1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
3 * New feature: Experimental support for bivalent streams: streams
4 opened with :element-type :default now allow character and binary
6 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
7 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
9 * bug fix: no more highly sporadic "couldn't check whether ~S is
10 readable" when reading a stream and an interrupt hits in the middle
11 of a select system call
12 * compiler better recognizes complex arrays (reported by Eduardo
14 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
15 for variables on SBCL built with linkage-tables. (reported by Luis
17 * various error reporting improvements.
18 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
19 (Thanks to Thiemo Seufer)
20 * bug fix: nested reader invokations work correctly
21 * bug fix: it is possible to have simultaneous references to foreign
22 code and foreign data with the same name.
24 ** added x86-64 support
25 ** incompatible change: the threading api now works with thread
26 objects instead of thread ids
27 ** bug fix: threads are protected from signals and interruption when
28 starting up or going down
29 ** bug fix: a race where an exiting thread could lose its stack to gc
30 ** bug fix: don't halt on infinite error in threads if possible
31 ** fixed numerous gc deadlocks introduced in the pthread merge
32 ** bug fix: fixed thread safety issues in read and print
33 ** bug fix: debugger doesn't hang on session lock if interrupted at
34 an inappropriate moment
35 ** bug fix: run-program is now thread safe(r)
36 ** bug fix: inner with-recursive-lock no longer releases the mutex
37 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
38 * fixed some bugs revealed by Paul Dietz' test suite:
39 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
40 EXPECTED-TYPE slots filled.
41 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
43 ** MAKE-CONDITION accepts classes as type-designators.
44 ** COMPILE may never return NIL.
45 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
46 range before calling Unix time functions
48 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
49 * numerous signal handling fixes to increase stability
50 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
51 * minor incompatible change: we now correctly canonize default
52 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
53 as per the MOP, rather than the historical (INITARG INITFUNCTION
54 INITFORM). (reported by Bruno Haible)
55 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
56 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
58 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
59 pathname is a directory pathname.
60 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
61 * dynamic space size on PPC has been increased to 768Mb. (thanks to
63 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
64 :METACLASS argument in addition to a class name. (reported by
65 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
66 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
67 (reported by Bruno Haible; patch from Gerd Moellmann)
68 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
70 * bug fix: more cleanups to the floating point exception handling on
71 x86-64 (thanks to James Knight)
72 * bug fix: the compiler does not try to constant fold calls of
73 COERCE and BIT-* functions when they return freshly constructed
74 objects. (reported by Nikodemus Siivola and Paul Dietz)
75 * optimization: improved the MIPS versions of generic (in the
76 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
77 * optimization: direct conversion of (unsigned-byte 32) to floats on
79 * optimization: structure instances with raw slots now use less
80 memory, and probably show better memory locality. (thanks to
82 * optimization: DYNAMIC-EXTENT declarations for lists and closures
83 are treated as requests for stack allocation on the x86-64,
84 Alpha-32, PPC and SPARC platforms.
85 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
86 the wrong file descriptor; implementation of SOCKET-OPEN-P.
87 (thanks to Tony Martinez)
89 ** gcing a dead thread can no longer lead to lockups
90 ** threads block signals until they are set up properly
91 ** errno is no longer shared by threads
92 ** interrupt-thread restores the eflags register on x86
93 ** fixed some lockups due to gc/thread interaction
94 * fixed some bugs revealed by Paul Dietz' test suite:
95 ** invalid dotted lists no longer raise a read error when
97 ** signal an error if a symbol that names a declaration is used
98 as the name of a type, or vice versa
99 ** allow using the (declare (typespec var*)) abbreviation for
100 (declare (type typespec var*)) with all type specifiers
101 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
102 ** always return NIL from PPRINT-POP when OBJECT is NIL
103 ** don't signal errors when pretty-printing malformed LABELS,
104 FLET or MACROLET forms
105 ** declarations in a DOLIST body are also in scope for the
107 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
109 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
112 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
113 * fixed cross-compiler leakages that prevented building a 32-bit
114 target with a 64-bit host compiler.
115 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
116 opened with :IF-EXISTS :OVERWRITE.
117 * fixed bug 281: error for an invalid qualifier in a short-form method
118 combination method is not signalled until the faulty method is called.
119 * bug fix: iteration variable type inferrer failed to deal with open
120 intervals. (reported by Alan Shields)
121 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
122 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
123 or more runtime options were provided to the sbcl binary.
124 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
125 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
126 * optimization: there's now a fast-path for fixnum arguments in the
127 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
128 * optimization: the code generated on x86-64 is more compact thanks
129 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
130 * bug fix: floating-point exception handling now partly works on
131 x86-64. (Thanks to James Knight)
132 * improvement to the MIPS backend from Thiemo Seufer: C-style
133 64-bit long long arguments and return values to alien functions
135 * fixed some bugs revealed by Paul Dietz' test suite:
136 ** the type-error signalled from WARN has a filled-in DATUM slot.
137 ** the type-error required when a stream is not associated with
138 a file has the stream as its datum.
139 ** type-errors on single-floats on x86-64 no longer have
140 :INVALID-OBJECT as the datum
141 ** the type-errors signalled for invalid function names now have
142 a correct expected type
143 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
144 for structures defined with a :TYPE. Documentation strings for
145 typed structures are no longer immediately discarded
146 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
147 broadcast streams and synonym streams. FILE-LENGTH now also works
148 on broadcast streams.
150 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
151 * incompatible change: the --noprogrammer option, deprecated since
152 version 0.7.5, has been removed. Please use the equivalent
153 --disable-debugger option instead.
154 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
156 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
157 inoperational for a while now) has been completely removed.
158 * null lexical environments are now printed as #<NULL-LEXENV>,
159 significantly reducing the amount of clutter in typical backtraces.
160 * documentation on weak pointers, finalization, and after GC hooks
161 has been added to the manual.
162 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
163 as other specialized array types, is much faster. SUBSEQ and
164 COPY-SEQ on such arrays have also been sped up.
165 * optimization: EQL is now more efficient when at least other argument
166 is known to be of type (OR FIXNUM (NOT NUMBER)).
167 * fixed bug: the runtime is now less vulnerable to changes in the
168 size of the SBCL object on OS X, and virtual memory is reserved for
169 all spaces that need to be at a fixed address.
170 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
171 * fixed bug: finalizers and after GC hooks that cause consing are now
173 * fixed bug: compiler error messages and summaries are now printed to
174 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
175 * fixed inference of the upper bound of an iteration variable.
176 (reported by Rajat Datta).
177 * fixed bug 211e: calling local functions with duplicated constant
178 keyword argument no longer causes a bogus style warning about an
180 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
181 ftype declaration to be disregarded. (reported by Dave Roberts)
182 * fixed bug 373: caused by erronous compilation of references to alien
183 variables in the runtime on ppc/darwin.
184 * fixed bug 376: CONJUGATE type deriver.
185 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
186 (reported by Baughn on #lisp)
187 * a cross-compiler bug on non-x86 platforms has been identified and
188 fixed. (thanks to Bruno Haible)
189 * improvements to the MIPS runtime code for increased stability.
190 (thanks to Thiemo Seufer)
191 * increased the maximimum compact environment size to allow
192 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
193 * partial workaround for bug 135: don't name the function we're
194 calling for hairy cases of EVAL, so as not to accumulate one environment
195 entry per call to EVAL. (reported by Kevin Reid)
196 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
197 * optimization: type testing for non-vector arrays should be faster.
198 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
200 * bug fix: setting 31st element of a bit vector to zero did not work
202 * bug fix: redefining a class definition which failed due to a
203 previous accessor / function clash now works (but see BUGS entry
204 #380 for more problems in this area). (thanks to Zach Beane)
205 * the long form of DEFINE-METHOD-COMBINATION disables method group
206 checking when given a single method group with pattern *. (thanks
208 * on x86 compiler supports stack allocation of results of simple
209 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
210 * support for building and running on Mac OS X 10.4 "Tiger" has been added
211 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
212 * fixed some bugs related to Unicode integration:
213 ** the restarts for recovering from input and output encoding
214 errors only appear when there is in fact such an error to
216 * fixed some bugs revealed by Paul Dietz' test suite:
217 ** MISC.549 and similar: late transformation of unsafe type
218 assertions into derived types caused unexpected code
220 ** SCALE-FLOAT type deriver is less wrong.
221 ** type derivers for EXP, LOG and similar functions compute result
222 types for complex arguments better.
223 ** (MISC.563) CONJUGATE type deriver works for very restricted
225 ** out-of-line type testers for character strings are available.
226 ** EQUAL compiler transform understands specialness of objects
228 ** accessing double-floats stored on the stack now works on x86-64.
229 ** debugger internals could sometimes create invalid lispobjs,
230 resulting in GC crashes.
231 ** MISC.548: type check weakening can convert required type into
233 ** initialization forms for bindings are not in scope of free special
236 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
237 * incompatible change: thread support for non-NPTL systems has
238 been removed - locking is buggy and unreliable. A threaded
239 SBCL build will now warn at startup and refuse to create
240 new threads, unless futex support is detected in the kernel
241 * incompatible change: the top level REPL now has only an ABORT
242 restart associated with it, not TOPLEVEL and ABORT as it used to.
243 TOP and TOPLEVEL are now available as debugger commands for
244 returning to the top level.
245 * incompatible change: forms evaluated in the REPL now use the
246 global optimization policy.
247 * incompatible change: user- and system-initialization files are
248 no longer processed with LOAD, but by READ and EVAL; hence the
249 global optimization policy, startup package, readtable, etc,
251 * The .fasl file format number has been incremented because of
252 various incompatible changes.
253 * internal entry point details and argument counts no longer appear
254 in backtraces unless explicitly requested by setting
255 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
256 * built-in and standard functions no longer have names like "top
257 level local call to FOO".
258 * fixed bug 32: functions defined in non-null lexical environments
259 now have more legible printed representation
260 * fixed bug 33: functions defined in non-null lexical environemnts
261 are now more amenable to inspection by INSPECT.
262 * workaround for bug 354: XEPs no longer appear in backtraces unless
263 explicitly requested.
264 * fixed bug: receiving the signal which results from attempting to
265 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
266 SIGBUS on other *BSDs) on architectures where the C stack is also
267 the Lisp stack (x86 and x86-64) from foreign code no longer leads
268 to debugger confusion or wild instability. (reported by Cheuksan
270 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
271 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
272 (reported by Lutz Euler)
273 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
274 compiled code. (reported by Paul Dietz)
275 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
276 specialized parameter to a method to something that is not TYPEP
277 the specializer is now possible.
278 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
279 face of package deletion.
280 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
281 pathname. (reported by David Tolpin on c.l.l)
282 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
283 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
284 STRING-STREAM and FILE-STREAM, respectively.
285 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
286 than previously; in addition, its test suite is now run on build.
287 (thanks to Luke Gorrie)
288 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
289 * optimization: added a immediate representation for single-floats
291 * fixed some bugs revealed by Paul Dietz' test suite:
292 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
293 ** attempting to create a package with a colliding nickname causes
294 correctable errors to be signalled.
295 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
296 ** several x86-64 backend bugs related to sign-extension of immediate
299 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
300 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
301 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
303 * fixed loading of multiply forward-referenced layouts.
304 (thanks to Cheuksan Wang)
305 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
306 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
307 (see COMPILE-FILE documentation for details of :PRINT option)
308 * fixed bugs 19 and 317: fixed-format floating point printing is
309 more accurate. This also fixes a bug reported by Adam Warner
310 related to the ~@F format directive.
311 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
313 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
314 dispatch table if the from-char is a dispatch macro character.
315 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
316 by the wordsize no longer ignore the last word. (reported by Lutz
318 * fixed bug in type checking of dynamic-extent variables. (reported
320 * optimization: sequence traversal functions use their freedom to
321 coerce function designators to functions.
322 * optimization: code with many calls to CLOS methods specialized on
323 CLOS classes has had redundant type checks removed.
324 * optimization: added declarations to speed up operations that access
325 the internal character database (for example STRING-UPCASE)
326 * optimization: comparison operations between floats and sufficiently small
327 fixnums no longer create extra rationals
328 * fixed some bugs related to Unicode integration:
329 ** portions of multibyte characters at the end of buffers for
330 character-based file input are correctly transferred to the
331 start of the buffer at the next read.
332 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
333 passing it through to OPEN.
334 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
335 argument given to internal calls to OPEN.
336 ** the built-in comment readers (introduced by character sequences
337 ";" and "#|") are more forgiving to encoding errors; they will
338 STYLE-WARN and then attempt to resync the stream at a character
339 boundary. (thanks to Teemu Kalvas)
340 * fixed some bugs in the x86-64 port:
341 ** Negative short int return values from c-calls are sign-extended
343 ** The stack is aligned to 16-bytes for c-calls, as required by
344 the ABI. (thanks to Cheuksan Wang)
345 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
346 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
347 for enums. (thanks to Vincent Arkesteijn)
348 ** Multiple small optimizations and bugfixes for floating point
350 * fixed some bugs revealed by Paul Dietz' test suite:
351 ** Space, Tab, Linefeed, Return and Page have the invalid
352 secondary constituent character trait.
353 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
355 ** WITH-INPUT-FROM-STRING should only update the index place on
357 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
359 ** Bit-array operations (BIT-AND and similar) worked incorrectly
360 with one-dimensional arrays with fill pointers.
361 ** TYPE-OF failed on a complex with an integer realpart and a
363 ** compiler failure during type inference for the code of form
364 (IF (EQL X (THE ...)) ...) (MISC.535).
366 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
367 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
368 platform. The port supports 61-bit fixnums, large memory spaces
369 and reloading shared object files.
370 * enhancement: saving cores with foreign code loaded is now
371 supported on ppc/Darwin in addition to the previously supported
373 * enhancement: the statistical profiler now walks deeper into the
374 call stack for more meaningful call-graphs and accrued time
375 reports (x86/x86-64 only). It also now reports time spent in
377 * enhancement: it is now possible to trace most individual methods
378 of a generic function in addition to tracing the generic function
380 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
382 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
383 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
384 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
385 produces an error. (thanks to Vincent Arkesteijn)
386 * bug fix: NAMESTRING on pathnames with :WILD components in their
387 directories works correctly. (thanks to Artem V. Andreev)
388 * fixed bug 125: compiler preserves identity of closures. (reported
390 * bug fixed: functions with &REST arguments sometimes failed with
391 "Undefined function" when compiled with (DEBUG 3). (reported by
393 * bug fix: overflow during compiling of setting element of a bit
394 vector with constant index and value. (reported by Timmy Douglas)
395 * build fix: fixed the dependence on *LOAD-PATHNAME* and
396 *COMPILE-FILE-PATHNAME* being absolute pathnames.
397 * on x86 compiler partially supports stack allocation of dynamic-extent
399 * GO and RETURN-FROM do not check the extent of their exit points
400 when compiled with SAFETY 0.
401 * fixed some bugs related to Unicode integration:
402 ** encoding and decoding errors are now much more robustly
403 handled; it should now be possible to recover even from invalid
404 input or output to the terminal. (thanks to Teemu Kalvas)
405 ** provided a first cut at implementing STRING-TO-OCTETS and
406 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
407 ** altered the SB-MD5 contributed module slightly, changing the
408 interface just enough for it to be supportable for builds where
409 lisp characters are not eight bits.
410 * fixed some bugs revealed by Paul Dietz' test suite:
411 ** the FORMATTER-generated functions for ~V[ conditionals require
412 the correct number of arguments.
413 ** READ-FROM-STRING returns the mandated second value when applied
414 to displaced strings.
415 ** the #\Rubout and #\Backspace characters are treated as invalid
416 constituent characters by the tokenizer.
418 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
419 * new feature: reloading changed shared object files with
420 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
421 * new feature: references to foreign variables and functions
422 can now be compiled and loaded before the corresponding shared
423 object file is loaded, as long as the foreign definitions are
424 available at runtime.
425 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
426 supported. (thanks to Dan Debertin)
427 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
428 just keywords. (thanks to Vincent Arkesteijn)
429 * fixed bug #331: structure-class instances corresponding to
430 DEFSTRUCT forms are now created eagerly.
431 * fixed bug #345: backtraces from calls to undefined functions work
432 on x86 as well. Related bug #61 is now also partially fixed on x86
433 (backtraces from throws to unknown catch tags.)
434 * bug fix: lambda-list parsing is now stricter vrt. order and number
435 of lambda-list keywords.
436 * bug fix: as specified by AMOP, an error is signalled if a
437 class-option appears multiple times in a DEFCLASS form. (reported
439 * bug fix: RANDOM can be compiled when the compiler derives the type
440 of its numeric argument as a disjoint set of small integers.
441 (reported by Paul Dietz)
442 * bug fix: starting a core saved with shared objects loaded when
443 those objects are not available no longer causes threaded SBCL to
444 hang. (reported by Sean Ross)
445 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
447 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
448 argument, which must always be NIL. (reported by Kalle Niemitalo)
449 * bug fix: printing 1.0d+23 no longer results in an error.
450 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
451 * bug fix: structure slot setters preserve evaluation order. (thanks
453 * fixed some bugs related to Unicode integration:
454 ** RUN-PROGRAM can allow its child to take input from a Lisp
455 stream. (reported by Stefan Scholl)
456 ** ASDF-INSTALL successfully downloads .tar.gz files in all
457 locales. (reported by Ken Causey)
458 * fixed some bugs revealed by Paul Dietz' test suite:
459 ** INCF, DECF and REMF evaluate their place form as specified in
461 ** FORMATTER expands ~{ ~} iteration directives with V or #
462 parameters correctly.
463 ** FORMATTER deals with the ~@[ ~] conditional directive where the
464 consequent uses no arguments correctly.
465 ** the system has a partial understanding of the (COMPLEX RATIO)
467 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
470 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
471 * new feature: a build-time option (controlled by the :SB-UNICODE
472 keyword feature, enabled by default) for building the system with
473 support for the entire 21-bit character space defined by the
475 * new feature: the system now has rudimentary external-format
476 support; the primary user-visible change at this time is that
477 characters with the high bit set (such as lower-case-e-acute) will
478 print correctly to a terminal in a UTF-8 environment.
479 * minor incompatible change: BASE-CHAR no longer names a class;
480 however, CHARACTER continues to do so, as required by ANSI.
481 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
482 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
483 has been moved to the SB-EXT package (temporarily re-exported from
485 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
486 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
488 * fixed bug #351: better error-handlind and reporting for malformed
490 * fixed bug #350: bignum-printing is now more memory-efficient,
491 allowing printing of very large bignums, eg. (expt 2 10000000).
492 (reported by Bruno Haible)
493 * fixed bug #302: better primitive-type selection for intersection
495 * fixed bug #308: non-graphic characters now all have names, as
496 required. (reported by Bruno Haible)
497 * bug fix: redefining a class with different superclasses now correctly
498 removes it from the direct-subclasses of its previous superclasses.
499 (reported by David Morse)
500 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
501 new value now works. (reported by Bruno Haible)
502 * bug fix: correct canonicalization of multiple non-standard slot
503 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
504 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
505 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
506 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
507 now exists, an signals an error.
508 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
509 during macroexpansion for non-iteration variables as well. (reported
510 by Bruno Haible for CMUCL)
511 * bug fix: Cyclic structures and unprintable objects in compiler
512 messages no longer cause errors. (reported by Bruno Haible)
513 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
514 and READ-FROM-STRING all now return a primary value of NIL if
515 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
516 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
517 (reported by Bruno Haible for CMUCL)
518 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
519 do not designate a condition. (reported by Bruno Haible for
521 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
522 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
523 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
524 permitting resolution in favour of any of the conflicting symbols.
525 (reported by Bruno Haible for CMUCL)
526 * bug fix: EQUAL compiler optimizations is less aggressive on
527 strings which can potentially compare true despite having distinct
528 specialized array element types.
529 * bug fix: unit enumerations can be defined without dividing by
530 zero. (thanks to Vincent Arkesteijn)
531 * FORMAT compile-time argument count checking has been enhanced.
532 (report from Bruno Haible for CMUCL)
533 * a partial workaround for the bug 262: the compiler does not try to
534 inline-expand a local function doing RETURN-FROM from a deleted
535 BLOCK. (thanks to Peter Denno for the bug report and to David
536 Wragg for the simple test case)
537 * fixed some bugs revealed by Paul Dietz' test suite:
538 ** RENAME-PACKAGE allows all package designators as new package
540 ** constraint propagation and lambda variable substitution are
541 more cautious in dealing with partially deleted code.
542 ** compiler sometimes lost reoptimization passes.
543 ** CERROR, given a condition as condition designator, passes its
544 remaining arguments to the continue format control without
546 ** Case-altering FORMAT directives work correctly on non-ASCII
548 ** The REAL type specifier handles bounds outside the floating
549 point ranges without signalling FLOATING-POINT-OVERFLOW.
550 ** Functions with IR1-transformations can create intercomponent
551 references to global functions.
552 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
554 ** FORMAT ~R treats a nil value for its first parameter correctly.
556 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
557 * enhancement: saving cores with foreign code loaded is now
558 supported on x86/NetBSD and sparc/Linux in addition to the previously
560 * bug fix: on some platforms repeated installations caused multiple
561 copies of HTML documentation to be installed -- should not happen
562 any more. (reported by Stefan Scholl)
563 * bug fix: parsing self-recursive alien record types multiple times
564 no longer causes infinite recursion. (reported by Thomas F. Burdick,
565 original patch by Helmut Eller for CMUCL)
566 * bug fix: stack-exhaustion detection works now on NetBSD as well.
567 (thanks to Richard Kreuter)
568 * bug fix: defining classes whose accessors are methods on existing
569 generic functions in other (locked) packages no longer signals
570 bogus package lock violations. (reported by François-René Rideau)
571 * bug fix: special variables as DEFMETHOD parameters no longer have
572 associated bogus type declarations. (reported by David Wragg and
574 * bug fix: read-write consistency on streams of element-type
575 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
576 * bug fix: redefiniton of the only method of a generic function with
577 no DEFGENERIC no longer emits a full WARNING. In addition,
578 redefinition of generic functions with no DEFGENERIC to an
579 incompatible lambda list now signals an error. (thanks to Zach
581 * bug fix: DEFGENERIC now works even when there's a function of the
582 same name in an enclosing lexical environment. (thanks to Zach
584 * fixed compiler failure, caused by instrumenting code during
585 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
586 * optimization: added loop analysis and improved register allocation
587 * fixed some bugs revealed by Paul Dietz' test suite:
588 ** POSITION on displaced vectors with non-zero displacement
589 returns the right answer.
590 ** (SIMPLE-STRING) is a valid type specifier for sequence
592 ** *PRINT-LEVEL* handling for slotless structures is pedantically
594 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
595 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
597 ** The justification version of the FORMAT ~< directive treats
598 non-zero minpad parameter correctly.
600 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
601 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
602 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
603 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
604 the supported interface.
605 * new feature: Single-stepping of code compiled with DEBUG 2 or
606 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
607 * new feature: saving cores with foreign code loaded is now
608 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
609 Timothy Moore's work for CMUCL)
610 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
611 parameters to * instead of NIL if no initform is supplied.
612 (reported by Johan Bockgaard)
613 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
614 correctly when FUNCALL appears as the car of the form. Note:
615 despite this FUNCALL forms are not currently subject to
616 compiler-macro expansion. (port of Raymond Toy's fix for the
617 same from CMUCL, reported by Johan Bockgaard)
618 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
619 (thanks for Teemu Kalvas)
620 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
621 case instead of terminating the iteration (thanks for Julian
622 Squires, Sean Champ and Raymond Toy)
623 * bug fix: incorrect expansion of defgeneric that caused a style
624 warning. (thanks for Zach Beane)
625 * bug fix: slot accessor effective method computation works properly
626 for classes with multiple non-standard applicable methods on
627 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
628 * on x86 compiler supports stack allocation of results of LIST and
629 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
630 CMUCL implementation by Gerd Moellmann)
631 * fixed some bugs revealed by Paul Dietz' test suite:
632 ** FORMAT strings with both the ~<~:;~> form of the justification
633 directive and pretty-printing directives cause an error.
635 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
636 * incompatible change: the internal functions
637 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
638 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
639 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
640 instead of the old functions.
641 * new feature: on platforms where "dladdr" is available foreign
642 function names now appear in backtraces. (based on Helmut Eller's
644 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
645 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
647 * documentation: ASDF manual is now distributed with SBCL.
648 * bug fix: SBCL can now load its contributed modules with REQUIRE
649 even if the system-provided entries have been removed from
650 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
652 * bug fix: backtraces involving undefined functions or assembly
653 routines are more informative. (thanks to Brian Downing)
654 * bug fix: mutually referent alien structures now work correctly.
655 (reported by Rick Taube)
656 * bug fix: structures defined by WITH-ALIEN can be referred to
657 within other definitions in the same WITH-ALIEN.
658 * bug fix: division operators (MOD, TRUNCATE and the like) with
659 constant zero divisors and integer dividends no longer generate
661 * bug fix: provide default methods for INPUT-STREAM-P and
662 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
663 * bug fix: improve the detection and resolution of MOP metacycles
664 (where effective method computation depends on the generic function
665 having its effective method computed). (reported by Bruno Haible)
666 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
667 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
668 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
669 * bug fix: pathnames with relative directory components can be
670 represented relative to default pathnames.
671 * optimization: in taking the GCD of bignums, reduce the two bignums
672 to approximately the same size (using Euclid's algorithm) before
673 applying the more sophisticated binary GCD. (thanks to Juho
675 * optimization: COUNT on bitvectors now operates word-at-a-time.
676 * optimization: ASH with a positive, but not necessarily constant,
677 (leftwards) shift, when in a modular context, is compiled to a
679 * fixed some bugs revealed by Paul Dietz' test suite:
680 ** FORMAT variable parameters ("~V<char>") are defaulted properly
681 if the corresponding argument is NIL.
682 ** FORMAT directives accepting a minpad parameter treat negative
684 ** literal commas as character parameters to FORMAT directives are
686 ** literal spaces directly after ~<Newline> directives within a
687 format-logical-block (~:< ~@:>) do not induce :FILL-style
688 conditional newlines.
689 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
690 as specified: it no longer includes conditional newlines).
691 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
693 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
694 :PER-LINE-PREFIX argument does not evaluate to a string.
696 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
697 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
698 the manual for details; for now, package locks can be disabled by
699 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
700 if no major problems are found then it is likely that they will be
701 compiled in unconditionally.
702 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
703 unsupported operators on all platforms. To load a shared library
704 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
705 object file, link it into a shared library outside of SBCL and
706 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
708 * bug workaround: ROOM T and the GC were not working together
709 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
710 reporting facility aren't preserved. That reporting has been
711 disabled, so now until and unless someone figures out how to make it
712 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
713 * minor incompatible change: as threatened around sbcl-0.8.0, the
714 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
715 The symbols are also exported from SB-PCL for backwards
716 compatibility, but more so than before SB-PCL should be treated as
717 an implementation-internal package.
718 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
720 * the SB-SPROF contrib now works on (most) non-x86 architectures.
721 It is known as of this release not to work on the Alpha, however.
722 * fixed bug #167: errors signalled due to illegal syntax in method
723 bodies are now more legible.
724 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
725 designators and can hence be used with TYPEP.
726 * fixed bug #333: CHECK-TYPE now ensures that the type error
727 signalled, if any, has the right object to be accessed by
728 TYPE-ERROR-DATUM. (reported by Tony Martinez)
729 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
730 5.1.2.3 for multiple-value place subforms. (reported by Kalle
732 * fixed bug #334: programmatic addition of slots using specialized
733 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
734 and :ALLOCATION :CLASS slots.
735 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
736 float exponents. (rereported by Peter Seibel)
737 * fixed a bug: #\Space (and other whitespace characters) are no
738 longer considered to be macro characters in standard syntax by
740 * fixed bug: initialization of condition class metaobjects no longer
741 causes an instance of the condition to be created. (reported by
743 * fixed bug: it is now possible to have slots such that
744 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
745 description is neither :INSTANCE nor :CLASS.
746 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
747 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
748 SLOT-VALUE-USING-CLASS), and no longer causes errors with
749 non-standard SLOT-DEFINITION-ALLOCATION values.
750 * fixed bugs: various race conditions handling exiting threads.
751 CL-PPCRE's thread torture test now passes.
752 * fixed arguably-a-bug: GC time proportional to number of threads in
753 system even when most of them are idle
754 * optimization: improved performance of BIT and SBIT on bit-vectors.
755 * .fasl file incompatibility: The fasl file version number has
756 been incremented because of changes associated with package locks.
758 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
759 * minor incompatible change: the system no longer provides
760 optimization hints (conditions which are TYPEP
761 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
762 mode; these hints are still emitted when higher SPEED optimization
764 * new contrib module: a sampling profiler (profiling by statistical
765 sampling, rather than by instrumenting functions) is available as
766 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
767 * the behaviour of the standard function ED is now customizeable by
768 third parties through a hook variable: see ED's documentation
769 string for information on the protocol.
770 * the compiler no longer emits efficiency notes for (FUNCALL X)
771 when the type of X is uncertain under default optimization
773 * fixed bug 276: mutating a binding of a specialized parameter to a
774 method to something that is not TYPEP the specializer is now
776 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
777 exhibit the required behaviour on the x86 platform. (thanks to
778 Peter van Eynde, Eric Marsden and Bruno Haible)
779 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
780 even for difficult arguments. (reported by Peter Graves)
781 * fixed bug 141a: the backquote printer now descends quoted
783 * fixed another bug in backquote printing: no more destructive
784 modification of the form's list structure. (reported by Brian
786 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
787 move between its address being taken and the call to
788 interrupt_thread, fixing a crashing race condition.
789 * the SB-POSIX contrib implementation has been adjusted so that it
790 no longer exhibits ridiculously poor performance when constructing
791 instances corresponding to C structs.
793 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
794 * minor incompatible change: the sb-grovel contrib now treats C
795 structures as alien (in the sense of SB-ALIEN) objects rather than
796 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
797 has implications for memory management of client code
798 (sb-grovel-returned objects must now be manually managed) and for
799 type safety (alien objects now have full types).
800 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
801 used to control emission of compiler diagnostics, rather than the
802 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
803 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
804 quality should be considered deprecated.
805 * install.sh now installs the user manual as well
806 * (not quite a new documentable feature, but worth considering in
807 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
808 beginnings of a semantically meaningful condition hierarchy is
809 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
811 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
812 designator as the defaults argument.
813 * fixed bug: Displaced arrays whose displaced-to array has become
814 too small now cause ARRAY-DIMENSION to signal an error, providing
815 for safer bounds-checking. (reported by Bruno Haible)
816 * fixed bug: DEFCLASS slot definitions with identical :READER and
817 :WRITER names now signal a reasonable error. (reported by Thomas
819 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
821 * fixed bug: Invalid slot specification errors now print correctly.
822 (thanks to Zach Beane)
823 * fixed bug 320: Shared to local slot value transfers in class
824 redefinitions now happen corrently from superclasses as
825 well. (reported by Bruno Haible)
826 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
828 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
829 improper lists correctly. (reported by Bruno Haible)
830 * fixed bug 313: source-transform for <fun-name> was erroneously
831 applied to a call of a value of a variable with name <fun-name>.
832 (reported by Antonio Menezes Leitao)
833 * fixed bug 307: The obsolete instance protocol ensures that
834 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
835 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
836 remove dead unknown-values globs from the middle of the stack.
837 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
838 easier to resume long computations after using *BREAK-ON-SIGNALS*
839 to diagnose and fix failures. (thanks to Nikodemus Siivola)
840 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
841 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
842 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
843 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
844 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
846 * fixed a (fairly theoretical) bug in string printing: if
847 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
848 does not have array-element-type equal to the most general string
850 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
851 function name. (thanks to Nikodemus Siivola)
852 * fixed bug: compile-time format string checker failed on
853 non-closed ~{. (reported by Thomas F Burdick)
854 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
855 don't warn on legal loop code involving more than one aggregate
856 boolean. (thanks to Nikodemus Siivola)
857 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
858 in keyword position to MAKE-PACKAGE was not being reported as
860 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
861 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
863 * fixed bugs 280 and 312: the checking for multiple definitions in a
864 file is less likely to become confused by uses of inline
866 * fixed bug: the #S reader macro performs the keyword coercion
867 specified for slot names. (reported by Kalle Niemitalo)
868 * fixed bug: lambda lists may contain symbols whose names start with
869 & but are not lambda-list-keywords; their occurrence triggers a
871 * fixed bug 321: define-method-combination argument lambda lists do
872 not cause invalid code to be generated when &optional or &aux
873 variables have default values. (reported by Bruno Haible)
874 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
875 have CLOS classes; this appears also to have fixed reported
876 instability in stack exhaustion detection.
877 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
878 MUFFLE-WARNING when no associated restart is present is now
880 * optimization: rearranged the expansion of various defining macros
881 so that each expands into only one top-level form in a
882 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
884 * optimization: used a previously little-used slot in symbols to
885 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
887 * fixed some bugs revealed by Paul Dietz' test suite:
888 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
890 ** Fixed an optimization bug related to inheritance of initforms
891 from local to shared slots.
892 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
893 ** CAST optimizer forgot to flush argument derived type.
894 ** print/read consistency on floats is now orders of magnitude
895 more likely. (thanks also to Bruno Haible for a similar report
897 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
898 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
899 for objects of type REAL. Make it so.
900 ** SXHASH is specified (infelicitously) to respect similarity,
901 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
902 it so. (thanks to Markus Ziegler)
903 ** on the Alpha, the compiler succeeds in compiling functions
904 returning a known number of arguments greater than 63.
905 ** fixed handling of invalid NIL arguments in keyword position
907 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
909 ** provided a readably-printable representation for RANDOM-STATE
911 ** ensured that pathnames, where they have a namestring, always
912 print using #P"..." syntax.
914 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
915 * Support for the forthcoming 2.0 version of the NetBSD kernel
916 running on x86 hardware has been added. (thanks to Perry
917 E. Metzger most immediately, and others for their past work)
918 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
919 current release version; SBCL's OpenBSD support had been broken
920 since about the time of OpenBSD's switch to ELF binary format.)
921 * [placeholder for DX summary]
922 ** user code with &REST lists declared dynamic-extent, under high
923 speed or space and low safety and debug optimization policy.
924 * The manual has been converted to Texinfo format and the debugger
925 chapter from the cmucl manual has been added.
926 * A facility has been added to extract documentation strings from
927 sbcl and store them as Texinfo-formatted snippets for inclusion in
928 the manual (via Texinfo's @include directive)
929 * bug fix: compiler emitted division in optimized DEREF. (thanks for
930 the test case to Dave Roberts)
931 * bug fix: multidimensional simple arrays loaded from FASLs had fill
932 pointers. (reported by Sean Ross)
933 * bug fix: PROFILE output is printed nicely even for large numerical
934 values. (thanks to Zach Beane)
935 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
936 greater than 32 handle EOF correctly.
937 * bug fix: on X86 an immediate argument of the IMUL instruction is
938 correctly printed in disassembly. (thanks to Lutz Euler)
939 * bug fix: class slots in redefined classes preserve their old
940 values. (thanks to Bruno Haible and Nikodemus Siivola)
941 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
942 succeeds. (reported by Marco Baringer)
943 * fixed some bugs revealed by Paul Dietz' test suite:
944 ** READ-SEQUENCE now works on ECHO-STREAMs.
945 ** RATIONALIZE works more according to its specification. (thanks
948 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
949 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
950 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
951 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
952 should matter to you only if you rebind the printer control
953 variables and then find you want different bindings in the
954 debugger than in the ordinary execution of your program.)
955 * The runtime build system has been tweaked to support building
956 (on SPARC/SunOS) using a C compiler which invokes Sun's own
957 assembler and linker. (thanks to Nikodemus Siivola)
958 * Unbound, undefined, undeclared variables now trigger full
959 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
960 more often programmer error than deliberate exploitation of undefined
962 * optimization: the hash algorithm for strings has changed to one
963 that is less vulnerable to spurious collisions. (thanks to Juho
965 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
966 needless bounds checking. (thanks to Juho Snellman)
967 * optimization: implemented multiplication as a modular
968 (UNSIGNED-BYTE 32) operation on the PPC backend.
969 * fixed some bugs revealed by Paul Dietz' test suite:
970 ** ADJUST-ARRAY now copies the datum in a zero rank array if
972 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
973 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
974 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
976 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
978 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
980 ** ECHO-STREAMs no longer attempt to echo the end of file value to
981 their output stream on EOF from read.
982 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
983 have been read to end-of-file.
984 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
986 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
987 description of determination of which consecutive characters
989 ** Printing the "Space" character with escaping on now yields "#\\ ",
990 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
991 ** Reading floating-point numbers with *READ-BASE* set to a number
992 less than 10 works correctly.
993 ** Reading floating-point numbers with *READ-BASE* set to a number
994 more than 10 works correctly.
995 ** Printing with *PRINT-READABLY* targets the standard readtable, not
996 the readtable currently in effect.
998 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
999 * minor incompatible change: parsing of namestrings on a physical
1000 (Unix) host has changed; numbers after the final #\. in a
1001 namestring are no longer interpreted as a version field. This is
1002 intented to be largely invisible to the user, except that the
1003 meaning of the namestring "*.*.*" has changed: it now refers to a
1004 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1005 should usually be replaced by
1006 (make-pathname :name :wild :type :wild :version :wild)
1007 with the added benefit that this is more likely to be portable.
1008 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1009 option to OPEN now signals an error if the file being opened
1010 exists; this may have an impact on existing code.
1011 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1012 signals received, on the PowerPC platforms (both Linux and
1013 Darwin). (thanks to Pierre Mai for pointing out the location of
1015 * several fixes on OS X: The system now builds and runs cleanly on
1016 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1017 (thanks to Brian Mastenbrook)
1018 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1019 second-resolution: integer multiples of 1/3600 between -24 and 24.
1020 (thanks to Vincent Arkesteijn)
1021 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1022 type when called with 1 argument; PEEK-CHAR checked type of
1023 PEEK-TYPE only after having read first character from a
1024 stream. (reported by Peter Graves)
1025 * bug fix: the garbage collector now has much better locality
1026 behaviour, and in particular no longer treats all memory as being
1027 exhausted when many small objects point to each other in a deeply
1029 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1030 recognized as being TYPEP their class.
1031 * bug fix: the PUSHNEW documentation string has been corrected.
1032 (thanks to Vincent Arkesteijn)
1033 * bug fix: defaulting of the value for the last of an atypically
1034 large number of multiple values being bound was not being
1035 performed correctly on the Alpha or PPC platforms
1036 * optimization: implemented multiplication as a modular
1037 (UNSIGNED-BYTE 32) operation on the x86 backend.
1038 * optimization: SEARCH on simple-base-strings can now be open-coded.
1039 (see also contrib/compiler-extras.lisp for inspiration for
1040 teaching the compiler about the Boyer-Moore algorithm).
1041 * value, returned by MAX (and MIN) called with several EQUALP, but
1042 not EQL, arguments now does not depend on compiler settings.
1043 * fixed some bugs revealed by Paul Dietz' test suite:
1044 ** in stack analysis liveness information is propagated from
1045 non-local entry points.
1046 ** pathwise CAST removing failed when the CAST node did not start
1048 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1049 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1051 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1052 host is already defined.
1053 ** RENAME-FILE works on streams instead of signalling an internal
1055 ** PEEK-CHAR uses the current readtable when determining whether
1056 or not a character is whitespace.
1057 ** MERGE-PATHNAMES handles the case when the pathname does not
1058 specify a name while the default-pathname specifies a version.
1059 ** Pathnames now stand a better chance of respecting print/read
1061 ** Attempting to use standardized file system operators with a
1062 pathname with invalid :DIRECTORY components signals a
1064 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1066 ** DIRECTORY on logical pathnames is more correct.
1067 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1068 signal a TYPE-ERROR if their argument is not a stream.
1069 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1070 designator argument does not designate a stream.
1071 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1072 examining the synonym.
1073 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1075 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1076 element-type larger than ([UN]SIGNED-BYTE 32).
1078 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1079 * When built with the :SB-FUTEX feature, threaded builds now take
1080 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1081 for faster/more reliable mutex and condition variable support.
1082 * Incompatible change (but one you probably shouldn't have been using
1083 anyway): the interface and code for arbitrating between multiple
1084 threads in the same user session has been redesigned.
1085 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1086 error when called without an explicit environment argument.
1087 (thanks to Nikodemus Siivola)
1088 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1089 confused about their position. (thanks to Adam Warner and Gerd
1091 * bug fix: Pretty printing backquoted forms with unquotations in the
1092 argument list position of various code constructs such as LAMBDA
1093 now works correctly. (reported by Paul Dietz)
1094 * bug fix: Pretty printing unquotations no longer loses all
1095 stream position information.
1096 * optimization: performance of string output streams is now less
1097 poor for multiple small sequence writes.
1098 * optimization: performance of CSUBTYPEP in the presence of complex
1099 expressions involving CONS and NOT many times has been improved.
1100 (reported by Paul Dietz)
1101 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1103 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1104 streams for the REPL, for the debugger, and so on. (thanks to
1106 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1107 * fixed some bugs revealed by Paul Dietz' test suite:
1108 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1109 is ignored if it is too small, rather than propagating through
1110 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1111 ** extremely complex negations of CONS types were not being
1112 sufficiently canonicalized, leading to inconsistencies in
1114 ** VALUES tranformer lost derived type.
1116 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1117 * fixed a bootstrapping bug: the build process no longer assumes
1118 that the various BOOLE-related constants have the same value in
1119 host and target lisps. (noted by Paul Dietz' test suite on an
1120 SBCL binary built from CLISP)
1121 * The system can now be dynamically linked on the MIPS platform,
1122 which enables dynamic loading of foreign code from Lisp. (thanks
1123 to Ralf Baechle for discussions on the MIPS ABI)
1124 * The system now records debugging information for its own source
1125 files in a filesystem-position-independent manner, relative to
1126 the "SYS" logical host.
1127 * fixed a compiler bug: MV-LET convertion did not check references
1128 to the "max args" entry point. (reported by Brian Downing)
1129 * tweaked disassembly notes to be less confident about proclaiming
1130 some instruction as an LRA. (thanks to Brian Downing)
1131 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1132 now each have their own history, command character, and other
1133 characteristics. (thanks to David Lichteblau)
1134 * fixed some bugs revealed by Paul Dietz' test suite:
1135 ** compiler failure in compiling LOGAND expressions including a
1137 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1138 shift greater than 32.
1139 ** FLUSH-DEST did not mark blocks for type check regeneration.
1140 ** HANDLER-CASE failed to accept declarations in handler clauses
1141 in some circumstances.
1143 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1144 * New code in contrib/sb-introspect (still probably not entirely
1145 stable yet) provides some support for smart Lisp development
1146 environments like SLIME.
1147 * The conditions signalled for errors occurring when loading .fasl
1148 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1149 in a way which should help ASDF recover gracefully.
1150 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1151 stuff has been cleaned up. If you code contrib/ stuff, this might
1152 affect you, and you can look at contrib/README, contrib/STANDARDS,
1153 and/or the 0.8.4.27 diff to check.
1154 * In full calls the compiler now does not generate checks for declared
1155 argument types for all arguments.
1156 * various threading fixes
1157 ** and some experimental patches which didn't make it into
1158 the main tree for this release, but which are shipped in
1159 contrib/experimental-thread.patch as a possible fix for some
1160 failures (deadlock, spinning...) in GC-intensive multithreaded
1162 * fixed PPC build problem (source code incompatibility of different
1163 library versions): added offsetof() hackery which attempts to divine
1164 where glibc maintainers put uc_mcontext today
1165 * fixed bug 282: compiler does not trust type assertions while passing
1166 arguments to a full call.
1167 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1168 &OPTIONAL VALUES type parameter.
1169 * fix bug 214: algorithm for noting rejected templates is now more
1170 similar to that of template seletion. (also reported by rydis on
1172 * fixed bug 141b: printing backquoted information readably and prettily
1173 inserts a space where necessary.
1174 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1175 (CHARACTER 10) are now reported as errors, rather than propagated
1176 as unknown types. (reported by piso on #lisp)
1177 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1178 with values NIL and :ERROR. (thanks to Milan Zamazal)
1179 * fixed bug 191c: CLOS now does proper keyword argument checking as
1180 described in CLHS 7.6.5 and 7.6.5.1.
1181 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1182 longer raise an error; further, using a list as a for-as-arithmetic
1183 counter now raises a meaningful error.
1184 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1185 now understood by sequence creation functions such as MAKE-SEQUENCE
1187 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1188 a STRING-INPUT-STREAM.
1189 * compiler enhancement: SIGNUM is now better able to derive the type
1191 * type declarations inside WITH-SLOTS are checked. (reported by
1193 * fixed some bugs revealed by Paul Dietz' test suite:
1194 ** incorrect optimization of TRUNCATE for a positive first
1195 argument and negative second.
1196 ** compiler failure in let-convertion during flushing dead code.
1197 ** compiler failure while deriving type of TRUNCATE on an
1198 interval, containing 0.
1199 ** ASH of a negative bignum by a negative bignum count now returns
1201 ** intersection of CONS types now canonicalizes properly, fixing
1202 inconsistencies in SUBTYPEP.
1204 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1205 * incompatible change: The --disable-debugger command line
1206 option now clobbers the debugger at a more fundamental
1207 level, by redefining #'INVOKE-DEBUGGER instead of by
1208 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1209 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1210 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1211 enter the debugger REPL and then suffer recursive errors
1212 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1213 while under the new system, BREAK in --disable-debugger mode
1214 terminates the system just as an unhandled error would.
1215 * fixed compiler performance when processing loops with a step >1;
1216 * bug fix: DOCUMENTATION now retrieves generic function
1217 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1218 support has been systematized, and now supports the methods
1219 specified by ANSI, along with a default method and a method for
1220 slot documentation. (reported by Nathan Froyd)
1221 * bug fix: effective methods associated with a generic function are
1222 no longer cached over a change of that generic function's method
1223 combination. (reported by Andreas Fuchs)
1224 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1225 names a non-existent directory. (thanks to Andreas Fuchs)
1226 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1227 the PPC platform, lead to stack corruption; this has been fixed.
1228 (reported by Rainer Joswig)
1229 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1230 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1231 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1232 arithmetic in general, is now much more reliable on the PPC
1234 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1235 platform now returns the right answer.
1236 * optimization: restored some effective method precomputation in
1237 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1238 precomputation is now tunable.
1239 * optimization: compiler-internal data structure use has been
1240 reviewed, and changes have been made that should improve the
1241 performance of the compiler by about 20%.
1242 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1243 simple-base-strings and simple-bit-vectors is improved.
1244 * optimization: the optimization of 32-bit logical and arithmetic
1245 functions introduced in version 0.8.3 on the x86 has been
1246 implemented on the mips, ppc and sparc platforms; an
1247 implementation of the same facility, but for 64-bit arithmetic,
1248 has been added for the alpha.
1249 * microoptimization: the compiler is better able to make use of the
1250 x86 LEA instruction for multiplication by constants.
1251 * bug fix: in some situations compiler did not report usage of
1252 generic arithmetic in (SPEED 3) policy.
1253 * bug 145b fix: compiler used wrong type specifier while converting
1254 MEMBER-types to numeric.
1255 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1257 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1258 index argument now works properly on the Alpha platform.
1259 * bug fix: floating point exception treatment on the Alpha platform
1261 * bug fix: FILE-POSITION works much better on string input and
1262 output streams. (thanks to Nikodemus Siivola)
1263 * bug fix: many threading/garbage collection symptoms sorted.
1264 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1265 might be pseudo-atomic.
1266 * internal change: Stopping for GC is now done with signals not
1267 ptrace. GC is now done in whichever thread wanted it, instead of
1269 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1271 * bug fix: result form in DO is not contained in the implicit
1273 * incompatible change: ICR structure is changed; the value part of
1274 CONTINUATION is now called LVAR; corresponding functions are
1275 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1276 * added type deriver for ISQRT (thanks to Robert E. Brown).
1277 * bug fix: better support for loading from the command line when an
1278 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1280 * fixed some bugs revealed by Paul Dietz' test suite:
1281 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1282 ** ROUND and FROUND now give the right answer when given very
1283 small float arguments.
1284 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1286 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1287 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1288 ** LCM should return a non-negative integer.
1289 ** PARSE-INTEGER returned the index of a terminator instead of the
1290 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1291 ** PARSE-INTEGER returned an incorrect index being applied to a
1293 ** LCM with two arguments of 0 returns 0 rather than signalling
1295 ** unsigned addition of a 32-bit constant with the high bit set no
1296 longer causes an internal compiler error on the x86.
1297 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1298 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1299 with negative last argument.
1300 ** byte specifiers with zero size and position no longer cause
1301 an error during type derivation.
1302 ** bignum multiplication on the Alpha platform now returns the
1304 * porting: The system now builds on SuSE AMD64, although it still
1305 generates a 32-bit binary.
1306 * .fasl file incompatibility: The fasl file version number has
1307 been incremented (because of the changes to internal compiler
1308 data structures referred to above).
1310 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1311 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1312 more accurately, on the Darwin kernel running on PowerPC hardware.
1313 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1314 * Compiler code deletion notes now signal a condition of type
1315 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1316 an associated MUFFLE-WARNING restart.
1317 * The compiler now performs limited argument count validation of
1318 constant format strings in FORMAT, and where appropriate in ERROR,
1319 CERROR and WARN. (thanks to Gerd Moellmann)
1320 * New ASDF-INSTALL contrib can be used for automatic download and
1321 installation of third-party Lisp code from CCLAN or other sites
1323 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1324 which forces another thread to execute a function supplied by the
1326 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1327 now accept and act upon their :ELEMENT-TYPE keyword argument.
1328 (reported by Martin Atzmueller, Edi Weitz)
1329 * bug fix: FILE-POSITION now accepts position designators up to
1330 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1331 is the greater. (thanks to Patrik Nordebo)
1332 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1333 Kalle Olavi Niemitalo)
1334 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1335 printable. (reported by Eric Marsden)
1336 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1337 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1338 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1339 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1340 (thanks to Stig Sandoe)
1341 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1342 UNREAD-CHAR now correctly includes the unread character in the
1343 target sequence. (thanks to Gerd Moellmann)
1344 * bug fix in threaded builds: the system can now be suspended and
1345 resumed by shell job control with minimal disruption.
1346 * bug fixes in times and timezones >2038AD
1347 * better handling of "where is GNU make?" problem in build scripts
1348 (thanks to Nikodemus Siivola)
1349 * new optimization: inside a named function any reference to a
1350 function with the same name is considered to be a self-reference;
1351 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1352 optimization quality.
1353 * new optimization on x86: logical functions and + now have
1354 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1355 used when the result is truncated to 32 bits.
1356 * VALUES declaration is partially enabled.
1357 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1358 * fixed some bugs revealed by Paul Dietz' test suite:
1359 ** The system now obeys the constraint imposed by
1360 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1361 types form a lattice under type intersection.
1362 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1363 ** ASSOC now ignores NIL elements in an alist.
1364 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1365 and (1+ MOST-POSITIVE-FIXNUM) answers.
1366 ** The addition of a method with invalid qualifiers to a generic
1367 function does not cause an error to be signalled immediately;
1368 a warning is signalled, and the error is generated only on
1369 calling the generic function.
1370 * changed .fasl file version number, in part to add type codes for
1371 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1372 obscure ANSI requirements
1374 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1375 * fixed bug 148: failure to inline-expand a local function left
1376 garbage, confusing the compiler.
1377 * fixed bugs 3cd: structure slot readers perform type check if the
1378 slot can have an invalid value (i.e. it is either not initialized
1379 or can be written with a less specific slot writer).
1380 * bug fix: the compiler now traps array references to elements off
1381 the end of an array; previously, the bounds checking in some
1382 circumstances could go off-by-one.
1383 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1385 * type declarations for array element types now obey the description
1386 on the CLHS page "Declaration TYPE", as per discussions on
1387 sbcl-help around 2003-05-08. This means that a declaration
1388 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1389 declaration, all references to BAR will be asserted or assumed
1390 (with THE, so dependent on compiler policy) to involve objects of
1391 type FOO. Note that no such declaration is implied in
1392 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1393 * declared types of functions from the "Conditions"
1394 chapter. (reported by Paul Dietz)
1395 * bug fix: CERROR accepts a function as its first argument.
1396 * bug fix: NTH an NTHCDR accept a bignum as index
1397 arguments. (reported by Adam Warner)
1398 * optimization: character compare routines now optimize comparing
1399 against a constant character. (reported by Gilbert Baumann)
1400 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1401 argument now works properly on the MIPS platform.
1402 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1403 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1404 declared to have a functional type, does not cause an error
1406 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1407 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1408 * fixed bug: if last continuation of a deleted block has a
1409 destination, this destination should be deleted too. (reported by
1411 * fixed a bug in the bootstrap process: the host compiler's values
1412 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1413 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1414 test case from Patrik Nordebo)
1415 * improved the ability of the disassembler on the PPC platform to
1416 provide helpful disassembly notes.
1417 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1418 the class in more cases than previously.
1419 * bug fix: FILE-POSITION now understands :START and :END for
1420 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1421 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1422 8) data. (thanks to David Lichteblau)
1423 * bug fix: it is possible to add a method to a generic function
1424 without lambda list.
1425 * bug fix: reader failed to signal END-OF-FILE inside an
1426 object representation. (reported by Nikodemus Siivola)
1427 * fixed some bugs revealed by Paul Dietz' test suite:
1428 ** LAST and [N]BUTLAST should accept a bignum.
1429 ** condition slot accessors are methods.
1430 ** (VECTOR NIL) is a subtype of STRING.
1432 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1433 * minor incompatible change: some nonsensical specialized lambda
1434 lists (used in DEFMETHOD) which were previously ignored now signal
1436 * minor incompatible change: the system is now aware of the types of
1437 variables in the COMMON-LISP package, and will signal errors for
1438 most violations of these type constraints (where previously they
1439 were silently accepted).
1440 * minor incompatible change: COMPILE-FILE now uses the freedom
1441 afforded (ANSI 3.2.2.3) to use derived function types for
1442 functions defined in the same file. This also permits the system
1443 to warn on static type mismatches and function
1444 redefinition. (Currently it does not work with high DEBUG level.)
1445 * minor incompatible change: VALUES declaration is disabled.
1446 * When issuing notes, the compiler now signals a condition of type
1447 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1448 restart for use in user handlers. It is expected that the
1449 COMPILER-NOTE condition will eventually become a condition
1450 supertype to a hierarchy of note types, which will then be
1451 handleable in a similar fashion. However, at the moment, no such
1452 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1453 but it's an implementation detail, not a classification for the
1455 * Changes in type checking closed the following bugs:
1456 ** type checking of unused values (192b, 194d, 203);
1457 ** template selection based on unsafe type assertions (192c, 236);
1458 ** type checking in branches (194bc).
1459 * A short form of VALUES type specifier has ANSI meaning (it has
1460 increased the number of situations when SBCL cannot perform type
1462 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1463 keywords or constants is permissible.
1464 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1465 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1466 classes. (thanks to Antonio Martinez)
1467 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1468 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1469 argument to a pretty-print stream. Instead, it leaves any such
1470 operations to DESCRIBE-OBJECT methods.
1471 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1473 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1474 lambda lists. (thanks to David Lichteblau)
1475 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1476 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1477 (thanks to Gerd Moellmann)
1478 * fixed bug 246: increased compilation speed of long
1479 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1481 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1483 * DEFCONSTANT now throws a condition of type
1484 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1485 constant to a non-EQL value; CONTINUE and ABORT restarts
1486 respectively change and preserve the value.
1487 * fixed bug 63: The code walker, part of the implementation of CLOS,
1488 is now better at handling symbol macros.
1489 * bug fix: There is no longer an internal implementation type named
1490 CL:LENGTH. (reported by Raymond Toy)
1491 * bug fix: In macro-like defining macros/special operators the
1492 implicit block does not enclose the lambda list.
1493 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1494 enclosed in parentheses) are not suitable as type specifiers, and
1495 their use properly signals an error now.
1496 * bug fix: An argument count mismatch for a type specifier in code
1497 being compiled no longer causes an unhandled error at compile
1498 time, but signals a compile-time warning.
1499 * fixed simple vector readable printing
1500 * bug fix: DESCRIBE takes more care over whether the class
1501 precedence list slot of a class is bound before accessing it.
1502 (reported by Markus Krummenacker)
1503 * bug fix: FORMATTER can successfully compile pretty-printer format
1504 strings which use variants of the ~* directive inside.
1505 * bug fix: SEARCH now applies its TEST predicate to the elements of
1506 the arguments in the correct order. (thanks to Wolfhard Buss)
1507 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1508 in type inference. (thanks to Robert E. Brown)
1509 * bug fix: Reading in symbols with an explicit package name of ""
1510 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1511 * fixed some bugs revealed by Paul Dietz' test suite:
1512 ** NIL is now allowed as a structure slot name.
1513 ** Arbitrary numbers, not just REALs, are allowed in certain
1514 circumstances in LOOP for-as-arithmetic clauses.
1515 ** Multiple class redefinitions before slot access no longer
1516 causes a type error.
1517 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1518 association between the name and a class.
1519 ** Generic functions with non-standard method-combination and over
1520 five methods all of which return constants no longer return NIL
1521 after the first few invocations. (thanks to Gerd Moellmann)
1522 ** CALL-NEXT-METHOD with no arguments now passes the original
1523 values of the arguments, even in the presence of assignment.
1524 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1526 ** DISASSEMBLE works with closures and funcallable instances.
1527 ** ADD-METHOD now returns the generic function, not the new method.
1528 ** FIND-METHOD signals an error if the lengths of the specializers
1529 is incompatible with the generic function, even if the ERRORP
1531 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1532 which its argument is a member.
1533 ** DEFCLASS only redefines the class named by its class-name
1534 argument if that name is the proper name of the class;
1535 otherwise, it creates a new class.
1536 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1537 of the UNBOUND-SLOT condition to the name of the slot.
1538 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1539 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1540 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1541 SLOT-MAKUNBOUND in the specified fashion.
1543 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1544 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1545 cross-compilation host. As a consequence, we can now bootstrap our
1546 way up to SBCL starting with a bare gcc toolchain and human-readable
1547 source code (first the source to CLISP, then the source to SBCL).
1548 * A contributed module containing a partial implementation of the
1549 simple-streams interface has been included. (thanks to Rudi
1551 * A contributed module implementing the RFC1321 Message Digest
1552 Algorithm, known as MD5, has been included.
1553 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1554 no longer has any effect, as the code controlled by this feature
1555 has been deleted. (As far as we know, no-one has ever built using
1556 this feature, and its semantics were confused in any case).
1557 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1558 work on conditions (as required by the ANSI specification),
1559 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1560 expected behaviour on conditions. Users should note, however,
1561 that such behaviour is not required by the ANSI specification,
1562 and so use of this behaviour may render their code unportable.
1563 * fixed some bugs revealed by Paul Dietz' test suite:
1564 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1566 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1567 ** SXHASH on condition objects no longer returns NIL.
1568 ** :ALLOCATION :CLASS slots are better treated; their values are
1569 updated on class redefinition, and initforms inherited from
1570 superclasses are applied.
1571 ** REMOVE-METHOD returns its generic function argument even when
1572 no method was removed.
1573 ** SHARED-INITIALIZE now initializes the values of the requested
1574 slots, including those with :ALLOCATION :CLASS.
1575 ** ALLOCATE-INSTANCE now works on structure classes defined via
1576 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1578 ** SLOT-EXISTS-P now works on conditions, as well as structures
1580 ** MAKE-LOAD-FORM now has the required methods on
1581 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1582 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1583 keyword, and now implements the SLOT-NAMES argument.
1584 ** methods with &OPTIONAL arguments no longer allow too many
1585 arguments to be passed in the call without error.
1586 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1587 option is consistent with the required arguments of the generic
1588 function lambda list.
1589 * bug fix: REQUIRE accepts a string designator. (Thanks to
1591 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1592 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1593 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1594 * bug fix: The compiler now checks for duplicated variables in macro
1596 * bug fix: SETQ on globals returns the correct value.
1597 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1598 not a COMPILER-ERROR (followed by some other strange error on
1599 choosing the CONTINUE restart).
1600 * bug fix: make.sh and friends are now more consistent in the way that
1601 they look for GNU "make".
1603 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1604 * experimental native threads support (on x86 Linux >=2.4 only).
1605 This is not compiled in by default: you need to add :SB-THREAD to
1606 the target features. See the "Beyond ANSI" chapter of the manual
1608 * fix for longstanding nonANSIism: The old distinction between
1609 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1610 The return value from CL:FIND-CLASS is now a CLOS class, and
1611 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1612 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1613 * An interface to the MetaObject Protocol, as described in Kiczales,
1614 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1615 MIT Press, 1991, is available from the SB-MOP package.
1616 * incompatible change: the SB-PCL package should now be considered
1617 a private implementation detail, and no longer a semi-private MOP
1619 * minor incompatible change: due to rearrangement for threads, the
1620 control stack and binding stack are now allocated at arbitrary
1621 addresses instead of being hardcoded per-port. Users affected by
1622 this probably have to be doing advanced things with shared
1623 libraries, and will know who they are.
1624 * minor incompatible change: Previously, all --eval forms used were
1625 processed with READ before any of them were processed with EVAL.
1626 Now each --eval form is processed with both READ and EVAL before
1627 the next --eval form is processed. (Thus package operations like
1628 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1629 work as the user might reasonably expect.)
1630 * minor incompatible change: *STANDARD-INPUT* is now only an
1631 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1633 * minor incompatible change: Y-OR-N-P is now character-oriented,
1634 not line oriented. Also, YES-OR-NO-P now works without errors.
1635 (thanks to Antonio Martinez)
1636 * sb-aclrepl module improvements: an integrated inspector, added
1637 repl features, and a bug fix to :trace command.
1638 * Known functions, which cannot be open coded by the backend, are
1639 considered to be able to check types of their arguments. (fixing
1640 a bug report by Nathan J. Froyd)
1641 * fixed a bug in computing method discriminating functions: It is
1642 now possible to define methods specialized on classes which have
1643 forward-referenced superclasses. (thanks to Gerd Moellmann)
1644 * fixed evaluation order in optional entries (reported by Gilbert
1646 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1647 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1648 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1649 required-by-AMOP NAME argument, as well as CLASS and
1650 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1651 * fixed bug 20: DEFMETHOD can define methods using names that are
1652 not the proper names of classes to designate class specializers.
1653 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1654 Unix file descriptors, instead of blowing up. (thanks to Antonio
1656 * Garbage collection refactoring: user-visible change is that a
1657 call to the GC function during WITHOUT-GCING will not do garbage
1658 collection until the end of the WITHOUT-GCING. If you were doing
1659 this you were probably losing anyway.
1660 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1661 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1662 TYPEP the latter but not the former.
1663 * The compiler issues a full WARNING for calls to undefined functions
1664 with names from the CL package.
1665 * MAP-INTO for a vector destination is open coded. (reported by
1666 Brian Downing on c.l.l)
1667 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1668 documentation string.
1669 * fixed some bugs revealed by Paul Dietz' test suite:
1670 ** COPY-ALIST now signals an error if its argument is a dotted
1672 ** Condition slots are now accessed more correctly in the presence
1673 of multiple initargs for a given slot.
1674 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1675 exclude restarts of the same name associated with a different
1677 ** DEFCLASS of forward-referenced classes with another
1678 forward-referenced class in the superclasses list no longer
1680 ** Condition slots are now initialized once each, not multiple
1681 times. (thanks to Gerd Moellmann)
1682 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1684 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1685 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1686 arguments contain duplicated elements.
1687 ** RESTART-CASE understands local macros.
1688 ** RESTART-CASE associates exactly its own restarts with a condition.
1689 ** ENDP in safe mode checks its argument to be of type LIST.
1690 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1691 in question is unbound.
1692 ** Optimized MAKE-INSTANCE functions no longer cause internal
1693 assertion failures in the presence of duplicate initargs.
1694 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1695 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1696 propagated even in the ctor optimized implementation of
1698 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1700 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1701 * a better implementation of SXHASH on (simple) bit vectors,
1702 measured both in execution speed and in distribution of results
1703 over the positive fixnums, has been installed. Likewise, a better
1704 implementation of EQUAL for simple bit vectors is now available.
1705 * fixed CEILING optimization for a divisor of form 2^k.
1706 * fixed bug 240 (emitting extra style warnings "using the lexical
1707 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1708 by Antonio Martinez)
1709 * fixed SXHASH, giving different results for NIL depending on type
1710 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1711 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1712 variable. (found by Rolf Wester)
1713 * SBCL does not ignore type declarations for special
1714 variables. (reported by rif on c.l.l 2003-03-05)
1715 * some bug fixes in contrib/sb-aclrepl/
1716 * fixed some bugs revealed by Paul Dietz' test suite:
1717 ** a bug in the CONS type specifier, whereby the CAR and CDR
1718 types got intertwined, has been fixed;
1719 ** the type system is now able to reason about the interaction
1720 between INTEGER and RATIO types more completely;
1721 ** APPEND, [N]REVERSE and NRECONC check that those their
1722 arguments, which must be proper lists, are really so;
1723 ** An array specialized to be unable to hold elements has been
1724 implemented, as required -- yes, really -- by ANSI;
1725 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1726 on malformed property lists;
1728 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1729 * incompatible packaging change: in line with Unix convention,
1730 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1731 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1732 * REQUIRE and PROVIDE are now optionally capable of doing something
1733 useful. See the documentation string for REQUIRE.
1734 * infrastructure for a managed SBCL contrib system: contributed
1735 modules in this release include:
1736 ** the ASDF system definition facility;
1737 ** an interface to the BSD Sockets API;
1738 ** an ACL-like convenience interface to the repl;
1739 (thanks to Kevin Rosenberg)
1740 ** an implementation of ROTATE-BYTE, with efficient implementation
1742 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1743 gives a closer approximation to the right answer than previously.
1744 (thanks to Raymond Toy)
1745 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1746 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1747 optional environment arguments, as required by ANSI.
1748 * fixed bugs in other functions taking environment objects, allowing
1749 calls with an explicit NIL environment argument to be compiled
1751 * fixed bug 228: primary return values from
1752 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1753 COMPILE or FUNCTION.
1754 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1755 :TYPE will no longer signal errors on innocuous objects.
1756 * fixed bug 231b: SETQ is better at respecting type declarations in
1757 the lexical environment.
1758 * fixed a bug in DEFCLASS: classes named by symbols with no or
1759 unprintable packages can now be defined.
1760 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1761 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1762 * fixed compiler failure related to checking types of functions.
1763 (reported by Robert E. Brown)
1764 * the compiler is now much more consistent in its error-checking
1765 treatment of bounding index arguments to sequence functions: in
1766 (SAFETY 3) code, errors will be signalled in almost all cases if
1767 invalid sequence bounding indices are passed to functions defined
1768 by ANSI to operate on sequences.
1769 * fixed a bug in the build procedure: documentation of SBCL-specific
1770 packages is now preserved and available in the final Lisp image.
1771 * lifted FDEFINITION lookup out of loops in the implementation of
1772 many list operations. (thanks to Robert E. Brown)
1773 * fixed a bug in the reader: the #n# reader macro now works for
1774 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1775 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1776 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1777 E. Brown and Tony Martinez respectively)
1778 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1779 SLOT-BOUNDP: in particular, it is now able to optimize them much
1780 better, and is now not vulnerable to having packages renamed.
1781 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1782 always called when a slot is not present in an instance. (thanks
1784 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1785 longer signal an error on unreading a character following EOF on
1786 the previous constituent stream. (thanks to Tony Martinez)
1787 * fixed some bugs revealed by Paul Dietz' test suite:
1788 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1789 not just nonnegative fixnums;
1790 ** the logical bit-array operators such as BIT-AND now accept an
1791 explicit NIL for their "opt-arg" argument (to indicate a
1792 freshly-consed result bit-array);
1793 ** ELT now signals an error on an invalid sequence index in safe
1795 ** the type system is now cleverer about negations of numeric
1796 types, and consequently understands the BIGNUM and RATIO types
1798 ** the type system is now cleverer about the interaction between
1799 INTEGER and RATIO types: while bugs still remain, many more
1800 cases are accurately computed;
1801 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1802 if it is in the last clause;
1803 ** CONSTANTLY now correctly returns a side-effect-free function in
1805 ** DECLARE is no longer treated as a special-operator; in
1806 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1807 * incremented fasl file version number due to the change in the
1808 DEFSTRUCT-SLOT-DESCRIPTION structure.
1810 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1811 * minor incompatible change: code processed by the "interpreter" or
1812 EVAL now has a compilation optimization policy of (DEBUG 2)
1813 (changed from (DEBUG 1)) to improve debuggability of interactive
1814 development, and to allow the use of the debug RETURN command in
1816 * an experimental implementation of the RETURN command for the
1817 debugger has been included. (thanks to Frederik Kuivinen)
1818 * fixed bug 62: constraints were not propagated into a loop.
1819 * fixed bug in embedded calls of SORT (reported and investigated by
1821 * fixed some bugs revealed by Paul Dietz' test suite:
1822 ** printing and reading of arrays with some dimensions having
1823 length 0 (thanks to Gerd Moellmann);
1824 ** BOA constructor with &AUX argument without a default value does
1825 not cause a type error;
1826 ** CONSTANTP now returns true for all self-evaluating objects.
1828 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1829 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1830 accessors that are related by inheritance, as specified in the
1831 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1832 Valtteri Vuorikoski)
1833 * The compiler is now able to inline functions that were defined in
1834 a complex lexical environment (e.g. inside a MACROLET).
1835 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1837 * Support for the upcoming FreeBSD-5.0 release has been included.
1838 (thanks to Dag-Erling Smorgrav)
1839 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1840 effect when it is not in a toplevel context.
1841 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1842 stand a better chance of being correct. (thanks to Gerd
1844 * fixed bug in COERCE, which now signals an error on coercing a
1845 rational to a bounded real type which excludes the expected
1847 * The compiler is now able to derive types more accurately from the
1848 COERCE and COMPILE functions.
1849 * fixed bug 223: functional binding is considered to be constant
1850 only for symbols in the CL package.
1851 * fixed bug 231: SETQ did not check the type of a variable being set
1852 (reported by Robert E. Brown)
1853 * A new optimization for MAKE-INSTANCE has been included, fixing
1854 various bugs (including relating to :ALLOCATION :CLASS slots and
1855 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1857 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1858 ** As required by ANSI, LOOP now disallows anonymous collection
1859 clauses such as COLLECT I in conjunction with aggregate boolean
1860 clauses such as THEREIS (= I 1);
1861 ** LOOP now signals an error when any variable is reused in the
1862 same loop (including the potentially useful construct analogous
1863 to WITH A = 1 WITH A = (1+ A);
1864 ** IT is only a special loop symbol within the first clause of a
1865 conditional loop clause;
1866 ** LOOP with a typed iteration variable over a hashtable now
1867 signals a type error iff it should.
1868 * fixed some other bugs revealed by Paul Dietz' test suite:
1869 ** FILE-STREAM now names the class previously known as FD-STREAM;
1870 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1871 argument) no longer signals an error;
1872 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1873 of the slot symbol, rather than using the current package
1874 ((:CONC-NAME "") continues to intern the slot's name in the
1876 * incremented fasl file version number, because of the incompatible
1877 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1878 of the new implementation of DEFINE-COMPILER-MACRO.
1880 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1881 * Support for building SBCL for MIPS platforms running in
1882 little-endian mode has now been checked in, and basic
1883 functionality on said platforms verified.
1884 * minor incompatible change: PCL now records the pathname of a file
1885 in which methods and the like are defined, rather than its
1887 * minor incompatible change: TRUENAME now considers the truename of
1888 a file naming a directory to be the pathname with :DIRECTORY
1889 component indicating that directory.
1890 * minor incompatible change: a NAMED clause in the extended form of
1891 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1892 reason for the previous behaviour is unclear.
1893 * more systematization and improvement of CLOS and MOP conformance
1894 in PCL (thanks to Gerd Moellman and Pierre Mai):
1895 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1897 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1898 primary methods with no specializers;
1899 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1901 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1902 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1903 FORWARD-REFERENCED-CLASSes; error reporting on
1904 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1906 ** SXHASH on CLOS instances now uses a slot internal to the
1907 instance to return different numbers on distinct instances,
1908 while preserving the same return value through invocations of
1910 ** DEFMETHOD signals errors when methods with longer incongruent
1911 lambda lists are added to generic functions;
1912 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1913 CLASS, as specified in AMOP;
1914 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1915 on the order returned by the primary method for classes of
1916 class STANDARD-CLASS;
1917 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1918 * fixed some bugs shown by Paul Dietz' test suite:
1919 ** DOLIST puts its body in TAGBODY;
1920 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1922 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1923 value producing form;
1924 ** if more variables are given to PROGV than values, extra
1925 variables are bound and made to have no value;
1926 ** NSUBSTITUTE on list arguments gets the right answer with
1928 ** ELT signals an error of type TYPE-ERROR when the index argument
1929 is not a valid sequence index;
1930 ** LOOP signals (at macroexpansion time) an error of type
1931 PROGRAM-ERROR when duplicate variable names are found;
1932 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1933 ** FUNCALL of special-operators now cause an error of type
1935 ** PSETQ now works as required in the presence of side-effecting
1936 symbol-macro places;
1937 ** NCONC accepts any object as its last argument;
1938 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1940 ** loop-for-as-package does not require a package to be explicitely
1942 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1944 * fixed bug 166: compiler preserves "there is a way to go"
1945 invariant when deleting code.
1946 * fixed bug 172: macro lambda lists with required arguments after
1947 &REST arguments now cause an error to be signalled. (thanks to
1949 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1950 bug. (thanks to Matthew Danish)
1951 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1953 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1954 arguments are lexically rebound. (thanks to Gerd Moellmann and
1956 * fixed bug 194: error messages are now more informative when there
1957 is no primary method applicable in a call to a generic
1958 function. (thanks to Gerd Moellmann)
1959 * fixed bug in command line argument checking (thanks to Julian
1961 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1962 (thanks to Matthew Danish)
1963 * incremented fasl file version number, because of the
1964 SXHASH-related changes in the layout of CLOS data structures
1966 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1967 * minor incompatible change: The runtime (the Unix executable named
1968 "sbcl") is now much pickier about the .core files it will load.
1969 Essentially it now requires .core files to descend from the same
1970 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1971 as the runtime does. (The intent is to prevent the crashes which
1972 can occur, and which can even be reported as mysterious failures,
1973 when people patch the sources or change the build parameters
1974 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1975 sbcl and .core files.)
1976 * fixed bug: VALUES-LIST is no longer optimized away.
1977 * fixed bug 142: The FFI conversion of C string values to Lisp
1978 string values no longer conses excessively. (thanks to Nathan
1979 Froyd porting Raymond Toy's fix to CMU CL)
1980 * began to systematize and improve MOP conformance in PCL (thanks to
1981 Nathan Froyd, Gerd Moellman and Pierre Mai):
1982 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1984 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1985 ** FINALIZE-INHERITANCE is now called on class finalization;
1986 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1987 argument precedence order.
1988 * fixed bug 202: The compiler no longer fails on functions whose
1989 derived types contradict their declared type.
1990 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1991 so it can be non-toplevel.
1992 * The fasl file version number has changed (because of the new
1993 implementation of DEFMACRO).
1994 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1995 safe code, that any length requirement by their type-specifier
1996 argument is valid. The exceptions to this are described in bug
1998 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1999 functions now check the types of their inputs as required by ANSI.
2000 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2001 introduced symbol is DECLAREd to be SPECIAL.
2002 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2003 * fixed bug: :COUNT argument to sequence functions may be negative
2004 * fixed bug: body of DO-SYMBOLS may contain declarations
2005 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2006 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2007 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2008 symbol macro only once
2009 * fixed printing of call frame when argument list is unavailable
2010 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2011 * compiler no longer signals WARNING on unknown keyword
2014 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2015 * A beta-quality port to the mips architecture running Linux,
2016 based on the old CMUCL backend, has been made. It has been tested
2017 on a big-endian kernel, and works sufficiently well to be able to
2018 rebuild itself; it has not been tested in little-endian mode.
2019 * fixed an inconsistency between gencgc.c and purify.c which made
2020 dumping/loading .core files unreliable
2021 * fixed bug 120a: The compiler now deals correctly with IFs where
2022 the consequent is the same as the alternative, instead of
2023 misderiving the return type. (thanks to Alexey Dejneka)
2024 * fixed bug 113: Logical pathnames are now dumpable (the logical
2025 host is resolved at load-time, throwing an error if it is not
2027 * fixed bug 174: FORMAT's error message is slightly clearer when a
2028 non-printing character is used in a format directive.
2029 * fixed several bugs in compiler checking of type declarations, i.e.
2030 violations of the Python "declarations are assertions" principle
2031 (thanks to Alexey Dejneka)
2032 * fixed several bugs in PCL's error checking (thanks to Gerd
2034 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2036 * fixed bug in compilation of functions as first class values
2037 (thanks to Antonio Martinez-Shotton)
2038 * The compiler's handling TYPE-ERRORs which it can prove will
2039 inevitably happen at runtime has been cleaned up and corrected
2040 in several ways. (thanks to Alexey Dejneka)
2041 * improved argument type checking for various basic arithmetic
2042 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2043 lavished on them in the past that they can be compiled in many
2044 ways in different special cases
2045 * fixed bug 181: compiler checks validity of user supplied type
2047 * cleaned up code flushing in optimization: Function calls which
2048 should signal errors for safety purposes (e.g. which ANSI says
2049 should signal errors when their arguments are of incorrect type)
2050 are no longer optimized away.
2051 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2052 * incremented fasl file version number, because changes in the
2053 implementation of sequence functions like COERCE caused
2054 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2055 in old inline expansions) to become undefined. (Actually these
2056 changes were later undone, so we might very well be binary
2057 compatible with 0.7.7 after all, but leaving the version number
2058 incremented seemed like the simplest and most conservative
2061 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2062 * An alpha-quality port to the parisc architecture running Linux,
2063 based on the old CMUCL backend, has been made. This, even more so
2064 than the other backends, should be considered still a work in
2065 progress; known problems include that the Linux kernel in 64-bit
2066 mode does not propagate the correct sigcontext structure to
2067 userspace, and consequently SBCL on a parisc64 kernel will not
2069 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2070 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2071 suppressed.) Also now that the compiler is looking at declarations
2072 in the environment, it checks optimization declarations as well,
2073 and suppresses inlining when (> DEBUG SPEED).
2074 * More fixes have been made to treatment of floating point exception
2075 treatment and other Unix signals. In particular, floating point
2076 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2077 * The detection and handling of control stack exhaustion (infinite
2078 or very deeply nested recursion) has changed. Stack exhaustion
2079 detection is now done by write-protecting pages at the OS level
2080 and applies at all optimization settings; when found, a
2081 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2082 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2083 be caught using IGNORE-ERRORS.
2084 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2085 correctly when transforming recursive LABELS functions to LETs.
2086 (thanks to Alexey Dejneka)
2087 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2088 that are names of constants or global variables.
2089 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2090 alien routines with docstrings.
2091 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2092 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2094 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2095 (thanks to Eric Marsden)
2096 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2097 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2098 * Bug fix: The SPARC backend can now compile functions involving
2099 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2100 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2101 to a functions expecting a PATHNAME-DESIGNATOR.
2102 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2103 lambda lists. (thanks to Alexey Dejneka)
2104 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2105 functions so that EXPORTing the name of the class doesn't cause
2106 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2107 to Antonio Martinez for reporting this)
2108 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2109 constant-folded without failing an assertion. (thanks to Einar
2110 Floystad Dorum for reporting this)
2111 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2112 types are now dealt with more correctly by the compiler.
2113 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2114 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2115 behaviour analogous to RENAME-FILE. This puts its behaviour more
2116 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2118 * The fasl file version number has changed again. (because of the
2119 bug fix involving the names of PCL MAKE-INSTANCE functions)
2121 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2122 * bug fix: Floating point exceptions are treated much more
2123 consistently on the x86/Linux and PPC/Linux platforms.
2124 * Array initialization with :INITIAL-ELEMENT is now much faster for
2125 cases when the compiler cannot open code the array creation, but
2126 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2127 array accesses have also seen a speed increase.
2128 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2129 specified. (This was at the root of some bad interactions between
2130 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2132 * bug fix: Internal error arguments for undefined functions are now
2133 computed correctly on the PPC/Linux platform.
2134 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2135 Raymond Toy's patch for CMU CL)
2136 * Support for the Solaris 9 operating environment has been included
2137 (thanks to Daniel Merritt)
2138 * A very ugly but hopefully complete draft of the missing FFI chapter
2139 of the manual has been created by reformatting the corresponding
2140 CMU CL manual chapter into (currently very ugly and incoherent)
2141 DocBook and bringing it up to date for SBCL behavior. Thus, the
2142 manual is now essentially complete, at least by my extreme
2143 once-and-only-once standards, whereby it's acceptable to refer to
2144 the doc strings of SB-EXT functions as the primary documentation.
2145 * The fasl file version number has changed again, due to cleanup of
2146 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2147 is no longer a static symbol.)
2149 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2150 * SBCL now builds with OpenMCL (version 0.12) as the
2151 cross-compilation host; also, more progress has been made toward
2152 bootstrapping under CLISP.
2153 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2155 * bug 158 fixed: The compiler can now deal with integer loop
2156 increments different from 1; fixing this turned out also to fix
2158 * bug 169 fixed: no more bogus warnings about using lexical bindings
2159 despite the presence of perfectly good SPECIAL declarations (thanks
2160 to David Lichteblau)
2161 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2162 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2163 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2165 * bug 180 fixed: Method combination specifications no longer ignore
2166 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2167 * bug fix: Structure type predicate functions now check their argument
2168 count as they should.
2169 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2170 correctly. (thanks to Pierre Mai)
2171 * minor incompatible change: The --noprogrammer option is deprecated
2172 in favor of the new --disable-debugger option, which is very similar.
2173 (The major difference is that it takes effect at a slightly different
2174 time at startup, causing handling of errors in --sysinit and
2175 --userinit files will be affected differently.) The
2176 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2177 been added to allow this functionality to be controlled from ordinary
2178 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2179 maintainers, who might want to run non-interactive scripts to
2180 build SBCL cores which will later be used interactively.)
2181 * minor incompatible change: The LOAD function no longer, when given
2182 a wild pathname to load, loads all files matching that pathname.
2183 Instead, an error of type FILE-ERROR is signalled.
2185 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2186 * bug 147 fixed: The compiler preserves its block link/count
2187 invariants more correctly now so that it doesn't crash. (thanks
2189 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2191 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2192 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2193 no-value-for-this-character as NIL (as specified by ANSI).
2194 * HOST-NAMESTRING on physical pathnames now returns a string that is
2195 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2196 (thanks to Christophe Rhodes)
2197 * The Alpha port handles icache flushing more correctly. (thanks to
2199 * More progress has been made toward bootstrapping under CLISP. (thanks
2200 to Christophe Rhodes)
2201 * The fasl file format has changed again, because dynamic loading
2202 on OpenBSD (which has non-ELF object files) motivated some cleanups
2203 in the way that foreign symbols are transformed and passed around.
2204 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2205 is no longer treated as whitespace by the reader, but instead as
2206 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2207 |A
\7fB|, instead of A as it used to.
2209 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2210 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2211 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2212 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2213 this as of 0.7.1.45, but was left out of the previous news section
2214 (thanks to Dan Barlow)
2215 * SBCL now runs on the Solaris operating system on SPARC architectures
2216 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2217 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2218 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2219 * SPARC backend cleanups, allowing builds of cores optimized for V8
2220 and V9 SPARCS, and also emission of code targeted to a particular
2221 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2223 * SBCL is closer to bootstrapping under CLISP, thanks to various
2224 fixes by Christophe Rhodes.
2225 * The fasl file format has changed again, to allow the compiler's
2226 INFO database to support symbol macros.
2227 * The user manual (in doc/) is formatted into HTML more nicely.
2228 (thanks to coreythomas)
2229 * The system is smarter about SUBTYPEP relationships, especially
2230 those involving NOT types (including types such as ATOM which are
2231 represented internally using NOT types). Thus SUBTYPEP is less
2232 likely to return (VALUES NIL NIL) in general, and in particular
2233 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2235 * The fasl file format has changed again, because the internal
2236 representation of types now includes a new slot to support the new
2237 SUBTYPEP-of-NOT-types logic.
2238 * (not a change in the main branch of SBCL, but a related prototype
2239 which can hopefully be merged into the main branch of SBCL in the
2240 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2241 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2242 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2243 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2244 dispatch macro character combinations. (thanks to Alexey Dejneka)
2245 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2246 that superficially look like logical namestrings correctly.
2247 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2249 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2250 default supplied-p) for &optional and &key arguments. (thanks to
2253 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2254 * incompatible change: The compiler is now less aggressive about
2255 tail call optimization, doing it only when (> SPACE DEBUG) or
2256 (> SPEED DEBUG). (This is an incompatible change because there are
2257 programs which relied on the old CMU-CL-style behavior to optimize
2258 away their unbounded recursion which will now die of stack overflow.)
2259 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2260 for non-GENCGC systems has been increased to 20M (since that
2261 seems much closer to the likely performance optimum for modern
2262 systems than the old 4M value was)
2263 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2264 and *DEBUG-PRINT-LEVEL*
2265 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2266 of CMU CL's support for SPARC, and various endianness and other
2267 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2268 * new syntactic sugar for the Unix command line: --load foo.bar is now
2269 an alternate notation for --eval '(load "foo.bar")'.
2271 ** The system now detects stack overflow and handles it gracefully,
2272 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2273 optimization settings. (This is a good thing in general, and
2274 its introduction in this version should be particularly timely
2275 for anyone whose code fails because of suppression of tail
2277 ** The system now hunts for the C variable "environ" in a more
2278 devious way, to avoid segfaults when the C library version differs
2279 between compile time and run time. (thanks to Christophe Rhodes)
2280 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2281 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2282 ** The compiler no longer issues bogus style warnings for undefined
2283 classes in the same source file as the DEFCLASSes which defined
2284 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2286 ** fixes in CONDITION class precedence list for undefined function
2287 errors (thanks to Alexei Dejneka)
2288 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2289 correctly. (thanks to Dan Barlow)
2290 ** portability fixes aiming at bootstrapping under CLISP (thanks
2291 to Dave McDonald and Christophe Rhodes)
2292 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2293 ** fixes in type translation and and type inference (thanks to
2295 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2296 ** various fixes in the new ports (thanks to Dan Barlow)
2297 * several changes related to debugging:
2298 ** suppression of tail recursion, as noted above
2299 ** stack overflow detection, as noted above
2300 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2301 is now the default. (For some time encapsulation has been more
2302 reliable than the breakpoint-based :ENCAPSULATE NIL
2303 implementation, at least on X86 systems; and I just noticed that
2304 encapsulation also seems closer to the spirit of the ANSI
2307 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2309 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2310 up properly again. (There was a packaging bug in 0.7.0 which
2311 left their definitions in SB-SYS::LOAD-FOREIGN and
2312 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2313 vital for most things which interface to C-level interfaces,
2314 like extensions working with sockets or databases or
2315 Perl-compatible regexes or whatever, and the need to fix
2316 this bug is the main reason that 0.7.1 was released so
2318 ** DEFGENERIC is now choosier about the methods it redefines, so that
2319 reLOADing a previously-LOADed file containing DEFGENERICs does
2320 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2321 cycle now works as it should. (thanks to Alexey Dejneka)
2322 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2323 fixed by Christophe Rhodes. (He actually submitted this patch
2324 months ago, and I delayed until after 0.7.0.)
2325 ** Bug 111 (internal compiler confusion about runtime checks on
2326 FUNCTION types) was fixed by Alexey Dejneka.
2327 * Some internal cleanups (getting rid of variables which aren't
2328 needed now that the byte interpreter is gone) caused the fasl
2329 file format number to change again.
2331 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2332 * major incompatible change: The default fasl file extension, i.e. the
2333 default extension for files produced by COMPILE-FILE, has changed
2334 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2336 ** There are many changes in the implementation of the compiler.
2337 SBCL is now essentially a compiler-only implementation of ANSI
2338 Common Lisp. EVAL still "interprets" a few special cases, but
2339 almost all the interesting cases are handled by creating
2340 a LAMBDA expression, calling COMPILE on it, then calling
2341 FUNCALL on the result.
2342 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2343 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2344 Since the code is newer, there might still be some new bugs
2345 (though not as many as before Martin Atzmueller's fixes:-). But
2346 the new code is substantially simpler and clearer, and hopefully
2347 any remaining bugs will be simpler, less fundamental, and more
2348 fixable then the bugs in the old code.
2349 ** The revised compiler is still a little unsteady on its feet.
2351 *** The debugging information it produces (particularly the names
2352 of FUNCTION objects) is sometimes much less useful than what
2353 the old compiler produced.
2354 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2355 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2356 within a MACROLET) has been temporarily weakened.
2357 ** There are new compiler optimizations for various functions:
2358 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2359 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2360 *** the math functions TRUNCATE, FLOOR, and CEILING
2361 *** the function-of-all-trades COERCE
2362 Mostly these should be transparent, but there's one
2363 potentially-annoying problem (bug 117): when the compiler
2364 inline-expands a function and does type analysis on the result,
2365 it can create control paths which have type mismatches, and
2366 when it can't prove that those control paths aren't taken,
2367 it will issue WARNINGs about the type mismatches. This is
2368 a particular problem in practice for the new sequence functions.
2369 It's not clear how this should be fixed, and for now, a
2370 workaround is given in the entry for 117 in the BUGS file.
2371 ** (Because of the interaction between the two previous items --
2372 occasional inlining problems and new inline expansions -- some
2373 of the new sequence function optimizations won't really kick in
2374 completely until debugging information, and then inlining, are
2375 straightened out in some future version.)
2376 * minor incompatible changes:
2377 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2378 DIRECTORY no longer implicitly promotes NIL slots of its
2379 pathname argument to :WILD. In particular, when you ask for the
2380 contents of a directory (which you used to be able to do without
2381 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2382 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2383 ** changes in behavior that ANSI explicitly defines to be
2384 implementation dependent:
2385 *** The new compiler-only implementation still conforms with ANSI,
2386 but acts a little different than before. Besides the obvious
2387 changes in performance tradeoffs (that the cost per form passed
2388 to EVAL has gone up, and the cost per form executed by EVAL
2389 has gone down), the behavior of the system changes a little
2390 because there are no longer any interpreted function objects.
2391 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2392 e.g. doing COMPILE on the output of interactive DEFUN is
2394 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2396 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2397 has changed from (:CL) to NIL.
2398 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2399 e.g. #\Newline and #\Tab, don't have names specified in the
2400 ANSI Common Lisp standard, is now based on their ASCII symbolic
2401 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2402 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2403 are no longer used for output.
2404 ** changes in internal implementation constants:
2405 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2406 4 million. (If your application spends a lot of time GCing and
2407 you have a lot of RAM, you might want to experiment with
2408 increasing it even more.)
2409 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2410 However, almost all old code should still continue to work without
2411 immediate update, as SB-C-CALL is now a (deprecated) nickname
2413 ** Old operator names in the style DEF-FOO are now deprecated in
2414 favor of new corresponding names DEFINE-FOO, for consistency with
2415 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2416 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2417 internal symbols, but a few supported extensions like
2418 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2419 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2420 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2421 etc. as you get deeper into recursive calls to the debugger
2422 command loop, instead of the old "5]", "5]]", "5]]]"
2423 sequence. (I was motivated to do this when squabbles between
2424 ILISP and SBCL left me very deeply nested in the debugger. In the
2425 short term, this change will probably provoke more ILISP/SBCL
2426 squabbles, but hopefully it will be an improvement in the long run.)
2427 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2428 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2429 function, since declaiming return types involving aliens is
2430 (1) annoyingly messy to do by hand and (2) vital to efficient
2431 compilation of code which calls such functions.
2432 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2433 longer reexported by the SB-EXT package. They're solely useful
2434 for alien code, so it seems more logical that you should get
2435 them from the SB-ALIEN package, not in SB-EXT.
2436 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2437 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2438 features. Instead, the code that they used to control is always
2439 built into the system.
2440 * many other bug fixes
2441 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2442 advantage of the new EVAL-WHEN stuff and to clean them up in
2443 general, and they are now more ANSI-compliant in a number of
2444 ways. Martin Atzmueller is responsible for a lot of this.
2445 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2447 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2448 *** correct ERROR type for various file operations
2449 *** some fixes for Lisp streams
2450 *** DEFMETHOD syntax checking
2451 *** changing old weird representation of debug information as
2452 strings (which, among their other deficiencies, don't transform
2453 correctly when you rename packages, and don't change their
2454 print representation when you change things like *PACKAGE*
2455 and *PRINT-LENGTH*) to symbols and lists of symbols
2456 He also made several improvements and fixed several bugs in DESCRIBE.
2457 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2459 *** misbehavior of WRITE-STRING/WRITE-LINE
2460 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2461 and several other LOOP problems as well
2462 *** DIRECTORY when similar filenames are present
2463 *** DEFGENERIC with :METHOD options
2464 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2465 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2466 *** argument ordering in FIND with :TEST option
2467 *** mishandled package designator argument in APROPOS-LIST
2468 *** various problems in the backquote readmacro
2469 *** a bug in APROPOS
2470 *** probably some others that I'm not describing very well here,
2471 since the CVS log documents them by reference to sbcl-devel
2472 messages, and the SourceForge archives aren't working well.:-(
2473 ** Dan Barlow improved the Alpha port (and is making progress on the
2474 PPC port, for those of you who think different).
2475 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2476 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2477 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2478 continues to work on the SPARC port (for those of you in a position
2479 to look down upon our little PC-compatible boxes from a great height).
2480 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2481 and uses that copy, rather than the current dynamic value, when
2482 it's trying to decide whether to truncate output. Thus e.g.
2483 (let ((*print-lines* 50))
2484 (pprint-logical-block (stream nil)
2486 (let ((*print-lines* 8))
2487 (print (aref possiblybigthings i) stream)))))
2488 should now truncate the logical block only at 50 lines, instead of
2489 often truncating it at 8 lines, as it did before.
2490 * The doc/cmucl/ directory, containing old CMU CL documentation
2491 from the time of the fork, is no longer part of the base system.
2492 SourceForge has shut down its anonymous FTP service, and with it
2493 my original plan for distributing the old CMU CL documentation
2494 there. For now, if you need these files you can download an old
2495 SBCL source release and extract them from it.
2496 * The fasl file version number changed again, for dozens of reasons,
2497 some of which are apparent above.
2499 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2500 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2501 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2502 patch, so that DISASSEMBLE output is much nicer.
2503 * The code in the SB-PROFILE package now seems reasonably stable.
2504 I still haven't decided what the final interface should look like
2505 (I'd like PROFILE to interact cleanly with TRACE, since both
2506 facilities use function encapsulation) but if you have a need
2507 for profiling now, you can probably use it successfully with
2508 the current CMU-CL-style interface.
2509 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2510 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2511 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2512 link now returns the dangling link itself, and for similar
2513 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2514 link itself. (In these cases the old code signalled an error and
2515 looped endlessly, respectively.) Thus, DIRECTORY now works even
2516 in the presence of dangling and cyclic symbolic links.
2517 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2518 is now a supported extension again, since the consensus on
2519 sbcl-devel was that it can be useful for ordinary development
2520 work, not just for debugging SBCL itself.
2521 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2522 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2523 that currently-defined functions might be redefined later with
2524 different return types.
2525 * Hash tables can be printed readably, as inspired by CMU CL code
2526 of Eric Marsden and SBCL code of Martin Atzmueller.
2527 * better error handling in CLOS method combination, thanks to
2528 Martin Atzmueller porting Pierre Mai's CMU CL patches
2529 * more overflow fixes for >16Mbyte I/O buffers
2530 * A bug in READ has been fixed, so that now a single Ctrl-D
2531 character suffices to cause end-of-file on character streams.
2532 In particular, now you only need one Ctrl-D at the command
2533 line (not two) to exit SBCL.
2534 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2536 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2537 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2538 does the right thing.
2539 * The implementation of some type tests, especially for CONDITION
2540 types, is now tidier and maybe faster, due to CMU CL code
2541 originally by Douglas Crosher, ported by Martin Atzmueller.
2542 * Some math functions have been fixed, and there are new
2543 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2544 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2545 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2546 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2547 likely to be built into sbcl-0.7.0.)
2548 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2549 CL, ported to SBCL by Martin Atzmueller)
2550 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2551 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2552 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2553 * There's a new slam.sh hack to shorten the edit/compile/debug
2554 cycle for low-level changes to SBCL itself, and a new
2555 :SB-AFTER-XC-CORE target feature to control the generation of
2556 the after-xc.core file needed by slam.sh.
2557 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2558 COMPILE-FILE is no longer supported, so that now every function
2559 gets an entry point, so that block compilation looks a little
2560 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2561 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2562 returns the number of bytes consed since the system started,
2563 rather than the number consed since the first time the function
2564 was called. (The new definition parallels ANSI functions like
2565 CL:GET-INTERNAL-RUN-TIME.)
2566 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2567 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2568 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2569 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2570 * Fasl file version numbers are now independent of the target CPU,
2571 since historically most system changes which required version
2572 number changes have affected all CPUs equally. Similarly,
2573 the byte fasl file version is now equal to the ordinary
2576 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2577 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2578 is no longer recognized. I apologize for this, because it was
2579 listed in SB-EXT as a supported extension, but I found that
2580 its existing behavior was poorly specified, as well as incorrectly
2581 specified, and it looked like too much of a mess to straighten it
2582 out. I have enough on my hands trying to get ANSI stuff to work..
2583 * many patches ported from CMU CL by Martin Atzmueller, with
2584 half a dozen bug fixes in pretty-printing and the debugger, and
2585 half a dozen others elsewhere
2586 * fixed bug 13: Floating point infinities are now supported again.
2587 They might still be a little bit flaky, but thanks to bug reports
2588 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2589 as flaky as they were.
2590 * The --noprogrammer command line option is now supported. (Its
2591 behavior is slightly different in detail from what the old man
2592 page claimed it would do, but it's still appropriate under the
2593 same circumstances that the man page talks about.)
2594 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2595 are now supported, and enabled by default. Thus, the compiler can
2596 handle many floating point and complex operations much less
2597 inefficiently. (Thus e.g. you can implement a complex FFT
2599 * The compiler now detects type mismatches between DECLAIM FTYPE
2600 and DEFUN better, and implements CHECK-TYPE more correctly, and
2601 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2602 thanks to patches from Martin Atzmueller.
2603 * various fixes to make the cross-compiler more portable to
2604 ANSI-conforming-but-different cross-compilation hosts (notably
2605 Lispworks for Windows, following bug reports from Arthur Lemmens)
2606 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2607 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2608 fixed thanks to Pierre Mai's CMU CL patches.
2609 * fixes to keep the system from overflowing internal counters when
2610 it tries to use i/o buffers larger than 16M bytes
2611 * fixed bug 45a: Various internal functions required to support
2612 complex special functions have been merged from CMU CL sources.
2613 (When I was first setting up SBCL, I misunderstood a compile-time
2614 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2615 * improved support for type intersection and union, fixing bug 12
2616 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2617 more obscure bugs as well
2618 * some steps toward byte-compiling non-performance-critical
2619 parts of the system, courtesy of patches from Martin Atzmueller
2620 * Christophe Rhodes has made some debian packages of sbcl at
2621 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2622 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2623 stable, but are nonetheless usable. When he's ready, I'd be happy
2624 to add them to the SourceForge "File Releases" section. (And if
2625 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2626 * new fasl file format version number (because of changes in
2627 internal representation of (OR ..) types to accommodate the new
2628 support for (AND ..) types, among other things)
2630 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2631 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2633 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2635 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2636 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2637 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2638 * DESCRIBE now works on structure objects again.
2639 * Most function call argument type mismatches are now handled as
2640 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2641 know whether the function will be redefined before the call is
2642 executed. (The compiler could flag local calls with full WARNINGs,
2643 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2644 it doesn't keep track of enough information to know whether calls
2645 are local in this sense.)
2646 * Compiler output is now more verbose, with messages truncated
2647 later than before. (There should be some supported way for users
2648 to override the default verbosity, but I haven't decided how to
2649 provide it yet, so this behavior is still controlled by the internal
2650 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2651 src/compiler/ir1util.lisp.)
2652 * Fasl file format version numbers have increased again, because
2653 support for the Gray streams extension changes the layout of the
2654 system's STREAM objects.
2655 * The Gray subclassable streams extension now works, thanks to a
2656 patch from Martin Atzmueller.
2657 * The full LOAD-FOREIGN extension (not just the primitive
2658 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2659 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2660 but like most other programs, it defaults to copying the Unix
2661 environment from the original process instead of starting the
2662 new process in an empty environment.
2663 * Extensions which manipulate the Unix environment now support
2664 an :ENVIRONMENT keyword option which doesn't smash case or
2665 do other bad things. The CMU-CL-style :ENV option is retained
2666 for porting convenience.
2667 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2668 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2670 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2672 * A patch from Martin Atzmueller seems to have solved the SIGINT
2673 problem, and as far as we know, signal-handling now works cleanly.
2674 (If you find any new bugs, please report them!)
2675 * The system no longer defaults Lisp source file names to types
2676 ".l", ".cl", or ".lsp", but only to ".lisp".
2677 * The compiler no longer uses special default file extensions for
2678 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2679 seems to expect a single default extension for all compiled code,
2680 and there's no compelling reason to try to stretch the standard
2681 to allow two different extensions.) Instead, byte-compiled files
2682 default to the same extension as native-compiled files.
2683 * Fasl file format version numbers have increased again, because
2684 a rearrangement of internal implementation packages made some
2685 dumped symbols in old fasl files unreadable in new cores.
2686 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2687 many fewer weird special cases.
2688 * Bug #17 (differing COMPILE-FILE behavior between logical and
2689 physical pathnames) has been fixed, and some related misbehavior too,
2690 thanks to a patch from Martin Atzmueller.
2691 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2692 by Tim Moore, ported to SBCL by Martin Atzmueller.
2693 * Martin Atzmueller fixed several filesystem-related problems,
2694 including bug #36, in part by porting CMU CL patches, which were
2695 written in part by Paul Werkowski.
2696 * More compiler warnings in src/runtime/ are gone, thanks to
2697 more patches from Martin Atzmueller.
2698 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2701 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2703 * DESCRIBE now works on CONDITION objects.
2704 * The debugger now handles errors which arise when trying to print
2705 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2707 * The build system now uses an additional file, customize-target-features.lisp,
2708 to allow local modifications to the target *FEATURES* list. (The point of
2709 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2710 debugging features enabled, without having to worry about propagating it
2711 into everyone's system when I do a "cvs update".) When no
2712 customize-target-features.lisp file exists, the target *FEATURES* list
2713 should be constructed the same way as before.
2714 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2715 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2716 warn about behavior which is undefined under the ANSI spec. Note: This
2717 is specified by ANSI, but it's not very popular with programmers.
2718 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2719 macro in the SBCL sources for an example of a workaround which you
2720 might use to make portable ANSI-standard code which does what you want.
2721 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2722 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2723 when it's used as a non-toplevel form. (This is required in order
2724 to implement the DEFCONSTANT-EQX macro.)
2725 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2726 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2727 funny things, due to bugs in the implementation of EVAL-WHEN
2728 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2729 non-toplevel DEFCONSTANTs are uncommon.)
2730 * The core file version number and fasl file version number have been
2731 incremented, because the old noncompliant DEFCONSTANT behavior involved
2732 calling functions which no longer exist, and because I also took the
2733 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2734 * fixed bug 1 (error handling before read-eval-print loop starts), and
2735 redid debugger restarts and related debugger commands somewhat while
2737 ** The QUIT debugger command is gone, since it did something
2738 rather different than the SB-EXT:QUIT command, and since it never
2739 worked properly outside the main toplevel read/eval/print loop.
2740 Invoking the new TOPLEVEL restart provides the same functionality.
2741 ** The GO debugger command is also gone, since you can just invoke
2742 the CONTINUE restart directly instead.
2743 ** The TOP debugger command is also gone, since it's redundant with the
2744 FRAME 0 command, and since it interfered with abbreviations for the
2746 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2748 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2749 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2750 some porting work by Martin Atzmueller)
2751 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2752 with a boatload of patches which he ported to SBCL. Now that those
2754 ** The system tries to make sure that its low-priority messages
2755 are prefixed by semicolons, to help people who like to use
2756 syntax highlighting in their ILISP buffer. (This patch
2757 was originally due to Raymond Toy.)
2758 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2759 patches originally written by Raymond Toy.
2760 ** The compiler understands coercion between single-value and
2761 multiple-VALUES type expressions better, getting rid of some very
2762 weird behavior, thanks to patches originally by Robert MacLachlan
2763 and Douglas Crosher.
2764 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2765 lambda lists, thanks to a patch originally by Pierre Mai.
2766 ** The system no longer bogusly warns about "abbreviated type
2768 ** The compiler gets less confused by inlining and RETURN-FROM,
2769 thanks to some patches originally by Tim Moore.
2770 ** The system no longer hangs when dumping circular lists to fasl
2771 files, thanks to a patch originally from Douglas Crosher.
2772 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2773 undefined function error.
2774 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2775 in structure slot initforms) for now, documented workaround instead:-|
2776 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2777 * fixed bug 5: added stubs for various Gray stream functions called
2778 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2779 installed, at least appropriate type errors are generated
2780 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2781 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2782 consistently in DEFMETHOD forms.
2783 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2784 it doesn't seem to affect SBCL after all
2785 * The C runtime system now builds with better optimization and many
2786 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2788 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2790 * The system is now under CVS at SourceForge (instead of the
2791 CVS repository on my home machine).
2792 * The new signal handling code has been tweaked to treat register
2793 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2794 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2795 downstream, e.g. in debug-int.lisp, has implicit dependencies
2796 on the unsignedness of integer representation of machine words,
2797 and that caused the system to bomb out with infinite regress
2798 when trying to recover from type errors involving signed values,
2799 e.g. (BUTLAST '(1 2 3) -1).)
2800 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2801 (This was one of the bugs Peter Van Eynde reported back in July.)
2802 * The system now uses code inspired by Colin Walters' O(N)
2803 implementation of MAP (from the cmucl-imp@cons.org mailing
2804 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2805 inline the MAP operation, and there is more than one
2806 sequence argument to the MAP call (so that it can't just
2807 do ETYPECASE once and for all based on the type of the
2808 single sequence argument). (The old non-inline implementation
2809 of the general M-argument sequence-of-length-N case required
2810 O(M*N*N) time when any of the sequence arguments were LISTs.)
2811 * The QUIT :UNIX-CODE keyword argument has been renamed to
2812 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2813 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2815 * Raymond Wiker's patches to port dynamic loading from Linux to
2816 FreeBSD have been added.
2817 * The BUGS file is now more nearly up to date, thanks in large part
2818 to Martin Atzmueller's review of it.
2819 * The debugger now flushes standard output streams before it begins
2820 its output ("debugger invoked" and so forth).
2821 * The core version number and fasl file version number have both
2822 been incremented, because of incompatible changes in the layout
2824 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2825 used to not be called for a saved Lisp image.)
2826 * Martin Atzmueller's version of a patch to fix a compiler crash,
2827 as posted on sbcl-devel 13 September 2000, has been installed.
2828 * Instead of installing Martin Atzmueller's patch for the
2829 compiler transform for SUBSEQ, I deleted the compiler transform,
2830 and transforms for some similar consing operations.
2831 * A bug in signal handling which kept TRACE from working on OpenBSD
2833 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2834 away by constant folding
2835 * The system now defines its address space constants in one place
2836 (in the Lisp sources), and propagates them automatically elsewhere
2837 (through GENESIS and the sbcl.h file). Therefore, patching the
2838 address map is less unnecessarily tedious and error-prone. The
2839 Lisp names of address space constants have also been systematized.
2840 * CVS tags like dollar-Header-dollar have been removed from
2841 the sources, because they have never saved me trouble and
2842 they've been source of trouble working with patches and other
2843 diff-related operations.
2844 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2845 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2847 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2849 * The system has been ported to OpenBSD.
2850 * The system now compiles with a simple "sh make.sh" on the systems
2851 that it's supported on. I.e., now you no longer need to tweak
2852 text in the target-features.lisp-expr and symlinks in src/runtime/
2853 by hand, the make.sh takes care of it for you.
2854 * The system is no longer so grossly inefficient when compiling code
2855 involving vectors implemented as general (not simple) vectors (VECTOR T),
2856 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2857 which dares to use the various sequence functions on non-simple
2858 vectors, takes less of a performance hit.
2859 * There is now a primitive type predicate VECTOR-T-P
2860 to test for the (VECTOR T) type, so that e.g.
2861 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2862 can now be compiled with some semblance of efficiency. (The old code
2863 turned the type declaration into a full call to %TYPEP at runtime!)
2864 * AREF on (VECTOR T) is still not fast, since it's still compiled
2865 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2866 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2867 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2868 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2869 without making full calls to SUBTYPEP at runtime.
2870 (Some analogous efficiency issues for non-simple vectors specialized to
2871 element types other than T, or for non-simple multidimensional arrays,
2872 have not been addressed. They could almost certainly be handled the
2873 same way if anyone is motivated to do so.)
2874 * The changes in array handling break binary compatibility, so
2875 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2876 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2877 instead of (VALUES T T).
2878 * By following the instructions that Dan Barlow posted to sbcl-devel
2879 on 2 July 2000, I was able to enable primitive dynamic object
2880 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2881 functionality is not implemented (since it calls ld to resolve
2882 library references automatically, requiring RUN-PROGRAM for its
2883 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2884 try to resolve library references) is now supported.
2885 * The system now flushes the standard output streams when it terminates,
2886 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2887 them at several other probably-convenient times, e.g. in each pass of
2888 the toplevel read-eval-print loop, and after evaluating a form given
2889 as an "--eval" command-line option. (These changes were motivated by a
2890 discussion of stream flushing issues on cmucl-imp in August 2000.)
2891 * The source transform for TYPEP of array types no longer assumes
2892 that an array whose element type is a not-yet-defined type
2893 is implemented as an array of T, but instead punts, so that the
2894 type will be interpreted at runtime.
2895 * There is now some support for cross-compiling in make.sh: each of
2896 the phases of make.sh has its own script. (This should be transparent
2897 to people doing ordinary, non-cross-compile builds.)
2898 * Since my laptop doesn't have hundreds of megabytes of memory like
2899 my desktop machine, I became more motivated to do some items on
2900 my to-do list in order to reduce the size of the system a little:
2901 ** Arrange for various needed-only-at-cold-init things to be
2902 uninterned after cold init. To support this, those things have
2903 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2904 symbols with such names are now uninterned after cold init).
2905 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2906 fasl files for cold load.
2907 ** Remove the old compiler structure pooling code (which used to
2908 be conditional on the target feature :SB-ALLOC) completely.
2909 ** Redo the representation of some data in cold init to be more compact.
2910 (I also looked into supporting byte compiled code at bootstrap time,
2911 which would probably reduce the size of the system a lot, but that
2912 looked too complicated, so I punted for now.)
2913 * The maximum signal nesting depth in the src/runtime/ support code has
2914 been reduced from 4096 to 256. (I don't know any reason for the very
2915 large old value. If the new smaller value turns out to break something,
2916 I'll probably just bump it back up.)
2917 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2919 * Many, many bugs reported by Peter Van Eynde have been added to
2920 the BUGS list; some have even been fixed.
2921 * While enabling dynamic object file loading, I tried to make the
2922 code easier to understand, renaming various functions and variables
2923 with less ambiguous names, and changing some function calling
2924 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2925 * While trying to figure out how to do the OpenBSD port, I tried to
2926 clean up some of the code in src/runtime/. In particular, I dropped
2927 support for non-POSIX signal handling, added various comments,
2928 tweaked the code to reduce the number of compilation warnings, and
2929 renamed some files to increase consistency.
2930 * To support the new automatic configuration functionality in make.sh,
2931 the source file target-features.lisp-expr has been replaced with the
2932 source file base-target-features.lisp-expr and the machine-generated
2933 file local-target-features.lisp-expr.
2934 * fixed a stupid quoting error in make.sh so that using CMU CL
2935 "lisp -batch" as cross-compilation host works again
2937 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2939 * DESCRIBE no longer tries to call itself recursively to describe
2940 bound/fbound values, so that it no longer fails on symbols which are
2941 bound to themselves (like keywords, T, and NIL).
2942 * DESCRIBE now works on generic functions.
2943 * The printer now prints less-screwed-up representations of closures
2944 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2945 * A private symbol is used instead of the :EMPTY keyword previously
2946 used to mark empty slots in hash tables. Thus
2947 (DEFVAR *HT* (MAKE-HASH-TABLE))
2948 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2949 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2950 now does what ANSI says that it should. (You can still get
2951 similar noncompliant behavior if bang on the hash table
2952 implementation with all the symbols you get back from
2953 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2954 This breaks binary compatibility, since tests for equality to
2955 :EMPTY are wired into things like the macroexpansion of
2956 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2958 * There's now a minimal placeholder implementation for CL:STEP,
2959 as required by ANSI.
2960 * An obscure bug in the interaction of the normal compiler, the byte
2961 compiler, inlining, and structure predicates has been patched
2962 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2963 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2965 * Missing ordinary arguments in a macro call are now detected even
2966 when the macro lambda list contains &KEY or &REST.
2967 * The debugger no longer complains about encountering the top of the
2968 stack when you type "FRAME 0" to explicitly instruct it to go to
2969 the top of the stack. And it now prints the frame you request even
2970 if it's the current frame (instead of saying "You are here.").
2971 * As specified by ANSI, the system now always prints keywords
2972 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2974 * The default initial SIZE of HASH-TABLEs is now smaller.
2975 * Type information from CLOS class dispatch is now propagated
2976 into DEFMETHOD bodies, so that e.g.
2977 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2979 is now basically equivalent to
2980 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2981 (DECLARE (TYPE SINGLE-FLOAT X))
2983 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2984 operation, without having to do run-time type dispatch.
2985 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2986 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2987 * Since I don't seem to be making big file reorganizations very often
2988 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2989 is overflowing my ability to conveniently back them up), I've finally
2990 checked the system into CVS. (The CVS repository is on my home system,
2991 not at SourceForge -- putting it on SourceForge might come later.)
2992 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2993 high-level GC-NOTIFY-FOO functions send their output. (There's
2994 still very little control of where low-level verbose GC functions
2995 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2996 less than it used to -- the GC-NOTIFY-FOO functions are now under
2997 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2998 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2999 in only one place in the source code, and propagates it automatically
3000 everywhere that it's needed. Thus e.g. when I bump the version from
3001 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3002 * The C source files now include boilerplate legalese and documentation
3003 at the head of each file (just as the Lisp source files already did).
3004 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3005 to his page will be replaced with a link to his new CLiki service.
3007 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3009 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3010 * The build process now looks for GNU make under the default name "gmake",
3011 instead of "make" as it used to. If GNU make is not available as "gmake"
3012 on your system, you can change this default behavior by setting the
3013 GNUMAKE environment variable.
3014 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3015 can build without error under CMU CL.
3017 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3019 * There is now a partial SBCL user manual (with some new text and some
3020 text cribbed from the CMU CL manual).
3021 * The beginnings of a profiler have been added (starting with the
3022 CMU CL profiler and simplifying and cleaning up). Eventually the
3023 main interface should be through the TRACE macro, but for now,
3024 it's still accessed through vaguely CMU-CL-style functions and macros
3025 exported from the package SB-PROFILE.
3026 * Some problems left over from porting CMU CL to the new
3027 cross-compilation bootstrap process have been cleaned up:
3028 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3029 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3030 part when getting rid of the compiler *BACKEND* stuff.)
3031 ** The value of *NULL-TYPE* was screwed up, because it was
3032 being initialized before the type system knew the final
3033 definition of the 'NULL type. This screwed up several key
3034 optimizations in the compiler, causing inefficiency in all sorts
3035 of places. (I found it because I wanted to understand why
3036 GET-INTERNAL-RUN-TIME was consing.)
3037 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3038 PROCLAIM FTYPE information. Unfortunately this broke binary
3039 compatibility again, since now the forms output by DEFGENERIC
3040 to refer to functions which didn't exist in 0.6.3.
3041 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3042 can use the new (as of 0.6.3) transform for SOME into MAP into
3044 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3045 values to (INTEGER 0 1000000), so that the time code will no longer
3046 occasionally get blown up by Linux returning 1000000 microseconds
3047 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3048 its output conform to the ANSI spec. (Alas, this makes its output
3049 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3050 * A full call to MAP NIL with a single sequence argument no longer conses.
3051 * fixes to problems pointed out by Martin Atzmueller:
3052 * The manual page no longer talks about multiprocessing as though
3053 it were currently supported.
3054 * The ILISP support patches have been removed from the distribution,
3055 because as of version 5.10.1, ILISP now supports SBCL without us
3056 having to maintain patches.
3057 * added a modified version of Raymond Toy's recent CMU CL patch for
3058 EQUALP comparison of HASH-TABLE
3060 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3062 * The system still can't cross-compile itself with
3063 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3064 entails), but at least it can compile more of itself that way
3065 than it used to be able to, and various buglets which were uncovered
3066 by trying to cross-compile itself that way have now been fixed.
3067 * This release breaks binary compatibility again. This time
3068 at least I've incremented the FASL file format version to 2, so that the
3069 problem can be detected reliably instead of just causing weird errors.
3070 * various new style warnings:
3071 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3072 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3073 ** using the lexical binding of a variable named in the *FOO* style
3074 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3075 as specified by ANSI.
3076 * *RANDOM-STATE* is no longer automatically initialized from
3077 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3078 default behavior of the system is to repeat its behavior every time
3079 it's run. If you'd like to change this behavior, you can always
3080 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3081 old convention there was no comparably easy way to get the system to
3082 repeat its behavior every time it was run.
3083 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3084 FTYPE declarations has been removed, in favor of their ANSI
3085 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3086 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3087 the types of their sequence arguments can be determined at compile time.
3088 This is done through a new open code expansion for MAP which eliminates
3089 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3090 argument types can be determined at compile time.
3091 * The optimizer now transforms COERCE into an identity operation when it
3092 can prove that the coerced object is already of the correct type. (This
3093 can be a win for machine generated code, including the output of other
3094 optimization transforms, such as the MAP transform above.)
3095 * Credit information has been moved from source file headers into CREDITS.
3096 * Source file headers have been made more standard.
3097 * The CASE macro now compiles without complaining even when it has
3100 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3102 * (Note that the way that the PCL macroexpansions were rewritten
3103 to accommodate the change in DEFGENERIC below breaks binary
3104 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3105 not run under sbcl-0.6.2. Once we get out of alpha releases,
3106 i.e. hit release 1.0.0, we'll probably try to maintain binary
3107 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3108 and sbcl-1.4.4. Until then, however, it might be fairly common
3109 for maintenance releases to break binary compatibility.)
3110 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3112 * The reporting of unhandled signals has been changed to print some
3113 explanatory text as well as the report form. (Previously only
3114 the report form was printed.)
3115 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3116 it defines, so that the compiler no longer issues undefined function
3117 warnings for compiled-but-not-yet-loaded generic functions.
3118 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3119 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3120 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3121 section of the ANSI spec are supported.
3122 * The "" nickname for the "KEYWORD" package has been removed.
3123 The reader still handles symbol tokens which begin with a package marker
3124 as keywords, but it doesn't expose its mechanism for doing so in the
3125 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3126 * The system now issues STYLE-WARNINGs for contradictory TYPE
3127 proclamations. (Warnings for contradictory FTYPE proclamations would
3128 be nice too, but those can't be done usefully unless the type system
3129 is made smarter about FUNCTION types.)
3130 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3131 systematized, so that "*target-*.lisp is supposed to exist only on the
3132 target and imply that there's a related file which exists on the
3133 host, and *host-*.lisp is supposed to exist only on the host and imply
3134 that there's a related file which exists on the target. This involves a
3135 lot of renaming. Hopefully the acute confusion caused by the renaming
3136 will be justified by the reduction in chronic confusion..
3137 ** runtime-type.lisp -> early-target-type.lisp
3138 ** target-type.lisp -> late-target-type.lisp
3139 ** early-host-format.lisp -> early-format.lisp
3140 ** late-host-format.lisp -> late-format.lisp
3141 ** host-error.lisp -> misc-error.lisp
3142 ** early-error.lisp -> early-target-error.lisp
3143 ** late-error.lisp -> late-target-error.lisp
3144 ** host-defboot.lisp -> early-defboot.lisp
3145 ** code/misc.lisp -> code/target-misc.lisp
3146 ** code/host-misc.lisp -> code/misc.lisp
3147 ** code/numbers.lisp -> code/target-numbers.lisp
3148 ** code/early-numbers.lisp -> numbers.lisp
3149 ** early-host-type.lisp -> early-type.lisp
3150 ** late-host-type.lisp -> late-type.lisp
3151 ** host-typep.lisp -> typep.lisp
3152 ** load.lisp -> target-load.lisp
3153 ** host-load.lisp -> load.lisp
3154 ** host-disassem.lisp -> disassem.lisp
3155 ** host-insts.lisp -> insts.lisp
3156 ** byte-comp.lisp -> target-byte-comp.lisp
3157 ** host-byte-comp.lisp -> byte-comp.lisp
3158 ** host-signal.lisp -> signal.lisp
3159 ** host-defstruct.lisp -> defstruct.lisp
3160 ** late-target-type.lisp -> deftypes-for-target.lisp
3161 Furthermore, several other previously target-only files foo.lisp (e.g.
3162 hash-table.lisp and random.lisp) have been split into a target-and-host
3163 foo.lisp file and a target-only target-foo.lisp file, with their key type
3164 definitions in the target-and-host part, so that the cross-compiler will
3165 know more about target types.
3166 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3167 gone away. In their place are various *BACKEND-FOO* variables
3168 corresponding to the slots of the old structure.
3169 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3170 into the target SBCL has been fixed.
3171 * The chill.lisp system for loading cold code into a running SBCL
3173 * Support for the CMU CL "scavenger hook" extension has been removed.
3174 (It was undocumented and unused in the CMU CL sources that SBCL was
3175 derived from, and stale in sbcl-0.6.1.)
3176 * Various errors in the cross-compiler type system were detected
3177 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3178 (enabling various consistency checks). Many of them were fixed,
3179 but some hard problems remain, so the compiler is back to
3180 running without *TYPE-SYSTEM-INITIALIZED* for now.
3181 * As part of the cross-compiler type system cleanup, I implemented
3182 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3183 * I have started adding UNCROSS calls throughout the type system
3184 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3185 on cross-compiler input files will be able to go away, and various
3187 * CONSTANTP now returns true for quoted forms (as explicitly required
3190 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3192 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3193 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3194 (MAKE-LIST -1). In the longer term, it ought to become true
3195 that declarations are assertions even at SAFETY 1. For now, it's not
3196 quite true even at SAFETY 3, but it's at least more nearly true..
3197 (Note that this change seems to increases the size of the system by
3198 O(5%) and to decrease the speed of the compiler by 20% or more.)
3199 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3200 for the problem of printing dozens of lines of distracting information
3201 about low-level system machinery as part of the top stack frame
3202 on entry to the debugger when an undefined function was called.
3203 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3204 is not reset to COMMON-LISP-USER.
3205 * Compilation of stuff related to dyncount.lisp has been made conditional
3206 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3207 smaller. The various dyncount-related symbols have been moved into
3208 a new "SB-DYNCOUNT" package.
3209 * tty-inspect.lisp has been renamed to inspect.lisp.
3210 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3211 feature has gone away. (When we eventually port to other flavors of
3212 libc and/or Unix, we'll try to make the differences between flavors
3213 invisible at the user level.)
3214 * Various other *FEATURES* tags, and/or their associated conditionals,
3215 have been removed if obsolescent, or given better documentation, or
3216 sometimes given more-mnemonic names.
3218 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3220 * tidied up "make.sh" script
3221 * tidied up system directory structure
3222 * better "clean.sh" behavior
3223 * added doc/FOR-CMUCL-DEVELOPERS
3224 * many many small tweaks to output format, e.g. removing possibly-confusing
3225 trailing #\. character in DESCRIBE-INSTANCE
3226 * (EQUALP #\A 'A) no longer signals an error.
3227 * new hashing code, including EQUALP hashing
3228 * tidied up Lisp initialization and toplevel
3229 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3230 * command line argument processing
3231 * added POSIX-GETENV function to deal with Unix-ish environment variables
3232 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3233 terminating SBCL on EOF
3234 * non-verbose GC by default
3235 * There is no more "sbcl" shell script; the sbcl file is now the C
3236 runtime executable (just like CMU CL).
3237 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3239 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3240 symbol and package information as Lisp native symbol and package objects
3241 instead of strings naming symbols and strings naming packages. This way,
3242 whenever packages are renamed (as in warm init), debug information is
3243 transformed along with everything else.
3244 * tweaked the optimization policy declarations which control the building
3245 of SBCL itself. Now, among other things, the system no longer saves
3246 source location debugging information. (This helps two problems at once
3247 by reducing SBCL size and by keeping SBCL from trying to look for its
3248 sources -- which may not exist -- when reporting errors.)
3249 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3250 debugging and testing purposes
3251 * cleaned up printing, making the printer call PRINT-OBJECT for
3252 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3253 methods, giving nearly-ANSI behavior
3254 * converted almost all special variables to use *FOO* naming convention
3255 * deleted PARSE-TIME functionality, since it can be done portably
3256 * moved some files out of cold init into warm init
3257 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3259 * regularized formatting of source files
3260 * added an install.sh script
3261 * fixed ridiculous memory usage of cross-compiler by making
3262 compiler/alloc.lisp not try to do pooling unless it can hook
3263 itself into the GC of the cross-compilation host. Now the system
3264 builds nicely on my old laptop.
3265 * added :SB-ALLOC in target-features.lisp-expr
3266 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3267 was not implemented)
3268 * re-did condition handling and note reporting in the compiler. Notes
3269 are no longer handled by signalling conditions. Style warnings
3270 and warnings are handled more correctly and reported in such a way
3271 that it's easy to find one or the other in your output (so that you
3272 can e.g. figure out which of many problems caused COMPILE-FILE to
3274 * changed the severity of several compiler warnings from full WARNING
3275 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3276 compiler note reporting so that it doesn't use the condition system
3277 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3278 COMPILE-FILE command)
3279 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3280 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3281 no longer supported, since their implementation was deeply intertwingled
3282 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3283 * removed bogus "support" for compiler macros named (SETF FOO), and
3284 removed the compiler macro for SETF INFO (but only after making a fool
3285 of myself on the cmucl-imp mailing list by posting a bogus patch for
3286 DEFINE-COMPILER-MACRO..)
3287 * Compiled files containing forms which have side effects on the Lisp
3288 reader (such as DEFPACKAGE forms) are now handled more correctly.
3289 (Compiler queuing of top level lambdas has been suppressed by setting
3290 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3291 * deleted various currently-unused source files, e.g. gengc.lisp. They
3292 may be added back at some point e.g. when porting to other architectures,
3293 but until they are it's distracting to distribute them and to try to
3295 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3296 there were so many of them they're just distractions, and UNCROSS is
3297 known to be able to handle the current sources
3298 * moved PROFILE functionality into TRACE, so that it will be clear
3299 how the wrapping and unwrapping of functions when you profile them
3300 interacts with the wrapping and unwrapping of functions when you
3301 trace them. (Actually, the functionality isn't there yet, but at least
3302 the interface specification is there. Hopefully, the functionality will
3303 arrive with some future maintenance release.)
3304 * removed host-oops.lisp
3305 * changed signature of QUIT function to allow UNIX-CODE argument
3306 * fixed READ-SEQUENCE bug
3307 * tweaked verbose GC output so that it looks more like the progress
3308 output that ANSI specifies for functions like LOAD
3309 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3310 * added <http://sbcl.sourceforge.com> to the banner information printed by