPRINT-POP when OBJECT is NIL
59 ** don't signal errors when pretty-printing malformed LABELS,
60 FLET or MACROLET forms
61 ** declarations in a DOLIST body are also in scope for the
63 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
65 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
68 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
69 * fixed cross-compiler leakages that prevented building a 32-bit
70 target with a 64-bit host compiler.
71 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
72 opened with :IF-EXISTS :OVERWRITE.
73 * fixed bug 281: error for an invalid qualifier in a short-form method
74 combination method is not signalled until the faulty method is called.
75 * bug fix: iteration variable type inferrer failed to deal with open
76 intervals. (reported by Alan Shields)
77 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
78 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
79 or more runtime options were provided to the sbcl binary.
80 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
81 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
82 * optimization: there's now a fast-path for fixnum arguments in the
83 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
84 * optimization: the code generated on x86-64 is more compact thanks
85 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
86 * bug fix: floating-point exception handling now partly works on
87 x86-64. (Thanks to James Knight)
88 * improvement to the MIPS backend from Thiemo Seufer: C-style
89 64-bit long long arguments and return values to alien functions
91 * fixed some bugs revealed by Paul Dietz' test suite:
92 ** the type-error signalled from WARN has a filled-in DATUM slot.
93 ** the type-error required when a stream is not associated with
94 a file has the stream as its datum.
95 ** type-errors on single-floats on x86-64 no longer have
96 :INVALID-OBJECT as the datum
97 ** the type-errors signalled for invalid function names now have
98 a correct expected type
99 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
100 for structures defined with a :TYPE. Documentation strings for
101 typed structures are no longer immediately discarded
102 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
103 broadcast streams and synonym streams. FILE-LENGTH now also works
104 on broadcast streams.
106 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
107 * incompatible change: the --noprogrammer option, deprecated since
108 version 0.7.5, has been removed. Please use the equivalent
109 --disable-debugger option instead.
110 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
112 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
113 inoperational for a while now) has been completely removed.
114 * null lexical environments are now printed as #<NULL-LEXENV>,
115 significantly reducing the amount of clutter in typical backtraces.
116 * documentation on weak pointers, finalization, and after GC hooks
117 has been added to the manual.
118 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
119 as other specialized array types, is much faster. SUBSEQ and
120 COPY-SEQ on such arrays have also been sped up.
121 * optimization: EQL is now more efficient when at least other argument
122 is known to be of type (OR FIXNUM (NOT NUMBER)).
123 * fixed bug: the runtime is now less vulnerable to changes in the
124 size of the SBCL object on OS X, and virtual memory is reserved for
125 all spaces that need to be at a fixed address.
126 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
127 * fixed bug: finalizers and after GC hooks that cause consing are now
129 * fixed bug: compiler error messages and summaries are now printed to
130 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
131 * fixed inference of the upper bound of an iteration variable.
132 (reported by Rajat Datta).
133 * fixed bug 211e: calling local functions with duplicated constant
134 keyword argument no longer causes a bogus style warning about an
136 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
137 ftype declaration to be disregarded. (reported by Dave Roberts)
138 * fixed bug 373: caused by erronous compilation of references to alien
139 variables in the runtime on ppc/darwin.
140 * fixed bug 376: CONJUGATE type deriver.
141 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
142 (reported by Baughn on #lisp)
143 * a cross-compiler bug on non-x86 platforms has been identified and
144 fixed. (thanks to Bruno Haible)
145 * improvements to the MIPS runtime code for increased stability.
146 (thanks to Thiemo Seufer)
147 * increased the maximimum compact environment size to allow
148 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
149 * partial workaround for bug 135: don't name the function we're
150 calling for hairy cases of EVAL, so as not to accumulate one environment
151 entry per call to EVAL. (reported by Kevin Reid)
152 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
153 * optimization: type testing for non-vector arrays should be faster.
154 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
156 * bug fix: setting 31st element of a bit vector to zero did not work
158 * bug fix: redefining a class definition which failed due to a
159 previous accessor / function clash now works (but see BUGS entry
160 #380 for more problems in this area). (thanks to Zach Beane)
161 * the long form of DEFINE-METHOD-COMBINATION disables method group
162 checking when given a single method group with pattern *. (thanks
164 * on x86 compiler supports stack allocation of results of simple
165 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
166 * support for building and running on Mac OS X 10.4 "Tiger" has been added
167 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
168 * fixed some bugs related to Unicode integration:
169 ** the restarts for recovering from input and output encoding
170 errors only appear when there is in fact such an error to
172 * fixed some bugs revealed by Paul Dietz' test suite:
173 ** MISC.549 and similar: late transformation of unsafe type
174 assertions into derived types caused unexpected code
176 ** SCALE-FLOAT type deriver is less wrong.
177 ** type derivers for EXP, LOG and similar functions compute result
178 types for complex arguments better.
179 ** (MISC.563) CONJUGATE type deriver works for very restricted
181 ** out-of-line type testers for character strings are available.
182 ** EQUAL compiler transform understands specialness of objects
184 ** accessing double-floats stored on the stack now works on x86-64.
185 ** debugger internals could sometimes create invalid lispobjs,
186 resulting in GC crashes.
187 ** MISC.548: type check weakening can convert required type into
189 ** initialization forms for bindings are not in scope of free special
192 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
193 * incompatible change: thread support for non-NPTL systems has
194 been removed - locking is buggy and unreliable. A threaded
195 SBCL build will now warn at startup and refuse to create
196 new threads, unless futex support is detected in the kernel
197 * incompatible change: the top level REPL now has only an ABORT
198 restart associated with it, not TOPLEVEL and ABORT as it used to.
199 TOP and TOPLEVEL are now available as debugger commands for
200 returning to the top level.
201 * incompatible change: forms evaluated in the REPL now use the
202 global optimization policy.
203 * incompatible change: user- and system-initialization files are
204 no longer processed with LOAD, but by READ and EVAL; hence the
205 global optimization policy, startup package, readtable, etc,
207 * The .fasl file format number has been incremented because of
208 various incompatible changes.
209 * internal entry point details and argument counts no longer appear
210 in backtraces unless explicitly requested by setting
211 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
212 * built-in and standard functions no longer have names like "top
213 level local call to FOO".
214 * fixed bug 32: functions defined in non-null lexical environments
215 now have more legible printed representation
216 * fixed bug 33: functions defined in non-null lexical environemnts
217 are now more amenable to inspection by INSPECT.
218 * workaround for bug 354: XEPs no longer appear in backtraces unless
219 explicitly requested.
220 * fixed bug: receiving the signal which results from attempting to
221 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
222 SIGBUS on other *BSDs) on architectures where the C stack is also
223 the Lisp stack (x86 and x86-64) from foreign code no longer leads
224 to debugger confusion or wild instability. (reported by Cheuksan
226 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
227 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
228 (reported by Lutz Euler)
229 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
230 compiled code. (reported by Paul Dietz)
231 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
232 specialized parameter to a method to something that is not TYPEP
233 the specializer is now possible.
234 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
235 face of package deletion.
236 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
237 pathname. (reported by David Tolpin on c.l.l)
238 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
239 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
240 STRING-STREAM and FILE-STREAM, respectively.
241 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
242 than previously; in addition, its test suite is now run on build.
243 (thanks to Luke Gorrie)
244 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
245 * optimization: added a immediate representation for single-floats
247 * fixed some bugs revealed by Paul Dietz' test suite:
248 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
249 ** attempting to create a package with a colliding nickname causes
250 correctable errors to be signalled.
251 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
252 ** several x86-64 backend bugs related to sign-extension of immediate
255 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
256 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
257 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
259 * fixed loading of multiply forward-referenced layouts.
260 (thanks to Cheuksan Wang)
261 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
262 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
263 (see COMPILE-FILE documentation for details of :PRINT option)
264 * fixed bugs 19 and 317: fixed-format floating point printing is
265 more accurate. This also fixes a bug reported by Adam Warner
266 related to the ~@F format directive.
267 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
269 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
270 dispatch table if the from-char is a dispatch macro character.
271 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
272 by the wordsize no longer ignore the last word. (reported by Lutz
274 * fixed bug in type checking of dynamic-extent variables. (reported
276 * optimization: sequence traversal functions use their freedom to
277 coerce function designators to functions.
278 * optimization: code with many calls to CLOS methods specialized on
279 CLOS classes has had redundant type checks removed.
280 * optimization: added declarations to speed up operations that access
281 the internal character database (for example STRING-UPCASE)
282 * optimization: comparison operations between floats and sufficiently small
283 fixnums no longer create extra rationals
284 * fixed some bugs related to Unicode integration:
285 ** portions of multibyte characters at the end of buffers for
286 character-based file input are correctly transferred to the
287 start of the buffer at the next read.
288 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
289 passing it through to OPEN.
290 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
291 argument given to internal calls to OPEN.
292 ** the built-in comment readers (introduced by character sequences
293 ";" and "#|") are more forgiving to encoding errors; they will
294 STYLE-WARN and then attempt to resync the stream at a character
295 boundary. (thanks to Teemu Kalvas)
296 * fixed some bugs in the x86-64 port:
297 ** Negative short int return values from c-calls are sign-extended
299 ** The stack is aligned to 16-bytes for c-calls, as required by
300 the ABI. (thanks to Cheuksan Wang)
301 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
302 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
303 for enums. (thanks to Vincent Arkesteijn)
304 ** Multiple small optimizations and bugfixes for floating point
306 * fixed some bugs revealed by Paul Dietz' test suite:
307 ** Space, Tab, Linefeed, Return and Page have the invalid
308 secondary constituent character trait.
309 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
311 ** WITH-INPUT-FROM-STRING should only update the index place on
313 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
315 ** Bit-array operations (BIT-AND and similar) worked incorrectly
316 with one-dimensional arrays with fill pointers.
317 ** TYPE-OF failed on a complex with an integer realpart and a
319 ** compiler failure during type inference for the code of form
320 (IF (EQL X (THE ...)) ...) (MISC.535).
322 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
323 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
324 platform. The port supports 61-bit fixnums, large memory spaces
325 and reloading shared object files.
326 * enhancement: saving cores with foreign code loaded is now
327 supported on ppc/Darwin in addition to the previously supported
329 * enhancement: the statistical profiler now walks deeper into the
330 call stack for more meaningful call-graphs and accrued time
331 reports (x86/x86-64 only). It also now reports time spent in
333 * enhancement: it is now possible to trace most individual methods
334 of a generic function in addition to tracing the generic function
336 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
338 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
339 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
340 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
341 produces an error. (thanks to Vincent Arkesteijn)
342 * bug fix: NAMESTRING on pathnames with :WILD components in their
343 directories works correctly. (thanks to Artem V. Andreev)
344 * fixed bug 125: compiler preserves identity of closures. (reported
346 * bug fixed: functions with &REST arguments sometimes failed with
347 "Undefined function" when compiled with (DEBUG 3). (reported by
349 * bug fix: overflow during compiling of setting element of a bit
350 vector with constant index and value. (reported by Timmy Douglas)
351 * build fix: fixed the dependence on *LOAD-PATHNAME* and
352 *COMPILE-FILE-PATHNAME* being absolute pathnames.
353 * on x86 compiler partially supports stack allocation of dynamic-extent
355 * GO and RETURN-FROM do not check the extent of their exit points
356 when compiled with SAFETY 0.
357 * fixed some bugs related to Unicode integration:
358 ** encoding and decoding errors are now much more robustly
359 handled; it should now be possible to recover even from invalid
360 input or output to the terminal. (thanks to Teemu Kalvas)
361 ** provided a first cut at implementing STRING-TO-OCTETS and
362 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
363 ** altered the SB-MD5 contributed module slightly, changing the
364 interface just enough for it to be supportable for builds where
365 lisp characters are not eight bits.
366 * fixed some bugs revealed by Paul Dietz' test suite:
367 ** the FORMATTER-generated functions for ~V[ conditionals require
368 the correct number of arguments.
369 ** READ-FROM-STRING returns the mandated second value when applied
370 to displaced strings.
371 ** the #\Rubout and #\Backspace characters are treated as invalid
372 constituent characters by the tokenizer.
374 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
375 * new feature: reloading changed shared object files with
376 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
377 * new feature: references to foreign variables and functions
378 can now be compiled and loaded before the corresponding shared
379 object file is loaded, as long as the foreign definitions are
380 available at runtime.
381 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
382 supported. (thanks to Dan Debertin)
383 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
384 just keywords. (thanks to Vincent Arkesteijn)
385 * fixed bug #331: structure-class instances corresponding to
386 DEFSTRUCT forms are now created eagerly.
387 * fixed bug #345: backtraces from calls to undefined functions work
388 on x86 as well. Related bug #61 is now also partially fixed on x86
389 (backtraces from throws to unknown catch tags.)
390 * bug fix: lambda-list parsing is now stricter vrt. order and number
391 of lambda-list keywords.
392 * bug fix: as specified by AMOP, an error is signalled if a
393 class-option appears multiple times in a DEFCLASS form. (reported
395 * bug fix: RANDOM can be compiled when the compiler derives the type
396 of its numeric argument as a disjoint set of small integers.
397 (reported by Paul Dietz)
398 * bug fix: starting a core saved with shared objects loaded when
399 those objects are not available no longer causes threaded SBCL to
400 hang. (reported by Sean Ross)
401 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
403 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
404 argument, which must always be NIL. (reported by Kalle Niemitalo)
405 * bug fix: printing 1.0d+23 no longer results in an error.
406 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
407 * bug fix: structure slot setters preserve evaluation order. (thanks
409 * fixed some bugs related to Unicode integration:
410 ** RUN-PROGRAM can allow its child to take input from a Lisp
411 stream. (reported by Stefan Scholl)
412 ** ASDF-INSTALL successfully downloads .tar.gz files in all
413 locales. (reported by Ken Causey)
414 * fixed some bugs revealed by Paul Dietz' test suite:
415 ** INCF, DECF and REMF evaluate their place form as specified in
417 ** FORMATTER expands ~{ ~} iteration directives with V or #
418 parameters correctly.
419 ** FORMATTER deals with the ~@[ ~] conditional directive where the
420 consequent uses no arguments correctly.
421 ** the system has a partial understanding of the (COMPLEX RATIO)
423 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
426 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
427 * new feature: a build-time option (controlled by the :SB-UNICODE
428 keyword feature, enabled by default) for building the system with
429 support for the entire 21-bit character space defined by the
431 * new feature: the system now has rudimentary external-format
432 support; the primary user-visible change at this time is that
433 characters with the high bit set (such as lower-case-e-acute) will
434 print correctly to a terminal in a UTF-8 environment.
435 * minor incompatible change: BASE-CHAR no longer names a class;
436 however, CHARACTER continues to do so, as required by ANSI.
437 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
438 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
439 has been moved to the SB-EXT package (temporarily re-exported from
441 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
442 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
444 * fixed bug #351: better error-handlind and reporting for malformed
446 * fixed bug #350: bignum-printing is now more memory-efficient,
447 allowing printing of very large bignums, eg. (expt 2 10000000).
448 (reported by Bruno Haible)
449 * fixed bug #302: better primitive-type selection for intersection
451 * fixed bug #308: non-graphic characters now all have names, as
452 required. (reported by Bruno Haible)
453 * bug fix: redefining a class with different superclasses now correctly
454 removes it from the direct-subclasses of its previous superclasses.
455 (reported by David Morse)
456 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
457 new value now works. (reported by Bruno Haible)
458 * bug fix: correct canonicalization of multiple non-standard slot
459 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
460 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
461 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
462 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
463 now exists, an signals an error.
464 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
465 during macroexpansion for non-iteration variables as well. (reported
466 by Bruno Haible for CMUCL)
467 * bug fix: Cyclic structures and unprintable objects in compiler
468 messages no longer cause errors. (reported by Bruno Haible)
469 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
470 and READ-FROM-STRING all now return a primary value of NIL if
471 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
472 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
473 (reported by Bruno Haible for CMUCL)
474 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
475 do not designate a condition. (reported by Bruno Haible for
477 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
478 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
479 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
480 permitting resolution in favour of any of the conflicting symbols.
481 (reported by Bruno Haible for CMUCL)
482 * bug fix: EQUAL compiler optimizations is less aggressive on
483 strings which can potentially compare true despite having distinct
484 specialized array element types.
485 * bug fix: unit enumerations can be defined without dividing by
486 zero. (thanks to Vincent Arkesteijn)
487 * FORMAT compile-time argument count checking has been enhanced.
488 (report from Bruno Haible for CMUCL)
489 * a partial workaround for the bug 262: the compiler does not try to
490 inline-expand a local function doing RETURN-FROM from a deleted
491 BLOCK. (thanks to Peter Denno for the bug report and to David
492 Wragg for the simple test case)
493 * fixed some bugs revealed by Paul Dietz' test suite:
494 ** RENAME-PACKAGE allows all package designators as new package
496 ** constraint propagation and lambda variable substitution are
497 more cautious in dealing with partially deleted code.
498 ** compiler sometimes lost reoptimization passes.
499 ** CERROR, given a condition as condition designator, passes its
500 remaining arguments to the continue format control without
502 ** Case-altering FORMAT directives work correctly on non-ASCII
504 ** The REAL type specifier handles bounds outside the floating
505 point ranges without signalling FLOATING-POINT-OVERFLOW.
506 ** Functions with IR1-transformations can create intercomponent
507 references to global functions.
508 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
510 ** FORMAT ~R treats a nil value for its first parameter correctly.
512 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
513 * enhancement: saving cores with foreign code loaded is now
514 supported on x86/NetBSD and sparc/Linux in addition to the previously
516 * bug fix: on some platforms repeated installations caused multiple
517 copies of HTML documentation to be installed -- should not happen
518 any more. (reported by Stefan Scholl)
519 * bug fix: parsing self-recursive alien record types multiple times
520 no longer causes infinite recursion. (reported by Thomas F. Burdick,
521 original patch by Helmut Eller for CMUCL)
522 * bug fix: stack-exhaustion detection works now on NetBSD as well.
523 (thanks to Richard Kreuter)
524 * bug fix: defining classes whose accessors are methods on existing
525 generic functions in other (locked) packages no longer signals
526 bogus package lock violations. (reported by François-René Rideau)
527 * bug fix: special variables as DEFMETHOD parameters no longer have
528 associated bogus type declarations. (reported by David Wragg and
530 * bug fix: read-write consistency on streams of element-type
531 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
532 * bug fix: redefiniton of the only method of a generic function with
533 no DEFGENERIC no longer emits a full WARNING. In addition,
534 redefinition of generic functions with no DEFGENERIC to an
535 incompatible lambda list now signals an error. (thanks to Zach
537 * bug fix: DEFGENERIC now works even when there's a function of the
538 same name in an enclosing lexical environment. (thanks to Zach
540 * fixed compiler failure, caused by instrumenting code during
541 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
542 * optimization: added loop analysis and improved register allocation
543 * fixed some bugs revealed by Paul Dietz' test suite:
544 ** POSITION on displaced vectors with non-zero displacement
545 returns the right answer.
546 ** (SIMPLE-STRING) is a valid type specifier for sequence
548 ** *PRINT-LEVEL* handling for slotless structures is pedantically
550 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
551 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
553 ** The justification version of the FORMAT ~< directive treats
554 non-zero minpad parameter correctly.
556 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
557 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
558 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
559 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
560 the supported interface.
561 * new feature: Single-stepping of code compiled with DEBUG 2 or
562 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
563 * new feature: saving cores with foreign code loaded is now
564 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
565 Timothy Moore's work for CMUCL)
566 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
567 parameters to * instead of NIL if no initform is supplied.
568 (reported by Johan Bockgaard)
569 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
570 correctly when FUNCALL appears as the car of the form. Note:
571 despite this FUNCALL forms are not currently subject to
572 compiler-macro expansion. (port of Raymond Toy's fix for the
573 same from CMUCL, reported by Johan Bockgaard)
574 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
575 (thanks for Teemu Kalvas)
576 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
577 case instead of terminating the iteration (thanks for Julian
578 Squires, Sean Champ and Raymond Toy)
579 * bug fix: incorrect expansion of defgeneric that caused a style
580 warning. (thanks for Zach Beane)
581 * bug fix: slot accessor effective method computation works properly
582 for classes with multiple non-standard applicable methods on
583 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
584 * on x86 compiler supports stack allocation of results of LIST and
585 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
586 CMUCL implementation by Gerd Moellmann)
587 * fixed some bugs revealed by Paul Dietz' test suite:
588 ** FORMAT strings with both the ~<~:;~> form of the justification
589 directive and pretty-printing directives cause an error.
591 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
592 * incompatible change: the internal functions
593 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
594 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
595 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
596 instead of the old functions.
597 * new feature: on platforms where "dladdr" is available foreign
598 function names now appear in backtraces. (based on Helmut Eller's
600 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
601 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
603 * documentation: ASDF manual is now distributed with SBCL.
604 * bug fix: SBCL can now load its contributed modules with REQUIRE
605 even if the system-provided entries have been removed from
606 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
608 * bug fix: backtraces involving undefined functions or assembly
609 routines are more informative. (thanks to Brian Downing)
610 * bug fix: mutually referent alien structures now work correctly.
611 (reported by Rick Taube)
612 * bug fix: structures defined by WITH-ALIEN can be referred to
613 within other definitions in the same WITH-ALIEN.
614 * bug fix: division operators (MOD, TRUNCATE and the like) with
615 constant zero divisors and integer dividends no longer generate
617 * bug fix: provide default methods for INPUT-STREAM-P and
618 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
619 * bug fix: improve the detection and resolution of MOP metacycles
620 (where effective method computation depends on the generic function
621 having its effective method computed). (reported by Bruno Haible)
622 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
623 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
624 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
625 * bug fix: pathnames with relative directory components can be
626 represented relative to default pathnames.
627 * optimization: in taking the GCD of bignums, reduce the two bignums
628 to approximately the same size (using Euclid's algorithm) before
629 applying the more sophisticated binary GCD. (thanks to Juho
631 * optimization: COUNT on bitvectors now operates word-at-a-time.
632 * optimization: ASH with a positive, but not necessarily constant,
633 (leftwards) shift, when in a modular context, is compiled to a
635 * fixed some bugs revealed by Paul Dietz' test suite:
636 ** FORMAT variable parameters ("~V<char>") are defaulted properly
637 if the corresponding argument is NIL.
638 ** FORMAT directives accepting a minpad parameter treat negative
640 ** literal commas as character parameters to FORMAT directives are
642 ** literal spaces directly after ~<Newline> directives within a
643 format-logical-block (~:< ~@:>) do not induce :FILL-style
644 conditional newlines.
645 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
646 as specified: it no longer includes conditional newlines).
647 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
649 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
650 :PER-LINE-PREFIX argument does not evaluate to a string.
652 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
653 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
654 the manual for details; for now, package locks can be disabled by
655 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
656 if no major problems are found then it is likely that they will be
657 compiled in unconditionally.
658 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
659 unsupported operators on all platforms. To load a shared library
660 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
661 object file, link it into a shared library outside of SBCL and
662 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
664 * bug workaround: ROOM T and the GC were not working together
665 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
666 reporting facility aren't preserved. That reporting has been
667 disabled, so now until and unless someone figures out how to make it
668 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
669 * minor incompatible change: as threatened around sbcl-0.8.0, the
670 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
671 The symbols are also exported from SB-PCL for backwards
672 compatibility, but more so than before SB-PCL should be treated as
673 an implementation-internal package.
674 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
676 * the SB-SPROF contrib now works on (most) non-x86 architectures.
677 It is known as of this release not to work on the Alpha, however.
678 * fixed bug #167: errors signalled due to illegal syntax in method
679 bodies are now more legible.
680 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
681 designators and can hence be used with TYPEP.
682 * fixed bug #333: CHECK-TYPE now ensures that the type error
683 signalled, if any, has the right object to be accessed by
684 TYPE-ERROR-DATUM. (reported by Tony Martinez)
685 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
686 5.1.2.3 for multiple-value place subforms. (reported by Kalle
688 * fixed bug #334: programmatic addition of slots using specialized
689 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
690 and :ALLOCATION :CLASS slots.
691 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
692 float exponents. (rereported by Peter Seibel)
693 * fixed a bug: #\Space (and other whitespace characters) are no
694 longer considered to be macro characters in standard syntax by
696 * fixed bug: initialization of condition class metaobjects no longer
697 causes an instance of the condition to be created. (reported by
699 * fixed bug: it is now possible to have slots such that
700 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
701 description is neither :INSTANCE nor :CLASS.
702 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
703 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
704 SLOT-VALUE-USING-CLASS), and no longer causes errors with
705 non-standard SLOT-DEFINITION-ALLOCATION values.
706 * fixed bugs: various race conditions handling exiting threads.
707 CL-PPCRE's thread torture test now passes.
708 * fixed arguably-a-bug: GC time proportional to number of threads in
709 system even when most of them are idle
710 * optimization: improved performance of BIT and SBIT on bit-vectors.
711 * .fasl file incompatibility: The fasl file version number has
712 been incremented because of changes associated with package locks.
714 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
715 * minor incompatible change: the system no longer provides
716 optimization hints (conditions which are TYPEP
717 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
718 mode; these hints are still emitted when higher SPEED optimization
720 * new contrib module: a sampling profiler (profiling by statistical
721 sampling, rather than by instrumenting functions) is available as
722 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
723 * the behaviour of the standard function ED is now customizeable by
724 third parties through a hook variable: see ED's documentation
725 string for information on the protocol.
726 * the compiler no longer emits efficiency notes for (FUNCALL X)
727 when the type of X is uncertain under default optimization
729 * fixed bug 276: mutating a binding of a specialized parameter to a
730 method to something that is not TYPEP the specializer is now
732 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
733 exhibit the required behaviour on the x86 platform. (thanks to
734 Peter van Eynde, Eric Marsden and Bruno Haible)
735 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
736 even for difficult arguments. (reported by Peter Graves)
737 * fixed bug 141a: the backquote printer now descends quoted
739 * fixed another bug in backquote printing: no more destructive
740 modification of the form's list structure. (reported by Brian
742 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
743 move between its address being taken and the call to
744 interrupt_thread, fixing a crashing race condition.
745 * the SB-POSIX contrib implementation has been adjusted so that it
746 no longer exhibits ridiculously poor performance when constructing
747 instances corresponding to C structs.
749 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
750 * minor incompatible change: the sb-grovel contrib now treats C
751 structures as alien (in the sense of SB-ALIEN) objects rather than
752 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
753 has implications for memory management of client code
754 (sb-grovel-returned objects must now be manually managed) and for
755 type safety (alien objects now have full types).
756 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
757 used to control emission of compiler diagnostics, rather than the
758 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
759 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
760 quality should be considered deprecated.
761 * install.sh now installs the user manual as well
762 * (not quite a new documentable feature, but worth considering in
763 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
764 beginnings of a semantically meaningful condition hierarchy is
765 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
767 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
768 designator as the defaults argument.
769 * fixed bug: Displaced arrays whose displaced-to array has become
770 too small now cause ARRAY-DIMENSION to signal an error, providing
771 for safer bounds-checking. (reported by Bruno Haible)
772 * fixed bug: DEFCLASS slot definitions with identical :READER and
773 :WRITER names now signal a reasonable error. (reported by Thomas
775 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
777 * fixed bug: Invalid slot specification errors now print correctly.
778 (thanks to Zach Beane)
779 * fixed bug 320: Shared to local slot value transfers in class
780 redefinitions now happen corrently from superclasses as
781 well. (reported by Bruno Haible)
782 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
784 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
785 improper lists correctly. (reported by Bruno Haible)
786 * fixed bug 313: source-transform for <fun-name> was erroneously
787 applied to a call of a value of a variable with name <fun-name>.
788 (reported by Antonio Menezes Leitao)
789 * fixed bug 307: The obsolete instance protocol ensures that
790 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
791 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
792 remove dead unknown-values globs from the middle of the stack.
793 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
794 easier to resume long computations after using *BREAK-ON-SIGNALS*
795 to diagnose and fix failures. (thanks to Nikodemus Siivola)
796 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
797 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
798 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
799 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
800 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
802 * fixed a (fairly theoretical) bug in string printing: if
803 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
804 does not have array-element-type equal to the most general string
806 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
807 function name. (thanks to Nikodemus Siivola)
808 * fixed bug: compile-time format string checker failed on
809 non-closed ~{. (reported by Thomas F Burdick)
810 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
811 don't warn on legal loop code involving more than one aggregate
812 boolean. (thanks to Nikodemus Siivola)
813 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
814 in keyword position to MAKE-PACKAGE was not being reported as
816 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
817 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
819 * fixed bugs 280 and 312: the checking for multiple definitions in a
820 file is less likely to become confused by uses of inline
822 * fixed bug: the #S reader macro performs the keyword coercion
823 specified for slot names. (reported by Kalle Niemitalo)
824 * fixed bug: lambda lists may contain symbols whose names start with
825 & but are not lambda-list-keywords; their occurrence triggers a
827 * fixed bug 321: define-method-combination argument lambda lists do
828 not cause invalid code to be generated when &optional or &aux
829 variables have default values. (reported by Bruno Haible)
830 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
831 have CLOS classes; this appears also to have fixed reported
832 instability in stack exhaustion detection.
833 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
834 MUFFLE-WARNING when no associated restart is present is now
836 * optimization: rearranged the expansion of various defining macros
837 so that each expands into only one top-level form in a
838 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
840 * optimization: used a previously little-used slot in symbols to
841 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
843 * fixed some bugs revealed by Paul Dietz' test suite:
844 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
846 ** Fixed an optimization bug related to inheritance of initforms
847 from local to shared slots.
848 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
849 ** CAST optimizer forgot to flush argument derived type.
850 ** print/read consistency on floats is now orders of magnitude
851 more likely. (thanks also to Bruno Haible for a similar report
853 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
854 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
855 for objects of type REAL. Make it so.
856 ** SXHASH is specified (infelicitously) to respect similarity,
857 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
858 it so. (thanks to Markus Ziegler)
859 ** on the Alpha, the compiler succeeds in compiling functions
860 returning a known number of arguments greater than 63.
861 ** fixed handling of invalid NIL arguments in keyword position
863 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
865 ** provided a readably-printable representation for RANDOM-STATE
867 ** ensured that pathnames, where they have a namestring, always
868 print using #P"..." syntax.
870 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
871 * Support for the forthcoming 2.0 version of the NetBSD kernel
872 running on x86 hardware has been added. (thanks to Perry
873 E. Metzger most immediately, and others for their past work)
874 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
875 current release version; SBCL's OpenBSD support had been broken
876 since about the time of OpenBSD's switch to ELF binary format.)
877 * [placeholder for DX summary]
878 ** user code with &REST lists declared dynamic-extent, under high
879 speed or space and low safety and debug optimization policy.
880 * The manual has been converted to Texinfo format and the debugger
881 chapter from the cmucl manual has been added.
882 * A facility has been added to extract documentation strings from
883 sbcl and store them as Texinfo-formatted snippets for inclusion in
884 the manual (via Texinfo's @include directive)
885 * bug fix: compiler emitted division in optimized DEREF. (thanks for
886 the test case to Dave Roberts)
887 * bug fix: multidimensional simple arrays loaded from FASLs had fill
888 pointers. (reported by Sean Ross)
889 * bug fix: PROFILE output is printed nicely even for large numerical
890 values. (thanks to Zach Beane)
891 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
892 greater than 32 handle EOF correctly.
893 * bug fix: on X86 an immediate argument of the IMUL instruction is
894 correctly printed in disassembly. (thanks to Lutz Euler)
895 * bug fix: class slots in redefined classes preserve their old
896 values. (thanks to Bruno Haible and Nikodemus Siivola)
897 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
898 succeeds. (reported by Marco Baringer)
899 * fixed some bugs revealed by Paul Dietz' test suite:
900 ** READ-SEQUENCE now works on ECHO-STREAMs.
901 ** RATIONALIZE works more according to its specification. (thanks
904 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
905 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
906 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
907 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
908 should matter to you only if you rebind the printer control
909 variables and then find you want different bindings in the
910 debugger than in the ordinary execution of your program.)
911 * The runtime build system has been tweaked to support building
912 (on SPARC/SunOS) using a C compiler which invokes Sun's own
913 assembler and linker. (thanks to Nikodemus Siivola)
914 * Unbound, undefined, undeclared variables now trigger full
915 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
916 more often programmer error than deliberate exploitation of undefined
918 * optimization: the hash algorithm for strings has changed to one
919 that is less vulnerable to spurious collisions. (thanks to Juho
921 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
922 needless bounds checking. (thanks to Juho Snellman)
923 * optimization: implemented multiplication as a modular
924 (UNSIGNED-BYTE 32) operation on the PPC backend.
925 * fixed some bugs revealed by Paul Dietz' test suite:
926 ** ADJUST-ARRAY now copies the datum in a zero rank array if
928 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
929 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
930 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
932 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
934 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
936 ** ECHO-STREAMs no longer attempt to echo the end of file value to
937 their output stream on EOF from read.
938 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
939 have been read to end-of-file.
940 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
942 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
943 description of determination of which consecutive characters
945 ** Printing the "Space" character with escaping on now yields "#\\ ",
946 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
947 ** Reading floating-point numbers with *READ-BASE* set to a number
948 less than 10 works correctly.
949 ** Reading floating-point numbers with *READ-BASE* set to a number
950 more than 10 works correctly.
951 ** Printing with *PRINT-READABLY* targets the standard readtable, not
952 the readtable currently in effect.
954 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
955 * minor incompatible change: parsing of namestrings on a physical
956 (Unix) host has changed; numbers after the final #\. in a
957 namestring are no longer interpreted as a version field. This is
958 intented to be largely invisible to the user, except that the
959 meaning of the namestring "*.*.*" has changed: it now refers to a
960 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
961 should usually be replaced by
962 (make-pathname :name :wild :type :wild :version :wild)
963 with the added benefit that this is more likely to be portable.
964 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
965 option to OPEN now signals an error if the file being opened
966 exists; this may have an impact on existing code.
967 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
968 signals received, on the PowerPC platforms (both Linux and
969 Darwin). (thanks to Pierre Mai for pointing out the location of
971 * several fixes on OS X: The system now builds and runs cleanly on
972 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
973 (thanks to Brian Mastenbrook)
974 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
975 second-resolution: integer multiples of 1/3600 between -24 and 24.
976 (thanks to Vincent Arkesteijn)
977 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
978 type when called with 1 argument; PEEK-CHAR checked type of
979 PEEK-TYPE only after having read first character from a
980 stream. (reported by Peter Graves)
981 * bug fix: the garbage collector now has much better locality
982 behaviour, and in particular no longer treats all memory as being
983 exhausted when many small objects point to each other in a deeply
985 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
986 recognized as being TYPEP their class.
987 * bug fix: the PUSHNEW documentation string has been corrected.
988 (thanks to Vincent Arkesteijn)
989 * bug fix: defaulting of the value for the last of an atypically
990 large number of multiple values being bound was not being
991 performed correctly on the Alpha or PPC platforms
992 * optimization: implemented multiplication as a modular
993 (UNSIGNED-BYTE 32) operation on the x86 backend.
994 * optimization: SEARCH on simple-base-strings can now be open-coded.
995 (see also contrib/compiler-extras.lisp for inspiration for
996 teaching the compiler about the Boyer-Moore algorithm).
997 * value, returned by MAX (and MIN) called with several EQUALP, but
998 not EQL, arguments now does not depend on compiler settings.
999 * fixed some bugs revealed by Paul Dietz' test suite:
1000 ** in stack analysis liveness information is propagated from
1001 non-local entry points.
1002 ** pathwise CAST removing failed when the CAST node did not start
1004 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1005 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1007 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1008 host is already defined.
1009 ** RENAME-FILE works on streams instead of signalling an internal
1011 ** PEEK-CHAR uses the current readtable when determining whether
1012 or not a character is whitespace.
1013 ** MERGE-PATHNAMES handles the case when the pathname does not
1014 specify a name while the default-pathname specifies a version.
1015 ** Pathnames now stand a better chance of respecting print/read
1017 ** Attempting to use standardized file system operators with a
1018 pathname with invalid :DIRECTORY components signals a
1020 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1022 ** DIRECTORY on logical pathnames is more correct.
1023 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1024 signal a TYPE-ERROR if their argument is not a stream.
1025 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1026 designator argument does not designate a stream.
1027 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1028 examining the synonym.
1029 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1031 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1032 element-type larger than ([UN]SIGNED-BYTE 32).
1034 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1035 * When built with the :SB-FUTEX feature, threaded builds now take
1036 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1037 for faster/more reliable mutex and condition variable support.
1038 * Incompatible change (but one you probably shouldn't have been using
1039 anyway): the interface and code for arbitrating between multiple
1040 threads in the same user session has been redesigned.
1041 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1042 error when called without an explicit environment argument.
1043 (thanks to Nikodemus Siivola)
1044 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1045 confused about their position. (thanks to Adam Warner and Gerd
1047 * bug fix: Pretty printing backquoted forms with unquotations in the
1048 argument list position of various code constructs such as LAMBDA
1049 now works correctly. (reported by Paul Dietz)
1050 * bug fix: Pretty printing unquotations no longer loses all
1051 stream position information.
1052 * optimization: performance of string output streams is now less
1053 poor for multiple small sequence writes.
1054 * optimization: performance of CSUBTYPEP in the presence of complex
1055 expressions involving CONS and NOT many times has been improved.
1056 (reported by Paul Dietz)
1057 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1059 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1060 streams for the REPL, for the debugger, and so on. (thanks to
1062 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1063 * fixed some bugs revealed by Paul Dietz' test suite:
1064 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1065 is ignored if it is too small, rather than propagating through
1066 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1067 ** extremely complex negations of CONS types were not being
1068 sufficiently canonicalized, leading to inconsistencies in
1070 ** VALUES tranformer lost derived type.
1072 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1073 * fixed a bootstrapping bug: the build process no longer assumes
1074 that the various BOOLE-related constants have the same value in
1075 host and target lisps. (noted by Paul Dietz' test suite on an
1076 SBCL binary built from CLISP)
1077 * The system can now be dynamically linked on the MIPS platform,
1078 which enables dynamic loading of foreign code from Lisp. (thanks
1079 to Ralf Baechle for discussions on the MIPS ABI)
1080 * The system now records debugging information for its own source
1081 files in a filesystem-position-independent manner, relative to
1082 the "SYS" logical host.
1083 * fixed a compiler bug: MV-LET convertion did not check references
1084 to the "max args" entry point. (reported by Brian Downing)
1085 * tweaked disassembly notes to be less confident about proclaiming
1086 some instruction as an LRA. (thanks to Brian Downing)
1087 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1088 now each have their own history, command character, and other
1089 characteristics. (thanks to David Lichteblau)
1090 * fixed some bugs revealed by Paul Dietz' test suite:
1091 ** compiler failure in compiling LOGAND expressions including a
1093 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1094 shift greater than 32.
1095 ** FLUSH-DEST did not mark blocks for type check regeneration.
1096 ** HANDLER-CASE failed to accept declarations in handler clauses
1097 in some circumstances.
1099 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1100 * New code in contrib/sb-introspect (still probably not entirely
1101 stable yet) provides some support for smart Lisp development
1102 environments like SLIME.
1103 * The conditions signalled for errors occurring when loading .fasl
1104 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1105 in a way which should help ASDF recover gracefully.
1106 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1107 stuff has been cleaned up. If you code contrib/ stuff, this might
1108 affect you, and you can look at contrib/README, contrib/STANDARDS,
1109 and/or the 0.8.4.27 diff to check.
1110 * In full calls the compiler now does not generate checks for declared
1111 argument types for all arguments.
1112 * various threading fixes
1113 ** and some experimental patches which didn't make it into
1114 the main tree for this release, but which are shipped in
1115 contrib/experimental-thread.patch as a possible fix for some
1116 failures (deadlock, spinning...) in GC-intensive multithreaded
1118 * fixed PPC build problem (source code incompatibility of different
1119 library versions): added offsetof() hackery which attempts to divine
1120 where glibc maintainers put uc_mcontext today
1121 * fixed bug 282: compiler does not trust type assertions while passing
1122 arguments to a full call.
1123 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1124 &OPTIONAL VALUES type parameter.
1125 * fix bug 214: algorithm for noting rejected templates is now more
1126 similar to that of template seletion. (also reported by rydis on
1128 * fixed bug 141b: printing backquoted information readably and prettily
1129 inserts a space where necessary.
1130 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1131 (CHARACTER 10) are now reported as errors, rather than propagated
1132 as unknown types. (reported by piso on #lisp)
1133 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1134 with values NIL and :ERROR. (thanks to Milan Zamazal)
1135 * fixed bug 191c: CLOS now does proper keyword argument checking as
1136 described in CLHS 7.6.5 and 7.6.5.1.
1137 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1138 longer raise an error; further, using a list as a for-as-arithmetic
1139 counter now raises a meaningful error.
1140 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1141 now understood by sequence creation functions such as MAKE-SEQUENCE
1143 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1144 a STRING-INPUT-STREAM.
1145 * compiler enhancement: SIGNUM is now better able to derive the type
1147 * type declarations inside WITH-SLOTS are checked. (reported by
1149 * fixed some bugs revealed by Paul Dietz' test suite:
1150 ** incorrect optimization of TRUNCATE for a positive first
1151 argument and negative second.
1152 ** compiler failure in let-convertion during flushing dead code.
1153 ** compiler failure while deriving type of TRUNCATE on an
1154 interval, containing 0.
1155 ** ASH of a negative bignum by a negative bignum count now returns
1157 ** intersection of CONS types now canonicalizes properly, fixing
1158 inconsistencies in SUBTYPEP.
1160 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1161 * incompatible change: The --disable-debugger command line
1162 option now clobbers the debugger at a more fundamental
1163 level, by redefining #'INVOKE-DEBUGGER instead of by
1164 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1165 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1166 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1167 enter the debugger REPL and then suffer recursive errors
1168 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1169 while under the new system, BREAK in --disable-debugger mode
1170 terminates the system just as an unhandled error would.
1171 * fixed compiler performance when processing loops with a step >1;
1172 * bug fix: DOCUMENTATION now retrieves generic function
1173 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1174 support has been systematized, and now supports the methods
1175 specified by ANSI, along with a default method and a method for
1176 slot documentation. (reported by Nathan Froyd)
1177 * bug fix: effective methods associated with a generic function are
1178 no longer cached over a change of that generic function's method
1179 combination. (reported by Andreas Fuchs)
1180 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1181 names a non-existent directory. (thanks to Andreas Fuchs)
1182 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1183 the PPC platform, lead to stack corruption; this has been fixed.
1184 (reported by Rainer Joswig)
1185 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1186 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1187 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1188 arithmetic in general, is now much more reliable on the PPC
1190 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1191 platform now returns the right answer.
1192 * optimization: restored some effective method precomputation in
1193 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1194 precomputation is now tunable.
1195 * optimization: compiler-internal data structure use has been
1196 reviewed, and changes have been made that should improve the
1197 performance of the compiler by about 20%.
1198 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1199 simple-base-strings and simple-bit-vectors is improved.
1200 * optimization: the optimization of 32-bit logical and arithmetic
1201 functions introduced in version 0.8.3 on the x86 has been
1202 implemented on the mips, ppc and sparc platforms; an
1203 implementation of the same facility, but for 64-bit arithmetic,
1204 has been added for the alpha.
1205 * microoptimization: the compiler is better able to make use of the
1206 x86 LEA instruction for multiplication by constants.
1207 * bug fix: in some situations compiler did not report usage of
1208 generic arithmetic in (SPEED 3) policy.
1209 * bug 145b fix: compiler used wrong type specifier while converting
1210 MEMBER-types to numeric.
1211 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1213 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1214 index argument now works properly on the Alpha platform.
1215 * bug fix: floating point exception treatment on the Alpha platform
1217 * bug fix: FILE-POSITION works much better on string input and
1218 output streams. (thanks to Nikodemus Siivola)
1219 * bug fix: many threading/garbage collection symptoms sorted.
1220 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1221 might be pseudo-atomic.
1222 * internal change: Stopping for GC is now done with signals not
1223 ptrace. GC is now done in whichever thread wanted it, instead of
1225 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1227 * bug fix: result form in DO is not contained in the implicit
1229 * incompatible change: ICR structure is changed; the value part of
1230 CONTINUATION is now called LVAR; corresponding functions are
1231 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1232 * added type deriver for ISQRT (thanks to Robert E. Brown).
1233 * bug fix: better support for loading from the command line when an
1234 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1236 * fixed some bugs revealed by Paul Dietz' test suite:
1237 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1238 ** ROUND and FROUND now give the right answer when given very
1239 small float arguments.
1240 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1242 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1243 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1244 ** LCM should return a non-negative integer.
1245 ** PARSE-INTEGER returned the index of a terminator instead of the
1246 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1247 ** PARSE-INTEGER returned an incorrect index being applied to a
1249 ** LCM with two arguments of 0 returns 0 rather than signalling
1251 ** unsigned addition of a 32-bit constant with the high bit set no
1252 longer causes an internal compiler error on the x86.
1253 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1254 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1255 with negative last argument.
1256 ** byte specifiers with zero size and position no longer cause
1257 an error during type derivation.
1258 ** bignum multiplication on the Alpha platform now returns the
1260 * porting: The system now builds on SuSE AMD64, although it still
1261 generates a 32-bit binary.
1262 * .fasl file incompatibility: The fasl file version number has
1263 been incremented (because of the changes to internal compiler
1264 data structures referred to above).
1266 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1267 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1268 more accurately, on the Darwin kernel running on PowerPC hardware.
1269 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1270 * Compiler code deletion notes now signal a condition of type
1271 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1272 an associated MUFFLE-WARNING restart.
1273 * The compiler now performs limited argument count validation of
1274 constant format strings in FORMAT, and where appropriate in ERROR,
1275 CERROR and WARN. (thanks to Gerd Moellmann)
1276 * New ASDF-INSTALL contrib can be used for automatic download and
1277 installation of third-party Lisp code from CCLAN or other sites
1279 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1280 which forces another thread to execute a function supplied by the
1282 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1283 now accept and act upon their :ELEMENT-TYPE keyword argument.
1284 (reported by Martin Atzmueller, Edi Weitz)
1285 * bug fix: FILE-POSITION now accepts position designators up to
1286 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1287 is the greater. (thanks to Patrik Nordebo)
1288 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1289 Kalle Olavi Niemitalo)
1290 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1291 printable. (reported by Eric Marsden)
1292 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1293 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1294 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1295 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1296 (thanks to Stig Sandoe)
1297 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1298 UNREAD-CHAR now correctly includes the unread character in the
1299 target sequence. (thanks to Gerd Moellmann)
1300 * bug fix in threaded builds: the system can now be suspended and
1301 resumed by shell job control with minimal disruption.
1302 * bug fixes in times and timezones >2038AD
1303 * better handling of "where is GNU make?" problem in build scripts
1304 (thanks to Nikodemus Siivola)
1305 * new optimization: inside a named function any reference to a
1306 function with the same name is considered to be a self-reference;
1307 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1308 optimization quality.
1309 * new optimization on x86: logical functions and + now have
1310 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1311 used when the result is truncated to 32 bits.
1312 * VALUES declaration is partially enabled.
1313 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1314 * fixed some bugs revealed by Paul Dietz' test suite:
1315 ** The system now obeys the constraint imposed by
1316 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1317 types form a lattice under type intersection.
1318 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1319 ** ASSOC now ignores NIL elements in an alist.
1320 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1321 and (1+ MOST-POSITIVE-FIXNUM) answers.
1322 ** The addition of a method with invalid qualifiers to a generic
1323 function does not cause an error to be signalled immediately;
1324 a warning is signalled, and the error is generated only on
1325 calling the generic function.
1326 * changed .fasl file version number, in part to add type codes for
1327 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1328 obscure ANSI requirements
1330 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1331 * fixed bug 148: failure to inline-expand a local function left
1332 garbage, confusing the compiler.
1333 * fixed bugs 3cd: structure slot readers perform type check if the
1334 slot can have an invalid value (i.e. it is either not initialized
1335 or can be written with a less specific slot writer).
1336 * bug fix: the compiler now traps array references to elements off
1337 the end of an array; previously, the bounds checking in some
1338 circumstances could go off-by-one.
1339 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1341 * type declarations for array element types now obey the description
1342 on the CLHS page "Declaration TYPE", as per discussions on
1343 sbcl-help around 2003-05-08. This means that a declaration
1344 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1345 declaration, all references to BAR will be asserted or assumed
1346 (with THE, so dependent on compiler policy) to involve objects of
1347 type FOO. Note that no such declaration is implied in
1348 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1349 * declared types of functions from the "Conditions"
1350 chapter. (reported by Paul Dietz)
1351 * bug fix: CERROR accepts a function as its first argument.
1352 * bug fix: NTH an NTHCDR accept a bignum as index
1353 arguments. (reported by Adam Warner)
1354 * optimization: character compare routines now optimize comparing
1355 against a constant character. (reported by Gilbert Baumann)
1356 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1357 argument now works properly on the MIPS platform.
1358 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1359 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1360 declared to have a functional type, does not cause an error
1362 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1363 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1364 * fixed bug: if last continuation of a deleted block has a
1365 destination, this destination should be deleted too. (reported by
1367 * fixed a bug in the bootstrap process: the host compiler's values
1368 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1369 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1370 test case from Patrik Nordebo)
1371 * improved the ability of the disassembler on the PPC platform to
1372 provide helpful disassembly notes.
1373 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1374 the class in more cases than previously.
1375 * bug fix: FILE-POSITION now understands :START and :END for
1376 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1377 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1378 8) data. (thanks to David Lichteblau)
1379 * bug fix: it is possible to add a method to a generic function
1380 without lambda list.
1381 * bug fix: reader failed to signal END-OF-FILE inside an
1382 object representation. (reported by Nikodemus Siivola)
1383 * fixed some bugs revealed by Paul Dietz' test suite:
1384 ** LAST and [N]BUTLAST should accept a bignum.
1385 ** condition slot accessors are methods.
1386 ** (VECTOR NIL) is a subtype of STRING.
1388 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1389 * minor incompatible change: some nonsensical specialized lambda
1390 lists (used in DEFMETHOD) which were previously ignored now signal
1392 * minor incompatible change: the system is now aware of the types of
1393 variables in the COMMON-LISP package, and will signal errors for
1394 most violations of these type constraints (where previously they
1395 were silently accepted).
1396 * minor incompatible change: COMPILE-FILE now uses the freedom
1397 afforded (ANSI 3.2.2.3) to use derived function types for
1398 functions defined in the same file. This also permits the system
1399 to warn on static type mismatches and function
1400 redefinition. (Currently it does not work with high DEBUG level.)
1401 * minor incompatible change: VALUES declaration is disabled.
1402 * When issuing notes, the compiler now signals a condition of type
1403 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1404 restart for use in user handlers. It is expected that the
1405 COMPILER-NOTE condition will eventually become a condition
1406 supertype to a hierarchy of note types, which will then be
1407 handleable in a similar fashion. However, at the moment, no such
1408 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1409 but it's an implementation detail, not a classification for the
1411 * Changes in type checking closed the following bugs:
1412 ** type checking of unused values (192b, 194d, 203);
1413 ** template selection based on unsafe type assertions (192c, 236);
1414 ** type checking in branches (194bc).
1415 * A short form of VALUES type specifier has ANSI meaning (it has
1416 increased the number of situations when SBCL cannot perform type
1418 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1419 keywords or constants is permissible.
1420 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1421 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1422 classes. (thanks to Antonio Martinez)
1423 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1424 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1425 argument to a pretty-print stream. Instead, it leaves any such
1426 operations to DESCRIBE-OBJECT methods.
1427 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1429 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1430 lambda lists. (thanks to David Lichteblau)
1431 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1432 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1433 (thanks to Gerd Moellmann)
1434 * fixed bug 246: increased compilation speed of long
1435 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1437 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1439 * DEFCONSTANT now throws a condition of type
1440 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1441 constant to a non-EQL value; CONTINUE and ABORT restarts
1442 respectively change and preserve the value.
1443 * fixed bug 63: The code walker, part of the implementation of CLOS,
1444 is now better at handling symbol macros.
1445 * bug fix: There is no longer an internal implementation type named
1446 CL:LENGTH. (reported by Raymond Toy)
1447 * bug fix: In macro-like defining macros/special operators the
1448 implicit block does not enclose the lambda list.
1449 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1450 enclosed in parentheses) are not suitable as type specifiers, and
1451 their use properly signals an error now.
1452 * bug fix: An argument count mismatch for a type specifier in code
1453 being compiled no longer causes an unhandled error at compile
1454 time, but signals a compile-time warning.
1455 * fixed simple vector readable printing
1456 * bug fix: DESCRIBE takes more care over whether the class
1457 precedence list slot of a class is bound before accessing it.
1458 (reported by Markus Krummenacker)
1459 * bug fix: FORMATTER can successfully compile pretty-printer format
1460 strings which use variants of the ~* directive inside.
1461 * bug fix: SEARCH now applies its TEST predicate to the elements of
1462 the arguments in the correct order. (thanks to Wolfhard Buss)
1463 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1464 in type inference. (thanks to Robert E. Brown)
1465 * bug fix: Reading in symbols with an explicit package name of ""
1466 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1467 * fixed some bugs revealed by Paul Dietz' test suite:
1468 ** NIL is now allowed as a structure slot name.
1469 ** Arbitrary numbers, not just REALs, are allowed in certain
1470 circumstances in LOOP for-as-arithmetic clauses.
1471 ** Multiple class redefinitions before slot access no longer
1472 causes a type error.
1473 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1474 association between the name and a class.
1475 ** Generic functions with non-standard method-combination and over
1476 five methods all of which return constants no longer return NIL
1477 after the first few invocations. (thanks to Gerd Moellmann)
1478 ** CALL-NEXT-METHOD with no arguments now passes the original
1479 values of the arguments, even in the presence of assignment.
1480 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1482 ** DISASSEMBLE works with closures and funcallable instances.
1483 ** ADD-METHOD now returns the generic function, not the new method.
1484 ** FIND-METHOD signals an error if the lengths of the specializers
1485 is incompatible with the generic function, even if the ERRORP
1487 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1488 which its argument is a member.
1489 ** DEFCLASS only redefines the class named by its class-name
1490 argument if that name is the proper name of the class;
1491 otherwise, it creates a new class.
1492 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1493 of the UNBOUND-SLOT condition to the name of the slot.
1494 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1495 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1496 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1497 SLOT-MAKUNBOUND in the specified fashion.
1499 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1500 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1501 cross-compilation host. As a consequence, we can now bootstrap our
1502 way up to SBCL starting with a bare gcc toolchain and human-readable
1503 source code (first the source to CLISP, then the source to SBCL).
1504 * A contributed module containing a partial implementation of the
1505 simple-streams interface has been included. (thanks to Rudi
1507 * A contributed module implementing the RFC1321 Message Digest
1508 Algorithm, known as MD5, has been included.
1509 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1510 no longer has any effect, as the code controlled by this feature
1511 has been deleted. (As far as we know, no-one has ever built using
1512 this feature, and its semantics were confused in any case).
1513 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1514 work on conditions (as required by the ANSI specification),
1515 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1516 expected behaviour on conditions. Users should note, however,
1517 that such behaviour is not required by the ANSI specification,
1518 and so use of this behaviour may render their code unportable.
1519 * fixed some bugs revealed by Paul Dietz' test suite:
1520 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1522 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1523 ** SXHASH on condition objects no longer returns NIL.
1524 ** :ALLOCATION :CLASS slots are better treated; their values are
1525 updated on class redefinition, and initforms inherited from
1526 superclasses are applied.
1527 ** REMOVE-METHOD returns its generic function argument even when
1528 no method was removed.
1529 ** SHARED-INITIALIZE now initializes the values of the requested
1530 slots, including those with :ALLOCATION :CLASS.
1531 ** ALLOCATE-INSTANCE now works on structure classes defined via
1532 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1534 ** SLOT-EXISTS-P now works on conditions, as well as structures
1536 ** MAKE-LOAD-FORM now has the required methods on
1537 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1538 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1539 keyword, and now implements the SLOT-NAMES argument.
1540 ** methods with &OPTIONAL arguments no longer allow too many
1541 arguments to be passed in the call without error.
1542 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1543 option is consistent with the required arguments of the generic
1544 function lambda list.
1545 * bug fix: REQUIRE accepts a string designator. (Thanks to
1547 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1548 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1549 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1550 * bug fix: The compiler now checks for duplicated variables in macro
1552 * bug fix: SETQ on globals returns the correct value.
1553 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1554 not a COMPILER-ERROR (followed by some other strange error on
1555 choosing the CONTINUE restart).
1556 * bug fix: make.sh and friends are now more consistent in the way that
1557 they look for GNU "make".
1559 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1560 * experimental native threads support (on x86 Linux >=2.4 only).
1561 This is not compiled in by default: you need to add :SB-THREAD to
1562 the target features. See the "Beyond ANSI" chapter of the manual
1564 * fix for longstanding nonANSIism: The old distinction between
1565 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1566 The return value from CL:FIND-CLASS is now a CLOS class, and
1567 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1568 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1569 * An interface to the MetaObject Protocol, as described in Kiczales,
1570 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1571 MIT Press, 1991, is available from the SB-MOP package.
1572 * incompatible change: the SB-PCL package should now be considered
1573 a private implementation detail, and no longer a semi-private MOP
1575 * minor incompatible change: due to rearrangement for threads, the
1576 control stack and binding stack are now allocated at arbitrary
1577 addresses instead of being hardcoded per-port. Users affected by
1578 this probably have to be doing advanced things with shared
1579 libraries, and will know who they are.
1580 * minor incompatible change: Previously, all --eval forms used were
1581 processed with READ before any of them were processed with EVAL.
1582 Now each --eval form is processed with both READ and EVAL before
1583 the next --eval form is processed. (Thus package operations like
1584 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1585 work as the user might reasonably expect.)
1586 * minor incompatible change: *STANDARD-INPUT* is now only an
1587 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1589 * minor incompatible change: Y-OR-N-P is now character-oriented,
1590 not line oriented. Also, YES-OR-NO-P now works without errors.
1591 (thanks to Antonio Martinez)
1592 * sb-aclrepl module improvements: an integrated inspector, added
1593 repl features, and a bug fix to :trace command.
1594 * Known functions, which cannot be open coded by the backend, are
1595 considered to be able to check types of their arguments. (fixing
1596 a bug report by Nathan J. Froyd)
1597 * fixed a bug in computing method discriminating functions: It is
1598 now possible to define methods specialized on classes which have
1599 forward-referenced superclasses. (thanks to Gerd Moellmann)
1600 * fixed evaluation order in optional entries (reported by Gilbert
1602 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1603 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1604 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1605 required-by-AMOP NAME argument, as well as CLASS and
1606 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1607 * fixed bug 20: DEFMETHOD can define methods using names that are
1608 not the proper names of classes to designate class specializers.
1609 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1610 Unix file descriptors, instead of blowing up. (thanks to Antonio
1612 * Garbage collection refactoring: user-visible change is that a
1613 call to the GC function during WITHOUT-GCING will not do garbage
1614 collection until the end of the WITHOUT-GCING. If you were doing
1615 this you were probably losing anyway.
1616 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1617 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1618 TYPEP the latter but not the former.
1619 * The compiler issues a full WARNING for calls to undefined functions
1620 with names from the CL package.
1621 * MAP-INTO for a vector destination is open coded. (reported by
1622 Brian Downing on c.l.l)
1623 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1624 documentation string.
1625 * fixed some bugs revealed by Paul Dietz' test suite:
1626 ** COPY-ALIST now signals an error if its argument is a dotted
1628 ** Condition slots are now accessed more correctly in the presence
1629 of multiple initargs for a given slot.
1630 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1631 exclude restarts of the same name associated with a different
1633 ** DEFCLASS of forward-referenced classes with another
1634 forward-referenced class in the superclasses list no longer
1636 ** Condition slots are now initialized once each, not multiple
1637 times. (thanks to Gerd Moellmann)
1638 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1640 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1641 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1642 arguments contain duplicated elements.
1643 ** RESTART-CASE understands local macros.
1644 ** RESTART-CASE associates exactly its own restarts with a condition.
1645 ** ENDP in safe mode checks its argument to be of type LIST.
1646 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1647 in question is unbound.
1648 ** Optimized MAKE-INSTANCE functions no longer cause internal
1649 assertion failures in the presence of duplicate initargs.
1650 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1651 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1652 propagated even in the ctor optimized implementation of
1654 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1656 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1657 * a better implementation of SXHASH on (simple) bit vectors,
1658 measured both in execution speed and in distribution of results
1659 over the positive fixnums, has been installed. Likewise, a better
1660 implementation of EQUAL for simple bit vectors is now available.
1661 * fixed CEILING optimization for a divisor of form 2^k.
1662 * fixed bug 240 (emitting extra style warnings "using the lexical
1663 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1664 by Antonio Martinez)
1665 * fixed SXHASH, giving different results for NIL depending on type
1666 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1667 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1668 variable. (found by Rolf Wester)
1669 * SBCL does not ignore type declarations for special
1670 variables. (reported by rif on c.l.l 2003-03-05)
1671 * some bug fixes in contrib/sb-aclrepl/
1672 * fixed some bugs revealed by Paul Dietz' test suite:
1673 ** a bug in the CONS type specifier, whereby the CAR and CDR
1674 types got intertwined, has been fixed;
1675 ** the type system is now able to reason about the interaction
1676 between INTEGER and RATIO types more completely;
1677 ** APPEND, [N]REVERSE and NRECONC check that those their
1678 arguments, which must be proper lists, are really so;
1679 ** An array specialized to be unable to hold elements has been
1680 implemented, as required -- yes, really -- by ANSI;
1681 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1682 on malformed property lists;
1684 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1685 * incompatible packaging change: in line with Unix convention,
1686 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1687 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1688 * REQUIRE and PROVIDE are now optionally capable of doing something
1689 useful. See the documentation string for REQUIRE.
1690 * infrastructure for a managed SBCL contrib system: contributed
1691 modules in this release include:
1692 ** the ASDF system definition facility;
1693 ** an interface to the BSD Sockets API;
1694 ** an ACL-like convenience interface to the repl;
1695 (thanks to Kevin Rosenberg)
1696 ** an implementation of ROTATE-BYTE, with efficient implementation
1698 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1699 gives a closer approximation to the right answer than previously.
1700 (thanks to Raymond Toy)
1701 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1702 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1703 optional environment arguments, as required by ANSI.
1704 * fixed bugs in other functions taking environment objects, allowing
1705 calls with an explicit NIL environment argument to be compiled
1707 * fixed bug 228: primary return values from
1708 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1709 COMPILE or FUNCTION.
1710 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1711 :TYPE will no longer signal errors on innocuous objects.
1712 * fixed bug 231b: SETQ is better at respecting type declarations in
1713 the lexical environment.
1714 * fixed a bug in DEFCLASS: classes named by symbols with no or
1715 unprintable packages can now be defined.
1716 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1717 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1718 * fixed compiler failure related to checking types of functions.
1719 (reported by Robert E. Brown)
1720 * the compiler is now much more consistent in its error-checking
1721 treatment of bounding index arguments to sequence functions: in
1722 (SAFETY 3) code, errors will be signalled in almost all cases if
1723 invalid sequence bounding indices are passed to functions defined
1724 by ANSI to operate on sequences.
1725 * fixed a bug in the build procedure: documentation of SBCL-specific
1726 packages is now preserved and available in the final Lisp image.
1727 * lifted FDEFINITION lookup out of loops in the implementation of
1728 many list operations. (thanks to Robert E. Brown)
1729 * fixed a bug in the reader: the #n# reader macro now works for
1730 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1731 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1732 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1733 E. Brown and Tony Martinez respectively)
1734 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1735 SLOT-BOUNDP: in particular, it is now able to optimize them much
1736 better, and is now not vulnerable to having packages renamed.
1737 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1738 always called when a slot is not present in an instance. (thanks
1740 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1741 longer signal an error on unreading a character following EOF on
1742 the previous constituent stream. (thanks to Tony Martinez)
1743 * fixed some bugs revealed by Paul Dietz' test suite:
1744 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1745 not just nonnegative fixnums;
1746 ** the logical bit-array operators such as BIT-AND now accept an
1747 explicit NIL for their "opt-arg" argument (to indicate a
1748 freshly-consed result bit-array);
1749 ** ELT now signals an error on an invalid sequence index in safe
1751 ** the type system is now cleverer about negations of numeric
1752 types, and consequently understands the BIGNUM and RATIO types
1754 ** the type system is now cleverer about the interaction between
1755 INTEGER and RATIO types: while bugs still remain, many more
1756 cases are accurately computed;
1757 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1758 if it is in the last clause;
1759 ** CONSTANTLY now correctly returns a side-effect-free function in
1761 ** DECLARE is no longer treated as a special-operator; in
1762 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1763 * incremented fasl file version number due to the change in the
1764 DEFSTRUCT-SLOT-DESCRIPTION structure.
1766 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1767 * minor incompatible change: code processed by the "interpreter" or
1768 EVAL now has a compilation optimization policy of (DEBUG 2)
1769 (changed from (DEBUG 1)) to improve debuggability of interactive
1770 development, and to allow the use of the debug RETURN command in
1772 * an experimental implementation of the RETURN command for the
1773 debugger has been included. (thanks to Frederik Kuivinen)
1774 * fixed bug 62: constraints were not propagated into a loop.
1775 * fixed bug in embedded calls of SORT (reported and investigated by
1777 * fixed some bugs revealed by Paul Dietz' test suite:
1778 ** printing and reading of arrays with some dimensions having
1779 length 0 (thanks to Gerd Moellmann);
1780 ** BOA constructor with &AUX argument without a default value does
1781 not cause a type error;
1782 ** CONSTANTP now returns true for all self-evaluating objects.
1784 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1785 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1786 accessors that are related by inheritance, as specified in the
1787 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1788 Valtteri Vuorikoski)
1789 * The compiler is now able to inline functions that were defined in
1790 a complex lexical environment (e.g. inside a MACROLET).
1791 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1793 * Support for the upcoming FreeBSD-5.0 release has been included.
1794 (thanks to Dag-Erling Smorgrav)
1795 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1796 effect when it is not in a toplevel context.
1797 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1798 stand a better chance of being correct. (thanks to Gerd
1800 * fixed bug in COERCE, which now signals an error on coercing a
1801 rational to a bounded real type which excludes the expected
1803 * The compiler is now able to derive types more accurately from the
1804 COERCE and COMPILE functions.
1805 * fixed bug 223: functional binding is considered to be constant
1806 only for symbols in the CL package.
1807 * fixed bug 231: SETQ did not check the type of a variable being set
1808 (reported by Robert E. Brown)
1809 * A new optimization for MAKE-INSTANCE has been included, fixing
1810 various bugs (including relating to :ALLOCATION :CLASS slots and
1811 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1813 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1814 ** As required by ANSI, LOOP now disallows anonymous collection
1815 clauses such as COLLECT I in conjunction with aggregate boolean
1816 clauses such as THEREIS (= I 1);
1817 ** LOOP now signals an error when any variable is reused in the
1818 same loop (including the potentially useful construct analogous
1819 to WITH A = 1 WITH A = (1+ A);
1820 ** IT is only a special loop symbol within the first clause of a
1821 conditional loop clause;
1822 ** LOOP with a typed iteration variable over a hashtable now
1823 signals a type error iff it should.
1824 * fixed some other bugs revealed by Paul Dietz' test suite:
1825 ** FILE-STREAM now names the class previously known as FD-STREAM;
1826 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1827 argument) no longer signals an error;
1828 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1829 of the slot symbol, rather than using the current package
1830 ((:CONC-NAME "") continues to intern the slot's name in the
1832 * incremented fasl file version number, because of the incompatible
1833 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1834 of the new implementation of DEFINE-COMPILER-MACRO.
1836 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1837 * Support for building SBCL for MIPS platforms running in
1838 little-endian mode has now been checked in, and basic
1839 functionality on said platforms verified.
1840 * minor incompatible change: PCL now records the pathname of a file
1841 in which methods and the like are defined, rather than its
1843 * minor incompatible change: TRUENAME now considers the truename of
1844 a file naming a directory to be the pathname with :DIRECTORY
1845 component indicating that directory.
1846 * minor incompatible change: a NAMED clause in the extended form of
1847 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1848 reason for the previous behaviour is unclear.
1849 * more systematization and improvement of CLOS and MOP conformance
1850 in PCL (thanks to Gerd Moellman and Pierre Mai):
1851 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1853 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1854 primary methods with no specializers;
1855 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1857 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1858 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1859 FORWARD-REFERENCED-CLASSes; error reporting on
1860 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1862 ** SXHASH on CLOS instances now uses a slot internal to the
1863 instance to return different numbers on distinct instances,
1864 while preserving the same return value through invocations of
1866 ** DEFMETHOD signals errors when methods with longer incongruent
1867 lambda lists are added to generic functions;
1868 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1869 CLASS, as specified in AMOP;
1870 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1871 on the order returned by the primary method for classes of
1872 class STANDARD-CLASS;
1873 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1874 * fixed some bugs shown by Paul Dietz' test suite:
1875 ** DOLIST puts its body in TAGBODY;
1876 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1878 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1879 value producing form;
1880 ** if more variables are given to PROGV than values, extra
1881 variables are bound and made to have no value;
1882 ** NSUBSTITUTE on list arguments gets the right answer with
1884 ** ELT signals an error of type TYPE-ERROR when the index argument
1885 is not a valid sequence index;
1886 ** LOOP signals (at macroexpansion time) an error of type
1887 PROGRAM-ERROR when duplicate variable names are found;
1888 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1889 ** FUNCALL of special-operators now cause an error of type
1891 ** PSETQ now works as required in the presence of side-effecting
1892 symbol-macro places;
1893 ** NCONC accepts any object as its last argument;
1894 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1896 ** loop-for-as-package does not require a package to be explicitely
1898 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1900 * fixed bug 166: compiler preserves "there is a way to go"
1901 invariant when deleting code.
1902 * fixed bug 172: macro lambda lists with required arguments after
1903 &REST arguments now cause an error to be signalled. (thanks to
1905 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1906 bug. (thanks to Matthew Danish)
1907 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1909 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1910 arguments are lexically rebound. (thanks to Gerd Moellmann and
1912 * fixed bug 194: error messages are now more informative when there
1913 is no primary method applicable in a call to a generic
1914 function. (thanks to Gerd Moellmann)
1915 * fixed bug in command line argument checking (thanks to Julian
1917 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1918 (thanks to Matthew Danish)
1919 * incremented fasl file version number, because of the
1920 SXHASH-related changes in the layout of CLOS data structures
1922 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1923 * minor incompatible change: The runtime (the Unix executable named
1924 "sbcl") is now much pickier about the .core files it will load.
1925 Essentially it now requires .core files to descend from the same
1926 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1927 as the runtime does. (The intent is to prevent the crashes which
1928 can occur, and which can even be reported as mysterious failures,
1929 when people patch the sources or change the build parameters
1930 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1931 sbcl and .core files.)
1932 * fixed bug: VALUES-LIST is no longer optimized away.
1933 * fixed bug 142: The FFI conversion of C string values to Lisp
1934 string values no longer conses excessively. (thanks to Nathan
1935 Froyd porting Raymond Toy's fix to CMU CL)
1936 * began to systematize and improve MOP conformance in PCL (thanks to
1937 Nathan Froyd, Gerd Moellman and Pierre Mai):
1938 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1940 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1941 ** FINALIZE-INHERITANCE is now called on class finalization;
1942 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1943 argument precedence order.
1944 * fixed bug 202: The compiler no longer fails on functions whose
1945 derived types contradict their declared type.
1946 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1947 so it can be non-toplevel.
1948 * The fasl file version number has changed (because of the new
1949 implementation of DEFMACRO).
1950 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1951 safe code, that any length requirement by their type-specifier
1952 argument is valid. The exceptions to this are described in bug
1954 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1955 functions now check the types of their inputs as required by ANSI.
1956 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1957 introduced symbol is DECLAREd to be SPECIAL.
1958 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1959 * fixed bug: :COUNT argument to sequence functions may be negative
1960 * fixed bug: body of DO-SYMBOLS may contain declarations
1961 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1962 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1963 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1964 symbol macro only once
1965 * fixed printing of call frame when argument list is unavailable
1966 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1967 * compiler no longer signals WARNING on unknown keyword
1970 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1971 * A beta-quality port to the mips architecture running Linux,
1972 based on the old CMUCL backend, has been made. It has been tested
1973 on a big-endian kernel, and works sufficiently well to be able to
1974 rebuild itself; it has not been tested in little-endian mode.
1975 * fixed an inconsistency between gencgc.c and purify.c which made
1976 dumping/loading .core files unreliable
1977 * fixed bug 120a: The compiler now deals correctly with IFs where
1978 the consequent is the same as the alternative, instead of
1979 misderiving the return type. (thanks to Alexey Dejneka)
1980 * fixed bug 113: Logical pathnames are now dumpable (the logical
1981 host is resolved at load-time, throwing an error if it is not
1983 * fixed bug 174: FORMAT's error message is slightly clearer when a
1984 non-printing character is used in a format directive.
1985 * fixed several bugs in compiler checking of type declarations, i.e.
1986 violations of the Python "declarations are assertions" principle
1987 (thanks to Alexey Dejneka)
1988 * fixed several bugs in PCL's error checking (thanks to Gerd
1990 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1992 * fixed bug in compilation of functions as first class values
1993 (thanks to Antonio Martinez-Shotton)
1994 * The compiler's handling TYPE-ERRORs which it can prove will
1995 inevitably happen at runtime has been cleaned up and corrected
1996 in several ways. (thanks to Alexey Dejneka)
1997 * improved argument type checking for various basic arithmetic
1998 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1999 lavished on them in the past that they can be compiled in many
2000 ways in different special cases
2001 * fixed bug 181: compiler checks validity of user supplied type
2003 * cleaned up code flushing in optimization: Function calls which
2004 should signal errors for safety purposes (e.g. which ANSI says
2005 should signal errors when their arguments are of incorrect type)
2006 are no longer optimized away.
2007 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2008 * incremented fasl file version number, because changes in the
2009 implementation of sequence functions like COERCE caused
2010 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2011 in old inline expansions) to become undefined. (Actually these
2012 changes were later undone, so we might very well be binary
2013 compatible with 0.7.7 after all, but leaving the version number
2014 incremented seemed like the simplest and most conservative
2017 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2018 * An alpha-quality port to the parisc architecture running Linux,
2019 based on the old CMUCL backend, has been made. This, even more so
2020 than the other backends, should be considered still a work in
2021 progress; known problems include that the Linux kernel in 64-bit
2022 mode does not propagate the correct sigcontext structure to
2023 userspace, and consequently SBCL on a parisc64 kernel will not
2025 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2026 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2027 suppressed.) Also now that the compiler is looking at declarations
2028 in the environment, it checks optimization declarations as well,
2029 and suppresses inlining when (> DEBUG SPEED).
2030 * More fixes have been made to treatment of floating point exception
2031 treatment and other Unix signals. In particular, floating point
2032 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2033 * The detection and handling of control stack exhaustion (infinite
2034 or very deeply nested recursion) has changed. Stack exhaustion
2035 detection is now done by write-protecting pages at the OS level
2036 and applies at all optimization settings; when found, a
2037 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2038 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2039 be caught using IGNORE-ERRORS.
2040 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2041 correctly when transforming recursive LABELS functions to LETs.
2042 (thanks to Alexey Dejneka)
2043 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2044 that are names of constants or global variables.
2045 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2046 alien routines with docstrings.
2047 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2048 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2050 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2051 (thanks to Eric Marsden)
2052 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2053 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2054 * Bug fix: The SPARC backend can now compile functions involving
2055 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2056 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2057 to a functions expecting a PATHNAME-DESIGNATOR.
2058 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2059 lambda lists. (thanks to Alexey Dejneka)
2060 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2061 functions so that EXPORTing the name of the class doesn't cause
2062 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2063 to Antonio Martinez for reporting this)
2064 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2065 constant-folded without failing an assertion. (thanks to Einar
2066 Floystad Dorum for reporting this)
2067 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2068 types are now dealt with more correctly by the compiler.
2069 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2070 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2071 behaviour analogous to RENAME-FILE. This puts its behaviour more
2072 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2074 * The fasl file version number has changed again. (because of the
2075 bug fix involving the names of PCL MAKE-INSTANCE functions)
2077 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2078 * bug fix: Floating point exceptions are treated much more
2079 consistently on the x86/Linux and PPC/Linux platforms.
2080 * Array initialization with :INITIAL-ELEMENT is now much faster for
2081 cases when the compiler cannot open code the array creation, but
2082 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2083 array accesses have also seen a speed increase.
2084 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2085 specified. (This was at the root of some bad interactions between
2086 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2088 * bug fix: Internal error arguments for undefined functions are now
2089 computed correctly on the PPC/Linux platform.
2090 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2091 Raymond Toy's patch for CMU CL)
2092 * Support for the Solaris 9 operating environment has been included
2093 (thanks to Daniel Merritt)
2094 * A very ugly but hopefully complete draft of the missing FFI chapter
2095 of the manual has been created by reformatting the corresponding
2096 CMU CL manual chapter into (currently very ugly and incoherent)
2097 DocBook and bringing it up to date for SBCL behavior. Thus, the
2098 manual is now essentially complete, at least by my extreme
2099 once-and-only-once standards, whereby it's acceptable to refer to
2100 the doc strings of SB-EXT functions as the primary documentation.
2101 * The fasl file version number has changed again, due to cleanup of
2102 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2103 is no longer a static symbol.)
2105 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2106 * SBCL now builds with OpenMCL (version 0.12) as the
2107 cross-compilation host; also, more progress has been made toward
2108 bootstrapping under CLISP.
2109 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2111 * bug 158 fixed: The compiler can now deal with integer loop
2112 increments different from 1; fixing this turned out also to fix
2114 * bug 169 fixed: no more bogus warnings about using lexical bindings
2115 despite the presence of perfectly good SPECIAL declarations (thanks
2116 to David Lichteblau)
2117 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2118 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2119 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2121 * bug 180 fixed: Method combination specifications no longer ignore
2122 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2123 * bug fix: Structure type predicate functions now check their argument
2124 count as they should.
2125 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2126 correctly. (thanks to Pierre Mai)
2127 * minor incompatible change: The --noprogrammer option is deprecated
2128 in favor of the new --disable-debugger option, which is very similar.
2129 (The major difference is that it takes effect at a slightly different
2130 time at startup, causing handling of errors in --sysinit and
2131 --userinit files will be affected differently.) The
2132 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2133 been added to allow this functionality to be controlled from ordinary
2134 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2135 maintainers, who might want to run non-interactive scripts to
2136 build SBCL cores which will later be used interactively.)
2137 * minor incompatible change: The LOAD function no longer, when given
2138 a wild pathname to load, loads all files matching that pathname.
2139 Instead, an error of type FILE-ERROR is signalled.
2141 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2142 * bug 147 fixed: The compiler preserves its block link/count
2143 invariants more correctly now so that it doesn't crash. (thanks
2145 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2147 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2148 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2149 no-value-for-this-character as NIL (as specified by ANSI).
2150 * HOST-NAMESTRING on physical pathnames now returns a string that is
2151 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2152 (thanks to Christophe Rhodes)
2153 * The Alpha port handles icache flushing more correctly. (thanks to
2155 * More progress has been made toward bootstrapping under CLISP. (thanks
2156 to Christophe Rhodes)
2157 * The fasl file format has changed again, because dynamic loading
2158 on OpenBSD (which has non-ELF object files) motivated some cleanups
2159 in the way that foreign symbols are transformed and passed around.
2160 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2161 is no longer treated as whitespace by the reader, but instead as
2162 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2163 |A
\7fB|, instead of A as it used to.
2165 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2166 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2167 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2168 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2169 this as of 0.7.1.45, but was left out of the previous news section
2170 (thanks to Dan Barlow)
2171 * SBCL now runs on the Solaris operating system on SPARC architectures
2172 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2173 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2174 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2175 * SPARC backend cleanups, allowing builds of cores optimized for V8
2176 and V9 SPARCS, and also emission of code targeted to a particular
2177 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2179 * SBCL is closer to bootstrapping under CLISP, thanks to various
2180 fixes by Christophe Rhodes.
2181 * The fasl file format has changed again, to allow the compiler's
2182 INFO database to support symbol macros.
2183 * The user manual (in doc/) is formatted into HTML more nicely.
2184 (thanks to coreythomas)
2185 * The system is smarter about SUBTYPEP relationships, especially
2186 those involving NOT types (including types such as ATOM which are
2187 represented internally using NOT types). Thus SUBTYPEP is less
2188 likely to return (VALUES NIL NIL) in general, and in particular
2189 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2191 * The fasl file format has changed again, because the internal
2192 representation of types now includes a new slot to support the new
2193 SUBTYPEP-of-NOT-types logic.
2194 * (not a change in the main branch of SBCL, but a related prototype
2195 which can hopefully be merged into the main branch of SBCL in the
2196 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2197 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2198 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2199 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2200 dispatch macro character combinations. (thanks to Alexey Dejneka)
2201 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2202 that superficially look like logical namestrings correctly.
2203 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2205 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2206 default supplied-p) for &optional and &key arguments. (thanks to
2209 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2210 * incompatible change: The compiler is now less aggressive about
2211 tail call optimization, doing it only when (> SPACE DEBUG) or
2212 (> SPEED DEBUG). (This is an incompatible change because there are
2213 programs which relied on the old CMU-CL-style behavior to optimize
2214 away their unbounded recursion which will now die of stack overflow.)
2215 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2216 for non-GENCGC systems has been increased to 20M (since that
2217 seems much closer to the likely performance optimum for modern
2218 systems than the old 4M value was)
2219 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2220 and *DEBUG-PRINT-LEVEL*
2221 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2222 of CMU CL's support for SPARC, and various endianness and other
2223 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2224 * new syntactic sugar for the Unix command line: --load foo.bar is now
2225 an alternate notation for --eval '(load "foo.bar")'.
2227 ** The system now detects stack overflow and handles it gracefully,
2228 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2229 optimization settings. (This is a good thing in general, and
2230 its introduction in this version should be particularly timely
2231 for anyone whose code fails because of suppression of tail
2233 ** The system now hunts for the C variable "environ" in a more
2234 devious way, to avoid segfaults when the C library version differs
2235 between compile time and run time. (thanks to Christophe Rhodes)
2236 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2237 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2238 ** The compiler no longer issues bogus style warnings for undefined
2239 classes in the same source file as the DEFCLASSes which defined
2240 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2242 ** fixes in CONDITION class precedence list for undefined function
2243 errors (thanks to Alexei Dejneka)
2244 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2245 correctly. (thanks to Dan Barlow)
2246 ** portability fixes aiming at bootstrapping under CLISP (thanks
2247 to Dave McDonald and Christophe Rhodes)
2248 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2249 ** fixes in type translation and and type inference (thanks to
2251 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2252 ** various fixes in the new ports (thanks to Dan Barlow)
2253 * several changes related to debugging:
2254 ** suppression of tail recursion, as noted above
2255 ** stack overflow detection, as noted above
2256 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2257 is now the default. (For some time encapsulation has been more
2258 reliable than the breakpoint-based :ENCAPSULATE NIL
2259 implementation, at least on X86 systems; and I just noticed that
2260 encapsulation also seems closer to the spirit of the ANSI
2263 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2265 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2266 up properly again. (There was a packaging bug in 0.7.0 which
2267 left their definitions in SB-SYS::LOAD-FOREIGN and
2268 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2269 vital for most things which interface to C-level interfaces,
2270 like extensions working with sockets or databases or
2271 Perl-compatible regexes or whatever, and the need to fix
2272 this bug is the main reason that 0.7.1 was released so
2274 ** DEFGENERIC is now choosier about the methods it redefines, so that
2275 reLOADing a previously-LOADed file containing DEFGENERICs does
2276 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2277 cycle now works as it should. (thanks to Alexey Dejneka)
2278 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2279 fixed by Christophe Rhodes. (He actually submitted this patch
2280 months ago, and I delayed until after 0.7.0.)
2281 ** Bug 111 (internal compiler confusion about runtime checks on
2282 FUNCTION types) was fixed by Alexey Dejneka.
2283 * Some internal cleanups (getting rid of variables which aren't
2284 needed now that the byte interpreter is gone) caused the fasl
2285 file format number to change again.
2287 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2288 * major incompatible change: The default fasl file extension, i.e. the
2289 default extension for files produced by COMPILE-FILE, has changed
2290 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2292 ** There are many changes in the implementation of the compiler.
2293 SBCL is now essentially a compiler-only implementation of ANSI
2294 Common Lisp. EVAL still "interprets" a few special cases, but
2295 almost all the interesting cases are handled by creating
2296 a LAMBDA expression, calling COMPILE on it, then calling
2297 FUNCALL on the result.
2298 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2299 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2300 Since the code is newer, there might still be some new bugs
2301 (though not as many as before Martin Atzmueller's fixes:-). But
2302 the new code is substantially simpler and clearer, and hopefully
2303 any remaining bugs will be simpler, less fundamental, and more
2304 fixable then the bugs in the old code.
2305 ** The revised compiler is still a little unsteady on its feet.
2307 *** The debugging information it produces (particularly the names
2308 of FUNCTION objects) is sometimes much less useful than what
2309 the old compiler produced.
2310 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2311 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2312 within a MACROLET) has been temporarily weakened.
2313 ** There are new compiler optimizations for various functions:
2314 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2315 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2316 *** the math functions TRUNCATE, FLOOR, and CEILING
2317 *** the function-of-all-trades COERCE
2318 Mostly these should be transparent, but there's one
2319 potentially-annoying problem (bug 117): when the compiler
2320 inline-expands a function and does type analysis on the result,
2321 it can create control paths which have type mismatches, and
2322 when it can't prove that those control paths aren't taken,
2323 it will issue WARNINGs about the type mismatches. This is
2324 a particular problem in practice for the new sequence functions.
2325 It's not clear how this should be fixed, and for now, a
2326 workaround is given in the entry for 117 in the BUGS file.
2327 ** (Because of the interaction between the two previous items --
2328 occasional inlining problems and new inline expansions -- some
2329 of the new sequence function optimizations won't really kick in
2330 completely until debugging information, and then inlining, are
2331 straightened out in some future version.)
2332 * minor incompatible changes:
2333 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2334 DIRECTORY no longer implicitly promotes NIL slots of its
2335 pathname argument to :WILD. In particular, when you ask for the
2336 contents of a directory (which you used to be able to do without
2337 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2338 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2339 ** changes in behavior that ANSI explicitly defines to be
2340 implementation dependent:
2341 *** The new compiler-only implementation still conforms with ANSI,
2342 but acts a little different than before. Besides the obvious
2343 changes in performance tradeoffs (that the cost per form passed
2344 to EVAL has gone up, and the cost per form executed by EVAL
2345 has gone down), the behavior of the system changes a little
2346 because there are no longer any interpreted function objects.
2347 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2348 e.g. doing COMPILE on the output of interactive DEFUN is
2350 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2352 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2353 has changed from (:CL) to NIL.
2354 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2355 e.g. #\Newline and #\Tab, don't have names specified in the
2356 ANSI Common Lisp standard, is now based on their ASCII symbolic
2357 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2358 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2359 are no longer used for output.
2360 ** changes in internal implementation constants:
2361 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2362 4 million. (If your application spends a lot of time GCing and
2363 you have a lot of RAM, you might want to experiment with
2364 increasing it even more.)
2365 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2366 However, almost all old code should still continue to work without
2367 immediate update, as SB-C-CALL is now a (deprecated) nickname
2369 ** Old operator names in the style DEF-FOO are now deprecated in
2370 favor of new corresponding names DEFINE-FOO, for consistency with
2371 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2372 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2373 internal symbols, but a few supported extensions like
2374 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2375 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2376 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2377 etc. as you get deeper into recursive calls to the debugger
2378 command loop, instead of the old "5]", "5]]", "5]]]"
2379 sequence. (I was motivated to do this when squabbles between
2380 ILISP and SBCL left me very deeply nested in the debugger. In the
2381 short term, this change will probably provoke more ILISP/SBCL
2382 squabbles, but hopefully it will be an improvement in the long run.)
2383 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2384 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2385 function, since declaiming return types involving aliens is
2386 (1) annoyingly messy to do by hand and (2) vital to efficient
2387 compilation of code which calls such functions.
2388 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2389 longer reexported by the SB-EXT package. They're solely useful
2390 for alien code, so it seems more logical that you should get
2391 them from the SB-ALIEN package, not in SB-EXT.
2392 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2393 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2394 features. Instead, the code that they used to control is always
2395 built into the system.
2396 * many other bug fixes
2397 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2398 advantage of the new EVAL-WHEN stuff and to clean them up in
2399 general, and they are now more ANSI-compliant in a number of
2400 ways. Martin Atzmueller is responsible for a lot of this.
2401 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2403 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2404 *** correct ERROR type for various file operations
2405 *** some fixes for Lisp streams
2406 *** DEFMETHOD syntax checking
2407 *** changing old weird representation of debug information as
2408 strings (which, among their other deficiencies, don't transform
2409 correctly when you rename packages, and don't change their
2410 print representation when you change things like *PACKAGE*
2411 and *PRINT-LENGTH*) to symbols and lists of symbols
2412 He also made several improvements and fixed several bugs in DESCRIBE.
2413 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2415 *** misbehavior of WRITE-STRING/WRITE-LINE
2416 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2417 and several other LOOP problems as well
2418 *** DIRECTORY when similar filenames are present
2419 *** DEFGENERIC with :METHOD options
2420 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2421 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2422 *** argument ordering in FIND with :TEST option
2423 *** mishandled package designator argument in APROPOS-LIST
2424 *** various problems in the backquote readmacro
2425 *** a bug in APROPOS
2426 *** probably some others that I'm not describing very well here,
2427 since the CVS log documents them by reference to sbcl-devel
2428 messages, and the SourceForge archives aren't working well.:-(
2429 ** Dan Barlow improved the Alpha port (and is making progress on the
2430 PPC port, for those of you who think different).
2431 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2432 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2433 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2434 continues to work on the SPARC port (for those of you in a position
2435 to look down upon our little PC-compatible boxes from a great height).
2436 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2437 and uses that copy, rather than the current dynamic value, when
2438 it's trying to decide whether to truncate output. Thus e.g.
2439 (let ((*print-lines* 50))
2440 (pprint-logical-block (stream nil)
2442 (let ((*print-lines* 8))
2443 (print (aref possiblybigthings i) stream)))))
2444 should now truncate the logical block only at 50 lines, instead of
2445 often truncating it at 8 lines, as it did before.
2446 * The doc/cmucl/ directory, containing old CMU CL documentation
2447 from the time of the fork, is no longer part of the base system.
2448 SourceForge has shut down its anonymous FTP service, and with it
2449 my original plan for distributing the old CMU CL documentation
2450 there. For now, if you need these files you can download an old
2451 SBCL source release and extract them from it.
2452 * The fasl file version number changed again, for dozens of reasons,
2453 some of which are apparent above.
2455 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2456 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2457 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2458 patch, so that DISASSEMBLE output is much nicer.
2459 * The code in the SB-PROFILE package now seems reasonably stable.
2460 I still haven't decided what the final interface should look like
2461 (I'd like PROFILE to interact cleanly with TRACE, since both
2462 facilities use function encapsulation) but if you have a need
2463 for profiling now, you can probably use it successfully with
2464 the current CMU-CL-style interface.
2465 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2466 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2467 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2468 link now returns the dangling link itself, and for similar
2469 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2470 link itself. (In these cases the old code signalled an error and
2471 looped endlessly, respectively.) Thus, DIRECTORY now works even
2472 in the presence of dangling and cyclic symbolic links.
2473 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2474 is now a supported extension again, since the consensus on
2475 sbcl-devel was that it can be useful for ordinary development
2476 work, not just for debugging SBCL itself.
2477 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2478 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2479 that currently-defined functions might be redefined later with
2480 different return types.
2481 * Hash tables can be printed readably, as inspired by CMU CL code
2482 of Eric Marsden and SBCL code of Martin Atzmueller.
2483 * better error handling in CLOS method combination, thanks to
2484 Martin Atzmueller porting Pierre Mai's CMU CL patches
2485 * more overflow fixes for >16Mbyte I/O buffers
2486 * A bug in READ has been fixed, so that now a single Ctrl-D
2487 character suffices to cause end-of-file on character streams.
2488 In particular, now you only need one Ctrl-D at the command
2489 line (not two) to exit SBCL.
2490 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2492 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2493 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2494 does the right thing.
2495 * The implementation of some type tests, especially for CONDITION
2496 types, is now tidier and maybe faster, due to CMU CL code
2497 originally by Douglas Crosher, ported by Martin Atzmueller.
2498 * Some math functions have been fixed, and there are new
2499 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2500 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2501 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2502 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2503 likely to be built into sbcl-0.7.0.)
2504 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2505 CL, ported to SBCL by Martin Atzmueller)
2506 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2507 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2508 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2509 * There's a new slam.sh hack to shorten the edit/compile/debug
2510 cycle for low-level changes to SBCL itself, and a new
2511 :SB-AFTER-XC-CORE target feature to control the generation of
2512 the after-xc.core file needed by slam.sh.
2513 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2514 COMPILE-FILE is no longer supported, so that now every function
2515 gets an entry point, so that block compilation looks a little
2516 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2517 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2518 returns the number of bytes consed since the system started,
2519 rather than the number consed since the first time the function
2520 was called. (The new definition parallels ANSI functions like
2521 CL:GET-INTERNAL-RUN-TIME.)
2522 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2523 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2524 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2525 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2526 * Fasl file version numbers are now independent of the target CPU,
2527 since historically most system changes which required version
2528 number changes have affected all CPUs equally. Similarly,
2529 the byte fasl file version is now equal to the ordinary
2532 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2533 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2534 is no longer recognized. I apologize for this, because it was
2535 listed in SB-EXT as a supported extension, but I found that
2536 its existing behavior was poorly specified, as well as incorrectly
2537 specified, and it looked like too much of a mess to straighten it
2538 out. I have enough on my hands trying to get ANSI stuff to work..
2539 * many patches ported from CMU CL by Martin Atzmueller, with
2540 half a dozen bug fixes in pretty-printing and the debugger, and
2541 half a dozen others elsewhere
2542 * fixed bug 13: Floating point infinities are now supported again.
2543 They might still be a little bit flaky, but thanks to bug reports
2544 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2545 as flaky as they were.
2546 * The --noprogrammer command line option is now supported. (Its
2547 behavior is slightly different in detail from what the old man
2548 page claimed it would do, but it's still appropriate under the
2549 same circumstances that the man page talks about.)
2550 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2551 are now supported, and enabled by default. Thus, the compiler can
2552 handle many floating point and complex operations much less
2553 inefficiently. (Thus e.g. you can implement a complex FFT
2555 * The compiler now detects type mismatches between DECLAIM FTYPE
2556 and DEFUN better, and implements CHECK-TYPE more correctly, and
2557 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2558 thanks to patches from Martin Atzmueller.
2559 * various fixes to make the cross-compiler more portable to
2560 ANSI-conforming-but-different cross-compilation hosts (notably
2561 Lispworks for Windows, following bug reports from Arthur Lemmens)
2562 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2563 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2564 fixed thanks to Pierre Mai's CMU CL patches.
2565 * fixes to keep the system from overflowing internal counters when
2566 it tries to use i/o buffers larger than 16M bytes
2567 * fixed bug 45a: Various internal functions required to support
2568 complex special functions have been merged from CMU CL sources.
2569 (When I was first setting up SBCL, I misunderstood a compile-time
2570 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2571 * improved support for type intersection and union, fixing bug 12
2572 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2573 more obscure bugs as well
2574 * some steps toward byte-compiling non-performance-critical
2575 parts of the system, courtesy of patches from Martin Atzmueller
2576 * Christophe Rhodes has made some debian packages of sbcl at
2577 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2578 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2579 stable, but are nonetheless usable. When he's ready, I'd be happy
2580 to add them to the SourceForge "File Releases" section. (And if
2581 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2582 * new fasl file format version number (because of changes in
2583 internal representation of (OR ..) types to accommodate the new
2584 support for (AND ..) types, among other things)
2586 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2587 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2589 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2591 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2592 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2593 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2594 * DESCRIBE now works on structure objects again.
2595 * Most function call argument type mismatches are now handled as
2596 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2597 know whether the function will be redefined before the call is
2598 executed. (The compiler could flag local calls with full WARNINGs,
2599 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2600 it doesn't keep track of enough information to know whether calls
2601 are local in this sense.)
2602 * Compiler output is now more verbose, with messages truncated
2603 later than before. (There should be some supported way for users
2604 to override the default verbosity, but I haven't decided how to
2605 provide it yet, so this behavior is still controlled by the internal
2606 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2607 src/compiler/ir1util.lisp.)
2608 * Fasl file format version numbers have increased again, because
2609 support for the Gray streams extension changes the layout of the
2610 system's STREAM objects.
2611 * The Gray subclassable streams extension now works, thanks to a
2612 patch from Martin Atzmueller.
2613 * The full LOAD-FOREIGN extension (not just the primitive
2614 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2615 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2616 but like most other programs, it defaults to copying the Unix
2617 environment from the original process instead of starting the
2618 new process in an empty environment.
2619 * Extensions which manipulate the Unix environment now support
2620 an :ENVIRONMENT keyword option which doesn't smash case or
2621 do other bad things. The CMU-CL-style :ENV option is retained
2622 for porting convenience.
2623 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2624 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2626 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2628 * A patch from Martin Atzmueller seems to have solved the SIGINT
2629 problem, and as far as we know, signal-handling now works cleanly.
2630 (If you find any new bugs, please report them!)
2631 * The system no longer defaults Lisp source file names to types
2632 ".l", ".cl", or ".lsp", but only to ".lisp".
2633 * The compiler no longer uses special default file extensions for
2634 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2635 seems to expect a single default extension for all compiled code,
2636 and there's no compelling reason to try to stretch the standard
2637 to allow two different extensions.) Instead, byte-compiled files
2638 default to the same extension as native-compiled files.
2639 * Fasl file format version numbers have increased again, because
2640 a rearrangement of internal implementation packages made some
2641 dumped symbols in old fasl files unreadable in new cores.
2642 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2643 many fewer weird special cases.
2644 * Bug #17 (differing COMPILE-FILE behavior between logical and
2645 physical pathnames) has been fixed, and some related misbehavior too,
2646 thanks to a patch from Martin Atzmueller.
2647 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2648 by Tim Moore, ported to SBCL by Martin Atzmueller.
2649 * Martin Atzmueller fixed several filesystem-related problems,
2650 including bug #36, in part by porting CMU CL patches, which were
2651 written in part by Paul Werkowski.
2652 * More compiler warnings in src/runtime/ are gone, thanks to
2653 more patches from Martin Atzmueller.
2654 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2657 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2659 * DESCRIBE now works on CONDITION objects.
2660 * The debugger now handles errors which arise when trying to print
2661 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2663 * The build system now uses an additional file, customize-target-features.lisp,
2664 to allow local modifications to the target *FEATURES* list. (The point of
2665 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2666 debugging features enabled, without having to worry about propagating it
2667 into everyone's system when I do a "cvs update".) When no
2668 customize-target-features.lisp file exists, the target *FEATURES* list
2669 should be constructed the same way as before.
2670 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2671 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2672 warn about behavior which is undefined under the ANSI spec. Note: This
2673 is specified by ANSI, but it's not very popular with programmers.
2674 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2675 macro in the SBCL sources for an example of a workaround which you
2676 might use to make portable ANSI-standard code which does what you want.
2677 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2678 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2679 when it's used as a non-toplevel form. (This is required in order
2680 to implement the DEFCONSTANT-EQX macro.)
2681 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2682 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2683 funny things, due to bugs in the implementation of EVAL-WHEN
2684 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2685 non-toplevel DEFCONSTANTs are uncommon.)
2686 * The core file version number and fasl file version number have been
2687 incremented, because the old noncompliant DEFCONSTANT behavior involved
2688 calling functions which no longer exist, and because I also took the
2689 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2690 * fixed bug 1 (error handling before read-eval-print loop starts), and
2691 redid debugger restarts and related debugger commands somewhat while
2693 ** The QUIT debugger command is gone, since it did something
2694 rather different than the SB-EXT:QUIT command, and since it never
2695 worked properly outside the main toplevel read/eval/print loop.
2696 Invoking the new TOPLEVEL restart provides the same functionality.
2697 ** The GO debugger command is also gone, since you can just invoke
2698 the CONTINUE restart directly instead.
2699 ** The TOP debugger command is also gone, since it's redundant with the
2700 FRAME 0 command, and since it interfered with abbreviations for the
2702 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2704 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2705 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2706 some porting work by Martin Atzmueller)
2707 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2708 with a boatload of patches which he ported to SBCL. Now that those
2710 ** The system tries to make sure that its low-priority messages
2711 are prefixed by semicolons, to help people who like to use
2712 syntax highlighting in their ILISP buffer. (This patch
2713 was originally due to Raymond Toy.)
2714 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2715 patches originally written by Raymond Toy.
2716 ** The compiler understands coercion between single-value and
2717 multiple-VALUES type expressions better, getting rid of some very
2718 weird behavior, thanks to patches originally by Robert MacLachlan
2719 and Douglas Crosher.
2720 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2721 lambda lists, thanks to a patch originally by Pierre Mai.
2722 ** The system no longer bogusly warns about "abbreviated type
2724 ** The compiler gets less confused by inlining and RETURN-FROM,
2725 thanks to some patches originally by Tim Moore.
2726 ** The system no longer hangs when dumping circular lists to fasl
2727 files, thanks to a patch originally from Douglas Crosher.
2728 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2729 undefined function error.
2730 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2731 in structure slot initforms) for now, documented workaround instead:-|
2732 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2733 * fixed bug 5: added stubs for various Gray stream functions called
2734 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2735 installed, at least appropriate type errors are generated
2736 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2737 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2738 consistently in DEFMETHOD forms.
2739 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2740 it doesn't seem to affect SBCL after all
2741 * The C runtime system now builds with better optimization and many
2742 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2744 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2746 * The system is now under CVS at SourceForge (instead of the
2747 CVS repository on my home machine).
2748 * The new signal handling code has been tweaked to treat register
2749 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2750 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2751 downstream, e.g. in debug-int.lisp, has implicit dependencies
2752 on the unsignedness of integer representation of machine words,
2753 and that caused the system to bomb out with infinite regress
2754 when trying to recover from type errors involving signed values,
2755 e.g. (BUTLAST '(1 2 3) -1).)
2756 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2757 (This was one of the bugs Peter Van Eynde reported back in July.)
2758 * The system now uses code inspired by Colin Walters' O(N)
2759 implementation of MAP (from the cmucl-imp@cons.org mailing
2760 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2761 inline the MAP operation, and there is more than one
2762 sequence argument to the MAP call (so that it can't just
2763 do ETYPECASE once and for all based on the type of the
2764 single sequence argument). (The old non-inline implementation
2765 of the general M-argument sequence-of-length-N case required
2766 O(M*N*N) time when any of the sequence arguments were LISTs.)
2767 * The QUIT :UNIX-CODE keyword argument has been renamed to
2768 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2769 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2771 * Raymond Wiker's patches to port dynamic loading from Linux to
2772 FreeBSD have been added.
2773 * The BUGS file is now more nearly up to date, thanks in large part
2774 to Martin Atzmueller's review of it.
2775 * The debugger now flushes standard output streams before it begins
2776 its output ("debugger invoked" and so forth).
2777 * The core version number and fasl file version number have both
2778 been incremented, because of incompatible changes in the layout
2780 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2781 used to not be called for a saved Lisp image.)
2782 * Martin Atzmueller's version of a patch to fix a compiler crash,
2783 as posted on sbcl-devel 13 September 2000, has been installed.
2784 * Instead of installing Martin Atzmueller's patch for the
2785 compiler transform for SUBSEQ, I deleted the compiler transform,
2786 and transforms for some similar consing operations.
2787 * A bug in signal handling which kept TRACE from working on OpenBSD
2789 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2790 away by constant folding
2791 * The system now defines its address space constants in one place
2792 (in the Lisp sources), and propagates them automatically elsewhere
2793 (through GENESIS and the sbcl.h file). Therefore, patching the
2794 address map is less unnecessarily tedious and error-prone. The
2795 Lisp names of address space constants have also been systematized.
2796 * CVS tags like dollar-Header-dollar have been removed from
2797 the sources, because they have never saved me trouble and
2798 they've been source of trouble working with patches and other
2799 diff-related operations.
2800 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2801 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2803 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2805 * The system has been ported to OpenBSD.
2806 * The system now compiles with a simple "sh make.sh" on the systems
2807 that it's supported on. I.e., now you no longer need to tweak
2808 text in the target-features.lisp-expr and symlinks in src/runtime/
2809 by hand, the make.sh takes care of it for you.
2810 * The system is no longer so grossly inefficient when compiling code
2811 involving vectors implemented as general (not simple) vectors (VECTOR T),
2812 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2813 which dares to use the various sequence functions on non-simple
2814 vectors, takes less of a performance hit.
2815 * There is now a primitive type predicate VECTOR-T-P
2816 to test for the (VECTOR T) type, so that e.g.
2817 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2818 can now be compiled with some semblance of efficiency. (The old code
2819 turned the type declaration into a full call to %TYPEP at runtime!)
2820 * AREF on (VECTOR T) is still not fast, since it's still compiled
2821 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2822 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2823 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2824 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2825 without making full calls to SUBTYPEP at runtime.
2826 (Some analogous efficiency issues for non-simple vectors specialized to
2827 element types other than T, or for non-simple multidimensional arrays,
2828 have not been addressed. They could almost certainly be handled the
2829 same way if anyone is motivated to do so.)
2830 * The changes in array handling break binary compatibility, so
2831 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2832 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2833 instead of (VALUES T T).
2834 * By following the instructions that Dan Barlow posted to sbcl-devel
2835 on 2 July 2000, I was able to enable primitive dynamic object
2836 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2837 functionality is not implemented (since it calls ld to resolve
2838 library references automatically, requiring RUN-PROGRAM for its
2839 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2840 try to resolve library references) is now supported.
2841 * The system now flushes the standard output streams when it terminates,
2842 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2843 them at several other probably-convenient times, e.g. in each pass of
2844 the toplevel read-eval-print loop, and after evaluating a form given
2845 as an "--eval" command-line option. (These changes were motivated by a
2846 discussion of stream flushing issues on cmucl-imp in August 2000.)
2847 * The source transform for TYPEP of array types no longer assumes
2848 that an array whose element type is a not-yet-defined type
2849 is implemented as an array of T, but instead punts, so that the
2850 type will be interpreted at runtime.
2851 * There is now some support for cross-compiling in make.sh: each of
2852 the phases of make.sh has its own script. (This should be transparent
2853 to people doing ordinary, non-cross-compile builds.)
2854 * Since my laptop doesn't have hundreds of megabytes of memory like
2855 my desktop machine, I became more motivated to do some items on
2856 my to-do list in order to reduce the size of the system a little:
2857 ** Arrange for various needed-only-at-cold-init things to be
2858 uninterned after cold init. To support this, those things have
2859 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2860 symbols with such names are now uninterned after cold init).
2861 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2862 fasl files for cold load.
2863 ** Remove the old compiler structure pooling code (which used to
2864 be conditional on the target feature :SB-ALLOC) completely.
2865 ** Redo the representation of some data in cold init to be more compact.
2866 (I also looked into supporting byte compiled code at bootstrap time,
2867 which would probably reduce the size of the system a lot, but that
2868 looked too complicated, so I punted for now.)
2869 * The maximum signal nesting depth in the src/runtime/ support code has
2870 been reduced from 4096 to 256. (I don't know any reason for the very
2871 large old value. If the new smaller value turns out to break something,
2872 I'll probably just bump it back up.)
2873 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2875 * Many, many bugs reported by Peter Van Eynde have been added to
2876 the BUGS list; some have even been fixed.
2877 * While enabling dynamic object file loading, I tried to make the
2878 code easier to understand, renaming various functions and variables
2879 with less ambiguous names, and changing some function calling
2880 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2881 * While trying to figure out how to do the OpenBSD port, I tried to
2882 clean up some of the code in src/runtime/. In particular, I dropped
2883 support for non-POSIX signal handling, added various comments,
2884 tweaked the code to reduce the number of compilation warnings, and
2885 renamed some files to increase consistency.
2886 * To support the new automatic configuration functionality in make.sh,
2887 the source file target-features.lisp-expr has been replaced with the
2888 source file base-target-features.lisp-expr and the machine-generated
2889 file local-target-features.lisp-expr.
2890 * fixed a stupid quoting error in make.sh so that using CMU CL
2891 "lisp -batch" as cross-compilation host works again
2893 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2895 * DESCRIBE no longer tries to call itself recursively to describe
2896 bound/fbound values, so that it no longer fails on symbols which are
2897 bound to themselves (like keywords, T, and NIL).
2898 * DESCRIBE now works on generic functions.
2899 * The printer now prints less-screwed-up representations of closures
2900 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2901 * A private symbol is used instead of the :EMPTY keyword previously
2902 used to mark empty slots in hash tables. Thus
2903 (DEFVAR *HT* (MAKE-HASH-TABLE))
2904 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2905 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2906 now does what ANSI says that it should. (You can still get
2907 similar noncompliant behavior if bang on the hash table
2908 implementation with all the symbols you get back from
2909 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2910 This breaks binary compatibility, since tests for equality to
2911 :EMPTY are wired into things like the macroexpansion of
2912 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2914 * There's now a minimal placeholder implementation for CL:STEP,
2915 as required by ANSI.
2916 * An obscure bug in the interaction of the normal compiler, the byte
2917 compiler, inlining, and structure predicates has been patched
2918 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2919 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2921 * Missing ordinary arguments in a macro call are now detected even
2922 when the macro lambda list contains &KEY or &REST.
2923 * The debugger no longer complains about encountering the top of the
2924 stack when you type "FRAME 0" to explicitly instruct it to go to
2925 the top of the stack. And it now prints the frame you request even
2926 if it's the current frame (instead of saying "You are here.").
2927 * As specified by ANSI, the system now always prints keywords
2928 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2930 * The default initial SIZE of HASH-TABLEs is now smaller.
2931 * Type information from CLOS class dispatch is now propagated
2932 into DEFMETHOD bodies, so that e.g.
2933 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2935 is now basically equivalent to
2936 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2937 (DECLARE (TYPE SINGLE-FLOAT X))
2939 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2940 operation, without having to do run-time type dispatch.
2941 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2942 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2943 * Since I don't seem to be making big file reorganizations very often
2944 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2945 is overflowing my ability to conveniently back them up), I've finally
2946 checked the system into CVS. (The CVS repository is on my home system,
2947 not at SourceForge -- putting it on SourceForge might come later.)
2948 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2949 high-level GC-NOTIFY-FOO functions send their output. (There's
2950 still very little control of where low-level verbose GC functions
2951 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2952 less than it used to -- the GC-NOTIFY-FOO functions are now under
2953 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2954 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2955 in only one place in the source code, and propagates it automatically
2956 everywhere that it's needed. Thus e.g. when I bump the version from
2957 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2958 * The C source files now include boilerplate legalese and documentation
2959 at the head of each file (just as the Lisp source files already did).
2960 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2961 to his page will be replaced with a link to his new CLiki service.
2963 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2965 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2966 * The build process now looks for GNU make under the default name "gmake",
2967 instead of "make" as it used to. If GNU make is not available as "gmake"
2968 on your system, you can change this default behavior by setting the
2969 GNUMAKE environment variable.
2970 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2971 can build without error under CMU CL.
2973 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2975 * There is now a partial SBCL user manual (with some new text and some
2976 text cribbed from the CMU CL manual).
2977 * The beginnings of a profiler have been added (starting with the
2978 CMU CL profiler and simplifying and cleaning up). Eventually the
2979 main interface should be through the TRACE macro, but for now,
2980 it's still accessed through vaguely CMU-CL-style functions and macros
2981 exported from the package SB-PROFILE.
2982 * Some problems left over from porting CMU CL to the new
2983 cross-compilation bootstrap process have been cleaned up:
2984 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2985 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2986 part when getting rid of the compiler *BACKEND* stuff.)
2987 ** The value of *NULL-TYPE* was screwed up, because it was
2988 being initialized before the type system knew the final
2989 definition of the 'NULL type. This screwed up several key
2990 optimizations in the compiler, causing inefficiency in all sorts
2991 of places. (I found it because I wanted to understand why
2992 GET-INTERNAL-RUN-TIME was consing.)
2993 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2994 PROCLAIM FTYPE information. Unfortunately this broke binary
2995 compatibility again, since now the forms output by DEFGENERIC
2996 to refer to functions which didn't exist in 0.6.3.
2997 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2998 can use the new (as of 0.6.3) transform for SOME into MAP into
3000 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3001 values to (INTEGER 0 1000000), so that the time code will no longer
3002 occasionally get blown up by Linux returning 1000000 microseconds
3003 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3004 its output conform to the ANSI spec. (Alas, this makes its output
3005 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3006 * A full call to MAP NIL with a single sequence argument no longer conses.
3007 * fixes to problems pointed out by Martin Atzmueller:
3008 * The manual page no longer talks about multiprocessing as though
3009 it were currently supported.
3010 * The ILISP support patches have been removed from the distribution,
3011 because as of version 5.10.1, ILISP now supports SBCL without us
3012 having to maintain patches.
3013 * added a modified version of Raymond Toy's recent CMU CL patch for
3014 EQUALP comparison of HASH-TABLE
3016 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3018 * The system still can't cross-compile itself with
3019 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3020 entails), but at least it can compile more of itself that way
3021 than it used to be able to, and various buglets which were uncovered
3022 by trying to cross-compile itself that way have now been fixed.
3023 * This release breaks binary compatibility again. This time
3024 at least I've incremented the FASL file format version to 2, so that the
3025 problem can be detected reliably instead of just causing weird errors.
3026 * various new style warnings:
3027 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3028 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3029 ** using the lexical binding of a variable named in the *FOO* style
3030 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3031 as specified by ANSI.
3032 * *RANDOM-STATE* is no longer automatically initialized from
3033 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3034 default behavior of the system is to repeat its behavior every time
3035 it's run. If you'd like to change this behavior, you can always
3036 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3037 old convention there was no comparably easy way to get the system to
3038 repeat its behavior every time it was run.
3039 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3040 FTYPE declarations has been removed, in favor of their ANSI
3041 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3042 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3043 the types of their sequence arguments can be determined at compile time.
3044 This is done through a new open code expansion for MAP which eliminates
3045 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3046 argument types can be determined at compile time.
3047 * The optimizer now transforms COERCE into an identity operation when it
3048 can prove that the coerced object is already of the correct type. (This
3049 can be a win for machine generated code, including the output of other
3050 optimization transforms, such as the MAP transform above.)
3051 * Credit information has been moved from source file headers into CREDITS.
3052 * Source file headers have been made more standard.
3053 * The CASE macro now compiles without complaining even when it has
3056 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3058 * (Note that the way that the PCL macroexpansions were rewritten
3059 to accommodate the change in DEFGENERIC below breaks binary
3060 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3061 not run under sbcl-0.6.2. Once we get out of alpha releases,
3062 i.e. hit release 1.0.0, we'll probably try to maintain binary
3063 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3064 and sbcl-1.4.4. Until then, however, it might be fairly common
3065 for maintenance releases to break binary compatibility.)
3066 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3068 * The reporting of unhandled signals has been changed to print some
3069 explanatory text as well as the report form. (Previously only
3070 the report form was printed.)
3071 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3072 it defines, so that the compiler no longer issues undefined function
3073 warnings for compiled-but-not-yet-loaded generic functions.
3074 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3075 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3076 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3077 section of the ANSI spec are supported.
3078 * The "" nickname for the "KEYWORD" package has been removed.
3079 The reader still handles symbol tokens which begin with a package marker
3080 as keywords, but it doesn't expose its mechanism for doing so in the
3081 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3082 * The system now issues STYLE-WARNINGs for contradictory TYPE
3083 proclamations. (Warnings for contradictory FTYPE proclamations would
3084 be nice too, but those can't be done usefully unless the type system
3085 is made smarter about FUNCTION types.)
3086 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3087 systematized, so that "*target-*.lisp is supposed to exist only on the
3088 target and imply that there's a related file which exists on the
3089 host, and *host-*.lisp is supposed to exist only on the host and imply
3090 that there's a related file which exists on the target. This involves a
3091 lot of renaming. Hopefully the acute confusion caused by the renaming
3092 will be justified by the reduction in chronic confusion..
3093 ** runtime-type.lisp -> early-target-type.lisp
3094 ** target-type.lisp -> late-target-type.lisp
3095 ** early-host-format.lisp -> early-format.lisp
3096 ** late-host-format.lisp -> late-format.lisp
3097 ** host-error.lisp -> misc-error.lisp
3098 ** early-error.lisp -> early-target-error.lisp
3099 ** late-error.lisp -> late-target-error.lisp
3100 ** host-defboot.lisp -> early-defboot.lisp
3101 ** code/misc.lisp -> code/target-misc.lisp
3102 ** code/host-misc.lisp -> code/misc.lisp
3103 ** code/numbers.lisp -> code/target-numbers.lisp
3104 ** code/early-numbers.lisp -> numbers.lisp
3105 ** early-host-type.lisp -> early-type.lisp
3106 ** late-host-type.lisp -> late-type.lisp
3107 ** host-typep.lisp -> typep.lisp
3108 ** load.lisp -> target-load.lisp
3109 ** host-load.lisp -> load.lisp
3110 ** host-disassem.lisp -> disassem.lisp
3111 ** host-insts.lisp -> insts.lisp
3112 ** byte-comp.lisp -> target-byte-comp.lisp
3113 ** host-byte-comp.lisp -> byte-comp.lisp
3114 ** host-signal.lisp -> signal.lisp
3115 ** host-defstruct.lisp -> defstruct.lisp
3116 ** late-target-type.lisp -> deftypes-for-target.lisp
3117 Furthermore, several other previously target-only files foo.lisp (e.g.
3118 hash-table.lisp and random.lisp) have been split into a target-and-host
3119 foo.lisp file and a target-only target-foo.lisp file, with their key type
3120 definitions in the target-and-host part, so that the cross-compiler will
3121 know more about target types.
3122 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3123 gone away. In their place are various *BACKEND-FOO* variables
3124 corresponding to the slots of the old structure.
3125 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3126 into the target SBCL has been fixed.
3127 * The chill.lisp system for loading cold code into a running SBCL
3129 * Support for the CMU CL "scavenger hook" extension has been removed.
3130 (It was undocumented and unused in the CMU CL sources that SBCL was
3131 derived from, and stale in sbcl-0.6.1.)
3132 * Various errors in the cross-compiler type system were detected
3133 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3134 (enabling various consistency checks). Many of them were fixed,
3135 but some hard problems remain, so the compiler is back to
3136 running without *TYPE-SYSTEM-INITIALIZED* for now.
3137 * As part of the cross-compiler type system cleanup, I implemented
3138 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3139 * I have started adding UNCROSS calls throughout the type system
3140 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3141 on cross-compiler input files will be able to go away, and various
3143 * CONSTANTP now returns true for quoted forms (as explicitly required
3146 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3148 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3149 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3150 (MAKE-LIST -1). In the longer term, it ought to become true
3151 that declarations are assertions even at SAFETY 1. For now, it's not
3152 quite true even at SAFETY 3, but it's at least more nearly true..
3153 (Note that this change seems to increases the size of the system by
3154 O(5%) and to decrease the speed of the compiler by 20% or more.)
3155 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3156 for the problem of printing dozens of lines of distracting information
3157 about low-level system machinery as part of the top stack frame
3158 on entry to the debugger when an undefined function was called.
3159 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3160 is not reset to COMMON-LISP-USER.
3161 * Compilation of stuff related to dyncount.lisp has been made conditional
3162 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3163 smaller. The various dyncount-related symbols have been moved into
3164 a new "SB-DYNCOUNT" package.
3165 * tty-inspect.lisp has been renamed to inspect.lisp.
3166 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3167 feature has gone away. (When we eventually port to other flavors of
3168 libc and/or Unix, we'll try to make the differences between flavors
3169 invisible at the user level.)
3170 * Various other *FEATURES* tags, and/or their associated conditionals,
3171 have been removed if obsolescent, or given better documentation, or
3172 sometimes given more-mnemonic names.
3174 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3176 * tidied up "make.sh" script
3177 * tidied up system directory structure
3178 * better "clean.sh" behavior
3179 * added doc/FOR-CMUCL-DEVELOPERS
3180 * many many small tweaks to output format, e.g. removing possibly-confusing
3181 trailing #\. character in DESCRIBE-INSTANCE
3182 * (EQUALP #\A 'A) no longer signals an error.
3183 * new hashing code, including EQUALP hashing
3184 * tidied up Lisp initialization and toplevel
3185 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3186 * command line argument processing
3187 * added POSIX-GETENV function to deal with Unix-ish environment variables
3188 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3189 terminating SBCL on EOF
3190 * non-verbose GC by default
3191 * There is no more "sbcl" shell script; the sbcl file is now the C
3192 runtime executable (just like CMU CL).
3193 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3195 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3196 symbol and package information as Lisp native symbol and package objects
3197 instead of strings naming symbols and strings naming packages. This way,
3198 whenever packages are renamed (as in warm init), debug information is
3199 transformed along with everything else.
3200 * tweaked the optimization policy declarations which control the building
3201 of SBCL itself. Now, among other things, the system no longer saves
3202 source location debugging information. (This helps two problems at once
3203 by reducing SBCL size and by keeping SBCL from trying to look for its
3204 sources -- which may not exist -- when reporting errors.)
3205 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3206 debugging and testing purposes
3207 * cleaned up printing, making the printer call PRINT-OBJECT for
3208 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3209 methods, giving nearly-ANSI behavior
3210 * converted almost all special variables to use *FOO* naming convention
3211 * deleted PARSE-TIME functionality, since it can be done portably
3212 * moved some files out of cold init into warm init
3213 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3215 * regularized formatting of source files
3216 * added an install.sh script
3217 * fixed ridiculous memory usage of cross-compiler by making
3218 compiler/alloc.lisp not try to do pooling unless it can hook
3219 itself into the GC of the cross-compilation host. Now the system
3220 builds nicely on my old laptop.
3221 * added :SB-ALLOC in target-features.lisp-expr
3222 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3223 was not implemented)
3224 * re-did condition handling and note reporting in the compiler. Notes
3225 are no longer handled by signalling conditions. Style warnings
3226 and warnings are handled more correctly and reported in such a way
3227 that it's easy to find one or the other in your output (so that you
3228 can e.g. figure out which of many problems caused COMPILE-FILE to
3230 * changed the severity of several compiler warnings from full WARNING
3231 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3232 compiler note reporting so that it doesn't use the condition system
3233 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3234 COMPILE-FILE command)
3235 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3236 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3237 no longer supported, since their implementation was deeply intertwingled
3238 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3239 * removed bogus "support" for compiler macros named (SETF FOO), and
3240 removed the compiler macro for SETF INFO (but only after making a fool
3241 of myself on the cmucl-imp mailing list by posting a bogus patch for
3242 DEFINE-COMPILER-MACRO..)
3243 * Compiled files containing forms which have side effects on the Lisp
3244 reader (such as DEFPACKAGE forms) are now handled more correctly.
3245 (Compiler queuing of top level lambdas has been suppressed by setting
3246 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3247 * deleted various currently-unused source files, e.g. gengc.lisp. They
3248 may be added back at some point e.g. when porting to other architectures,
3249 but until they are it's distracting to distribute them and to try to
3251 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3252 there were so many of them they're just distractions, and UNCROSS is
3253 known to be able to handle the current sources
3254 * moved PROFILE functionality into TRACE, so that it will be clear
3255 how the wrapping and unwrapping of functions when you profile them
3256 interacts with the wrapping and unwrapping of functions when you
3257 trace them. (Actually, the functionality isn't there yet, but at least
3258 the interface specification is there. Hopefully, the functionality will
3259 arrive with some future maintenance release.)
3260 * removed host-oops.lisp
3261 * changed signature of QUIT function to allow UNIX-CODE argument
3262 * fixed READ-SEQUENCE bug
3263 * tweaked verbose GC output so that it looks more like the progress
3264 output that ANSI specifies for functions like LOAD
3265 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3266 * added <http://sbcl.sourceforge.com> to the banner information printed by