1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
3 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
4 space for its heap. (thanks to Thiemo Seufer)
5 * bug fix: discriminating functions for generic function classes
6 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
7 make invalid assumptions about method precedence order. (reported
9 * optimizations: REMOVE-DUPLICATES now runs in linear time on
10 lists in some cases. This partially fixes bug 384.
11 * flush all standard streams before prompting in the REPL and the
13 * eof selects abort in the debugger
14 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
15 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
17 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
19 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
20 * New feature: Experimental support for bivalent streams: streams
21 opened with :element-type :default now allow character and binary
23 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
24 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
26 * bug fix: no more highly sporadic "couldn't check whether ~S is
27 readable" when reading a stream and an interrupt hits in the middle
28 of a select system call
29 * compiler better recognizes complex arrays (reported by Eduardo
31 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
32 for variables on SBCL built with linkage-tables. (reported by Luis
34 * various error reporting improvements.
35 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
36 (Thanks to Thiemo Seufer)
37 * bug fix: nested reader invokations work correctly
38 * bug fix: it is possible to have simultaneous references to foreign
39 code and foreign data with the same name.
41 ** added x86-64 support
42 ** incompatible change: the threading api now works with thread
43 objects instead of thread ids
44 ** bug fix: threads are protected from signals and interruption when
45 starting up or going down
46 ** bug fix: a race where an exiting thread could lose its stack to gc
47 ** bug fix: don't halt on infinite error in threads if possible
48 ** fixed numerous gc deadlocks introduced in the pthread merge
49 ** bug fix: fixed thread safety issues in read and print
50 ** bug fix: debugger doesn't hang on session lock if interrupted at
51 an inappropriate moment
52 ** bug fix: run-program is now thread safe(r)
53 ** bug fix: inner with-recursive-lock no longer releases the mutex
54 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
55 * fixed some bugs revealed by Paul Dietz' test suite:
56 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
57 EXPECTED-TYPE slots filled.
58 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
60 ** MAKE-CONDITION accepts classes as type-designators.
61 ** COMPILE may never return NIL.
62 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
63 range before calling Unix time functions
65 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
66 * numerous signal handling fixes to increase stability
67 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
68 * minor incompatible change: we now correctly canonize default
69 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
70 as per the MOP, rather than the historical (INITARG INITFUNCTION
71 INITFORM). (reported by Bruno Haible)
72 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
73 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
75 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
76 pathname is a directory pathname.
77 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
78 * dynamic space size on PPC has been increased to 768Mb. (thanks to
80 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
81 :METACLASS argument in addition to a class name. (reported by
82 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
83 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
84 (reported by Bruno Haible; patch from Gerd Moellmann)
85 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
87 * bug fix: more cleanups to the floating point exception handling on
88 x86-64 (thanks to James Knight)
89 * bug fix: the compiler does not try to constant fold calls of
90 COERCE and BIT-* functions when they return freshly constructed
91 objects. (reported by Nikodemus Siivola and Paul Dietz)
92 * optimization: improved the MIPS versions of generic (in the
93 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
94 * optimization: direct conversion of (unsigned-byte 32) to floats on
96 * optimization: structure instances with raw slots now use less
97 memory, and probably show better memory locality. (thanks to
99 * optimization: DYNAMIC-EXTENT declarations for lists and closures
100 are treated as requests for stack allocation on the x86-64,
101 Alpha-32, PPC and SPARC platforms.
102 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
103 the wrong file descriptor; implementation of SOCKET-OPEN-P.
104 (thanks to Tony Martinez)
106 ** gcing a dead thread can no longer lead to lockups
107 ** threads block signals until they are set up properly
108 ** errno is no longer shared by threads
109 ** interrupt-thread restores the eflags register on x86
110 ** fixed some lockups due to gc/thread interaction
111 * fixed some bugs revealed by Paul Dietz' test suite:
112 ** invalid dotted lists no longer raise a read error when
114 ** signal an error if a symbol that names a declaration is used
115 as the name of a type, or vice versa
116 ** allow using the (declare (typespec var*)) abbreviation for
117 (declare (type typespec var*)) with all type specifiers
118 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
119 ** always return NIL from PPRINT-POP when OBJECT is NIL
120 ** don't signal errors when pretty-printing malformed LABELS,
121 FLET or MACROLET forms
122 ** declarations in a DOLIST body are also in scope for the
124 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
126 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
129 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
130 * fixed cross-compiler leakages that prevented building a 32-bit
131 target with a 64-bit host compiler.
132 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
133 opened with :IF-EXISTS :OVERWRITE.
134 * fixed bug 281: error for an invalid qualifier in a short-form method
135 combination method is not signalled until the faulty method is called.
136 * bug fix: iteration variable type inferrer failed to deal with open
137 intervals. (reported by Alan Shields)
138 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
139 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
140 or more runtime options were provided to the sbcl binary.
141 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
142 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
143 * optimization: there's now a fast-path for fixnum arguments in the
144 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
145 * optimization: the code generated on x86-64 is more compact thanks
146 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
147 * bug fix: floating-point exception handling now partly works on
148 x86-64. (Thanks to James Knight)
149 * improvement to the MIPS backend from Thiemo Seufer: C-style
150 64-bit long long arguments and return values to alien functions
152 * fixed some bugs revealed by Paul Dietz' test suite:
153 ** the type-error signalled from WARN has a filled-in DATUM slot.
154 ** the type-error required when a stream is not associated with
155 a file has the stream as its datum.
156 ** type-errors on single-floats on x86-64 no longer have
157 :INVALID-OBJECT as the datum
158 ** the type-errors signalled for invalid function names now have
159 a correct expected type
160 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
161 for structures defined with a :TYPE. Documentation strings for
162 typed structures are no longer immediately discarded
163 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
164 broadcast streams and synonym streams. FILE-LENGTH now also works
165 on broadcast streams.
167 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
168 * incompatible change: the --noprogrammer option, deprecated since
169 version 0.7.5, has been removed. Please use the equivalent
170 --disable-debugger option instead.
171 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
173 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
174 inoperational for a while now) has been completely removed.
175 * null lexical environments are now printed as #<NULL-LEXENV>,
176 significantly reducing the amount of clutter in typical backtraces.
177 * documentation on weak pointers, finalization, and after GC hooks
178 has been added to the manual.
179 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
180 as other specialized array types, is much faster. SUBSEQ and
181 COPY-SEQ on such arrays have also been sped up.
182 * optimization: EQL is now more efficient when at least other argument
183 is known to be of type (OR FIXNUM (NOT NUMBER)).
184 * fixed bug: the runtime is now less vulnerable to changes in the
185 size of the SBCL object on OS X, and virtual memory is reserved for
186 all spaces that need to be at a fixed address.
187 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
188 * fixed bug: finalizers and after GC hooks that cause consing are now
190 * fixed bug: compiler error messages and summaries are now printed to
191 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
192 * fixed inference of the upper bound of an iteration variable.
193 (reported by Rajat Datta).
194 * fixed bug 211e: calling local functions with duplicated constant
195 keyword argument no longer causes a bogus style warning about an
197 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
198 ftype declaration to be disregarded. (reported by Dave Roberts)
199 * fixed bug 373: caused by erronous compilation of references to alien
200 variables in the runtime on ppc/darwin.
201 * fixed bug 376: CONJUGATE type deriver.
202 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
203 (reported by Baughn on #lisp)
204 * a cross-compiler bug on non-x86 platforms has been identified and
205 fixed. (thanks to Bruno Haible)
206 * improvements to the MIPS runtime code for increased stability.
207 (thanks to Thiemo Seufer)
208 * increased the maximimum compact environment size to allow
209 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
210 * partial workaround for bug 135: don't name the function we're
211 calling for hairy cases of EVAL, so as not to accumulate one environment
212 entry per call to EVAL. (reported by Kevin Reid)
213 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
214 * optimization: type testing for non-vector arrays should be faster.
215 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
217 * bug fix: setting 31st element of a bit vector to zero did not work
219 * bug fix: redefining a class definition which failed due to a
220 previous accessor / function clash now works (but see BUGS entry
221 #380 for more problems in this area). (thanks to Zach Beane)
222 * the long form of DEFINE-METHOD-COMBINATION disables method group
223 checking when given a single method group with pattern *. (thanks
225 * on x86 compiler supports stack allocation of results of simple
226 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
227 * support for building and running on Mac OS X 10.4 "Tiger" has been added
228 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
229 * fixed some bugs related to Unicode integration:
230 ** the restarts for recovering from input and output encoding
231 errors only appear when there is in fact such an error to
233 * fixed some bugs revealed by Paul Dietz' test suite:
234 ** MISC.549 and similar: late transformation of unsafe type
235 assertions into derived types caused unexpected code
237 ** SCALE-FLOAT type deriver is less wrong.
238 ** type derivers for EXP, LOG and similar functions compute result
239 types for complex arguments better.
240 ** (MISC.563) CONJUGATE type deriver works for very restricted
242 ** out-of-line type testers for character strings are available.
243 ** EQUAL compiler transform understands specialness of objects
245 ** accessing double-floats stored on the stack now works on x86-64.
246 ** debugger internals could sometimes create invalid lispobjs,
247 resulting in GC crashes.
248 ** MISC.548: type check weakening can convert required type into
250 ** initialization forms for bindings are not in scope of free special
253 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
254 * incompatible change: thread support for non-NPTL systems has
255 been removed - locking is buggy and unreliable. A threaded
256 SBCL build will now warn at startup and refuse to create
257 new threads, unless futex support is detected in the kernel
258 * incompatible change: the top level REPL now has only an ABORT
259 restart associated with it, not TOPLEVEL and ABORT as it used to.
260 TOP and TOPLEVEL are now available as debugger commands for
261 returning to the top level.
262 * incompatible change: forms evaluated in the REPL now use the
263 global optimization policy.
264 * incompatible change: user- and system-initialization files are
265 no longer processed with LOAD, but by READ and EVAL; hence the
266 global optimization policy, startup package, readtable, etc,
268 * The .fasl file format number has been incremented because of
269 various incompatible changes.
270 * internal entry point details and argument counts no longer appear
271 in backtraces unless explicitly requested by setting
272 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
273 * built-in and standard functions no longer have names like "top
274 level local call to FOO".
275 * fixed bug 32: functions defined in non-null lexical environments
276 now have more legible printed representation
277 * fixed bug 33: functions defined in non-null lexical environemnts
278 are now more amenable to inspection by INSPECT.
279 * workaround for bug 354: XEPs no longer appear in backtraces unless
280 explicitly requested.
281 * fixed bug: receiving the signal which results from attempting to
282 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
283 SIGBUS on other *BSDs) on architectures where the C stack is also
284 the Lisp stack (x86 and x86-64) from foreign code no longer leads
285 to debugger confusion or wild instability. (reported by Cheuksan
287 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
288 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
289 (reported by Lutz Euler)
290 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
291 compiled code. (reported by Paul Dietz)
292 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
293 specialized parameter to a method to something that is not TYPEP
294 the specializer is now possible.
295 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
296 face of package deletion.
297 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
298 pathname. (reported by David Tolpin on c.l.l)
299 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
300 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
301 STRING-STREAM and FILE-STREAM, respectively.
302 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
303 than previously; in addition, its test suite is now run on build.
304 (thanks to Luke Gorrie)
305 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
306 * optimization: added a immediate representation for single-floats
308 * fixed some bugs revealed by Paul Dietz' test suite:
309 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
310 ** attempting to create a package with a colliding nickname causes
311 correctable errors to be signalled.
312 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
313 ** several x86-64 backend bugs related to sign-extension of immediate
316 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
317 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
318 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
320 * fixed loading of multiply forward-referenced layouts.
321 (thanks to Cheuksan Wang)
322 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
323 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
324 (see COMPILE-FILE documentation for details of :PRINT option)
325 * fixed bugs 19 and 317: fixed-format floating point printing is
326 more accurate. This also fixes a bug reported by Adam Warner
327 related to the ~@F format directive.
328 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
330 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
331 dispatch table if the from-char is a dispatch macro character.
332 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
333 by the wordsize no longer ignore the last word. (reported by Lutz
335 * fixed bug in type checking of dynamic-extent variables. (reported
337 * optimization: sequence traversal functions use their freedom to
338 coerce function designators to functions.
339 * optimization: code with many calls to CLOS methods specialized on
340 CLOS classes has had redundant type checks removed.
341 * optimization: added declarations to speed up operations that access
342 the internal character database (for example STRING-UPCASE)
343 * optimization: comparison operations between floats and sufficiently small
344 fixnums no longer create extra rationals
345 * fixed some bugs related to Unicode integration:
346 ** portions of multibyte characters at the end of buffers for
347 character-based file input are correctly transferred to the
348 start of the buffer at the next read.
349 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
350 passing it through to OPEN.
351 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
352 argument given to internal calls to OPEN.
353 ** the built-in comment readers (introduced by character sequences
354 ";" and "#|") are more forgiving to encoding errors; they will
355 STYLE-WARN and then attempt to resync the stream at a character
356 boundary. (thanks to Teemu Kalvas)
357 * fixed some bugs in the x86-64 port:
358 ** Negative short int return values from c-calls are sign-extended
360 ** The stack is aligned to 16-bytes for c-calls, as required by
361 the ABI. (thanks to Cheuksan Wang)
362 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
363 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
364 for enums. (thanks to Vincent Arkesteijn)
365 ** Multiple small optimizations and bugfixes for floating point
367 * fixed some bugs revealed by Paul Dietz' test suite:
368 ** Space, Tab, Linefeed, Return and Page have the invalid
369 secondary constituent character trait.
370 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
372 ** WITH-INPUT-FROM-STRING should only update the index place on
374 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
376 ** Bit-array operations (BIT-AND and similar) worked incorrectly
377 with one-dimensional arrays with fill pointers.
378 ** TYPE-OF failed on a complex with an integer realpart and a
380 ** compiler failure during type inference for the code of form
381 (IF (EQL X (THE ...)) ...) (MISC.535).
383 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
384 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
385 platform. The port supports 61-bit fixnums, large memory spaces
386 and reloading shared object files.
387 * enhancement: saving cores with foreign code loaded is now
388 supported on ppc/Darwin in addition to the previously supported
390 * enhancement: the statistical profiler now walks deeper into the
391 call stack for more meaningful call-graphs and accrued time
392 reports (x86/x86-64 only). It also now reports time spent in
394 * enhancement: it is now possible to trace most individual methods
395 of a generic function in addition to tracing the generic function
397 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
399 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
400 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
401 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
402 produces an error. (thanks to Vincent Arkesteijn)
403 * bug fix: NAMESTRING on pathnames with :WILD components in their
404 directories works correctly. (thanks to Artem V. Andreev)
405 * fixed bug 125: compiler preserves identity of closures. (reported
407 * bug fixed: functions with &REST arguments sometimes failed with
408 "Undefined function" when compiled with (DEBUG 3). (reported by
410 * bug fix: overflow during compiling of setting element of a bit
411 vector with constant index and value. (reported by Timmy Douglas)
412 * build fix: fixed the dependence on *LOAD-PATHNAME* and
413 *COMPILE-FILE-PATHNAME* being absolute pathnames.
414 * on x86 compiler partially supports stack allocation of dynamic-extent
416 * GO and RETURN-FROM do not check the extent of their exit points
417 when compiled with SAFETY 0.
418 * fixed some bugs related to Unicode integration:
419 ** encoding and decoding errors are now much more robustly
420 handled; it should now be possible to recover even from invalid
421 input or output to the terminal. (thanks to Teemu Kalvas)
422 ** provided a first cut at implementing STRING-TO-OCTETS and
423 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
424 ** altered the SB-MD5 contributed module slightly, changing the
425 interface just enough for it to be supportable for builds where
426 lisp characters are not eight bits.
427 * fixed some bugs revealed by Paul Dietz' test suite:
428 ** the FORMATTER-generated functions for ~V[ conditionals require
429 the correct number of arguments.
430 ** READ-FROM-STRING returns the mandated second value when applied
431 to displaced strings.
432 ** the #\Rubout and #\Backspace characters are treated as invalid
433 constituent characters by the tokenizer.
435 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
436 * new feature: reloading changed shared object files with
437 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
438 * new feature: references to foreign variables and functions
439 can now be compiled and loaded before the corresponding shared
440 object file is loaded, as long as the foreign definitions are
441 available at runtime.
442 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
443 supported. (thanks to Dan Debertin)
444 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
445 just keywords. (thanks to Vincent Arkesteijn)
446 * fixed bug #331: structure-class instances corresponding to
447 DEFSTRUCT forms are now created eagerly.
448 * fixed bug #345: backtraces from calls to undefined functions work
449 on x86 as well. Related bug #61 is now also partially fixed on x86
450 (backtraces from throws to unknown catch tags.)
451 * bug fix: lambda-list parsing is now stricter vrt. order and number
452 of lambda-list keywords.
453 * bug fix: as specified by AMOP, an error is signalled if a
454 class-option appears multiple times in a DEFCLASS form. (reported
456 * bug fix: RANDOM can be compiled when the compiler derives the type
457 of its numeric argument as a disjoint set of small integers.
458 (reported by Paul Dietz)
459 * bug fix: starting a core saved with shared objects loaded when
460 those objects are not available no longer causes threaded SBCL to
461 hang. (reported by Sean Ross)
462 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
464 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
465 argument, which must always be NIL. (reported by Kalle Niemitalo)
466 * bug fix: printing 1.0d+23 no longer results in an error.
467 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
468 * bug fix: structure slot setters preserve evaluation order. (thanks
470 * fixed some bugs related to Unicode integration:
471 ** RUN-PROGRAM can allow its child to take input from a Lisp
472 stream. (reported by Stefan Scholl)
473 ** ASDF-INSTALL successfully downloads .tar.gz files in all
474 locales. (reported by Ken Causey)
475 * fixed some bugs revealed by Paul Dietz' test suite:
476 ** INCF, DECF and REMF evaluate their place form as specified in
478 ** FORMATTER expands ~{ ~} iteration directives with V or #
479 parameters correctly.
480 ** FORMATTER deals with the ~@[ ~] conditional directive where the
481 consequent uses no arguments correctly.
482 ** the system has a partial understanding of the (COMPLEX RATIO)
484 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
487 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
488 * new feature: a build-time option (controlled by the :SB-UNICODE
489 keyword feature, enabled by default) for building the system with
490 support for the entire 21-bit character space defined by the
492 * new feature: the system now has rudimentary external-format
493 support; the primary user-visible change at this time is that
494 characters with the high bit set (such as lower-case-e-acute) will
495 print correctly to a terminal in a UTF-8 environment.
496 * minor incompatible change: BASE-CHAR no longer names a class;
497 however, CHARACTER continues to do so, as required by ANSI.
498 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
499 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
500 has been moved to the SB-EXT package (temporarily re-exported from
502 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
503 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
505 * fixed bug #351: better error-handlind and reporting for malformed
507 * fixed bug #350: bignum-printing is now more memory-efficient,
508 allowing printing of very large bignums, eg. (expt 2 10000000).
509 (reported by Bruno Haible)
510 * fixed bug #302: better primitive-type selection for intersection
512 * fixed bug #308: non-graphic characters now all have names, as
513 required. (reported by Bruno Haible)
514 * bug fix: redefining a class with different superclasses now correctly
515 removes it from the direct-subclasses of its previous superclasses.
516 (reported by David Morse)
517 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
518 new value now works. (reported by Bruno Haible)
519 * bug fix: correct canonicalization of multiple non-standard slot
520 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
521 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
522 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
523 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
524 now exists, an signals an error.
525 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
526 during macroexpansion for non-iteration variables as well. (reported
527 by Bruno Haible for CMUCL)
528 * bug fix: Cyclic structures and unprintable objects in compiler
529 messages no longer cause errors. (reported by Bruno Haible)
530 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
531 and READ-FROM-STRING all now return a primary value of NIL if
532 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
533 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
534 (reported by Bruno Haible for CMUCL)
535 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
536 do not designate a condition. (reported by Bruno Haible for
538 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
539 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
540 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
541 permitting resolution in favour of any of the conflicting symbols.
542 (reported by Bruno Haible for CMUCL)
543 * bug fix: EQUAL compiler optimizations is less aggressive on
544 strings which can potentially compare true despite having distinct
545 specialized array element types.
546 * bug fix: unit enumerations can be defined without dividing by
547 zero. (thanks to Vincent Arkesteijn)
548 * FORMAT compile-time argument count checking has been enhanced.
549 (report from Bruno Haible for CMUCL)
550 * a partial workaround for the bug 262: the compiler does not try to
551 inline-expand a local function doing RETURN-FROM from a deleted
552 BLOCK. (thanks to Peter Denno for the bug report and to David
553 Wragg for the simple test case)
554 * fixed some bugs revealed by Paul Dietz' test suite:
555 ** RENAME-PACKAGE allows all package designators as new package
557 ** constraint propagation and lambda variable substitution are
558 more cautious in dealing with partially deleted code.
559 ** compiler sometimes lost reoptimization passes.
560 ** CERROR, given a condition as condition designator, passes its
561 remaining arguments to the continue format control without
563 ** Case-altering FORMAT directives work correctly on non-ASCII
565 ** The REAL type specifier handles bounds outside the floating
566 point ranges without signalling FLOATING-POINT-OVERFLOW.
567 ** Functions with IR1-transformations can create intercomponent
568 references to global functions.
569 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
571 ** FORMAT ~R treats a nil value for its first parameter correctly.
573 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
574 * enhancement: saving cores with foreign code loaded is now
575 supported on x86/NetBSD and sparc/Linux in addition to the previously
577 * bug fix: on some platforms repeated installations caused multiple
578 copies of HTML documentation to be installed -- should not happen
579 any more. (reported by Stefan Scholl)
580 * bug fix: parsing self-recursive alien record types multiple times
581 no longer causes infinite recursion. (reported by Thomas F. Burdick,
582 original patch by Helmut Eller for CMUCL)
583 * bug fix: stack-exhaustion detection works now on NetBSD as well.
584 (thanks to Richard Kreuter)
585 * bug fix: defining classes whose accessors are methods on existing
586 generic functions in other (locked) packages no longer signals
587 bogus package lock violations. (reported by François-René Rideau)
588 * bug fix: special variables as DEFMETHOD parameters no longer have
589 associated bogus type declarations. (reported by David Wragg and
591 * bug fix: read-write consistency on streams of element-type
592 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
593 * bug fix: redefiniton of the only method of a generic function with
594 no DEFGENERIC no longer emits a full WARNING. In addition,
595 redefinition of generic functions with no DEFGENERIC to an
596 incompatible lambda list now signals an error. (thanks to Zach
598 * bug fix: DEFGENERIC now works even when there's a function of the
599 same name in an enclosing lexical environment. (thanks to Zach
601 * fixed compiler failure, caused by instrumenting code during
602 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
603 * optimization: added loop analysis and improved register allocation
604 * fixed some bugs revealed by Paul Dietz' test suite:
605 ** POSITION on displaced vectors with non-zero displacement
606 returns the right answer.
607 ** (SIMPLE-STRING) is a valid type specifier for sequence
609 ** *PRINT-LEVEL* handling for slotless structures is pedantically
611 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
612 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
614 ** The justification version of the FORMAT ~< directive treats
615 non-zero minpad parameter correctly.
617 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
618 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
619 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
620 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
621 the supported interface.
622 * new feature: Single-stepping of code compiled with DEBUG 2 or
623 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
624 * new feature: saving cores with foreign code loaded is now
625 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
626 Timothy Moore's work for CMUCL)
627 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
628 parameters to * instead of NIL if no initform is supplied.
629 (reported by Johan Bockgaard)
630 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
631 correctly when FUNCALL appears as the car of the form. Note:
632 despite this FUNCALL forms are not currently subject to
633 compiler-macro expansion. (port of Raymond Toy's fix for the
634 same from CMUCL, reported by Johan Bockgaard)
635 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
636 (thanks for Teemu Kalvas)
637 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
638 case instead of terminating the iteration (thanks for Julian
639 Squires, Sean Champ and Raymond Toy)
640 * bug fix: incorrect expansion of defgeneric that caused a style
641 warning. (thanks for Zach Beane)
642 * bug fix: slot accessor effective method computation works properly
643 for classes with multiple non-standard applicable methods on
644 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
645 * on x86 compiler supports stack allocation of results of LIST and
646 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
647 CMUCL implementation by Gerd Moellmann)
648 * fixed some bugs revealed by Paul Dietz' test suite:
649 ** FORMAT strings with both the ~<~:;~> form of the justification
650 directive and pretty-printing directives cause an error.
652 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
653 * incompatible change: the internal functions
654 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
655 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
656 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
657 instead of the old functions.
658 * new feature: on platforms where "dladdr" is available foreign
659 function names now appear in backtraces. (based on Helmut Eller's
661 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
662 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
664 * documentation: ASDF manual is now distributed with SBCL.
665 * bug fix: SBCL can now load its contributed modules with REQUIRE
666 even if the system-provided entries have been removed from
667 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
669 * bug fix: backtraces involving undefined functions or assembly
670 routines are more informative. (thanks to Brian Downing)
671 * bug fix: mutually referent alien structures now work correctly.
672 (reported by Rick Taube)
673 * bug fix: structures defined by WITH-ALIEN can be referred to
674 within other definitions in the same WITH-ALIEN.
675 * bug fix: division operators (MOD, TRUNCATE and the like) with
676 constant zero divisors and integer dividends no longer generate
678 * bug fix: provide default methods for INPUT-STREAM-P and
679 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
680 * bug fix: improve the detection and resolution of MOP metacycles
681 (where effective method computation depends on the generic function
682 having its effective method computed). (reported by Bruno Haible)
683 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
684 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
685 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
686 * bug fix: pathnames with relative directory components can be
687 represented relative to default pathnames.
688 * optimization: in taking the GCD of bignums, reduce the two bignums
689 to approximately the same size (using Euclid's algorithm) before
690 applying the more sophisticated binary GCD. (thanks to Juho
692 * optimization: COUNT on bitvectors now operates word-at-a-time.
693 * optimization: ASH with a positive, but not necessarily constant,
694 (leftwards) shift, when in a modular context, is compiled to a
696 * fixed some bugs revealed by Paul Dietz' test suite:
697 ** FORMAT variable parameters ("~V<char>") are defaulted properly
698 if the corresponding argument is NIL.
699 ** FORMAT directives accepting a minpad parameter treat negative
701 ** literal commas as character parameters to FORMAT directives are
703 ** literal spaces directly after ~<Newline> directives within a
704 format-logical-block (~:< ~@:>) do not induce :FILL-style
705 conditional newlines.
706 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
707 as specified: it no longer includes conditional newlines).
708 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
710 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
711 :PER-LINE-PREFIX argument does not evaluate to a string.
713 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
714 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
715 the manual for details; for now, package locks can be disabled by
716 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
717 if no major problems are found then it is likely that they will be
718 compiled in unconditionally.
719 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
720 unsupported operators on all platforms. To load a shared library
721 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
722 object file, link it into a shared library outside of SBCL and
723 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
725 * bug workaround: ROOM T and the GC were not working together
726 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
727 reporting facility aren't preserved. That reporting has been
728 disabled, so now until and unless someone figures out how to make it
729 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
730 * minor incompatible change: as threatened around sbcl-0.8.0, the
731 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
732 The symbols are also exported from SB-PCL for backwards
733 compatibility, but more so than before SB-PCL should be treated as
734 an implementation-internal package.
735 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
737 * the SB-SPROF contrib now works on (most) non-x86 architectures.
738 It is known as of this release not to work on the Alpha, however.
739 * fixed bug #167: errors signalled due to illegal syntax in method
740 bodies are now more legible.
741 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
742 designators and can hence be used with TYPEP.
743 * fixed bug #333: CHECK-TYPE now ensures that the type error
744 signalled, if any, has the right object to be accessed by
745 TYPE-ERROR-DATUM. (reported by Tony Martinez)
746 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
747 5.1.2.3 for multiple-value place subforms. (reported by Kalle
749 * fixed bug #334: programmatic addition of slots using specialized
750 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
751 and :ALLOCATION :CLASS slots.
752 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
753 float exponents. (rereported by Peter Seibel)
754 * fixed a bug: #\Space (and other whitespace characters) are no
755 longer considered to be macro characters in standard syntax by
757 * fixed bug: initialization of condition class metaobjects no longer
758 causes an instance of the condition to be created. (reported by
760 * fixed bug: it is now possible to have slots such that
761 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
762 description is neither :INSTANCE nor :CLASS.
763 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
764 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
765 SLOT-VALUE-USING-CLASS), and no longer causes errors with
766 non-standard SLOT-DEFINITION-ALLOCATION values.
767 * fixed bugs: various race conditions handling exiting threads.
768 CL-PPCRE's thread torture test now passes.
769 * fixed arguably-a-bug: GC time proportional to number of threads in
770 system even when most of them are idle
771 * optimization: improved performance of BIT and SBIT on bit-vectors.
772 * .fasl file incompatibility: The fasl file version number has
773 been incremented because of changes associated with package locks.
775 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
776 * minor incompatible change: the system no longer provides
777 optimization hints (conditions which are TYPEP
778 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
779 mode; these hints are still emitted when higher SPEED optimization
781 * new contrib module: a sampling profiler (profiling by statistical
782 sampling, rather than by instrumenting functions) is available as
783 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
784 * the behaviour of the standard function ED is now customizeable by
785 third parties through a hook variable: see ED's documentation
786 string for information on the protocol.
787 * the compiler no longer emits efficiency notes for (FUNCALL X)
788 when the type of X is uncertain under default optimization
790 * fixed bug 276: mutating a binding of a specialized parameter to a
791 method to something that is not TYPEP the specializer is now
793 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
794 exhibit the required behaviour on the x86 platform. (thanks to
795 Peter van Eynde, Eric Marsden and Bruno Haible)
796 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
797 even for difficult arguments. (reported by Peter Graves)
798 * fixed bug 141a: the backquote printer now descends quoted
800 * fixed another bug in backquote printing: no more destructive
801 modification of the form's list structure. (reported by Brian
803 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
804 move between its address being taken and the call to
805 interrupt_thread, fixing a crashing race condition.
806 * the SB-POSIX contrib implementation has been adjusted so that it
807 no longer exhibits ridiculously poor performance when constructing
808 instances corresponding to C structs.
810 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
811 * minor incompatible change: the sb-grovel contrib now treats C
812 structures as alien (in the sense of SB-ALIEN) objects rather than
813 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
814 has implications for memory management of client code
815 (sb-grovel-returned objects must now be manually managed) and for
816 type safety (alien objects now have full types).
817 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
818 used to control emission of compiler diagnostics, rather than the
819 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
820 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
821 quality should be considered deprecated.
822 * install.sh now installs the user manual as well
823 * (not quite a new documentable feature, but worth considering in
824 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
825 beginnings of a semantically meaningful condition hierarchy is
826 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
828 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
829 designator as the defaults argument.
830 * fixed bug: Displaced arrays whose displaced-to array has become
831 too small now cause ARRAY-DIMENSION to signal an error, providing
832 for safer bounds-checking. (reported by Bruno Haible)
833 * fixed bug: DEFCLASS slot definitions with identical :READER and
834 :WRITER names now signal a reasonable error. (reported by Thomas
836 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
838 * fixed bug: Invalid slot specification errors now print correctly.
839 (thanks to Zach Beane)
840 * fixed bug 320: Shared to local slot value transfers in class
841 redefinitions now happen corrently from superclasses as
842 well. (reported by Bruno Haible)
843 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
845 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
846 improper lists correctly. (reported by Bruno Haible)
847 * fixed bug 313: source-transform for <fun-name> was erroneously
848 applied to a call of a value of a variable with name <fun-name>.
849 (reported by Antonio Menezes Leitao)
850 * fixed bug 307: The obsolete instance protocol ensures that
851 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
852 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
853 remove dead unknown-values globs from the middle of the stack.
854 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
855 easier to resume long computations after using *BREAK-ON-SIGNALS*
856 to diagnose and fix failures. (thanks to Nikodemus Siivola)
857 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
858 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
859 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
860 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
861 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
863 * fixed a (fairly theoretical) bug in string printing: if
864 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
865 does not have array-element-type equal to the most general string
867 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
868 function name. (thanks to Nikodemus Siivola)
869 * fixed bug: compile-time format string checker failed on
870 non-closed ~{. (reported by Thomas F Burdick)
871 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
872 don't warn on legal loop code involving more than one aggregate
873 boolean. (thanks to Nikodemus Siivola)
874 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
875 in keyword position to MAKE-PACKAGE was not being reported as
877 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
878 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
880 * fixed bugs 280 and 312: the checking for multiple definitions in a
881 file is less likely to become confused by uses of inline
883 * fixed bug: the #S reader macro performs the keyword coercion
884 specified for slot names. (reported by Kalle Niemitalo)
885 * fixed bug: lambda lists may contain symbols whose names start with
886 & but are not lambda-list-keywords; their occurrence triggers a
888 * fixed bug 321: define-method-combination argument lambda lists do
889 not cause invalid code to be generated when &optional or &aux
890 variables have default values. (reported by Bruno Haible)
891 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
892 have CLOS classes; this appears also to have fixed reported
893 instability in stack exhaustion detection.
894 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
895 MUFFLE-WARNING when no associated restart is present is now
897 * optimization: rearranged the expansion of various defining macros
898 so that each expands into only one top-level form in a
899 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
901 * optimization: used a previously little-used slot in symbols to
902 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
904 * fixed some bugs revealed by Paul Dietz' test suite:
905 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
907 ** Fixed an optimization bug related to inheritance of initforms
908 from local to shared slots.
909 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
910 ** CAST optimizer forgot to flush argument derived type.
911 ** print/read consistency on floats is now orders of magnitude
912 more likely. (thanks also to Bruno Haible for a similar report
914 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
915 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
916 for objects of type REAL. Make it so.
917 ** SXHASH is specified (infelicitously) to respect similarity,
918 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
919 it so. (thanks to Markus Ziegler)
920 ** on the Alpha, the compiler succeeds in compiling functions
921 returning a known number of arguments greater than 63.
922 ** fixed handling of invalid NIL arguments in keyword position
924 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
926 ** provided a readably-printable representation for RANDOM-STATE
928 ** ensured that pathnames, where they have a namestring, always
929 print using #P"..." syntax.
931 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
932 * Support for the forthcoming 2.0 version of the NetBSD kernel
933 running on x86 hardware has been added. (thanks to Perry
934 E. Metzger most immediately, and others for their past work)
935 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
936 current release version; SBCL's OpenBSD support had been broken
937 since about the time of OpenBSD's switch to ELF binary format.)
938 * [placeholder for DX summary]
939 ** user code with &REST lists declared dynamic-extent, under high
940 speed or space and low safety and debug optimization policy.
941 * The manual has been converted to Texinfo format and the debugger
942 chapter from the cmucl manual has been added.
943 * A facility has been added to extract documentation strings from
944 sbcl and store them as Texinfo-formatted snippets for inclusion in
945 the manual (via Texinfo's @include directive)
946 * bug fix: compiler emitted division in optimized DEREF. (thanks for
947 the test case to Dave Roberts)
948 * bug fix: multidimensional simple arrays loaded from FASLs had fill
949 pointers. (reported by Sean Ross)
950 * bug fix: PROFILE output is printed nicely even for large numerical
951 values. (thanks to Zach Beane)
952 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
953 greater than 32 handle EOF correctly.
954 * bug fix: on X86 an immediate argument of the IMUL instruction is
955 correctly printed in disassembly. (thanks to Lutz Euler)
956 * bug fix: class slots in redefined classes preserve their old
957 values. (thanks to Bruno Haible and Nikodemus Siivola)
958 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
959 succeeds. (reported by Marco Baringer)
960 * fixed some bugs revealed by Paul Dietz' test suite:
961 ** READ-SEQUENCE now works on ECHO-STREAMs.
962 ** RATIONALIZE works more according to its specification. (thanks
965 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
966 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
967 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
968 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
969 should matter to you only if you rebind the printer control
970 variables and then find you want different bindings in the
971 debugger than in the ordinary execution of your program.)
972 * The runtime build system has been tweaked to support building
973 (on SPARC/SunOS) using a C compiler which invokes Sun's own
974 assembler and linker. (thanks to Nikodemus Siivola)
975 * Unbound, undefined, undeclared variables now trigger full
976 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
977 more often programmer error than deliberate exploitation of undefined
979 * optimization: the hash algorithm for strings has changed to one
980 that is less vulnerable to spurious collisions. (thanks to Juho
982 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
983 needless bounds checking. (thanks to Juho Snellman)
984 * optimization: implemented multiplication as a modular
985 (UNSIGNED-BYTE 32) operation on the PPC backend.
986 * fixed some bugs revealed by Paul Dietz' test suite:
987 ** ADJUST-ARRAY now copies the datum in a zero rank array if
989 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
990 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
991 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
993 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
995 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
997 ** ECHO-STREAMs no longer attempt to echo the end of file value to
998 their output stream on EOF from read.
999 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1000 have been read to end-of-file.
1001 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1003 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1004 description of determination of which consecutive characters
1006 ** Printing the "Space" character with escaping on now yields "#\\ ",
1007 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1008 ** Reading floating-point numbers with *READ-BASE* set to a number
1009 less than 10 works correctly.
1010 ** Reading floating-point numbers with *READ-BASE* set to a number
1011 more than 10 works correctly.
1012 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1013 the readtable currently in effect.
1015 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1016 * minor incompatible change: parsing of namestrings on a physical
1017 (Unix) host has changed; numbers after the final #\. in a
1018 namestring are no longer interpreted as a version field. This is
1019 intented to be largely invisible to the user, except that the
1020 meaning of the namestring "*.*.*" has changed: it now refers to a
1021 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1022 should usually be replaced by
1023 (make-pathname :name :wild :type :wild :version :wild)
1024 with the added benefit that this is more likely to be portable.
1025 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1026 option to OPEN now signals an error if the file being opened
1027 exists; this may have an impact on existing code.
1028 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1029 signals received, on the PowerPC platforms (both Linux and
1030 Darwin). (thanks to Pierre Mai for pointing out the location of
1032 * several fixes on OS X: The system now builds and runs cleanly on
1033 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1034 (thanks to Brian Mastenbrook)
1035 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1036 second-resolution: integer multiples of 1/3600 between -24 and 24.
1037 (thanks to Vincent Arkesteijn)
1038 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1039 type when called with 1 argument; PEEK-CHAR checked type of
1040 PEEK-TYPE only after having read first character from a
1041 stream. (reported by Peter Graves)
1042 * bug fix: the garbage collector now has much better locality
1043 behaviour, and in particular no longer treats all memory as being
1044 exhausted when many small objects point to each other in a deeply
1046 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1047 recognized as being TYPEP their class.
1048 * bug fix: the PUSHNEW documentation string has been corrected.
1049 (thanks to Vincent Arkesteijn)
1050 * bug fix: defaulting of the value for the last of an atypically
1051 large number of multiple values being bound was not being
1052 performed correctly on the Alpha or PPC platforms
1053 * optimization: implemented multiplication as a modular
1054 (UNSIGNED-BYTE 32) operation on the x86 backend.
1055 * optimization: SEARCH on simple-base-strings can now be open-coded.
1056 (see also contrib/compiler-extras.lisp for inspiration for
1057 teaching the compiler about the Boyer-Moore algorithm).
1058 * value, returned by MAX (and MIN) called with several EQUALP, but
1059 not EQL, arguments now does not depend on compiler settings.
1060 * fixed some bugs revealed by Paul Dietz' test suite:
1061 ** in stack analysis liveness information is propagated from
1062 non-local entry points.
1063 ** pathwise CAST removing failed when the CAST node did not start
1065 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1066 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1068 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1069 host is already defined.
1070 ** RENAME-FILE works on streams instead of signalling an internal
1072 ** PEEK-CHAR uses the current readtable when determining whether
1073 or not a character is whitespace.
1074 ** MERGE-PATHNAMES handles the case when the pathname does not
1075 specify a name while the default-pathname specifies a version.
1076 ** Pathnames now stand a better chance of respecting print/read
1078 ** Attempting to use standardized file system operators with a
1079 pathname with invalid :DIRECTORY components signals a
1081 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1083 ** DIRECTORY on logical pathnames is more correct.
1084 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1085 signal a TYPE-ERROR if their argument is not a stream.
1086 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1087 designator argument does not designate a stream.
1088 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1089 examining the synonym.
1090 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1092 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1093 element-type larger than ([UN]SIGNED-BYTE 32).
1095 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1096 * When built with the :SB-FUTEX feature, threaded builds now take
1097 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1098 for faster/more reliable mutex and condition variable support.
1099 * Incompatible change (but one you probably shouldn't have been using
1100 anyway): the interface and code for arbitrating between multiple
1101 threads in the same user session has been redesigned.
1102 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1103 error when called without an explicit environment argument.
1104 (thanks to Nikodemus Siivola)
1105 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1106 confused about their position. (thanks to Adam Warner and Gerd
1108 * bug fix: Pretty printing backquoted forms with unquotations in the
1109 argument list position of various code constructs such as LAMBDA
1110 now works correctly. (reported by Paul Dietz)
1111 * bug fix: Pretty printing unquotations no longer loses all
1112 stream position information.
1113 * optimization: performance of string output streams is now less
1114 poor for multiple small sequence writes.
1115 * optimization: performance of CSUBTYPEP in the presence of complex
1116 expressions involving CONS and NOT many times has been improved.
1117 (reported by Paul Dietz)
1118 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1120 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1121 streams for the REPL, for the debugger, and so on. (thanks to
1123 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1124 * fixed some bugs revealed by Paul Dietz' test suite:
1125 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1126 is ignored if it is too small, rather than propagating through
1127 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1128 ** extremely complex negations of CONS types were not being
1129 sufficiently canonicalized, leading to inconsistencies in
1131 ** VALUES tranformer lost derived type.
1133 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1134 * fixed a bootstrapping bug: the build process no longer assumes
1135 that the various BOOLE-related constants have the same value in
1136 host and target lisps. (noted by Paul Dietz' test suite on an
1137 SBCL binary built from CLISP)
1138 * The system can now be dynamically linked on the MIPS platform,
1139 which enables dynamic loading of foreign code from Lisp. (thanks
1140 to Ralf Baechle for discussions on the MIPS ABI)
1141 * The system now records debugging information for its own source
1142 files in a filesystem-position-independent manner, relative to
1143 the "SYS" logical host.
1144 * fixed a compiler bug: MV-LET convertion did not check references
1145 to the "max args" entry point. (reported by Brian Downing)
1146 * tweaked disassembly notes to be less confident about proclaiming
1147 some instruction as an LRA. (thanks to Brian Downing)
1148 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1149 now each have their own history, command character, and other
1150 characteristics. (thanks to David Lichteblau)
1151 * fixed some bugs revealed by Paul Dietz' test suite:
1152 ** compiler failure in compiling LOGAND expressions including a
1154 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1155 shift greater than 32.
1156 ** FLUSH-DEST did not mark blocks for type check regeneration.
1157 ** HANDLER-CASE failed to accept declarations in handler clauses
1158 in some circumstances.
1160 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1161 * New code in contrib/sb-introspect (still probably not entirely
1162 stable yet) provides some support for smart Lisp development
1163 environments like SLIME.
1164 * The conditions signalled for errors occurring when loading .fasl
1165 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1166 in a way which should help ASDF recover gracefully.
1167 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1168 stuff has been cleaned up. If you code contrib/ stuff, this might
1169 affect you, and you can look at contrib/README, contrib/STANDARDS,
1170 and/or the 0.8.4.27 diff to check.
1171 * In full calls the compiler now does not generate checks for declared
1172 argument types for all arguments.
1173 * various threading fixes
1174 ** and some experimental patches which didn't make it into
1175 the main tree for this release, but which are shipped in
1176 contrib/experimental-thread.patch as a possible fix for some
1177 failures (deadlock, spinning...) in GC-intensive multithreaded
1179 * fixed PPC build problem (source code incompatibility of different
1180 library versions): added offsetof() hackery which attempts to divine
1181 where glibc maintainers put uc_mcontext today
1182 * fixed bug 282: compiler does not trust type assertions while passing
1183 arguments to a full call.
1184 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1185 &OPTIONAL VALUES type parameter.
1186 * fix bug 214: algorithm for noting rejected templates is now more
1187 similar to that of template seletion. (also reported by rydis on
1189 * fixed bug 141b: printing backquoted information readably and prettily
1190 inserts a space where necessary.
1191 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1192 (CHARACTER 10) are now reported as errors, rather than propagated
1193 as unknown types. (reported by piso on #lisp)
1194 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1195 with values NIL and :ERROR. (thanks to Milan Zamazal)
1196 * fixed bug 191c: CLOS now does proper keyword argument checking as
1197 described in CLHS 7.6.5 and 7.6.5.1.
1198 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1199 longer raise an error; further, using a list as a for-as-arithmetic
1200 counter now raises a meaningful error.
1201 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1202 now understood by sequence creation functions such as MAKE-SEQUENCE
1204 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1205 a STRING-INPUT-STREAM.
1206 * compiler enhancement: SIGNUM is now better able to derive the type
1208 * type declarations inside WITH-SLOTS are checked. (reported by
1210 * fixed some bugs revealed by Paul Dietz' test suite:
1211 ** incorrect optimization of TRUNCATE for a positive first
1212 argument and negative second.
1213 ** compiler failure in let-convertion during flushing dead code.
1214 ** compiler failure while deriving type of TRUNCATE on an
1215 interval, containing 0.
1216 ** ASH of a negative bignum by a negative bignum count now returns
1218 ** intersection of CONS types now canonicalizes properly, fixing
1219 inconsistencies in SUBTYPEP.
1221 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1222 * incompatible change: The --disable-debugger command line
1223 option now clobbers the debugger at a more fundamental
1224 level, by redefining #'INVOKE-DEBUGGER instead of by
1225 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1226 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1227 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1228 enter the debugger REPL and then suffer recursive errors
1229 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1230 while under the new system, BREAK in --disable-debugger mode
1231 terminates the system just as an unhandled error would.
1232 * fixed compiler performance when processing loops with a step >1;
1233 * bug fix: DOCUMENTATION now retrieves generic function
1234 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1235 support has been systematized, and now supports the methods
1236 specified by ANSI, along with a default method and a method for
1237 slot documentation. (reported by Nathan Froyd)
1238 * bug fix: effective methods associated with a generic function are
1239 no longer cached over a change of that generic function's method
1240 combination. (reported by Andreas Fuchs)
1241 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1242 names a non-existent directory. (thanks to Andreas Fuchs)
1243 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1244 the PPC platform, lead to stack corruption; this has been fixed.
1245 (reported by Rainer Joswig)
1246 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1247 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1248 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1249 arithmetic in general, is now much more reliable on the PPC
1251 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1252 platform now returns the right answer.
1253 * optimization: restored some effective method precomputation in
1254 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1255 precomputation is now tunable.
1256 * optimization: compiler-internal data structure use has been
1257 reviewed, and changes have been made that should improve the
1258 performance of the compiler by about 20%.
1259 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1260 simple-base-strings and simple-bit-vectors is improved.
1261 * optimization: the optimization of 32-bit logical and arithmetic
1262 functions introduced in version 0.8.3 on the x86 has been
1263 implemented on the mips, ppc and sparc platforms; an
1264 implementation of the same facility, but for 64-bit arithmetic,
1265 has been added for the alpha.
1266 * microoptimization: the compiler is better able to make use of the
1267 x86 LEA instruction for multiplication by constants.
1268 * bug fix: in some situations compiler did not report usage of
1269 generic arithmetic in (SPEED 3) policy.
1270 * bug 145b fix: compiler used wrong type specifier while converting
1271 MEMBER-types to numeric.
1272 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1274 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1275 index argument now works properly on the Alpha platform.
1276 * bug fix: floating point exception treatment on the Alpha platform
1278 * bug fix: FILE-POSITION works much better on string input and
1279 output streams. (thanks to Nikodemus Siivola)
1280 * bug fix: many threading/garbage collection symptoms sorted.
1281 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1282 might be pseudo-atomic.
1283 * internal change: Stopping for GC is now done with signals not
1284 ptrace. GC is now done in whichever thread wanted it, instead of
1286 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1288 * bug fix: result form in DO is not contained in the implicit
1290 * incompatible change: ICR structure is changed; the value part of
1291 CONTINUATION is now called LVAR; corresponding functions are
1292 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1293 * added type deriver for ISQRT (thanks to Robert E. Brown).
1294 * bug fix: better support for loading from the command line when an
1295 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1297 * fixed some bugs revealed by Paul Dietz' test suite:
1298 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1299 ** ROUND and FROUND now give the right answer when given very
1300 small float arguments.
1301 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1303 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1304 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1305 ** LCM should return a non-negative integer.
1306 ** PARSE-INTEGER returned the index of a terminator instead of the
1307 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1308 ** PARSE-INTEGER returned an incorrect index being applied to a
1310 ** LCM with two arguments of 0 returns 0 rather than signalling
1312 ** unsigned addition of a 32-bit constant with the high bit set no
1313 longer causes an internal compiler error on the x86.
1314 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1315 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1316 with negative last argument.
1317 ** byte specifiers with zero size and position no longer cause
1318 an error during type derivation.
1319 ** bignum multiplication on the Alpha platform now returns the
1321 * porting: The system now builds on SuSE AMD64, although it still
1322 generates a 32-bit binary.
1323 * .fasl file incompatibility: The fasl file version number has
1324 been incremented (because of the changes to internal compiler
1325 data structures referred to above).
1327 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1328 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1329 more accurately, on the Darwin kernel running on PowerPC hardware.
1330 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1331 * Compiler code deletion notes now signal a condition of type
1332 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1333 an associated MUFFLE-WARNING restart.
1334 * The compiler now performs limited argument count validation of
1335 constant format strings in FORMAT, and where appropriate in ERROR,
1336 CERROR and WARN. (thanks to Gerd Moellmann)
1337 * New ASDF-INSTALL contrib can be used for automatic download and
1338 installation of third-party Lisp code from CCLAN or other sites
1340 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1341 which forces another thread to execute a function supplied by the
1343 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1344 now accept and act upon their :ELEMENT-TYPE keyword argument.
1345 (reported by Martin Atzmueller, Edi Weitz)
1346 * bug fix: FILE-POSITION now accepts position designators up to
1347 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1348 is the greater. (thanks to Patrik Nordebo)
1349 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1350 Kalle Olavi Niemitalo)
1351 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1352 printable. (reported by Eric Marsden)
1353 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1354 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1355 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1356 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1357 (thanks to Stig Sandoe)
1358 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1359 UNREAD-CHAR now correctly includes the unread character in the
1360 target sequence. (thanks to Gerd Moellmann)
1361 * bug fix in threaded builds: the system can now be suspended and
1362 resumed by shell job control with minimal disruption.
1363 * bug fixes in times and timezones >2038AD
1364 * better handling of "where is GNU make?" problem in build scripts
1365 (thanks to Nikodemus Siivola)
1366 * new optimization: inside a named function any reference to a
1367 function with the same name is considered to be a self-reference;
1368 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1369 optimization quality.
1370 * new optimization on x86: logical functions and + now have
1371 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1372 used when the result is truncated to 32 bits.
1373 * VALUES declaration is partially enabled.
1374 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1375 * fixed some bugs revealed by Paul Dietz' test suite:
1376 ** The system now obeys the constraint imposed by
1377 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1378 types form a lattice under type intersection.
1379 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1380 ** ASSOC now ignores NIL elements in an alist.
1381 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1382 and (1+ MOST-POSITIVE-FIXNUM) answers.
1383 ** The addition of a method with invalid qualifiers to a generic
1384 function does not cause an error to be signalled immediately;
1385 a warning is signalled, and the error is generated only on
1386 calling the generic function.
1387 * changed .fasl file version number, in part to add type codes for
1388 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1389 obscure ANSI requirements
1391 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1392 * fixed bug 148: failure to inline-expand a local function left
1393 garbage, confusing the compiler.
1394 * fixed bugs 3cd: structure slot readers perform type check if the
1395 slot can have an invalid value (i.e. it is either not initialized
1396 or can be written with a less specific slot writer).
1397 * bug fix: the compiler now traps array references to elements off
1398 the end of an array; previously, the bounds checking in some
1399 circumstances could go off-by-one.
1400 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1402 * type declarations for array element types now obey the description
1403 on the CLHS page "Declaration TYPE", as per discussions on
1404 sbcl-help around 2003-05-08. This means that a declaration
1405 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1406 declaration, all references to BAR will be asserted or assumed
1407 (with THE, so dependent on compiler policy) to involve objects of
1408 type FOO. Note that no such declaration is implied in
1409 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1410 * declared types of functions from the "Conditions"
1411 chapter. (reported by Paul Dietz)
1412 * bug fix: CERROR accepts a function as its first argument.
1413 * bug fix: NTH an NTHCDR accept a bignum as index
1414 arguments. (reported by Adam Warner)
1415 * optimization: character compare routines now optimize comparing
1416 against a constant character. (reported by Gilbert Baumann)
1417 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1418 argument now works properly on the MIPS platform.
1419 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1420 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1421 declared to have a functional type, does not cause an error
1423 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1424 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1425 * fixed bug: if last continuation of a deleted block has a
1426 destination, this destination should be deleted too. (reported by
1428 * fixed a bug in the bootstrap process: the host compiler's values
1429 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1430 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1431 test case from Patrik Nordebo)
1432 * improved the ability of the disassembler on the PPC platform to
1433 provide helpful disassembly notes.
1434 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1435 the class in more cases than previously.
1436 * bug fix: FILE-POSITION now understands :START and :END for
1437 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1438 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1439 8) data. (thanks to David Lichteblau)
1440 * bug fix: it is possible to add a method to a generic function
1441 without lambda list.
1442 * bug fix: reader failed to signal END-OF-FILE inside an
1443 object representation. (reported by Nikodemus Siivola)
1444 * fixed some bugs revealed by Paul Dietz' test suite:
1445 ** LAST and [N]BUTLAST should accept a bignum.
1446 ** condition slot accessors are methods.
1447 ** (VECTOR NIL) is a subtype of STRING.
1449 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1450 * minor incompatible change: some nonsensical specialized lambda
1451 lists (used in DEFMETHOD) which were previously ignored now signal
1453 * minor incompatible change: the system is now aware of the types of
1454 variables in the COMMON-LISP package, and will signal errors for
1455 most violations of these type constraints (where previously they
1456 were silently accepted).
1457 * minor incompatible change: COMPILE-FILE now uses the freedom
1458 afforded (ANSI 3.2.2.3) to use derived function types for
1459 functions defined in the same file. This also permits the system
1460 to warn on static type mismatches and function
1461 redefinition. (Currently it does not work with high DEBUG level.)
1462 * minor incompatible change: VALUES declaration is disabled.
1463 * When issuing notes, the compiler now signals a condition of type
1464 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1465 restart for use in user handlers. It is expected that the
1466 COMPILER-NOTE condition will eventually become a condition
1467 supertype to a hierarchy of note types, which will then be
1468 handleable in a similar fashion. However, at the moment, no such
1469 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1470 but it's an implementation detail, not a classification for the
1472 * Changes in type checking closed the following bugs:
1473 ** type checking of unused values (192b, 194d, 203);
1474 ** template selection based on unsafe type assertions (192c, 236);
1475 ** type checking in branches (194bc).
1476 * A short form of VALUES type specifier has ANSI meaning (it has
1477 increased the number of situations when SBCL cannot perform type
1479 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1480 keywords or constants is permissible.
1481 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1482 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1483 classes. (thanks to Antonio Martinez)
1484 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1485 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1486 argument to a pretty-print stream. Instead, it leaves any such
1487 operations to DESCRIBE-OBJECT methods.
1488 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1490 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1491 lambda lists. (thanks to David Lichteblau)
1492 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1493 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1494 (thanks to Gerd Moellmann)
1495 * fixed bug 246: increased compilation speed of long
1496 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1498 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1500 * DEFCONSTANT now throws a condition of type
1501 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1502 constant to a non-EQL value; CONTINUE and ABORT restarts
1503 respectively change and preserve the value.
1504 * fixed bug 63: The code walker, part of the implementation of CLOS,
1505 is now better at handling symbol macros.
1506 * bug fix: There is no longer an internal implementation type named
1507 CL:LENGTH. (reported by Raymond Toy)
1508 * bug fix: In macro-like defining macros/special operators the
1509 implicit block does not enclose the lambda list.
1510 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1511 enclosed in parentheses) are not suitable as type specifiers, and
1512 their use properly signals an error now.
1513 * bug fix: An argument count mismatch for a type specifier in code
1514 being compiled no longer causes an unhandled error at compile
1515 time, but signals a compile-time warning.
1516 * fixed simple vector readable printing
1517 * bug fix: DESCRIBE takes more care over whether the class
1518 precedence list slot of a class is bound before accessing it.
1519 (reported by Markus Krummenacker)
1520 * bug fix: FORMATTER can successfully compile pretty-printer format
1521 strings which use variants of the ~* directive inside.
1522 * bug fix: SEARCH now applies its TEST predicate to the elements of
1523 the arguments in the correct order. (thanks to Wolfhard Buss)
1524 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1525 in type inference. (thanks to Robert E. Brown)
1526 * bug fix: Reading in symbols with an explicit package name of ""
1527 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1528 * fixed some bugs revealed by Paul Dietz' test suite:
1529 ** NIL is now allowed as a structure slot name.
1530 ** Arbitrary numbers, not just REALs, are allowed in certain
1531 circumstances in LOOP for-as-arithmetic clauses.
1532 ** Multiple class redefinitions before slot access no longer
1533 causes a type error.
1534 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1535 association between the name and a class.
1536 ** Generic functions with non-standard method-combination and over
1537 five methods all of which return constants no longer return NIL
1538 after the first few invocations. (thanks to Gerd Moellmann)
1539 ** CALL-NEXT-METHOD with no arguments now passes the original
1540 values of the arguments, even in the presence of assignment.
1541 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1543 ** DISASSEMBLE works with closures and funcallable instances.
1544 ** ADD-METHOD now returns the generic function, not the new method.
1545 ** FIND-METHOD signals an error if the lengths of the specializers
1546 is incompatible with the generic function, even if the ERRORP
1548 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1549 which its argument is a member.
1550 ** DEFCLASS only redefines the class named by its class-name
1551 argument if that name is the proper name of the class;
1552 otherwise, it creates a new class.
1553 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1554 of the UNBOUND-SLOT condition to the name of the slot.
1555 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1556 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1557 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1558 SLOT-MAKUNBOUND in the specified fashion.
1560 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1561 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1562 cross-compilation host. As a consequence, we can now bootstrap our
1563 way up to SBCL starting with a bare gcc toolchain and human-readable
1564 source code (first the source to CLISP, then the source to SBCL).
1565 * A contributed module containing a partial implementation of the
1566 simple-streams interface has been included. (thanks to Rudi
1568 * A contributed module implementing the RFC1321 Message Digest
1569 Algorithm, known as MD5, has been included.
1570 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1571 no longer has any effect, as the code controlled by this feature
1572 has been deleted. (As far as we know, no-one has ever built using
1573 this feature, and its semantics were confused in any case).
1574 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1575 work on conditions (as required by the ANSI specification),
1576 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1577 expected behaviour on conditions. Users should note, however,
1578 that such behaviour is not required by the ANSI specification,
1579 and so use of this behaviour may render their code unportable.
1580 * fixed some bugs revealed by Paul Dietz' test suite:
1581 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1583 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1584 ** SXHASH on condition objects no longer returns NIL.
1585 ** :ALLOCATION :CLASS slots are better treated; their values are
1586 updated on class redefinition, and initforms inherited from
1587 superclasses are applied.
1588 ** REMOVE-METHOD returns its generic function argument even when
1589 no method was removed.
1590 ** SHARED-INITIALIZE now initializes the values of the requested
1591 slots, including those with :ALLOCATION :CLASS.
1592 ** ALLOCATE-INSTANCE now works on structure classes defined via
1593 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1595 ** SLOT-EXISTS-P now works on conditions, as well as structures
1597 ** MAKE-LOAD-FORM now has the required methods on
1598 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1599 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1600 keyword, and now implements the SLOT-NAMES argument.
1601 ** methods with &OPTIONAL arguments no longer allow too many
1602 arguments to be passed in the call without error.
1603 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1604 option is consistent with the required arguments of the generic
1605 function lambda list.
1606 * bug fix: REQUIRE accepts a string designator. (Thanks to
1608 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1609 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1610 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1611 * bug fix: The compiler now checks for duplicated variables in macro
1613 * bug fix: SETQ on globals returns the correct value.
1614 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1615 not a COMPILER-ERROR (followed by some other strange error on
1616 choosing the CONTINUE restart).
1617 * bug fix: make.sh and friends are now more consistent in the way that
1618 they look for GNU "make".
1620 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1621 * experimental native threads support (on x86 Linux >=2.4 only).
1622 This is not compiled in by default: you need to add :SB-THREAD to
1623 the target features. See the "Beyond ANSI" chapter of the manual
1625 * fix for longstanding nonANSIism: The old distinction between
1626 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1627 The return value from CL:FIND-CLASS is now a CLOS class, and
1628 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1629 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1630 * An interface to the MetaObject Protocol, as described in Kiczales,
1631 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1632 MIT Press, 1991, is available from the SB-MOP package.
1633 * incompatible change: the SB-PCL package should now be considered
1634 a private implementation detail, and no longer a semi-private MOP
1636 * minor incompatible change: due to rearrangement for threads, the
1637 control stack and binding stack are now allocated at arbitrary
1638 addresses instead of being hardcoded per-port. Users affected by
1639 this probably have to be doing advanced things with shared
1640 libraries, and will know who they are.
1641 * minor incompatible change: Previously, all --eval forms used were
1642 processed with READ before any of them were processed with EVAL.
1643 Now each --eval form is processed with both READ and EVAL before
1644 the next --eval form is processed. (Thus package operations like
1645 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1646 work as the user might reasonably expect.)
1647 * minor incompatible change: *STANDARD-INPUT* is now only an
1648 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1650 * minor incompatible change: Y-OR-N-P is now character-oriented,
1651 not line oriented. Also, YES-OR-NO-P now works without errors.
1652 (thanks to Antonio Martinez)
1653 * sb-aclrepl module improvements: an integrated inspector, added
1654 repl features, and a bug fix to :trace command.
1655 * Known functions, which cannot be open coded by the backend, are
1656 considered to be able to check types of their arguments. (fixing
1657 a bug report by Nathan J. Froyd)
1658 * fixed a bug in computing method discriminating functions: It is
1659 now possible to define methods specialized on classes which have
1660 forward-referenced superclasses. (thanks to Gerd Moellmann)
1661 * fixed evaluation order in optional entries (reported by Gilbert
1663 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1664 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1665 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1666 required-by-AMOP NAME argument, as well as CLASS and
1667 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1668 * fixed bug 20: DEFMETHOD can define methods using names that are
1669 not the proper names of classes to designate class specializers.
1670 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1671 Unix file descriptors, instead of blowing up. (thanks to Antonio
1673 * Garbage collection refactoring: user-visible change is that a
1674 call to the GC function during WITHOUT-GCING will not do garbage
1675 collection until the end of the WITHOUT-GCING. If you were doing
1676 this you were probably losing anyway.
1677 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1678 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1679 TYPEP the latter but not the former.
1680 * The compiler issues a full WARNING for calls to undefined functions
1681 with names from the CL package.
1682 * MAP-INTO for a vector destination is open coded. (reported by
1683 Brian Downing on c.l.l)
1684 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1685 documentation string.
1686 * fixed some bugs revealed by Paul Dietz' test suite:
1687 ** COPY-ALIST now signals an error if its argument is a dotted
1689 ** Condition slots are now accessed more correctly in the presence
1690 of multiple initargs for a given slot.
1691 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1692 exclude restarts of the same name associated with a different
1694 ** DEFCLASS of forward-referenced classes with another
1695 forward-referenced class in the superclasses list no longer
1697 ** Condition slots are now initialized once each, not multiple
1698 times. (thanks to Gerd Moellmann)
1699 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1701 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1702 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1703 arguments contain duplicated elements.
1704 ** RESTART-CASE understands local macros.
1705 ** RESTART-CASE associates exactly its own restarts with a condition.
1706 ** ENDP in safe mode checks its argument to be of type LIST.
1707 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1708 in question is unbound.
1709 ** Optimized MAKE-INSTANCE functions no longer cause internal
1710 assertion failures in the presence of duplicate initargs.
1711 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1712 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1713 propagated even in the ctor optimized implementation of
1715 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1717 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1718 * a better implementation of SXHASH on (simple) bit vectors,
1719 measured both in execution speed and in distribution of results
1720 over the positive fixnums, has been installed. Likewise, a better
1721 implementation of EQUAL for simple bit vectors is now available.
1722 * fixed CEILING optimization for a divisor of form 2^k.
1723 * fixed bug 240 (emitting extra style warnings "using the lexical
1724 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1725 by Antonio Martinez)
1726 * fixed SXHASH, giving different results for NIL depending on type
1727 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1728 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1729 variable. (found by Rolf Wester)
1730 * SBCL does not ignore type declarations for special
1731 variables. (reported by rif on c.l.l 2003-03-05)
1732 * some bug fixes in contrib/sb-aclrepl/
1733 * fixed some bugs revealed by Paul Dietz' test suite:
1734 ** a bug in the CONS type specifier, whereby the CAR and CDR
1735 types got intertwined, has been fixed;
1736 ** the type system is now able to reason about the interaction
1737 between INTEGER and RATIO types more completely;
1738 ** APPEND, [N]REVERSE and NRECONC check that those their
1739 arguments, which must be proper lists, are really so;
1740 ** An array specialized to be unable to hold elements has been
1741 implemented, as required -- yes, really -- by ANSI;
1742 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1743 on malformed property lists;
1745 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1746 * incompatible packaging change: in line with Unix convention,
1747 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1748 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1749 * REQUIRE and PROVIDE are now optionally capable of doing something
1750 useful. See the documentation string for REQUIRE.
1751 * infrastructure for a managed SBCL contrib system: contributed
1752 modules in this release include:
1753 ** the ASDF system definition facility;
1754 ** an interface to the BSD Sockets API;
1755 ** an ACL-like convenience interface to the repl;
1756 (thanks to Kevin Rosenberg)
1757 ** an implementation of ROTATE-BYTE, with efficient implementation
1759 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1760 gives a closer approximation to the right answer than previously.
1761 (thanks to Raymond Toy)
1762 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1763 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1764 optional environment arguments, as required by ANSI.
1765 * fixed bugs in other functions taking environment objects, allowing
1766 calls with an explicit NIL environment argument to be compiled
1768 * fixed bug 228: primary return values from
1769 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1770 COMPILE or FUNCTION.
1771 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1772 :TYPE will no longer signal errors on innocuous objects.
1773 * fixed bug 231b: SETQ is better at respecting type declarations in
1774 the lexical environment.
1775 * fixed a bug in DEFCLASS: classes named by symbols with no or
1776 unprintable packages can now be defined.
1777 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1778 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1779 * fixed compiler failure related to checking types of functions.
1780 (reported by Robert E. Brown)
1781 * the compiler is now much more consistent in its error-checking
1782 treatment of bounding index arguments to sequence functions: in
1783 (SAFETY 3) code, errors will be signalled in almost all cases if
1784 invalid sequence bounding indices are passed to functions defined
1785 by ANSI to operate on sequences.
1786 * fixed a bug in the build procedure: documentation of SBCL-specific
1787 packages is now preserved and available in the final Lisp image.
1788 * lifted FDEFINITION lookup out of loops in the implementation of
1789 many list operations. (thanks to Robert E. Brown)
1790 * fixed a bug in the reader: the #n# reader macro now works for
1791 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1792 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1793 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1794 E. Brown and Tony Martinez respectively)
1795 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1796 SLOT-BOUNDP: in particular, it is now able to optimize them much
1797 better, and is now not vulnerable to having packages renamed.
1798 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1799 always called when a slot is not present in an instance. (thanks
1801 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1802 longer signal an error on unreading a character following EOF on
1803 the previous constituent stream. (thanks to Tony Martinez)
1804 * fixed some bugs revealed by Paul Dietz' test suite:
1805 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1806 not just nonnegative fixnums;
1807 ** the logical bit-array operators such as BIT-AND now accept an
1808 explicit NIL for their "opt-arg" argument (to indicate a
1809 freshly-consed result bit-array);
1810 ** ELT now signals an error on an invalid sequence index in safe
1812 ** the type system is now cleverer about negations of numeric
1813 types, and consequently understands the BIGNUM and RATIO types
1815 ** the type system is now cleverer about the interaction between
1816 INTEGER and RATIO types: while bugs still remain, many more
1817 cases are accurately computed;
1818 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1819 if it is in the last clause;
1820 ** CONSTANTLY now correctly returns a side-effect-free function in
1822 ** DECLARE is no longer treated as a special-operator; in
1823 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1824 * incremented fasl file version number due to the change in the
1825 DEFSTRUCT-SLOT-DESCRIPTION structure.
1827 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1828 * minor incompatible change: code processed by the "interpreter" or
1829 EVAL now has a compilation optimization policy of (DEBUG 2)
1830 (changed from (DEBUG 1)) to improve debuggability of interactive
1831 development, and to allow the use of the debug RETURN command in
1833 * an experimental implementation of the RETURN command for the
1834 debugger has been included. (thanks to Frederik Kuivinen)
1835 * fixed bug 62: constraints were not propagated into a loop.
1836 * fixed bug in embedded calls of SORT (reported and investigated by
1838 * fixed some bugs revealed by Paul Dietz' test suite:
1839 ** printing and reading of arrays with some dimensions having
1840 length 0 (thanks to Gerd Moellmann);
1841 ** BOA constructor with &AUX argument without a default value does
1842 not cause a type error;
1843 ** CONSTANTP now returns true for all self-evaluating objects.
1845 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1846 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1847 accessors that are related by inheritance, as specified in the
1848 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1849 Valtteri Vuorikoski)
1850 * The compiler is now able to inline functions that were defined in
1851 a complex lexical environment (e.g. inside a MACROLET).
1852 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1854 * Support for the upcoming FreeBSD-5.0 release has been included.
1855 (thanks to Dag-Erling Smorgrav)
1856 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1857 effect when it is not in a toplevel context.
1858 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1859 stand a better chance of being correct. (thanks to Gerd
1861 * fixed bug in COERCE, which now signals an error on coercing a
1862 rational to a bounded real type which excludes the expected
1864 * The compiler is now able to derive types more accurately from the
1865 COERCE and COMPILE functions.
1866 * fixed bug 223: functional binding is considered to be constant
1867 only for symbols in the CL package.
1868 * fixed bug 231: SETQ did not check the type of a variable being set
1869 (reported by Robert E. Brown)
1870 * A new optimization for MAKE-INSTANCE has been included, fixing
1871 various bugs (including relating to :ALLOCATION :CLASS slots and
1872 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1874 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1875 ** As required by ANSI, LOOP now disallows anonymous collection
1876 clauses such as COLLECT I in conjunction with aggregate boolean
1877 clauses such as THEREIS (= I 1);
1878 ** LOOP now signals an error when any variable is reused in the
1879 same loop (including the potentially useful construct analogous
1880 to WITH A = 1 WITH A = (1+ A);
1881 ** IT is only a special loop symbol within the first clause of a
1882 conditional loop clause;
1883 ** LOOP with a typed iteration variable over a hashtable now
1884 signals a type error iff it should.
1885 * fixed some other bugs revealed by Paul Dietz' test suite:
1886 ** FILE-STREAM now names the class previously known as FD-STREAM;
1887 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1888 argument) no longer signals an error;
1889 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1890 of the slot symbol, rather than using the current package
1891 ((:CONC-NAME "") continues to intern the slot's name in the
1893 * incremented fasl file version number, because of the incompatible
1894 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1895 of the new implementation of DEFINE-COMPILER-MACRO.
1897 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1898 * Support for building SBCL for MIPS platforms running in
1899 little-endian mode has now been checked in, and basic
1900 functionality on said platforms verified.
1901 * minor incompatible change: PCL now records the pathname of a file
1902 in which methods and the like are defined, rather than its
1904 * minor incompatible change: TRUENAME now considers the truename of
1905 a file naming a directory to be the pathname with :DIRECTORY
1906 component indicating that directory.
1907 * minor incompatible change: a NAMED clause in the extended form of
1908 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1909 reason for the previous behaviour is unclear.
1910 * more systematization and improvement of CLOS and MOP conformance
1911 in PCL (thanks to Gerd Moellman and Pierre Mai):
1912 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1914 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1915 primary methods with no specializers;
1916 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1918 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1919 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1920 FORWARD-REFERENCED-CLASSes; error reporting on
1921 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1923 ** SXHASH on CLOS instances now uses a slot internal to the
1924 instance to return different numbers on distinct instances,
1925 while preserving the same return value through invocations of
1927 ** DEFMETHOD signals errors when methods with longer incongruent
1928 lambda lists are added to generic functions;
1929 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1930 CLASS, as specified in AMOP;
1931 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1932 on the order returned by the primary method for classes of
1933 class STANDARD-CLASS;
1934 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1935 * fixed some bugs shown by Paul Dietz' test suite:
1936 ** DOLIST puts its body in TAGBODY;
1937 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1939 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1940 value producing form;
1941 ** if more variables are given to PROGV than values, extra
1942 variables are bound and made to have no value;
1943 ** NSUBSTITUTE on list arguments gets the right answer with
1945 ** ELT signals an error of type TYPE-ERROR when the index argument
1946 is not a valid sequence index;
1947 ** LOOP signals (at macroexpansion time) an error of type
1948 PROGRAM-ERROR when duplicate variable names are found;
1949 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1950 ** FUNCALL of special-operators now cause an error of type
1952 ** PSETQ now works as required in the presence of side-effecting
1953 symbol-macro places;
1954 ** NCONC accepts any object as its last argument;
1955 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1957 ** loop-for-as-package does not require a package to be explicitely
1959 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1961 * fixed bug 166: compiler preserves "there is a way to go"
1962 invariant when deleting code.
1963 * fixed bug 172: macro lambda lists with required arguments after
1964 &REST arguments now cause an error to be signalled. (thanks to
1966 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1967 bug. (thanks to Matthew Danish)
1968 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1970 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1971 arguments are lexically rebound. (thanks to Gerd Moellmann and
1973 * fixed bug 194: error messages are now more informative when there
1974 is no primary method applicable in a call to a generic
1975 function. (thanks to Gerd Moellmann)
1976 * fixed bug in command line argument checking (thanks to Julian
1978 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1979 (thanks to Matthew Danish)
1980 * incremented fasl file version number, because of the
1981 SXHASH-related changes in the layout of CLOS data structures
1983 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1984 * minor incompatible change: The runtime (the Unix executable named
1985 "sbcl") is now much pickier about the .core files it will load.
1986 Essentially it now requires .core files to descend from the same
1987 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1988 as the runtime does. (The intent is to prevent the crashes which
1989 can occur, and which can even be reported as mysterious failures,
1990 when people patch the sources or change the build parameters
1991 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1992 sbcl and .core files.)
1993 * fixed bug: VALUES-LIST is no longer optimized away.
1994 * fixed bug 142: The FFI conversion of C string values to Lisp
1995 string values no longer conses excessively. (thanks to Nathan
1996 Froyd porting Raymond Toy's fix to CMU CL)
1997 * began to systematize and improve MOP conformance in PCL (thanks to
1998 Nathan Froyd, Gerd Moellman and Pierre Mai):
1999 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2001 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2002 ** FINALIZE-INHERITANCE is now called on class finalization;
2003 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2004 argument precedence order.
2005 * fixed bug 202: The compiler no longer fails on functions whose
2006 derived types contradict their declared type.
2007 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2008 so it can be non-toplevel.
2009 * The fasl file version number has changed (because of the new
2010 implementation of DEFMACRO).
2011 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2012 safe code, that any length requirement by their type-specifier
2013 argument is valid. The exceptions to this are described in bug
2015 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2016 functions now check the types of their inputs as required by ANSI.
2017 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2018 introduced symbol is DECLAREd to be SPECIAL.
2019 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2020 * fixed bug: :COUNT argument to sequence functions may be negative
2021 * fixed bug: body of DO-SYMBOLS may contain declarations
2022 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2023 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2024 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2025 symbol macro only once
2026 * fixed printing of call frame when argument list is unavailable
2027 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2028 * compiler no longer signals WARNING on unknown keyword
2031 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2032 * A beta-quality port to the mips architecture running Linux,
2033 based on the old CMUCL backend, has been made. It has been tested
2034 on a big-endian kernel, and works sufficiently well to be able to
2035 rebuild itself; it has not been tested in little-endian mode.
2036 * fixed an inconsistency between gencgc.c and purify.c which made
2037 dumping/loading .core files unreliable
2038 * fixed bug 120a: The compiler now deals correctly with IFs where
2039 the consequent is the same as the alternative, instead of
2040 misderiving the return type. (thanks to Alexey Dejneka)
2041 * fixed bug 113: Logical pathnames are now dumpable (the logical
2042 host is resolved at load-time, throwing an error if it is not
2044 * fixed bug 174: FORMAT's error message is slightly clearer when a
2045 non-printing character is used in a format directive.
2046 * fixed several bugs in compiler checking of type declarations, i.e.
2047 violations of the Python "declarations are assertions" principle
2048 (thanks to Alexey Dejneka)
2049 * fixed several bugs in PCL's error checking (thanks to Gerd
2051 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2053 * fixed bug in compilation of functions as first class values
2054 (thanks to Antonio Martinez-Shotton)
2055 * The compiler's handling TYPE-ERRORs which it can prove will
2056 inevitably happen at runtime has been cleaned up and corrected
2057 in several ways. (thanks to Alexey Dejneka)
2058 * improved argument type checking for various basic arithmetic
2059 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2060 lavished on them in the past that they can be compiled in many
2061 ways in different special cases
2062 * fixed bug 181: compiler checks validity of user supplied type
2064 * cleaned up code flushing in optimization: Function calls which
2065 should signal errors for safety purposes (e.g. which ANSI says
2066 should signal errors when their arguments are of incorrect type)
2067 are no longer optimized away.
2068 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2069 * incremented fasl file version number, because changes in the
2070 implementation of sequence functions like COERCE caused
2071 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2072 in old inline expansions) to become undefined. (Actually these
2073 changes were later undone, so we might very well be binary
2074 compatible with 0.7.7 after all, but leaving the version number
2075 incremented seemed like the simplest and most conservative
2078 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2079 * An alpha-quality port to the parisc architecture running Linux,
2080 based on the old CMUCL backend, has been made. This, even more so
2081 than the other backends, should be considered still a work in
2082 progress; known problems include that the Linux kernel in 64-bit
2083 mode does not propagate the correct sigcontext structure to
2084 userspace, and consequently SBCL on a parisc64 kernel will not
2086 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2087 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2088 suppressed.) Also now that the compiler is looking at declarations
2089 in the environment, it checks optimization declarations as well,
2090 and suppresses inlining when (> DEBUG SPEED).
2091 * More fixes have been made to treatment of floating point exception
2092 treatment and other Unix signals. In particular, floating point
2093 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2094 * The detection and handling of control stack exhaustion (infinite
2095 or very deeply nested recursion) has changed. Stack exhaustion
2096 detection is now done by write-protecting pages at the OS level
2097 and applies at all optimization settings; when found, a
2098 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2099 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2100 be caught using IGNORE-ERRORS.
2101 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2102 correctly when transforming recursive LABELS functions to LETs.
2103 (thanks to Alexey Dejneka)
2104 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2105 that are names of constants or global variables.
2106 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2107 alien routines with docstrings.
2108 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2109 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2111 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2112 (thanks to Eric Marsden)
2113 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2114 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2115 * Bug fix: The SPARC backend can now compile functions involving
2116 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2117 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2118 to a functions expecting a PATHNAME-DESIGNATOR.
2119 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2120 lambda lists. (thanks to Alexey Dejneka)
2121 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2122 functions so that EXPORTing the name of the class doesn't cause
2123 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2124 to Antonio Martinez for reporting this)
2125 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2126 constant-folded without failing an assertion. (thanks to Einar
2127 Floystad Dorum for reporting this)
2128 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2129 types are now dealt with more correctly by the compiler.
2130 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2131 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2132 behaviour analogous to RENAME-FILE. This puts its behaviour more
2133 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2135 * The fasl file version number has changed again. (because of the
2136 bug fix involving the names of PCL MAKE-INSTANCE functions)
2138 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2139 * bug fix: Floating point exceptions are treated much more
2140 consistently on the x86/Linux and PPC/Linux platforms.
2141 * Array initialization with :INITIAL-ELEMENT is now much faster for
2142 cases when the compiler cannot open code the array creation, but
2143 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2144 array accesses have also seen a speed increase.
2145 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2146 specified. (This was at the root of some bad interactions between
2147 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2149 * bug fix: Internal error arguments for undefined functions are now
2150 computed correctly on the PPC/Linux platform.
2151 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2152 Raymond Toy's patch for CMU CL)
2153 * Support for the Solaris 9 operating environment has been included
2154 (thanks to Daniel Merritt)
2155 * A very ugly but hopefully complete draft of the missing FFI chapter
2156 of the manual has been created by reformatting the corresponding
2157 CMU CL manual chapter into (currently very ugly and incoherent)
2158 DocBook and bringing it up to date for SBCL behavior. Thus, the
2159 manual is now essentially complete, at least by my extreme
2160 once-and-only-once standards, whereby it's acceptable to refer to
2161 the doc strings of SB-EXT functions as the primary documentation.
2162 * The fasl file version number has changed again, due to cleanup of
2163 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2164 is no longer a static symbol.)
2166 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2167 * SBCL now builds with OpenMCL (version 0.12) as the
2168 cross-compilation host; also, more progress has been made toward
2169 bootstrapping under CLISP.
2170 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2172 * bug 158 fixed: The compiler can now deal with integer loop
2173 increments different from 1; fixing this turned out also to fix
2175 * bug 169 fixed: no more bogus warnings about using lexical bindings
2176 despite the presence of perfectly good SPECIAL declarations (thanks
2177 to David Lichteblau)
2178 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2179 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2180 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2182 * bug 180 fixed: Method combination specifications no longer ignore
2183 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2184 * bug fix: Structure type predicate functions now check their argument
2185 count as they should.
2186 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2187 correctly. (thanks to Pierre Mai)
2188 * minor incompatible change: The --noprogrammer option is deprecated
2189 in favor of the new --disable-debugger option, which is very similar.
2190 (The major difference is that it takes effect at a slightly different
2191 time at startup, causing handling of errors in --sysinit and
2192 --userinit files will be affected differently.) The
2193 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2194 been added to allow this functionality to be controlled from ordinary
2195 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2196 maintainers, who might want to run non-interactive scripts to
2197 build SBCL cores which will later be used interactively.)
2198 * minor incompatible change: The LOAD function no longer, when given
2199 a wild pathname to load, loads all files matching that pathname.
2200 Instead, an error of type FILE-ERROR is signalled.
2202 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2203 * bug 147 fixed: The compiler preserves its block link/count
2204 invariants more correctly now so that it doesn't crash. (thanks
2206 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2208 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2209 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2210 no-value-for-this-character as NIL (as specified by ANSI).
2211 * HOST-NAMESTRING on physical pathnames now returns a string that is
2212 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2213 (thanks to Christophe Rhodes)
2214 * The Alpha port handles icache flushing more correctly. (thanks to
2216 * More progress has been made toward bootstrapping under CLISP. (thanks
2217 to Christophe Rhodes)
2218 * The fasl file format has changed again, because dynamic loading
2219 on OpenBSD (which has non-ELF object files) motivated some cleanups
2220 in the way that foreign symbols are transformed and passed around.
2221 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2222 is no longer treated as whitespace by the reader, but instead as
2223 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2224 |A
\7fB|, instead of A as it used to.
2226 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2227 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2228 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2229 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2230 this as of 0.7.1.45, but was left out of the previous news section
2231 (thanks to Dan Barlow)
2232 * SBCL now runs on the Solaris operating system on SPARC architectures
2233 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2234 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2235 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2236 * SPARC backend cleanups, allowing builds of cores optimized for V8
2237 and V9 SPARCS, and also emission of code targeted to a particular
2238 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2240 * SBCL is closer to bootstrapping under CLISP, thanks to various
2241 fixes by Christophe Rhodes.
2242 * The fasl file format has changed again, to allow the compiler's
2243 INFO database to support symbol macros.
2244 * The user manual (in doc/) is formatted into HTML more nicely.
2245 (thanks to coreythomas)
2246 * The system is smarter about SUBTYPEP relationships, especially
2247 those involving NOT types (including types such as ATOM which are
2248 represented internally using NOT types). Thus SUBTYPEP is less
2249 likely to return (VALUES NIL NIL) in general, and in particular
2250 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2252 * The fasl file format has changed again, because the internal
2253 representation of types now includes a new slot to support the new
2254 SUBTYPEP-of-NOT-types logic.
2255 * (not a change in the main branch of SBCL, but a related prototype
2256 which can hopefully be merged into the main branch of SBCL in the
2257 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2258 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2259 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2260 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2261 dispatch macro character combinations. (thanks to Alexey Dejneka)
2262 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2263 that superficially look like logical namestrings correctly.
2264 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2266 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2267 default supplied-p) for &optional and &key arguments. (thanks to
2270 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2271 * incompatible change: The compiler is now less aggressive about
2272 tail call optimization, doing it only when (> SPACE DEBUG) or
2273 (> SPEED DEBUG). (This is an incompatible change because there are
2274 programs which relied on the old CMU-CL-style behavior to optimize
2275 away their unbounded recursion which will now die of stack overflow.)
2276 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2277 for non-GENCGC systems has been increased to 20M (since that
2278 seems much closer to the likely performance optimum for modern
2279 systems than the old 4M value was)
2280 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2281 and *DEBUG-PRINT-LEVEL*
2282 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2283 of CMU CL's support for SPARC, and various endianness and other
2284 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2285 * new syntactic sugar for the Unix command line: --load foo.bar is now
2286 an alternate notation for --eval '(load "foo.bar")'.
2288 ** The system now detects stack overflow and handles it gracefully,
2289 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2290 optimization settings. (This is a good thing in general, and
2291 its introduction in this version should be particularly timely
2292 for anyone whose code fails because of suppression of tail
2294 ** The system now hunts for the C variable "environ" in a more
2295 devious way, to avoid segfaults when the C library version differs
2296 between compile time and run time. (thanks to Christophe Rhodes)
2297 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2298 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2299 ** The compiler no longer issues bogus style warnings for undefined
2300 classes in the same source file as the DEFCLASSes which defined
2301 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2303 ** fixes in CONDITION class precedence list for undefined function
2304 errors (thanks to Alexei Dejneka)
2305 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2306 correctly. (thanks to Dan Barlow)
2307 ** portability fixes aiming at bootstrapping under CLISP (thanks
2308 to Dave McDonald and Christophe Rhodes)
2309 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2310 ** fixes in type translation and and type inference (thanks to
2312 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2313 ** various fixes in the new ports (thanks to Dan Barlow)
2314 * several changes related to debugging:
2315 ** suppression of tail recursion, as noted above
2316 ** stack overflow detection, as noted above
2317 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2318 is now the default. (For some time encapsulation has been more
2319 reliable than the breakpoint-based :ENCAPSULATE NIL
2320 implementation, at least on X86 systems; and I just noticed that
2321 encapsulation also seems closer to the spirit of the ANSI
2324 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2326 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2327 up properly again. (There was a packaging bug in 0.7.0 which
2328 left their definitions in SB-SYS::LOAD-FOREIGN and
2329 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2330 vital for most things which interface to C-level interfaces,
2331 like extensions working with sockets or databases or
2332 Perl-compatible regexes or whatever, and the need to fix
2333 this bug is the main reason that 0.7.1 was released so
2335 ** DEFGENERIC is now choosier about the methods it redefines, so that
2336 reLOADing a previously-LOADed file containing DEFGENERICs does
2337 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2338 cycle now works as it should. (thanks to Alexey Dejneka)
2339 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2340 fixed by Christophe Rhodes. (He actually submitted this patch
2341 months ago, and I delayed until after 0.7.0.)
2342 ** Bug 111 (internal compiler confusion about runtime checks on
2343 FUNCTION types) was fixed by Alexey Dejneka.
2344 * Some internal cleanups (getting rid of variables which aren't
2345 needed now that the byte interpreter is gone) caused the fasl
2346 file format number to change again.
2348 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2349 * major incompatible change: The default fasl file extension, i.e. the
2350 default extension for files produced by COMPILE-FILE, has changed
2351 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2353 ** There are many changes in the implementation of the compiler.
2354 SBCL is now essentially a compiler-only implementation of ANSI
2355 Common Lisp. EVAL still "interprets" a few special cases, but
2356 almost all the interesting cases are handled by creating
2357 a LAMBDA expression, calling COMPILE on it, then calling
2358 FUNCALL on the result.
2359 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2360 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2361 Since the code is newer, there might still be some new bugs
2362 (though not as many as before Martin Atzmueller's fixes:-). But
2363 the new code is substantially simpler and clearer, and hopefully
2364 any remaining bugs will be simpler, less fundamental, and more
2365 fixable then the bugs in the old code.
2366 ** The revised compiler is still a little unsteady on its feet.
2368 *** The debugging information it produces (particularly the names
2369 of FUNCTION objects) is sometimes much less useful than what
2370 the old compiler produced.
2371 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2372 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2373 within a MACROLET) has been temporarily weakened.
2374 ** There are new compiler optimizations for various functions:
2375 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2376 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2377 *** the math functions TRUNCATE, FLOOR, and CEILING
2378 *** the function-of-all-trades COERCE
2379 Mostly these should be transparent, but there's one
2380 potentially-annoying problem (bug 117): when the compiler
2381 inline-expands a function and does type analysis on the result,
2382 it can create control paths which have type mismatches, and
2383 when it can't prove that those control paths aren't taken,
2384 it will issue WARNINGs about the type mismatches. This is
2385 a particular problem in practice for the new sequence functions.
2386 It's not clear how this should be fixed, and for now, a
2387 workaround is given in the entry for 117 in the BUGS file.
2388 ** (Because of the interaction between the two previous items --
2389 occasional inlining problems and new inline expansions -- some
2390 of the new sequence function optimizations won't really kick in
2391 completely until debugging information, and then inlining, are
2392 straightened out in some future version.)
2393 * minor incompatible changes:
2394 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2395 DIRECTORY no longer implicitly promotes NIL slots of its
2396 pathname argument to :WILD. In particular, when you ask for the
2397 contents of a directory (which you used to be able to do without
2398 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2399 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2400 ** changes in behavior that ANSI explicitly defines to be
2401 implementation dependent:
2402 *** The new compiler-only implementation still conforms with ANSI,
2403 but acts a little different than before. Besides the obvious
2404 changes in performance tradeoffs (that the cost per form passed
2405 to EVAL has gone up, and the cost per form executed by EVAL
2406 has gone down), the behavior of the system changes a little
2407 because there are no longer any interpreted function objects.
2408 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2409 e.g. doing COMPILE on the output of interactive DEFUN is
2411 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2413 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2414 has changed from (:CL) to NIL.
2415 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2416 e.g. #\Newline and #\Tab, don't have names specified in the
2417 ANSI Common Lisp standard, is now based on their ASCII symbolic
2418 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2419 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2420 are no longer used for output.
2421 ** changes in internal implementation constants:
2422 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2423 4 million. (If your application spends a lot of time GCing and
2424 you have a lot of RAM, you might want to experiment with
2425 increasing it even more.)
2426 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2427 However, almost all old code should still continue to work without
2428 immediate update, as SB-C-CALL is now a (deprecated) nickname
2430 ** Old operator names in the style DEF-FOO are now deprecated in
2431 favor of new corresponding names DEFINE-FOO, for consistency with
2432 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2433 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2434 internal symbols, but a few supported extensions like
2435 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2436 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2437 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2438 etc. as you get deeper into recursive calls to the debugger
2439 command loop, instead of the old "5]", "5]]", "5]]]"
2440 sequence. (I was motivated to do this when squabbles between
2441 ILISP and SBCL left me very deeply nested in the debugger. In the
2442 short term, this change will probably provoke more ILISP/SBCL
2443 squabbles, but hopefully it will be an improvement in the long run.)
2444 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2445 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2446 function, since declaiming return types involving aliens is
2447 (1) annoyingly messy to do by hand and (2) vital to efficient
2448 compilation of code which calls such functions.
2449 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2450 longer reexported by the SB-EXT package. They're solely useful
2451 for alien code, so it seems more logical that you should get
2452 them from the SB-ALIEN package, not in SB-EXT.
2453 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2454 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2455 features. Instead, the code that they used to control is always
2456 built into the system.
2457 * many other bug fixes
2458 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2459 advantage of the new EVAL-WHEN stuff and to clean them up in
2460 general, and they are now more ANSI-compliant in a number of
2461 ways. Martin Atzmueller is responsible for a lot of this.
2462 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2464 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2465 *** correct ERROR type for various file operations
2466 *** some fixes for Lisp streams
2467 *** DEFMETHOD syntax checking
2468 *** changing old weird representation of debug information as
2469 strings (which, among their other deficiencies, don't transform
2470 correctly when you rename packages, and don't change their
2471 print representation when you change things like *PACKAGE*
2472 and *PRINT-LENGTH*) to symbols and lists of symbols
2473 He also made several improvements and fixed several bugs in DESCRIBE.
2474 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2476 *** misbehavior of WRITE-STRING/WRITE-LINE
2477 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2478 and several other LOOP problems as well
2479 *** DIRECTORY when similar filenames are present
2480 *** DEFGENERIC with :METHOD options
2481 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2482 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2483 *** argument ordering in FIND with :TEST option
2484 *** mishandled package designator argument in APROPOS-LIST
2485 *** various problems in the backquote readmacro
2486 *** a bug in APROPOS
2487 *** probably some others that I'm not describing very well here,
2488 since the CVS log documents them by reference to sbcl-devel
2489 messages, and the SourceForge archives aren't working well.:-(
2490 ** Dan Barlow improved the Alpha port (and is making progress on the
2491 PPC port, for those of you who think different).
2492 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2493 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2494 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2495 continues to work on the SPARC port (for those of you in a position
2496 to look down upon our little PC-compatible boxes from a great height).
2497 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2498 and uses that copy, rather than the current dynamic value, when
2499 it's trying to decide whether to truncate output. Thus e.g.
2500 (let ((*print-lines* 50))
2501 (pprint-logical-block (stream nil)
2503 (let ((*print-lines* 8))
2504 (print (aref possiblybigthings i) stream)))))
2505 should now truncate the logical block only at 50 lines, instead of
2506 often truncating it at 8 lines, as it did before.
2507 * The doc/cmucl/ directory, containing old CMU CL documentation
2508 from the time of the fork, is no longer part of the base system.
2509 SourceForge has shut down its anonymous FTP service, and with it
2510 my original plan for distributing the old CMU CL documentation
2511 there. For now, if you need these files you can download an old
2512 SBCL source release and extract them from it.
2513 * The fasl file version number changed again, for dozens of reasons,
2514 some of which are apparent above.
2516 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2517 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2518 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2519 patch, so that DISASSEMBLE output is much nicer.
2520 * The code in the SB-PROFILE package now seems reasonably stable.
2521 I still haven't decided what the final interface should look like
2522 (I'd like PROFILE to interact cleanly with TRACE, since both
2523 facilities use function encapsulation) but if you have a need
2524 for profiling now, you can probably use it successfully with
2525 the current CMU-CL-style interface.
2526 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2527 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2528 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2529 link now returns the dangling link itself, and for similar
2530 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2531 link itself. (In these cases the old code signalled an error and
2532 looped endlessly, respectively.) Thus, DIRECTORY now works even
2533 in the presence of dangling and cyclic symbolic links.
2534 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2535 is now a supported extension again, since the consensus on
2536 sbcl-devel was that it can be useful for ordinary development
2537 work, not just for debugging SBCL itself.
2538 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2539 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2540 that currently-defined functions might be redefined later with
2541 different return types.
2542 * Hash tables can be printed readably, as inspired by CMU CL code
2543 of Eric Marsden and SBCL code of Martin Atzmueller.
2544 * better error handling in CLOS method combination, thanks to
2545 Martin Atzmueller porting Pierre Mai's CMU CL patches
2546 * more overflow fixes for >16Mbyte I/O buffers
2547 * A bug in READ has been fixed, so that now a single Ctrl-D
2548 character suffices to cause end-of-file on character streams.
2549 In particular, now you only need one Ctrl-D at the command
2550 line (not two) to exit SBCL.
2551 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2553 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2554 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2555 does the right thing.
2556 * The implementation of some type tests, especially for CONDITION
2557 types, is now tidier and maybe faster, due to CMU CL code
2558 originally by Douglas Crosher, ported by Martin Atzmueller.
2559 * Some math functions have been fixed, and there are new
2560 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2561 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2562 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2563 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2564 likely to be built into sbcl-0.7.0.)
2565 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2566 CL, ported to SBCL by Martin Atzmueller)
2567 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2568 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2569 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2570 * There's a new slam.sh hack to shorten the edit/compile/debug
2571 cycle for low-level changes to SBCL itself, and a new
2572 :SB-AFTER-XC-CORE target feature to control the generation of
2573 the after-xc.core file needed by slam.sh.
2574 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2575 COMPILE-FILE is no longer supported, so that now every function
2576 gets an entry point, so that block compilation looks a little
2577 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2578 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2579 returns the number of bytes consed since the system started,
2580 rather than the number consed since the first time the function
2581 was called. (The new definition parallels ANSI functions like
2582 CL:GET-INTERNAL-RUN-TIME.)
2583 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2584 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2585 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2586 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2587 * Fasl file version numbers are now independent of the target CPU,
2588 since historically most system changes which required version
2589 number changes have affected all CPUs equally. Similarly,
2590 the byte fasl file version is now equal to the ordinary
2593 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2594 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2595 is no longer recognized. I apologize for this, because it was
2596 listed in SB-EXT as a supported extension, but I found that
2597 its existing behavior was poorly specified, as well as incorrectly
2598 specified, and it looked like too much of a mess to straighten it
2599 out. I have enough on my hands trying to get ANSI stuff to work..
2600 * many patches ported from CMU CL by Martin Atzmueller, with
2601 half a dozen bug fixes in pretty-printing and the debugger, and
2602 half a dozen others elsewhere
2603 * fixed bug 13: Floating point infinities are now supported again.
2604 They might still be a little bit flaky, but thanks to bug reports
2605 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2606 as flaky as they were.
2607 * The --noprogrammer command line option is now supported. (Its
2608 behavior is slightly different in detail from what the old man
2609 page claimed it would do, but it's still appropriate under the
2610 same circumstances that the man page talks about.)
2611 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2612 are now supported, and enabled by default. Thus, the compiler can
2613 handle many floating point and complex operations much less
2614 inefficiently. (Thus e.g. you can implement a complex FFT
2616 * The compiler now detects type mismatches between DECLAIM FTYPE
2617 and DEFUN better, and implements CHECK-TYPE more correctly, and
2618 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2619 thanks to patches from Martin Atzmueller.
2620 * various fixes to make the cross-compiler more portable to
2621 ANSI-conforming-but-different cross-compilation hosts (notably
2622 Lispworks for Windows, following bug reports from Arthur Lemmens)
2623 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2624 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2625 fixed thanks to Pierre Mai's CMU CL patches.
2626 * fixes to keep the system from overflowing internal counters when
2627 it tries to use i/o buffers larger than 16M bytes
2628 * fixed bug 45a: Various internal functions required to support
2629 complex special functions have been merged from CMU CL sources.
2630 (When I was first setting up SBCL, I misunderstood a compile-time
2631 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2632 * improved support for type intersection and union, fixing bug 12
2633 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2634 more obscure bugs as well
2635 * some steps toward byte-compiling non-performance-critical
2636 parts of the system, courtesy of patches from Martin Atzmueller
2637 * Christophe Rhodes has made some debian packages of sbcl at
2638 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2639 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2640 stable, but are nonetheless usable. When he's ready, I'd be happy
2641 to add them to the SourceForge "File Releases" section. (And if
2642 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2643 * new fasl file format version number (because of changes in
2644 internal representation of (OR ..) types to accommodate the new
2645 support for (AND ..) types, among other things)
2647 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2648 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2650 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2652 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2653 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2654 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2655 * DESCRIBE now works on structure objects again.
2656 * Most function call argument type mismatches are now handled as
2657 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2658 know whether the function will be redefined before the call is
2659 executed. (The compiler could flag local calls with full WARNINGs,
2660 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2661 it doesn't keep track of enough information to know whether calls
2662 are local in this sense.)
2663 * Compiler output is now more verbose, with messages truncated
2664 later than before. (There should be some supported way for users
2665 to override the default verbosity, but I haven't decided how to
2666 provide it yet, so this behavior is still controlled by the internal
2667 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2668 src/compiler/ir1util.lisp.)
2669 * Fasl file format version numbers have increased again, because
2670 support for the Gray streams extension changes the layout of the
2671 system's STREAM objects.
2672 * The Gray subclassable streams extension now works, thanks to a
2673 patch from Martin Atzmueller.
2674 * The full LOAD-FOREIGN extension (not just the primitive
2675 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2676 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2677 but like most other programs, it defaults to copying the Unix
2678 environment from the original process instead of starting the
2679 new process in an empty environment.
2680 * Extensions which manipulate the Unix environment now support
2681 an :ENVIRONMENT keyword option which doesn't smash case or
2682 do other bad things. The CMU-CL-style :ENV option is retained
2683 for porting convenience.
2684 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2685 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2687 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2689 * A patch from Martin Atzmueller seems to have solved the SIGINT
2690 problem, and as far as we know, signal-handling now works cleanly.
2691 (If you find any new bugs, please report them!)
2692 * The system no longer defaults Lisp source file names to types
2693 ".l", ".cl", or ".lsp", but only to ".lisp".
2694 * The compiler no longer uses special default file extensions for
2695 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2696 seems to expect a single default extension for all compiled code,
2697 and there's no compelling reason to try to stretch the standard
2698 to allow two different extensions.) Instead, byte-compiled files
2699 default to the same extension as native-compiled files.
2700 * Fasl file format version numbers have increased again, because
2701 a rearrangement of internal implementation packages made some
2702 dumped symbols in old fasl files unreadable in new cores.
2703 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2704 many fewer weird special cases.
2705 * Bug #17 (differing COMPILE-FILE behavior between logical and
2706 physical pathnames) has been fixed, and some related misbehavior too,
2707 thanks to a patch from Martin Atzmueller.
2708 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2709 by Tim Moore, ported to SBCL by Martin Atzmueller.
2710 * Martin Atzmueller fixed several filesystem-related problems,
2711 including bug #36, in part by porting CMU CL patches, which were
2712 written in part by Paul Werkowski.
2713 * More compiler warnings in src/runtime/ are gone, thanks to
2714 more patches from Martin Atzmueller.
2715 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2718 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2720 * DESCRIBE now works on CONDITION objects.
2721 * The debugger now handles errors which arise when trying to print
2722 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2724 * The build system now uses an additional file, customize-target-features.lisp,
2725 to allow local modifications to the target *FEATURES* list. (The point of
2726 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2727 debugging features enabled, without having to worry about propagating it
2728 into everyone's system when I do a "cvs update".) When no
2729 customize-target-features.lisp file exists, the target *FEATURES* list
2730 should be constructed the same way as before.
2731 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2732 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2733 warn about behavior which is undefined under the ANSI spec. Note: This
2734 is specified by ANSI, but it's not very popular with programmers.
2735 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2736 macro in the SBCL sources for an example of a workaround which you
2737 might use to make portable ANSI-standard code which does what you want.
2738 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2739 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2740 when it's used as a non-toplevel form. (This is required in order
2741 to implement the DEFCONSTANT-EQX macro.)
2742 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2743 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2744 funny things, due to bugs in the implementation of EVAL-WHEN
2745 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2746 non-toplevel DEFCONSTANTs are uncommon.)
2747 * The core file version number and fasl file version number have been
2748 incremented, because the old noncompliant DEFCONSTANT behavior involved
2749 calling functions which no longer exist, and because I also took the
2750 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2751 * fixed bug 1 (error handling before read-eval-print loop starts), and
2752 redid debugger restarts and related debugger commands somewhat while
2754 ** The QUIT debugger command is gone, since it did something
2755 rather different than the SB-EXT:QUIT command, and since it never
2756 worked properly outside the main toplevel read/eval/print loop.
2757 Invoking the new TOPLEVEL restart provides the same functionality.
2758 ** The GO debugger command is also gone, since you can just invoke
2759 the CONTINUE restart directly instead.
2760 ** The TOP debugger command is also gone, since it's redundant with the
2761 FRAME 0 command, and since it interfered with abbreviations for the
2763 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2765 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2766 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2767 some porting work by Martin Atzmueller)
2768 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2769 with a boatload of patches which he ported to SBCL. Now that those
2771 ** The system tries to make sure that its low-priority messages
2772 are prefixed by semicolons, to help people who like to use
2773 syntax highlighting in their ILISP buffer. (This patch
2774 was originally due to Raymond Toy.)
2775 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2776 patches originally written by Raymond Toy.
2777 ** The compiler understands coercion between single-value and
2778 multiple-VALUES type expressions better, getting rid of some very
2779 weird behavior, thanks to patches originally by Robert MacLachlan
2780 and Douglas Crosher.
2781 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2782 lambda lists, thanks to a patch originally by Pierre Mai.
2783 ** The system no longer bogusly warns about "abbreviated type
2785 ** The compiler gets less confused by inlining and RETURN-FROM,
2786 thanks to some patches originally by Tim Moore.
2787 ** The system no longer hangs when dumping circular lists to fasl
2788 files, thanks to a patch originally from Douglas Crosher.
2789 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2790 undefined function error.
2791 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2792 in structure slot initforms) for now, documented workaround instead:-|
2793 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2794 * fixed bug 5: added stubs for various Gray stream functions called
2795 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2796 installed, at least appropriate type errors are generated
2797 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2798 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2799 consistently in DEFMETHOD forms.
2800 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2801 it doesn't seem to affect SBCL after all
2802 * The C runtime system now builds with better optimization and many
2803 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2805 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2807 * The system is now under CVS at SourceForge (instead of the
2808 CVS repository on my home machine).
2809 * The new signal handling code has been tweaked to treat register
2810 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2811 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2812 downstream, e.g. in debug-int.lisp, has implicit dependencies
2813 on the unsignedness of integer representation of machine words,
2814 and that caused the system to bomb out with infinite regress
2815 when trying to recover from type errors involving signed values,
2816 e.g. (BUTLAST '(1 2 3) -1).)
2817 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2818 (This was one of the bugs Peter Van Eynde reported back in July.)
2819 * The system now uses code inspired by Colin Walters' O(N)
2820 implementation of MAP (from the cmucl-imp@cons.org mailing
2821 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2822 inline the MAP operation, and there is more than one
2823 sequence argument to the MAP call (so that it can't just
2824 do ETYPECASE once and for all based on the type of the
2825 single sequence argument). (The old non-inline implementation
2826 of the general M-argument sequence-of-length-N case required
2827 O(M*N*N) time when any of the sequence arguments were LISTs.)
2828 * The QUIT :UNIX-CODE keyword argument has been renamed to
2829 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2830 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2832 * Raymond Wiker's patches to port dynamic loading from Linux to
2833 FreeBSD have been added.
2834 * The BUGS file is now more nearly up to date, thanks in large part
2835 to Martin Atzmueller's review of it.
2836 * The debugger now flushes standard output streams before it begins
2837 its output ("debugger invoked" and so forth).
2838 * The core version number and fasl file version number have both
2839 been incremented, because of incompatible changes in the layout
2841 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2842 used to not be called for a saved Lisp image.)
2843 * Martin Atzmueller's version of a patch to fix a compiler crash,
2844 as posted on sbcl-devel 13 September 2000, has been installed.
2845 * Instead of installing Martin Atzmueller's patch for the
2846 compiler transform for SUBSEQ, I deleted the compiler transform,
2847 and transforms for some similar consing operations.
2848 * A bug in signal handling which kept TRACE from working on OpenBSD
2850 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2851 away by constant folding
2852 * The system now defines its address space constants in one place
2853 (in the Lisp sources), and propagates them automatically elsewhere
2854 (through GENESIS and the sbcl.h file). Therefore, patching the
2855 address map is less unnecessarily tedious and error-prone. The
2856 Lisp names of address space constants have also been systematized.
2857 * CVS tags like dollar-Header-dollar have been removed from
2858 the sources, because they have never saved me trouble and
2859 they've been source of trouble working with patches and other
2860 diff-related operations.
2861 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2862 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2864 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2866 * The system has been ported to OpenBSD.
2867 * The system now compiles with a simple "sh make.sh" on the systems
2868 that it's supported on. I.e., now you no longer need to tweak
2869 text in the target-features.lisp-expr and symlinks in src/runtime/
2870 by hand, the make.sh takes care of it for you.
2871 * The system is no longer so grossly inefficient when compiling code
2872 involving vectors implemented as general (not simple) vectors (VECTOR T),
2873 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2874 which dares to use the various sequence functions on non-simple
2875 vectors, takes less of a performance hit.
2876 * There is now a primitive type predicate VECTOR-T-P
2877 to test for the (VECTOR T) type, so that e.g.
2878 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2879 can now be compiled with some semblance of efficiency. (The old code
2880 turned the type declaration into a full call to %TYPEP at runtime!)
2881 * AREF on (VECTOR T) is still not fast, since it's still compiled
2882 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2883 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2884 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2885 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2886 without making full calls to SUBTYPEP at runtime.
2887 (Some analogous efficiency issues for non-simple vectors specialized to
2888 element types other than T, or for non-simple multidimensional arrays,
2889 have not been addressed. They could almost certainly be handled the
2890 same way if anyone is motivated to do so.)
2891 * The changes in array handling break binary compatibility, so
2892 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2893 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2894 instead of (VALUES T T).
2895 * By following the instructions that Dan Barlow posted to sbcl-devel
2896 on 2 July 2000, I was able to enable primitive dynamic object
2897 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2898 functionality is not implemented (since it calls ld to resolve
2899 library references automatically, requiring RUN-PROGRAM for its
2900 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2901 try to resolve library references) is now supported.
2902 * The system now flushes the standard output streams when it terminates,
2903 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2904 them at several other probably-convenient times, e.g. in each pass of
2905 the toplevel read-eval-print loop, and after evaluating a form given
2906 as an "--eval" command-line option. (These changes were motivated by a
2907 discussion of stream flushing issues on cmucl-imp in August 2000.)
2908 * The source transform for TYPEP of array types no longer assumes
2909 that an array whose element type is a not-yet-defined type
2910 is implemented as an array of T, but instead punts, so that the
2911 type will be interpreted at runtime.
2912 * There is now some support for cross-compiling in make.sh: each of
2913 the phases of make.sh has its own script. (This should be transparent
2914 to people doing ordinary, non-cross-compile builds.)
2915 * Since my laptop doesn't have hundreds of megabytes of memory like
2916 my desktop machine, I became more motivated to do some items on
2917 my to-do list in order to reduce the size of the system a little:
2918 ** Arrange for various needed-only-at-cold-init things to be
2919 uninterned after cold init. To support this, those things have
2920 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2921 symbols with such names are now uninterned after cold init).
2922 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2923 fasl files for cold load.
2924 ** Remove the old compiler structure pooling code (which used to
2925 be conditional on the target feature :SB-ALLOC) completely.
2926 ** Redo the representation of some data in cold init to be more compact.
2927 (I also looked into supporting byte compiled code at bootstrap time,
2928 which would probably reduce the size of the system a lot, but that
2929 looked too complicated, so I punted for now.)
2930 * The maximum signal nesting depth in the src/runtime/ support code has
2931 been reduced from 4096 to 256. (I don't know any reason for the very
2932 large old value. If the new smaller value turns out to break something,
2933 I'll probably just bump it back up.)
2934 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2936 * Many, many bugs reported by Peter Van Eynde have been added to
2937 the BUGS list; some have even been fixed.
2938 * While enabling dynamic object file loading, I tried to make the
2939 code easier to understand, renaming various functions and variables
2940 with less ambiguous names, and changing some function calling
2941 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2942 * While trying to figure out how to do the OpenBSD port, I tried to
2943 clean up some of the code in src/runtime/. In particular, I dropped
2944 support for non-POSIX signal handling, added various comments,
2945 tweaked the code to reduce the number of compilation warnings, and
2946 renamed some files to increase consistency.
2947 * To support the new automatic configuration functionality in make.sh,
2948 the source file target-features.lisp-expr has been replaced with the
2949 source file base-target-features.lisp-expr and the machine-generated
2950 file local-target-features.lisp-expr.
2951 * fixed a stupid quoting error in make.sh so that using CMU CL
2952 "lisp -batch" as cross-compilation host works again
2954 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2956 * DESCRIBE no longer tries to call itself recursively to describe
2957 bound/fbound values, so that it no longer fails on symbols which are
2958 bound to themselves (like keywords, T, and NIL).
2959 * DESCRIBE now works on generic functions.
2960 * The printer now prints less-screwed-up representations of closures
2961 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2962 * A private symbol is used instead of the :EMPTY keyword previously
2963 used to mark empty slots in hash tables. Thus
2964 (DEFVAR *HT* (MAKE-HASH-TABLE))
2965 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2966 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2967 now does what ANSI says that it should. (You can still get
2968 similar noncompliant behavior if bang on the hash table
2969 implementation with all the symbols you get back from
2970 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2971 This breaks binary compatibility, since tests for equality to
2972 :EMPTY are wired into things like the macroexpansion of
2973 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2975 * There's now a minimal placeholder implementation for CL:STEP,
2976 as required by ANSI.
2977 * An obscure bug in the interaction of the normal compiler, the byte
2978 compiler, inlining, and structure predicates has been patched
2979 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2980 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2982 * Missing ordinary arguments in a macro call are now detected even
2983 when the macro lambda list contains &KEY or &REST.
2984 * The debugger no longer complains about encountering the top of the
2985 stack when you type "FRAME 0" to explicitly instruct it to go to
2986 the top of the stack. And it now prints the frame you request even
2987 if it's the current frame (instead of saying "You are here.").
2988 * As specified by ANSI, the system now always prints keywords
2989 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2991 * The default initial SIZE of HASH-TABLEs is now smaller.
2992 * Type information from CLOS class dispatch is now propagated
2993 into DEFMETHOD bodies, so that e.g.
2994 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2996 is now basically equivalent to
2997 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2998 (DECLARE (TYPE SINGLE-FLOAT X))
3000 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3001 operation, without having to do run-time type dispatch.
3002 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3003 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3004 * Since I don't seem to be making big file reorganizations very often
3005 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3006 is overflowing my ability to conveniently back them up), I've finally
3007 checked the system into CVS. (The CVS repository is on my home system,
3008 not at SourceForge -- putting it on SourceForge might come later.)
3009 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3010 high-level GC-NOTIFY-FOO functions send their output. (There's
3011 still very little control of where low-level verbose GC functions
3012 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3013 less than it used to -- the GC-NOTIFY-FOO functions are now under
3014 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3015 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3016 in only one place in the source code, and propagates it automatically
3017 everywhere that it's needed. Thus e.g. when I bump the version from
3018 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3019 * The C source files now include boilerplate legalese and documentation
3020 at the head of each file (just as the Lisp source files already did).
3021 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3022 to his page will be replaced with a link to his new CLiki service.
3024 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3026 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3027 * The build process now looks for GNU make under the default name "gmake",
3028 instead of "make" as it used to. If GNU make is not available as "gmake"
3029 on your system, you can change this default behavior by setting the
3030 GNUMAKE environment variable.
3031 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3032 can build without error under CMU CL.
3034 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3036 * There is now a partial SBCL user manual (with some new text and some
3037 text cribbed from the CMU CL manual).
3038 * The beginnings of a profiler have been added (starting with the
3039 CMU CL profiler and simplifying and cleaning up). Eventually the
3040 main interface should be through the TRACE macro, but for now,
3041 it's still accessed through vaguely CMU-CL-style functions and macros
3042 exported from the package SB-PROFILE.
3043 * Some problems left over from porting CMU CL to the new
3044 cross-compilation bootstrap process have been cleaned up:
3045 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3046 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3047 part when getting rid of the compiler *BACKEND* stuff.)
3048 ** The value of *NULL-TYPE* was screwed up, because it was
3049 being initialized before the type system knew the final
3050 definition of the 'NULL type. This screwed up several key
3051 optimizations in the compiler, causing inefficiency in all sorts
3052 of places. (I found it because I wanted to understand why
3053 GET-INTERNAL-RUN-TIME was consing.)
3054 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3055 PROCLAIM FTYPE information. Unfortunately this broke binary
3056 compatibility again, since now the forms output by DEFGENERIC
3057 to refer to functions which didn't exist in 0.6.3.
3058 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3059 can use the new (as of 0.6.3) transform for SOME into MAP into
3061 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3062 values to (INTEGER 0 1000000), so that the time code will no longer
3063 occasionally get blown up by Linux returning 1000000 microseconds
3064 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3065 its output conform to the ANSI spec. (Alas, this makes its output
3066 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3067 * A full call to MAP NIL with a single sequence argument no longer conses.
3068 * fixes to problems pointed out by Martin Atzmueller:
3069 * The manual page no longer talks about multiprocessing as though
3070 it were currently supported.
3071 * The ILISP support patches have been removed from the distribution,
3072 because as of version 5.10.1, ILISP now supports SBCL without us
3073 having to maintain patches.
3074 * added a modified version of Raymond Toy's recent CMU CL patch for
3075 EQUALP comparison of HASH-TABLE
3077 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3079 * The system still can't cross-compile itself with
3080 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3081 entails), but at least it can compile more of itself that way
3082 than it used to be able to, and various buglets which were uncovered
3083 by trying to cross-compile itself that way have now been fixed.
3084 * This release breaks binary compatibility again. This time
3085 at least I've incremented the FASL file format version to 2, so that the
3086 problem can be detected reliably instead of just causing weird errors.
3087 * various new style warnings:
3088 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3089 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3090 ** using the lexical binding of a variable named in the *FOO* style
3091 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3092 as specified by ANSI.
3093 * *RANDOM-STATE* is no longer automatically initialized from
3094 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3095 default behavior of the system is to repeat its behavior every time
3096 it's run. If you'd like to change this behavior, you can always
3097 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3098 old convention there was no comparably easy way to get the system to
3099 repeat its behavior every time it was run.
3100 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3101 FTYPE declarations has been removed, in favor of their ANSI
3102 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3103 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3104 the types of their sequence arguments can be determined at compile time.
3105 This is done through a new open code expansion for MAP which eliminates
3106 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3107 argument types can be determined at compile time.
3108 * The optimizer now transforms COERCE into an identity operation when it
3109 can prove that the coerced object is already of the correct type. (This
3110 can be a win for machine generated code, including the output of other
3111 optimization transforms, such as the MAP transform above.)
3112 * Credit information has been moved from source file headers into CREDITS.
3113 * Source file headers have been made more standard.
3114 * The CASE macro now compiles without complaining even when it has
3117 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3119 * (Note that the way that the PCL macroexpansions were rewritten
3120 to accommodate the change in DEFGENERIC below breaks binary
3121 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3122 not run under sbcl-0.6.2. Once we get out of alpha releases,
3123 i.e. hit release 1.0.0, we'll probably try to maintain binary
3124 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3125 and sbcl-1.4.4. Until then, however, it might be fairly common
3126 for maintenance releases to break binary compatibility.)
3127 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3129 * The reporting of unhandled signals has been changed to print some
3130 explanatory text as well as the report form. (Previously only
3131 the report form was printed.)
3132 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3133 it defines, so that the compiler no longer issues undefined function
3134 warnings for compiled-but-not-yet-loaded generic functions.
3135 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3136 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3137 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3138 section of the ANSI spec are supported.
3139 * The "" nickname for the "KEYWORD" package has been removed.
3140 The reader still handles symbol tokens which begin with a package marker
3141 as keywords, but it doesn't expose its mechanism for doing so in the
3142 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3143 * The system now issues STYLE-WARNINGs for contradictory TYPE
3144 proclamations. (Warnings for contradictory FTYPE proclamations would
3145 be nice too, but those can't be done usefully unless the type system
3146 is made smarter about FUNCTION types.)
3147 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3148 systematized, so that "*target-*.lisp is supposed to exist only on the
3149 target and imply that there's a related file which exists on the
3150 host, and *host-*.lisp is supposed to exist only on the host and imply
3151 that there's a related file which exists on the target. This involves a
3152 lot of renaming. Hopefully the acute confusion caused by the renaming
3153 will be justified by the reduction in chronic confusion..
3154 ** runtime-type.lisp -> early-target-type.lisp
3155 ** target-type.lisp -> late-target-type.lisp
3156 ** early-host-format.lisp -> early-format.lisp
3157 ** late-host-format.lisp -> late-format.lisp
3158 ** host-error.lisp -> misc-error.lisp
3159 ** early-error.lisp -> early-target-error.lisp
3160 ** late-error.lisp -> late-target-error.lisp
3161 ** host-defboot.lisp -> early-defboot.lisp
3162 ** code/misc.lisp -> code/target-misc.lisp
3163 ** code/host-misc.lisp -> code/misc.lisp
3164 ** code/numbers.lisp -> code/target-numbers.lisp
3165 ** code/early-numbers.lisp -> numbers.lisp
3166 ** early-host-type.lisp -> early-type.lisp
3167 ** late-host-type.lisp -> late-type.lisp
3168 ** host-typep.lisp -> typep.lisp
3169 ** load.lisp -> target-load.lisp
3170 ** host-load.lisp -> load.lisp
3171 ** host-disassem.lisp -> disassem.lisp
3172 ** host-insts.lisp -> insts.lisp
3173 ** byte-comp.lisp -> target-byte-comp.lisp
3174 ** host-byte-comp.lisp -> byte-comp.lisp
3175 ** host-signal.lisp -> signal.lisp
3176 ** host-defstruct.lisp -> defstruct.lisp
3177 ** late-target-type.lisp -> deftypes-for-target.lisp
3178 Furthermore, several other previously target-only files foo.lisp (e.g.
3179 hash-table.lisp and random.lisp) have been split into a target-and-host
3180 foo.lisp file and a target-only target-foo.lisp file, with their key type
3181 definitions in the target-and-host part, so that the cross-compiler will
3182 know more about target types.
3183 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3184 gone away. In their place are various *BACKEND-FOO* variables
3185 corresponding to the slots of the old structure.
3186 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3187 into the target SBCL has been fixed.
3188 * The chill.lisp system for loading cold code into a running SBCL
3190 * Support for the CMU CL "scavenger hook" extension has been removed.
3191 (It was undocumented and unused in the CMU CL sources that SBCL was
3192 derived from, and stale in sbcl-0.6.1.)
3193 * Various errors in the cross-compiler type system were detected
3194 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3195 (enabling various consistency checks). Many of them were fixed,
3196 but some hard problems remain, so the compiler is back to
3197 running without *TYPE-SYSTEM-INITIALIZED* for now.
3198 * As part of the cross-compiler type system cleanup, I implemented
3199 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3200 * I have started adding UNCROSS calls throughout the type system
3201 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3202 on cross-compiler input files will be able to go away, and various
3204 * CONSTANTP now returns true for quoted forms (as explicitly required
3207 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3209 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3210 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3211 (MAKE-LIST -1). In the longer term, it ought to become true
3212 that declarations are assertions even at SAFETY 1. For now, it's not
3213 quite true even at SAFETY 3, but it's at least more nearly true..
3214 (Note that this change seems to increases the size of the system by
3215 O(5%) and to decrease the speed of the compiler by 20% or more.)
3216 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3217 for the problem of printing dozens of lines of distracting information
3218 about low-level system machinery as part of the top stack frame
3219 on entry to the debugger when an undefined function was called.
3220 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3221 is not reset to COMMON-LISP-USER.
3222 * Compilation of stuff related to dyncount.lisp has been made conditional
3223 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3224 smaller. The various dyncount-related symbols have been moved into
3225 a new "SB-DYNCOUNT" package.
3226 * tty-inspect.lisp has been renamed to inspect.lisp.
3227 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3228 feature has gone away. (When we eventually port to other flavors of
3229 libc and/or Unix, we'll try to make the differences between flavors
3230 invisible at the user level.)
3231 * Various other *FEATURES* tags, and/or their associated conditionals,
3232 have been removed if obsolescent, or given better documentation, or
3233 sometimes given more-mnemonic names.
3235 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3237 * tidied up "make.sh" script
3238 * tidied up system directory structure
3239 * better "clean.sh" behavior
3240 * added doc/FOR-CMUCL-DEVELOPERS
3241 * many many small tweaks to output format, e.g. removing possibly-confusing
3242 trailing #\. character in DESCRIBE-INSTANCE
3243 * (EQUALP #\A 'A) no longer signals an error.
3244 * new hashing code, including EQUALP hashing
3245 * tidied up Lisp initialization and toplevel
3246 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3247 * command line argument processing
3248 * added POSIX-GETENV function to deal with Unix-ish environment variables
3249 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3250 terminating SBCL on EOF
3251 * non-verbose GC by default
3252 * There is no more "sbcl" shell script; the sbcl file is now the C
3253 runtime executable (just like CMU CL).
3254 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3256 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3257 symbol and package information as Lisp native symbol and package objects
3258 instead of strings naming symbols and strings naming packages. This way,
3259 whenever packages are renamed (as in warm init), debug information is
3260 transformed along with everything else.
3261 * tweaked the optimization policy declarations which control the building
3262 of SBCL itself. Now, among other things, the system no longer saves
3263 source location debugging information. (This helps two problems at once
3264 by reducing SBCL size and by keeping SBCL from trying to look for its
3265 sources -- which may not exist -- when reporting errors.)
3266 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3267 debugging and testing purposes
3268 * cleaned up printing, making the printer call PRINT-OBJECT for
3269 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3270 methods, giving nearly-ANSI behavior
3271 * converted almost all special variables to use *FOO* naming convention
3272 * deleted PARSE-TIME functionality, since it can be done portably
3273 * moved some files out of cold init into warm init
3274 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3276 * regularized formatting of source files
3277 * added an install.sh script
3278 * fixed ridiculous memory usage of cross-compiler by making
3279 compiler/alloc.lisp not try to do pooling unless it can hook
3280 itself into the GC of the cross-compilation host. Now the system
3281 builds nicely on my old laptop.
3282 * added :SB-ALLOC in target-features.lisp-expr
3283 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3284 was not implemented)
3285 * re-did condition handling and note reporting in the compiler. Notes
3286 are no longer handled by signalling conditions. Style warnings
3287 and warnings are handled more correctly and reported in such a way
3288 that it's easy to find one or the other in your output (so that you
3289 can e.g. figure out which of many problems caused COMPILE-FILE to
3291 * changed the severity of several compiler warnings from full WARNING
3292 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3293 compiler note reporting so that it doesn't use the condition system
3294 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3295 COMPILE-FILE command)
3296 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3297 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3298 no longer supported, since their implementation was deeply intertwingled
3299 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3300 * removed bogus "support" for compiler macros named (SETF FOO), and
3301 removed the compiler macro for SETF INFO (but only after making a fool
3302 of myself on the cmucl-imp mailing list by posting a bogus patch for
3303 DEFINE-COMPILER-MACRO..)
3304 * Compiled files containing forms which have side effects on the Lisp
3305 reader (such as DEFPACKAGE forms) are now handled more correctly.
3306 (Compiler queuing of top level lambdas has been suppressed by setting
3307 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3308 * deleted various currently-unused source files, e.g. gengc.lisp. They
3309 may be added back at some point e.g. when porting to other architectures,
3310 but until they are it's distracting to distribute them and to try to
3312 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3313 there were so many of them they're just distractions, and UNCROSS is
3314 known to be able to handle the current sources
3315 * moved PROFILE functionality into TRACE, so that it will be clear
3316 how the wrapping and unwrapping of functions when you profile them
3317 interacts with the wrapping and unwrapping of functions when you
3318 trace them. (Actually, the functionality isn't there yet, but at least
3319 the interface specification is there. Hopefully, the functionality will
3320 arrive with some future maintenance release.)
3321 * removed host-oops.lisp
3322 * changed signature of QUIT function to allow UNIX-CODE argument
3323 * fixed READ-SEQUENCE bug
3324 * tweaked verbose GC output so that it looks more like the progress
3325 output that ANSI specifies for functions like LOAD
3326 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3327 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob