1 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2 * New feature: Experimental support for bivalent streams: streams
3 opened with :element-type :default now allow character and binary
5 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
6 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
9 ** incompatible change: the threading api now works with thread
10 objects instead of thread ids
11 ** bug fix: threads are protected from signals and interruption when
12 starting up or going down
13 ** bug fix: a race where an exiting thread could lose its stack to gc
14 ** fixed numerous gc deadlocks introduced in the pthread merge
16 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
17 * numerous signal handling fixes to increase stability
18 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
19 * minor incompatible change: we now correctly canonize default
20 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
21 as per the MOP, rather than the historical (INITARG INITFUNCTION
22 INITFORM). (reported by Bruno Haible)
23 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
24 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
26 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
27 pathname is a directory pathname.
28 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
29 * dynamic space size on PPC has been increased to 768Mb. (thanks to
31 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
32 :METACLASS argument in addition to a class name. (reported by
33 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
34 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
35 (reported by Bruno Haible; patch from Gerd Moellmann)
36 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
38 * bug fix: more cleanups to the floating point exception handling on
39 x86-64 (thanks to James Knight)
40 * bug fix: the compiler does not try to constant fold calls of
41 COERCE and BIT-* functions when they return freshly constructed
42 objects. (reported by Nikodemus Siivola and Paul Dietz)
43 * optimization: improved the MIPS versions of generic (in the
44 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
45 * optimization: direct conversion of (unsigned-byte 32) to floats on
47 * optimization: structure instances with raw slots now use less
48 memory, and probably show better memory locality. (thanks to
50 * optimization: DYNAMIC-EXTENT declarations for lists and closures
51 are treated as requests for stack allocation on the x86-64,
52 Alpha-32, PPC and SPARC platforms.
53 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
54 the wrong file descriptor; implementation of SOCKET-OPEN-P.
55 (thanks to Tony Martinez)
57 ** gcing a dead thread can no longer lead to lockups
58 ** threads block signals until they are set up properly
59 ** errno is no longer shared by threads
60 ** interrupt-thread restores the eflags register on x86
61 ** fixed some lockups due to gc/thread interaction
62 * fixed some bugs revealed by Paul Dietz' test suite:
63 ** invalid dotted lists no longer raise a read error when
65 ** signal an error if a symbol that names a declaration is used
66 as the name of a type, or vice versa
67 ** allow using the (declare (typespec var*)) abbreviation for
68 (declare (type typespec var*)) with all type specifiers
69 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
70 ** always return NIL from PPRINT-POP when OBJECT is NIL
71 ** don't signal errors when pretty-printing malformed LABELS,
72 FLET or MACROLET forms
73 ** declarations in a DOLIST body are also in scope for the
75 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
77 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
80 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
81 * fixed cross-compiler leakages that prevented building a 32-bit
82 target with a 64-bit host compiler.
83 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
84 opened with :IF-EXISTS :OVERWRITE.
85 * fixed bug 281: error for an invalid qualifier in a short-form method
86 combination method is not signalled until the faulty method is called.
87 * bug fix: iteration variable type inferrer failed to deal with open
88 intervals. (reported by Alan Shields)
89 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
90 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
91 or more runtime options were provided to the sbcl binary.
92 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
93 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
94 * optimization: there's now a fast-path for fixnum arguments in the
95 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
96 * optimization: the code generated on x86-64 is more compact thanks
97 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
98 * bug fix: floating-point exception handling now partly works on
99 x86-64. (Thanks to James Knight)
100 * improvement to the MIPS backend from Thiemo Seufer: C-style
101 64-bit long long arguments and return values to alien functions
103 * fixed some bugs revealed by Paul Dietz' test suite:
104 ** the type-error signalled from WARN has a filled-in DATUM slot.
105 ** the type-error required when a stream is not associated with
106 a file has the stream as its datum.
107 ** type-errors on single-floats on x86-64 no longer have
108 :INVALID-OBJECT as the datum
109 ** the type-errors signalled for invalid function names now have
110 a correct expected type
111 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
112 for structures defined with a :TYPE. Documentation strings for
113 typed structures are no longer immediately discarded
114 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
115 broadcast streams and synonym streams. FILE-LENGTH now also works
116 on broadcast streams.
118 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
119 * incompatible change: the --noprogrammer option, deprecated since
120 version 0.7.5, has been removed. Please use the equivalent
121 --disable-debugger option instead.
122 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
124 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
125 inoperational for a while now) has been completely removed.
126 * null lexical environments are now printed as #<NULL-LEXENV>,
127 significantly reducing the amount of clutter in typical backtraces.
128 * documentation on weak pointers, finalization, and after GC hooks
129 has been added to the manual.
130 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
131 as other specialized array types, is much faster. SUBSEQ and
132 COPY-SEQ on such arrays have also been sped up.
133 * optimization: EQL is now more efficient when at least other argument
134 is known to be of type (OR FIXNUM (NOT NUMBER)).
135 * fixed bug: the runtime is now less vulnerable to changes in the
136 size of the SBCL object on OS X, and virtual memory is reserved for
137 all spaces that need to be at a fixed address.
138 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
139 * fixed bug: finalizers and after GC hooks that cause consing are now
141 * fixed bug: compiler error messages and summaries are now printed to
142 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
143 * fixed inference of the upper bound of an iteration variable.
144 (reported by Rajat Datta).
145 * fixed bug 211e: calling local functions with duplicated constant
146 keyword argument no longer causes a bogus style warning about an
148 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
149 ftype declaration to be disregarded. (reported by Dave Roberts)
150 * fixed bug 373: caused by erronous compilation of references to alien
151 variables in the runtime on ppc/darwin.
152 * fixed bug 376: CONJUGATE type deriver.
153 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
154 (reported by Baughn on #lisp)
155 * a cross-compiler bug on non-x86 platforms has been identified and
156 fixed. (thanks to Bruno Haible)
157 * improvements to the MIPS runtime code for increased stability.
158 (thanks to Thiemo Seufer)
159 * increased the maximimum compact environment size to allow
160 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
161 * partial workaround for bug 135: don't name the function we're
162 calling for hairy cases of EVAL, so as not to accumulate one environment
163 entry per call to EVAL. (reported by Kevin Reid)
164 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
165 * optimization: type testing for non-vector arrays should be faster.
166 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
168 * bug fix: setting 31st element of a bit vector to zero did not work
170 * bug fix: redefining a class definition which failed due to a
171 previous accessor / function clash now works (but see BUGS entry
172 #380 for more problems in this area). (thanks to Zach Beane)
173 * the long form of DEFINE-METHOD-COMBINATION disables method group
174 checking when given a single method group with pattern *. (thanks
176 * on x86 compiler supports stack allocation of results of simple
177 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
178 * support for building and running on Mac OS X 10.4 "Tiger" has been added
179 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
180 * fixed some bugs related to Unicode integration:
181 ** the restarts for recovering from input and output encoding
182 errors only appear when there is in fact such an error to
184 * fixed some bugs revealed by Paul Dietz' test suite:
185 ** MISC.549 and similar: late transformation of unsafe type
186 assertions into derived types caused unexpected code
188 ** SCALE-FLOAT type deriver is less wrong.
189 ** type derivers for EXP, LOG and similar functions compute result
190 types for complex arguments better.
191 ** (MISC.563) CONJUGATE type deriver works for very restricted
193 ** out-of-line type testers for character strings are available.
194 ** EQUAL compiler transform understands specialness of objects
196 ** accessing double-floats stored on the stack now works on x86-64.
197 ** debugger internals could sometimes create invalid lispobjs,
198 resulting in GC crashes.
199 ** MISC.548: type check weakening can convert required type into
201 ** initialization forms for bindings are not in scope of free special
204 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
205 * incompatible change: thread support for non-NPTL systems has
206 been removed - locking is buggy and unreliable. A threaded
207 SBCL build will now warn at startup and refuse to create
208 new threads, unless futex support is detected in the kernel
209 * incompatible change: the top level REPL now has only an ABORT
210 restart associated with it, not TOPLEVEL and ABORT as it used to.
211 TOP and TOPLEVEL are now available as debugger commands for
212 returning to the top level.
213 * incompatible change: forms evaluated in the REPL now use the
214 global optimization policy.
215 * incompatible change: user- and system-initialization files are
216 no longer processed with LOAD, but by READ and EVAL; hence the
217 global optimization policy, startup package, readtable, etc,
219 * The .fasl file format number has been incremented because of
220 various incompatible changes.
221 * internal entry point details and argument counts no longer appear
222 in backtraces unless explicitly requested by setting
223 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
224 * built-in and standard functions no longer have names like "top
225 level local call to FOO".
226 * fixed bug 32: functions defined in non-null lexical environments
227 now have more legible printed representation
228 * fixed bug 33: functions defined in non-null lexical environemnts
229 are now more amenable to inspection by INSPECT.
230 * workaround for bug 354: XEPs no longer appear in backtraces unless
231 explicitly requested.
232 * fixed bug: receiving the signal which results from attempting to
233 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
234 SIGBUS on other *BSDs) on architectures where the C stack is also
235 the Lisp stack (x86 and x86-64) from foreign code no longer leads
236 to debugger confusion or wild instability. (reported by Cheuksan
238 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
239 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
240 (reported by Lutz Euler)
241 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
242 compiled code. (reported by Paul Dietz)
243 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
244 specialized parameter to a method to something that is not TYPEP
245 the specializer is now possible.
246 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
247 face of package deletion.
248 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
249 pathname. (reported by David Tolpin on c.l.l)
250 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
251 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
252 STRING-STREAM and FILE-STREAM, respectively.
253 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
254 than previously; in addition, its test suite is now run on build.
255 (thanks to Luke Gorrie)
256 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
257 * optimization: added a immediate representation for single-floats
259 * fixed some bugs revealed by Paul Dietz' test suite:
260 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
261 ** attempting to create a package with a colliding nickname causes
262 correctable errors to be signalled.
263 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
264 ** several x86-64 backend bugs related to sign-extension of immediate
267 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
268 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
269 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
271 * fixed loading of multiply forward-referenced layouts.
272 (thanks to Cheuksan Wang)
273 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
274 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
275 (see COMPILE-FILE documentation for details of :PRINT option)
276 * fixed bugs 19 and 317: fixed-format floating point printing is
277 more accurate. This also fixes a bug reported by Adam Warner
278 related to the ~@F format directive.
279 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
281 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
282 dispatch table if the from-char is a dispatch macro character.
283 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
284 by the wordsize no longer ignore the last word. (reported by Lutz
286 * fixed bug in type checking of dynamic-extent variables. (reported
288 * optimization: sequence traversal functions use their freedom to
289 coerce function designators to functions.
290 * optimization: code with many calls to CLOS methods specialized on
291 CLOS classes has had redundant type checks removed.
292 * optimization: added declarations to speed up operations that access
293 the internal character database (for example STRING-UPCASE)
294 * optimization: comparison operations between floats and sufficiently small
295 fixnums no longer create extra rationals
296 * fixed some bugs related to Unicode integration:
297 ** portions of multibyte characters at the end of buffers for
298 character-based file input are correctly transferred to the
299 start of the buffer at the next read.
300 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
301 passing it through to OPEN.
302 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
303 argument given to internal calls to OPEN.
304 ** the built-in comment readers (introduced by character sequences
305 ";" and "#|") are more forgiving to encoding errors; they will
306 STYLE-WARN and then attempt to resync the stream at a character
307 boundary. (thanks to Teemu Kalvas)
308 * fixed some bugs in the x86-64 port:
309 ** Negative short int return values from c-calls are sign-extended
311 ** The stack is aligned to 16-bytes for c-calls, as required by
312 the ABI. (thanks to Cheuksan Wang)
313 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
314 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
315 for enums. (thanks to Vincent Arkesteijn)
316 ** Multiple small optimizations and bugfixes for floating point
318 * fixed some bugs revealed by Paul Dietz' test suite:
319 ** Space, Tab, Linefeed, Return and Page have the invalid
320 secondary constituent character trait.
321 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
323 ** WITH-INPUT-FROM-STRING should only update the index place on
325 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
327 ** Bit-array operations (BIT-AND and similar) worked incorrectly
328 with one-dimensional arrays with fill pointers.
329 ** TYPE-OF failed on a complex with an integer realpart and a
331 ** compiler failure during type inference for the code of form
332 (IF (EQL X (THE ...)) ...) (MISC.535).
334 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
335 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
336 platform. The port supports 61-bit fixnums, large memory spaces
337 and reloading shared object files.
338 * enhancement: saving cores with foreign code loaded is now
339 supported on ppc/Darwin in addition to the previously supported
341 * enhancement: the statistical profiler now walks deeper into the
342 call stack for more meaningful call-graphs and accrued time
343 reports (x86/x86-64 only). It also now reports time spent in
345 * enhancement: it is now possible to trace most individual methods
346 of a generic function in addition to tracing the generic function
348 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
350 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
351 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
352 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
353 produces an error. (thanks to Vincent Arkesteijn)
354 * bug fix: NAMESTRING on pathnames with :WILD components in their
355 directories works correctly. (thanks to Artem V. Andreev)
356 * fixed bug 125: compiler preserves identity of closures. (reported
358 * bug fixed: functions with &REST arguments sometimes failed with
359 "Undefined function" when compiled with (DEBUG 3). (reported by
361 * bug fix: overflow during compiling of setting element of a bit
362 vector with constant index and value. (reported by Timmy Douglas)
363 * build fix: fixed the dependence on *LOAD-PATHNAME* and
364 *COMPILE-FILE-PATHNAME* being absolute pathnames.
365 * on x86 compiler partially supports stack allocation of dynamic-extent
367 * GO and RETURN-FROM do not check the extent of their exit points
368 when compiled with SAFETY 0.
369 * fixed some bugs related to Unicode integration:
370 ** encoding and decoding errors are now much more robustly
371 handled; it should now be possible to recover even from invalid
372 input or output to the terminal. (thanks to Teemu Kalvas)
373 ** provided a first cut at implementing STRING-TO-OCTETS and
374 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
375 ** altered the SB-MD5 contributed module slightly, changing the
376 interface just enough for it to be supportable for builds where
377 lisp characters are not eight bits.
378 * fixed some bugs revealed by Paul Dietz' test suite:
379 ** the FORMATTER-generated functions for ~V[ conditionals require
380 the correct number of arguments.
381 ** READ-FROM-STRING returns the mandated second value when applied
382 to displaced strings.
383 ** the #\Rubout and #\Backspace characters are treated as invalid
384 constituent characters by the tokenizer.
386 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
387 * new feature: reloading changed shared object files with
388 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
389 * new feature: references to foreign variables and functions
390 can now be compiled and loaded before the corresponding shared
391 object file is loaded, as long as the foreign definitions are
392 available at runtime.
393 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
394 supported. (thanks to Dan Debertin)
395 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
396 just keywords. (thanks to Vincent Arkesteijn)
397 * fixed bug #331: structure-class instances corresponding to
398 DEFSTRUCT forms are now created eagerly.
399 * fixed bug #345: backtraces from calls to undefined functions work
400 on x86 as well. Related bug #61 is now also partially fixed on x86
401 (backtraces from throws to unknown catch tags.)
402 * bug fix: lambda-list parsing is now stricter vrt. order and number
403 of lambda-list keywords.
404 * bug fix: as specified by AMOP, an error is signalled if a
405 class-option appears multiple times in a DEFCLASS form. (reported
407 * bug fix: RANDOM can be compiled when the compiler derives the type
408 of its numeric argument as a disjoint set of small integers.
409 (reported by Paul Dietz)
410 * bug fix: starting a core saved with shared objects loaded when
411 those objects are not available no longer causes threaded SBCL to
412 hang. (reported by Sean Ross)
413 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
415 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
416 argument, which must always be NIL. (reported by Kalle Niemitalo)
417 * bug fix: printing 1.0d+23 no longer results in an error.
418 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
419 * bug fix: structure slot setters preserve evaluation order. (thanks
421 * fixed some bugs related to Unicode integration:
422 ** RUN-PROGRAM can allow its child to take input from a Lisp
423 stream. (reported by Stefan Scholl)
424 ** ASDF-INSTALL successfully downloads .tar.gz files in all
425 locales. (reported by Ken Causey)
426 * fixed some bugs revealed by Paul Dietz' test suite:
427 ** INCF, DECF and REMF evaluate their place form as specified in
429 ** FORMATTER expands ~{ ~} iteration directives with V or #
430 parameters correctly.
431 ** FORMATTER deals with the ~@[ ~] conditional directive where the
432 consequent uses no arguments correctly.
433 ** the system has a partial understanding of the (COMPLEX RATIO)
435 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
438 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
439 * new feature: a build-time option (controlled by the :SB-UNICODE
440 keyword feature, enabled by default) for building the system with
441 support for the entire 21-bit character space defined by the
443 * new feature: the system now has rudimentary external-format
444 support; the primary user-visible change at this time is that
445 characters with the high bit set (such as lower-case-e-acute) will
446 print correctly to a terminal in a UTF-8 environment.
447 * minor incompatible change: BASE-CHAR no longer names a class;
448 however, CHARACTER continues to do so, as required by ANSI.
449 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
450 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
451 has been moved to the SB-EXT package (temporarily re-exported from
453 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
454 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
456 * fixed bug #351: better error-handlind and reporting for malformed
458 * fixed bug #350: bignum-printing is now more memory-efficient,
459 allowing printing of very large bignums, eg. (expt 2 10000000).
460 (reported by Bruno Haible)
461 * fixed bug #302: better primitive-type selection for intersection
463 * fixed bug #308: non-graphic characters now all have names, as
464 required. (reported by Bruno Haible)
465 * bug fix: redefining a class with different superclasses now correctly
466 removes it from the direct-subclasses of its previous superclasses.
467 (reported by David Morse)
468 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
469 new value now works. (reported by Bruno Haible)
470 * bug fix: correct canonicalization of multiple non-standard slot
471 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
472 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
473 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
474 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
475 now exists, an signals an error.
476 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
477 during macroexpansion for non-iteration variables as well. (reported
478 by Bruno Haible for CMUCL)
479 * bug fix: Cyclic structures and unprintable objects in compiler
480 messages no longer cause errors. (reported by Bruno Haible)
481 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
482 and READ-FROM-STRING all now return a primary value of NIL if
483 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
484 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
485 (reported by Bruno Haible for CMUCL)
486 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
487 do not designate a condition. (reported by Bruno Haible for
489 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
490 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
491 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
492 permitting resolution in favour of any of the conflicting symbols.
493 (reported by Bruno Haible for CMUCL)
494 * bug fix: EQUAL compiler optimizations is less aggressive on
495 strings which can potentially compare true despite having distinct
496 specialized array element types.
497 * bug fix: unit enumerations can be defined without dividing by
498 zero. (thanks to Vincent Arkesteijn)
499 * FORMAT compile-time argument count checking has been enhanced.
500 (report from Bruno Haible for CMUCL)
501 * a partial workaround for the bug 262: the compiler does not try to
502 inline-expand a local function doing RETURN-FROM from a deleted
503 BLOCK. (thanks to Peter Denno for the bug report and to David
504 Wragg for the simple test case)
505 * fixed some bugs revealed by Paul Dietz' test suite:
506 ** RENAME-PACKAGE allows all package designators as new package
508 ** constraint propagation and lambda variable substitution are
509 more cautious in dealing with partially deleted code.
510 ** compiler sometimes lost reoptimization passes.
511 ** CERROR, given a condition as condition designator, passes its
512 remaining arguments to the continue format control without
514 ** Case-altering FORMAT directives work correctly on non-ASCII
516 ** The REAL type specifier handles bounds outside the floating
517 point ranges without signalling FLOATING-POINT-OVERFLOW.
518 ** Functions with IR1-transformations can create intercomponent
519 references to global functions.
520 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
522 ** FORMAT ~R treats a nil value for its first parameter correctly.
524 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
525 * enhancement: saving cores with foreign code loaded is now
526 supported on x86/NetBSD and sparc/Linux in addition to the previously
528 * bug fix: on some platforms repeated installations caused multiple
529 copies of HTML documentation to be installed -- should not happen
530 any more. (reported by Stefan Scholl)
531 * bug fix: parsing self-recursive alien record types multiple times
532 no longer causes infinite recursion. (reported by Thomas F. Burdick,
533 original patch by Helmut Eller for CMUCL)
534 * bug fix: stack-exhaustion detection works now on NetBSD as well.
535 (thanks to Richard Kreuter)
536 * bug fix: defining classes whose accessors are methods on existing
537 generic functions in other (locked) packages no longer signals
538 bogus package lock violations. (reported by François-René Rideau)
539 * bug fix: special variables as DEFMETHOD parameters no longer have
540 associated bogus type declarations. (reported by David Wragg and
542 * bug fix: read-write consistency on streams of element-type
543 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
544 * bug fix: redefiniton of the only method of a generic function with
545 no DEFGENERIC no longer emits a full WARNING. In addition,
546 redefinition of generic functions with no DEFGENERIC to an
547 incompatible lambda list now signals an error. (thanks to Zach
549 * bug fix: DEFGENERIC now works even when there's a function of the
550 same name in an enclosing lexical environment. (thanks to Zach
552 * fixed compiler failure, caused by instrumenting code during
553 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
554 * optimization: added loop analysis and improved register allocation
555 * fixed some bugs revealed by Paul Dietz' test suite:
556 ** POSITION on displaced vectors with non-zero displacement
557 returns the right answer.
558 ** (SIMPLE-STRING) is a valid type specifier for sequence
560 ** *PRINT-LEVEL* handling for slotless structures is pedantically
562 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
563 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
565 ** The justification version of the FORMAT ~< directive treats
566 non-zero minpad parameter correctly.
568 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
569 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
570 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
571 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
572 the supported interface.
573 * new feature: Single-stepping of code compiled with DEBUG 2 or
574 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
575 * new feature: saving cores with foreign code loaded is now
576 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
577 Timothy Moore's work for CMUCL)
578 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
579 parameters to * instead of NIL if no initform is supplied.
580 (reported by Johan Bockgaard)
581 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
582 correctly when FUNCALL appears as the car of the form. Note:
583 despite this FUNCALL forms are not currently subject to
584 compiler-macro expansion. (port of Raymond Toy's fix for the
585 same from CMUCL, reported by Johan Bockgaard)
586 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
587 (thanks for Teemu Kalvas)
588 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
589 case instead of terminating the iteration (thanks for Julian
590 Squires, Sean Champ and Raymond Toy)
591 * bug fix: incorrect expansion of defgeneric that caused a style
592 warning. (thanks for Zach Beane)
593 * bug fix: slot accessor effective method computation works properly
594 for classes with multiple non-standard applicable methods on
595 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
596 * on x86 compiler supports stack allocation of results of LIST and
597 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
598 CMUCL implementation by Gerd Moellmann)
599 * fixed some bugs revealed by Paul Dietz' test suite:
600 ** FORMAT strings with both the ~<~:;~> form of the justification
601 directive and pretty-printing directives cause an error.
603 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
604 * incompatible change: the internal functions
605 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
606 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
607 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
608 instead of the old functions.
609 * new feature: on platforms where "dladdr" is available foreign
610 function names now appear in backtraces. (based on Helmut Eller's
612 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
613 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
615 * documentation: ASDF manual is now distributed with SBCL.
616 * bug fix: SBCL can now load its contributed modules with REQUIRE
617 even if the system-provided entries have been removed from
618 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
620 * bug fix: backtraces involving undefined functions or assembly
621 routines are more informative. (thanks to Brian Downing)
622 * bug fix: mutually referent alien structures now work correctly.
623 (reported by Rick Taube)
624 * bug fix: structures defined by WITH-ALIEN can be referred to
625 within other definitions in the same WITH-ALIEN.
626 * bug fix: division operators (MOD, TRUNCATE and the like) with
627 constant zero divisors and integer dividends no longer generate
629 * bug fix: provide default methods for INPUT-STREAM-P and
630 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
631 * bug fix: improve the detection and resolution of MOP metacycles
632 (where effective method computation depends on the generic function
633 having its effective method computed). (reported by Bruno Haible)
634 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
635 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
636 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
637 * bug fix: pathnames with relative directory components can be
638 represented relative to default pathnames.
639 * optimization: in taking the GCD of bignums, reduce the two bignums
640 to approximately the same size (using Euclid's algorithm) before
641 applying the more sophisticated binary GCD. (thanks to Juho
643 * optimization: COUNT on bitvectors now operates word-at-a-time.
644 * optimization: ASH with a positive, but not necessarily constant,
645 (leftwards) shift, when in a modular context, is compiled to a
647 * fixed some bugs revealed by Paul Dietz' test suite:
648 ** FORMAT variable parameters ("~V<char>") are defaulted properly
649 if the corresponding argument is NIL.
650 ** FORMAT directives accepting a minpad parameter treat negative
652 ** literal commas as character parameters to FORMAT directives are
654 ** literal spaces directly after ~<Newline> directives within a
655 format-logical-block (~:< ~@:>) do not induce :FILL-style
656 conditional newlines.
657 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
658 as specified: it no longer includes conditional newlines).
659 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
661 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
662 :PER-LINE-PREFIX argument does not evaluate to a string.
664 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
665 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
666 the manual for details; for now, package locks can be disabled by
667 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
668 if no major problems are found then it is likely that they will be
669 compiled in unconditionally.
670 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
671 unsupported operators on all platforms. To load a shared library
672 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
673 object file, link it into a shared library outside of SBCL and
674 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
676 * bug workaround: ROOM T and the GC were not working together
677 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
678 reporting facility aren't preserved. That reporting has been
679 disabled, so now until and unless someone figures out how to make it
680 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
681 * minor incompatible change: as threatened around sbcl-0.8.0, the
682 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
683 The symbols are also exported from SB-PCL for backwards
684 compatibility, but more so than before SB-PCL should be treated as
685 an implementation-internal package.
686 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
688 * the SB-SPROF contrib now works on (most) non-x86 architectures.
689 It is known as of this release not to work on the Alpha, however.
690 * fixed bug #167: errors signalled due to illegal syntax in method
691 bodies are now more legible.
692 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
693 designators and can hence be used with TYPEP.
694 * fixed bug #333: CHECK-TYPE now ensures that the type error
695 signalled, if any, has the right object to be accessed by
696 TYPE-ERROR-DATUM. (reported by Tony Martinez)
697 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
698 5.1.2.3 for multiple-value place subforms. (reported by Kalle
700 * fixed bug #334: programmatic addition of slots using specialized
701 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
702 and :ALLOCATION :CLASS slots.
703 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
704 float exponents. (rereported by Peter Seibel)
705 * fixed a bug: #\Space (and other whitespace characters) are no
706 longer considered to be macro characters in standard syntax by
708 * fixed bug: initialization of condition class metaobjects no longer
709 causes an instance of the condition to be created. (reported by
711 * fixed bug: it is now possible to have slots such that
712 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
713 description is neither :INSTANCE nor :CLASS.
714 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
715 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
716 SLOT-VALUE-USING-CLASS), and no longer causes errors with
717 non-standard SLOT-DEFINITION-ALLOCATION values.
718 * fixed bugs: various race conditions handling exiting threads.
719 CL-PPCRE's thread torture test now passes.
720 * fixed arguably-a-bug: GC time proportional to number of threads in
721 system even when most of them are idle
722 * optimization: improved performance of BIT and SBIT on bit-vectors.
723 * .fasl file incompatibility: The fasl file version number has
724 been incremented because of changes associated with package locks.
726 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
727 * minor incompatible change: the system no longer provides
728 optimization hints (conditions which are TYPEP
729 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
730 mode; these hints are still emitted when higher SPEED optimization
732 * new contrib module: a sampling profiler (profiling by statistical
733 sampling, rather than by instrumenting functions) is available as
734 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
735 * the behaviour of the standard function ED is now customizeable by
736 third parties through a hook variable: see ED's documentation
737 string for information on the protocol.
738 * the compiler no longer emits efficiency notes for (FUNCALL X)
739 when the type of X is uncertain under default optimization
741 * fixed bug 276: mutating a binding of a specialized parameter to a
742 method to something that is not TYPEP the specializer is now
744 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
745 exhibit the required behaviour on the x86 platform. (thanks to
746 Peter van Eynde, Eric Marsden and Bruno Haible)
747 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
748 even for difficult arguments. (reported by Peter Graves)
749 * fixed bug 141a: the backquote printer now descends quoted
751 * fixed another bug in backquote printing: no more destructive
752 modification of the form's list structure. (reported by Brian
754 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
755 move between its address being taken and the call to
756 interrupt_thread, fixing a crashing race condition.
757 * the SB-POSIX contrib implementation has been adjusted so that it
758 no longer exhibits ridiculously poor performance when constructing
759 instances corresponding to C structs.
761 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
762 * minor incompatible change: the sb-grovel contrib now treats C
763 structures as alien (in the sense of SB-ALIEN) objects rather than
764 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
765 has implications for memory management of client code
766 (sb-grovel-returned objects must now be manually managed) and for
767 type safety (alien objects now have full types).
768 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
769 used to control emission of compiler diagnostics, rather than the
770 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
771 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
772 quality should be considered deprecated.
773 * install.sh now installs the user manual as well
774 * (not quite a new documentable feature, but worth considering in
775 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
776 beginnings of a semantically meaningful condition hierarchy is
777 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
779 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
780 designator as the defaults argument.
781 * fixed bug: Displaced arrays whose displaced-to array has become
782 too small now cause ARRAY-DIMENSION to signal an error, providing
783 for safer bounds-checking. (reported by Bruno Haible)
784 * fixed bug: DEFCLASS slot definitions with identical :READER and
785 :WRITER names now signal a reasonable error. (reported by Thomas
787 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
789 * fixed bug: Invalid slot specification errors now print correctly.
790 (thanks to Zach Beane)
791 * fixed bug 320: Shared to local slot value transfers in class
792 redefinitions now happen corrently from superclasses as
793 well. (reported by Bruno Haible)
794 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
796 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
797 improper lists correctly. (reported by Bruno Haible)
798 * fixed bug 313: source-transform for <fun-name> was erroneously
799 applied to a call of a value of a variable with name <fun-name>.
800 (reported by Antonio Menezes Leitao)
801 * fixed bug 307: The obsolete instance protocol ensures that
802 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
803 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
804 remove dead unknown-values globs from the middle of the stack.
805 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
806 easier to resume long computations after using *BREAK-ON-SIGNALS*
807 to diagnose and fix failures. (thanks to Nikodemus Siivola)
808 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
809 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
810 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
811 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
812 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
814 * fixed a (fairly theoretical) bug in string printing: if
815 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
816 does not have array-element-type equal to the most general string
818 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
819 function name. (thanks to Nikodemus Siivola)
820 * fixed bug: compile-time format string checker failed on
821 non-closed ~{. (reported by Thomas F Burdick)
822 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
823 don't warn on legal loop code involving more than one aggregate
824 boolean. (thanks to Nikodemus Siivola)
825 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
826 in keyword position to MAKE-PACKAGE was not being reported as
828 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
829 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
831 * fixed bugs 280 and 312: the checking for multiple definitions in a
832 file is less likely to become confused by uses of inline
834 * fixed bug: the #S reader macro performs the keyword coercion
835 specified for slot names. (reported by Kalle Niemitalo)
836 * fixed bug: lambda lists may contain symbols whose names start with
837 & but are not lambda-list-keywords; their occurrence triggers a
839 * fixed bug 321: define-method-combination argument lambda lists do
840 not cause invalid code to be generated when &optional or &aux
841 variables have default values. (reported by Bruno Haible)
842 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
843 have CLOS classes; this appears also to have fixed reported
844 instability in stack exhaustion detection.
845 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
846 MUFFLE-WARNING when no associated restart is present is now
848 * optimization: rearranged the expansion of various defining macros
849 so that each expands into only one top-level form in a
850 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
852 * optimization: used a previously little-used slot in symbols to
853 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
855 * fixed some bugs revealed by Paul Dietz' test suite:
856 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
858 ** Fixed an optimization bug related to inheritance of initforms
859 from local to shared slots.
860 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
861 ** CAST optimizer forgot to flush argument derived type.
862 ** print/read consistency on floats is now orders of magnitude
863 more likely. (thanks also to Bruno Haible for a similar report
865 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
866 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
867 for objects of type REAL. Make it so.
868 ** SXHASH is specified (infelicitously) to respect similarity,
869 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
870 it so. (thanks to Markus Ziegler)
871 ** on the Alpha, the compiler succeeds in compiling functions
872 returning a known number of arguments greater than 63.
873 ** fixed handling of invalid NIL arguments in keyword position
875 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
877 ** provided a readably-printable representation for RANDOM-STATE
879 ** ensured that pathnames, where they have a namestring, always
880 print using #P"..." syntax.
882 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
883 * Support for the forthcoming 2.0 version of the NetBSD kernel
884 running on x86 hardware has been added. (thanks to Perry
885 E. Metzger most immediately, and others for their past work)
886 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
887 current release version; SBCL's OpenBSD support had been broken
888 since about the time of OpenBSD's switch to ELF binary format.)
889 * [placeholder for DX summary]
890 ** user code with &REST lists declared dynamic-extent, under high
891 speed or space and low safety and debug optimization policy.
892 * The manual has been converted to Texinfo format and the debugger
893 chapter from the cmucl manual has been added.
894 * A facility has been added to extract documentation strings from
895 sbcl and store them as Texinfo-formatted snippets for inclusion in
896 the manual (via Texinfo's @include directive)
897 * bug fix: compiler emitted division in optimized DEREF. (thanks for
898 the test case to Dave Roberts)
899 * bug fix: multidimensional simple arrays loaded from FASLs had fill
900 pointers. (reported by Sean Ross)
901 * bug fix: PROFILE output is printed nicely even for large numerical
902 values. (thanks to Zach Beane)
903 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
904 greater than 32 handle EOF correctly.
905 * bug fix: on X86 an immediate argument of the IMUL instruction is
906 correctly printed in disassembly. (thanks to Lutz Euler)
907 * bug fix: class slots in redefined classes preserve their old
908 values. (thanks to Bruno Haible and Nikodemus Siivola)
909 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
910 succeeds. (reported by Marco Baringer)
911 * fixed some bugs revealed by Paul Dietz' test suite:
912 ** READ-SEQUENCE now works on ECHO-STREAMs.
913 ** RATIONALIZE works more according to its specification. (thanks
916 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
917 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
918 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
919 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
920 should matter to you only if you rebind the printer control
921 variables and then find you want different bindings in the
922 debugger than in the ordinary execution of your program.)
923 * The runtime build system has been tweaked to support building
924 (on SPARC/SunOS) using a C compiler which invokes Sun's own
925 assembler and linker. (thanks to Nikodemus Siivola)
926 * Unbound, undefined, undeclared variables now trigger full
927 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
928 more often programmer error than deliberate exploitation of undefined
930 * optimization: the hash algorithm for strings has changed to one
931 that is less vulnerable to spurious collisions. (thanks to Juho
933 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
934 needless bounds checking. (thanks to Juho Snellman)
935 * optimization: implemented multiplication as a modular
936 (UNSIGNED-BYTE 32) operation on the PPC backend.
937 * fixed some bugs revealed by Paul Dietz' test suite:
938 ** ADJUST-ARRAY now copies the datum in a zero rank array if
940 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
941 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
942 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
944 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
946 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
948 ** ECHO-STREAMs no longer attempt to echo the end of file value to
949 their output stream on EOF from read.
950 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
951 have been read to end-of-file.
952 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
954 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
955 description of determination of which consecutive characters
957 ** Printing the "Space" character with escaping on now yields "#\\ ",
958 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
959 ** Reading floating-point numbers with *READ-BASE* set to a number
960 less than 10 works correctly.
961 ** Reading floating-point numbers with *READ-BASE* set to a number
962 more than 10 works correctly.
963 ** Printing with *PRINT-READABLY* targets the standard readtable, not
964 the readtable currently in effect.
966 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
967 * minor incompatible change: parsing of namestrings on a physical
968 (Unix) host has changed; numbers after the final #\. in a
969 namestring are no longer interpreted as a version field. This is
970 intented to be largely invisible to the user, except that the
971 meaning of the namestring "*.*.*" has changed: it now refers to a
972 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
973 should usually be replaced by
974 (make-pathname :name :wild :type :wild :version :wild)
975 with the added benefit that this is more likely to be portable.
976 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
977 option to OPEN now signals an error if the file being opened
978 exists; this may have an impact on existing code.
979 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
980 signals received, on the PowerPC platforms (both Linux and
981 Darwin). (thanks to Pierre Mai for pointing out the location of
983 * several fixes on OS X: The system now builds and runs cleanly on
984 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
985 (thanks to Brian Mastenbrook)
986 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
987 second-resolution: integer multiples of 1/3600 between -24 and 24.
988 (thanks to Vincent Arkesteijn)
989 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
990 type when called with 1 argument; PEEK-CHAR checked type of
991 PEEK-TYPE only after having read first character from a
992 stream. (reported by Peter Graves)
993 * bug fix: the garbage collector now has much better locality
994 behaviour, and in particular no longer treats all memory as being
995 exhausted when many small objects point to each other in a deeply
997 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
998 recognized as being TYPEP their class.
999 * bug fix: the PUSHNEW documentation string has been corrected.
1000 (thanks to Vincent Arkesteijn)
1001 * bug fix: defaulting of the value for the last of an atypically
1002 large number of multiple values being bound was not being
1003 performed correctly on the Alpha or PPC platforms
1004 * optimization: implemented multiplication as a modular
1005 (UNSIGNED-BYTE 32) operation on the x86 backend.
1006 * optimization: SEARCH on simple-base-strings can now be open-coded.
1007 (see also contrib/compiler-extras.lisp for inspiration for
1008 teaching the compiler about the Boyer-Moore algorithm).
1009 * value, returned by MAX (and MIN) called with several EQUALP, but
1010 not EQL, arguments now does not depend on compiler settings.
1011 * fixed some bugs revealed by Paul Dietz' test suite:
1012 ** in stack analysis liveness information is propagated from
1013 non-local entry points.
1014 ** pathwise CAST removing failed when the CAST node did not start
1016 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1017 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1019 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1020 host is already defined.
1021 ** RENAME-FILE works on streams instead of signalling an internal
1023 ** PEEK-CHAR uses the current readtable when determining whether
1024 or not a character is whitespace.
1025 ** MERGE-PATHNAMES handles the case when the pathname does not
1026 specify a name while the default-pathname specifies a version.
1027 ** Pathnames now stand a better chance of respecting print/read
1029 ** Attempting to use standardized file system operators with a
1030 pathname with invalid :DIRECTORY components signals a
1032 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1034 ** DIRECTORY on logical pathnames is more correct.
1035 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1036 signal a TYPE-ERROR if their argument is not a stream.
1037 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1038 designator argument does not designate a stream.
1039 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1040 examining the synonym.
1041 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1043 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1044 element-type larger than ([UN]SIGNED-BYTE 32).
1046 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1047 * When built with the :SB-FUTEX feature, threaded builds now take
1048 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1049 for faster/more reliable mutex and condition variable support.
1050 * Incompatible change (but one you probably shouldn't have been using
1051 anyway): the interface and code for arbitrating between multiple
1052 threads in the same user session has been redesigned.
1053 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1054 error when called without an explicit environment argument.
1055 (thanks to Nikodemus Siivola)
1056 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1057 confused about their position. (thanks to Adam Warner and Gerd
1059 * bug fix: Pretty printing backquoted forms with unquotations in the
1060 argument list position of various code constructs such as LAMBDA
1061 now works correctly. (reported by Paul Dietz)
1062 * bug fix: Pretty printing unquotations no longer loses all
1063 stream position information.
1064 * optimization: performance of string output streams is now less
1065 poor for multiple small sequence writes.
1066 * optimization: performance of CSUBTYPEP in the presence of complex
1067 expressions involving CONS and NOT many times has been improved.
1068 (reported by Paul Dietz)
1069 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1071 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1072 streams for the REPL, for the debugger, and so on. (thanks to
1074 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1075 * fixed some bugs revealed by Paul Dietz' test suite:
1076 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1077 is ignored if it is too small, rather than propagating through
1078 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1079 ** extremely complex negations of CONS types were not being
1080 sufficiently canonicalized, leading to inconsistencies in
1082 ** VALUES tranformer lost derived type.
1084 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1085 * fixed a bootstrapping bug: the build process no longer assumes
1086 that the various BOOLE-related constants have the same value in
1087 host and target lisps. (noted by Paul Dietz' test suite on an
1088 SBCL binary built from CLISP)
1089 * The system can now be dynamically linked on the MIPS platform,
1090 which enables dynamic loading of foreign code from Lisp. (thanks
1091 to Ralf Baechle for discussions on the MIPS ABI)
1092 * The system now records debugging information for its own source
1093 files in a filesystem-position-independent manner, relative to
1094 the "SYS" logical host.
1095 * fixed a compiler bug: MV-LET convertion did not check references
1096 to the "max args" entry point. (reported by Brian Downing)
1097 * tweaked disassembly notes to be less confident about proclaiming
1098 some instruction as an LRA. (thanks to Brian Downing)
1099 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1100 now each have their own history, command character, and other
1101 characteristics. (thanks to David Lichteblau)
1102 * fixed some bugs revealed by Paul Dietz' test suite:
1103 ** compiler failure in compiling LOGAND expressions including a
1105 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1106 shift greater than 32.
1107 ** FLUSH-DEST did not mark blocks for type check regeneration.
1108 ** HANDLER-CASE failed to accept declarations in handler clauses
1109 in some circumstances.
1111 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1112 * New code in contrib/sb-introspect (still probably not entirely
1113 stable yet) provides some support for smart Lisp development
1114 environments like SLIME.
1115 * The conditions signalled for errors occurring when loading .fasl
1116 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1117 in a way which should help ASDF recover gracefully.
1118 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1119 stuff has been cleaned up. If you code contrib/ stuff, this might
1120 affect you, and you can look at contrib/README, contrib/STANDARDS,
1121 and/or the 0.8.4.27 diff to check.
1122 * In full calls the compiler now does not generate checks for declared
1123 argument types for all arguments.
1124 * various threading fixes
1125 ** and some experimental patches which didn't make it into
1126 the main tree for this release, but which are shipped in
1127 contrib/experimental-thread.patch as a possible fix for some
1128 failures (deadlock, spinning...) in GC-intensive multithreaded
1130 * fixed PPC build problem (source code incompatibility of different
1131 library versions): added offsetof() hackery which attempts to divine
1132 where glibc maintainers put uc_mcontext today
1133 * fixed bug 282: compiler does not trust type assertions while passing
1134 arguments to a full call.
1135 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1136 &OPTIONAL VALUES type parameter.
1137 * fix bug 214: algorithm for noting rejected templates is now more
1138 similar to that of template seletion. (also reported by rydis on
1140 * fixed bug 141b: printing backquoted information readably and prettily
1141 inserts a space where necessary.
1142 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1143 (CHARACTER 10) are now reported as errors, rather than propagated
1144 as unknown types. (reported by piso on #lisp)
1145 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1146 with values NIL and :ERROR. (thanks to Milan Zamazal)
1147 * fixed bug 191c: CLOS now does proper keyword argument checking as
1148 described in CLHS 7.6.5 and 7.6.5.1.
1149 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1150 longer raise an error; further, using a list as a for-as-arithmetic
1151 counter now raises a meaningful error.
1152 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1153 now understood by sequence creation functions such as MAKE-SEQUENCE
1155 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1156 a STRING-INPUT-STREAM.
1157 * compiler enhancement: SIGNUM is now better able to derive the type
1159 * type declarations inside WITH-SLOTS are checked. (reported by
1161 * fixed some bugs revealed by Paul Dietz' test suite:
1162 ** incorrect optimization of TRUNCATE for a positive first
1163 argument and negative second.
1164 ** compiler failure in let-convertion during flushing dead code.
1165 ** compiler failure while deriving type of TRUNCATE on an
1166 interval, containing 0.
1167 ** ASH of a negative bignum by a negative bignum count now returns
1169 ** intersection of CONS types now canonicalizes properly, fixing
1170 inconsistencies in SUBTYPEP.
1172 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1173 * incompatible change: The --disable-debugger command line
1174 option now clobbers the debugger at a more fundamental
1175 level, by redefining #'INVOKE-DEBUGGER instead of by
1176 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1177 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1178 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1179 enter the debugger REPL and then suffer recursive errors
1180 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1181 while under the new system, BREAK in --disable-debugger mode
1182 terminates the system just as an unhandled error would.
1183 * fixed compiler performance when processing loops with a step >1;
1184 * bug fix: DOCUMENTATION now retrieves generic function
1185 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1186 support has been systematized, and now supports the methods
1187 specified by ANSI, along with a default method and a method for
1188 slot documentation. (reported by Nathan Froyd)
1189 * bug fix: effective methods associated with a generic function are
1190 no longer cached over a change of that generic function's method
1191 combination. (reported by Andreas Fuchs)
1192 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1193 names a non-existent directory. (thanks to Andreas Fuchs)
1194 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1195 the PPC platform, lead to stack corruption; this has been fixed.
1196 (reported by Rainer Joswig)
1197 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1198 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1199 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1200 arithmetic in general, is now much more reliable on the PPC
1202 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1203 platform now returns the right answer.
1204 * optimization: restored some effective method precomputation in
1205 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1206 precomputation is now tunable.
1207 * optimization: compiler-internal data structure use has been
1208 reviewed, and changes have been made that should improve the
1209 performance of the compiler by about 20%.
1210 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1211 simple-base-strings and simple-bit-vectors is improved.
1212 * optimization: the optimization of 32-bit logical and arithmetic
1213 functions introduced in version 0.8.3 on the x86 has been
1214 implemented on the mips, ppc and sparc platforms; an
1215 implementation of the same facility, but for 64-bit arithmetic,
1216 has been added for the alpha.
1217 * microoptimization: the compiler is better able to make use of the
1218 x86 LEA instruction for multiplication by constants.
1219 * bug fix: in some situations compiler did not report usage of
1220 generic arithmetic in (SPEED 3) policy.
1221 * bug 145b fix: compiler used wrong type specifier while converting
1222 MEMBER-types to numeric.
1223 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1225 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1226 index argument now works properly on the Alpha platform.
1227 * bug fix: floating point exception treatment on the Alpha platform
1229 * bug fix: FILE-POSITION works much better on string input and
1230 output streams. (thanks to Nikodemus Siivola)
1231 * bug fix: many threading/garbage collection symptoms sorted.
1232 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1233 might be pseudo-atomic.
1234 * internal change: Stopping for GC is now done with signals not
1235 ptrace. GC is now done in whichever thread wanted it, instead of
1237 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1239 * bug fix: result form in DO is not contained in the implicit
1241 * incompatible change: ICR structure is changed; the value part of
1242 CONTINUATION is now called LVAR; corresponding functions are
1243 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1244 * added type deriver for ISQRT (thanks to Robert E. Brown).
1245 * bug fix: better support for loading from the command line when an
1246 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1248 * fixed some bugs revealed by Paul Dietz' test suite:
1249 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1250 ** ROUND and FROUND now give the right answer when given very
1251 small float arguments.
1252 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1254 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1255 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1256 ** LCM should return a non-negative integer.
1257 ** PARSE-INTEGER returned the index of a terminator instead of the
1258 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1259 ** PARSE-INTEGER returned an incorrect index being applied to a
1261 ** LCM with two arguments of 0 returns 0 rather than signalling
1263 ** unsigned addition of a 32-bit constant with the high bit set no
1264 longer causes an internal compiler error on the x86.
1265 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1266 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1267 with negative last argument.
1268 ** byte specifiers with zero size and position no longer cause
1269 an error during type derivation.
1270 ** bignum multiplication on the Alpha platform now returns the
1272 * porting: The system now builds on SuSE AMD64, although it still
1273 generates a 32-bit binary.
1274 * .fasl file incompatibility: The fasl file version number has
1275 been incremented (because of the changes to internal compiler
1276 data structures referred to above).
1278 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1279 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1280 more accurately, on the Darwin kernel running on PowerPC hardware.
1281 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1282 * Compiler code deletion notes now signal a condition of type
1283 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1284 an associated MUFFLE-WARNING restart.
1285 * The compiler now performs limited argument count validation of
1286 constant format strings in FORMAT, and where appropriate in ERROR,
1287 CERROR and WARN. (thanks to Gerd Moellmann)
1288 * New ASDF-INSTALL contrib can be used for automatic download and
1289 installation of third-party Lisp code from CCLAN or other sites
1291 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1292 which forces another thread to execute a function supplied by the
1294 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1295 now accept and act upon their :ELEMENT-TYPE keyword argument.
1296 (reported by Martin Atzmueller, Edi Weitz)
1297 * bug fix: FILE-POSITION now accepts position designators up to
1298 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1299 is the greater. (thanks to Patrik Nordebo)
1300 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1301 Kalle Olavi Niemitalo)
1302 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1303 printable. (reported by Eric Marsden)
1304 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1305 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1306 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1307 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1308 (thanks to Stig Sandoe)
1309 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1310 UNREAD-CHAR now correctly includes the unread character in the
1311 target sequence. (thanks to Gerd Moellmann)
1312 * bug fix in threaded builds: the system can now be suspended and
1313 resumed by shell job control with minimal disruption.
1314 * bug fixes in times and timezones >2038AD
1315 * better handling of "where is GNU make?" problem in build scripts
1316 (thanks to Nikodemus Siivola)
1317 * new optimization: inside a named function any reference to a
1318 function with the same name is considered to be a self-reference;
1319 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1320 optimization quality.
1321 * new optimization on x86: logical functions and + now have
1322 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1323 used when the result is truncated to 32 bits.
1324 * VALUES declaration is partially enabled.
1325 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1326 * fixed some bugs revealed by Paul Dietz' test suite:
1327 ** The system now obeys the constraint imposed by
1328 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1329 types form a lattice under type intersection.
1330 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1331 ** ASSOC now ignores NIL elements in an alist.
1332 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1333 and (1+ MOST-POSITIVE-FIXNUM) answers.
1334 ** The addition of a method with invalid qualifiers to a generic
1335 function does not cause an error to be signalled immediately;
1336 a warning is signalled, and the error is generated only on
1337 calling the generic function.
1338 * changed .fasl file version number, in part to add type codes for
1339 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1340 obscure ANSI requirements
1342 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1343 * fixed bug 148: failure to inline-expand a local function left
1344 garbage, confusing the compiler.
1345 * fixed bugs 3cd: structure slot readers perform type check if the
1346 slot can have an invalid value (i.e. it is either not initialized
1347 or can be written with a less specific slot writer).
1348 * bug fix: the compiler now traps array references to elements off
1349 the end of an array; previously, the bounds checking in some
1350 circumstances could go off-by-one.
1351 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1353 * type declarations for array element types now obey the description
1354 on the CLHS page "Declaration TYPE", as per discussions on
1355 sbcl-help around 2003-05-08. This means that a declaration
1356 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1357 declaration, all references to BAR will be asserted or assumed
1358 (with THE, so dependent on compiler policy) to involve objects of
1359 type FOO. Note that no such declaration is implied in
1360 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1361 * declared types of functions from the "Conditions"
1362 chapter. (reported by Paul Dietz)
1363 * bug fix: CERROR accepts a function as its first argument.
1364 * bug fix: NTH an NTHCDR accept a bignum as index
1365 arguments. (reported by Adam Warner)
1366 * optimization: character compare routines now optimize comparing
1367 against a constant character. (reported by Gilbert Baumann)
1368 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1369 argument now works properly on the MIPS platform.
1370 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1371 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1372 declared to have a functional type, does not cause an error
1374 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1375 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1376 * fixed bug: if last continuation of a deleted block has a
1377 destination, this destination should be deleted too. (reported by
1379 * fixed a bug in the bootstrap process: the host compiler's values
1380 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1381 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1382 test case from Patrik Nordebo)
1383 * improved the ability of the disassembler on the PPC platform to
1384 provide helpful disassembly notes.
1385 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1386 the class in more cases than previously.
1387 * bug fix: FILE-POSITION now understands :START and :END for
1388 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1389 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1390 8) data. (thanks to David Lichteblau)
1391 * bug fix: it is possible to add a method to a generic function
1392 without lambda list.
1393 * bug fix: reader failed to signal END-OF-FILE inside an
1394 object representation. (reported by Nikodemus Siivola)
1395 * fixed some bugs revealed by Paul Dietz' test suite:
1396 ** LAST and [N]BUTLAST should accept a bignum.
1397 ** condition slot accessors are methods.
1398 ** (VECTOR NIL) is a subtype of STRING.
1400 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1401 * minor incompatible change: some nonsensical specialized lambda
1402 lists (used in DEFMETHOD) which were previously ignored now signal
1404 * minor incompatible change: the system is now aware of the types of
1405 variables in the COMMON-LISP package, and will signal errors for
1406 most violations of these type constraints (where previously they
1407 were silently accepted).
1408 * minor incompatible change: COMPILE-FILE now uses the freedom
1409 afforded (ANSI 3.2.2.3) to use derived function types for
1410 functions defined in the same file. This also permits the system
1411 to warn on static type mismatches and function
1412 redefinition. (Currently it does not work with high DEBUG level.)
1413 * minor incompatible change: VALUES declaration is disabled.
1414 * When issuing notes, the compiler now signals a condition of type
1415 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1416 restart for use in user handlers. It is expected that the
1417 COMPILER-NOTE condition will eventually become a condition
1418 supertype to a hierarchy of note types, which will then be
1419 handleable in a similar fashion. However, at the moment, no such
1420 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1421 but it's an implementation detail, not a classification for the
1423 * Changes in type checking closed the following bugs:
1424 ** type checking of unused values (192b, 194d, 203);
1425 ** template selection based on unsafe type assertions (192c, 236);
1426 ** type checking in branches (194bc).
1427 * A short form of VALUES type specifier has ANSI meaning (it has
1428 increased the number of situations when SBCL cannot perform type
1430 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1431 keywords or constants is permissible.
1432 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1433 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1434 classes. (thanks to Antonio Martinez)
1435 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1436 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1437 argument to a pretty-print stream. Instead, it leaves any such
1438 operations to DESCRIBE-OBJECT methods.
1439 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1441 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1442 lambda lists. (thanks to David Lichteblau)
1443 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1444 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1445 (thanks to Gerd Moellmann)
1446 * fixed bug 246: increased compilation speed of long
1447 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1449 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1451 * DEFCONSTANT now throws a condition of type
1452 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1453 constant to a non-EQL value; CONTINUE and ABORT restarts
1454 respectively change and preserve the value.
1455 * fixed bug 63: The code walker, part of the implementation of CLOS,
1456 is now better at handling symbol macros.
1457 * bug fix: There is no longer an internal implementation type named
1458 CL:LENGTH. (reported by Raymond Toy)
1459 * bug fix: In macro-like defining macros/special operators the
1460 implicit block does not enclose the lambda list.
1461 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1462 enclosed in parentheses) are not suitable as type specifiers, and
1463 their use properly signals an error now.
1464 * bug fix: An argument count mismatch for a type specifier in code
1465 being compiled no longer causes an unhandled error at compile
1466 time, but signals a compile-time warning.
1467 * fixed simple vector readable printing
1468 * bug fix: DESCRIBE takes more care over whether the class
1469 precedence list slot of a class is bound before accessing it.
1470 (reported by Markus Krummenacker)
1471 * bug fix: FORMATTER can successfully compile pretty-printer format
1472 strings which use variants of the ~* directive inside.
1473 * bug fix: SEARCH now applies its TEST predicate to the elements of
1474 the arguments in the correct order. (thanks to Wolfhard Buss)
1475 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1476 in type inference. (thanks to Robert E. Brown)
1477 * bug fix: Reading in symbols with an explicit package name of ""
1478 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1479 * fixed some bugs revealed by Paul Dietz' test suite:
1480 ** NIL is now allowed as a structure slot name.
1481 ** Arbitrary numbers, not just REALs, are allowed in certain
1482 circumstances in LOOP for-as-arithmetic clauses.
1483 ** Multiple class redefinitions before slot access no longer
1484 causes a type error.
1485 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1486 association between the name and a class.
1487 ** Generic functions with non-standard method-combination and over
1488 five methods all of which return constants no longer return NIL
1489 after the first few invocations. (thanks to Gerd Moellmann)
1490 ** CALL-NEXT-METHOD with no arguments now passes the original
1491 values of the arguments, even in the presence of assignment.
1492 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1494 ** DISASSEMBLE works with closures and funcallable instances.
1495 ** ADD-METHOD now returns the generic function, not the new method.
1496 ** FIND-METHOD signals an error if the lengths of the specializers
1497 is incompatible with the generic function, even if the ERRORP
1499 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1500 which its argument is a member.
1501 ** DEFCLASS only redefines the class named by its class-name
1502 argument if that name is the proper name of the class;
1503 otherwise, it creates a new class.
1504 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1505 of the UNBOUND-SLOT condition to the name of the slot.
1506 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1507 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1508 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1509 SLOT-MAKUNBOUND in the specified fashion.
1511 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1512 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1513 cross-compilation host. As a consequence, we can now bootstrap our
1514 way up to SBCL starting with a bare gcc toolchain and human-readable
1515 source code (first the source to CLISP, then the source to SBCL).
1516 * A contributed module containing a partial implementation of the
1517 simple-streams interface has been included. (thanks to Rudi
1519 * A contributed module implementing the RFC1321 Message Digest
1520 Algorithm, known as MD5, has been included.
1521 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1522 no longer has any effect, as the code controlled by this feature
1523 has been deleted. (As far as we know, no-one has ever built using
1524 this feature, and its semantics were confused in any case).
1525 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1526 work on conditions (as required by the ANSI specification),
1527 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1528 expected behaviour on conditions. Users should note, however,
1529 that such behaviour is not required by the ANSI specification,
1530 and so use of this behaviour may render their code unportable.
1531 * fixed some bugs revealed by Paul Dietz' test suite:
1532 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1534 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1535 ** SXHASH on condition objects no longer returns NIL.
1536 ** :ALLOCATION :CLASS slots are better treated; their values are
1537 updated on class redefinition, and initforms inherited from
1538 superclasses are applied.
1539 ** REMOVE-METHOD returns its generic function argument even when
1540 no method was removed.
1541 ** SHARED-INITIALIZE now initializes the values of the requested
1542 slots, including those with :ALLOCATION :CLASS.
1543 ** ALLOCATE-INSTANCE now works on structure classes defined via
1544 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1546 ** SLOT-EXISTS-P now works on conditions, as well as structures
1548 ** MAKE-LOAD-FORM now has the required methods on
1549 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1550 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1551 keyword, and now implements the SLOT-NAMES argument.
1552 ** methods with &OPTIONAL arguments no longer allow too many
1553 arguments to be passed in the call without error.
1554 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1555 option is consistent with the required arguments of the generic
1556 function lambda list.
1557 * bug fix: REQUIRE accepts a string designator. (Thanks to
1559 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1560 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1561 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1562 * bug fix: The compiler now checks for duplicated variables in macro
1564 * bug fix: SETQ on globals returns the correct value.
1565 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1566 not a COMPILER-ERROR (followed by some other strange error on
1567 choosing the CONTINUE restart).
1568 * bug fix: make.sh and friends are now more consistent in the way that
1569 they look for GNU "make".
1571 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1572 * experimental native threads support (on x86 Linux >=2.4 only).
1573 This is not compiled in by default: you need to add :SB-THREAD to
1574 the target features. See the "Beyond ANSI" chapter of the manual
1576 * fix for longstanding nonANSIism: The old distinction between
1577 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1578 The return value from CL:FIND-CLASS is now a CLOS class, and
1579 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1580 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1581 * An interface to the MetaObject Protocol, as described in Kiczales,
1582 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1583 MIT Press, 1991, is available from the SB-MOP package.
1584 * incompatible change: the SB-PCL package should now be considered
1585 a private implementation detail, and no longer a semi-private MOP
1587 * minor incompatible change: due to rearrangement for threads, the
1588 control stack and binding stack are now allocated at arbitrary
1589 addresses instead of being hardcoded per-port. Users affected by
1590 this probably have to be doing advanced things with shared
1591 libraries, and will know who they are.
1592 * minor incompatible change: Previously, all --eval forms used were
1593 processed with READ before any of them were processed with EVAL.
1594 Now each --eval form is processed with both READ and EVAL before
1595 the next --eval form is processed. (Thus package operations like
1596 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1597 work as the user might reasonably expect.)
1598 * minor incompatible change: *STANDARD-INPUT* is now only an
1599 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1601 * minor incompatible change: Y-OR-N-P is now character-oriented,
1602 not line oriented. Also, YES-OR-NO-P now works without errors.
1603 (thanks to Antonio Martinez)
1604 * sb-aclrepl module improvements: an integrated inspector, added
1605 repl features, and a bug fix to :trace command.
1606 * Known functions, which cannot be open coded by the backend, are
1607 considered to be able to check types of their arguments. (fixing
1608 a bug report by Nathan J. Froyd)
1609 * fixed a bug in computing method discriminating functions: It is
1610 now possible to define methods specialized on classes which have
1611 forward-referenced superclasses. (thanks to Gerd Moellmann)
1612 * fixed evaluation order in optional entries (reported by Gilbert
1614 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1615 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1616 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1617 required-by-AMOP NAME argument, as well as CLASS and
1618 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1619 * fixed bug 20: DEFMETHOD can define methods using names that are
1620 not the proper names of classes to designate class specializers.
1621 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1622 Unix file descriptors, instead of blowing up. (thanks to Antonio
1624 * Garbage collection refactoring: user-visible change is that a
1625 call to the GC function during WITHOUT-GCING will not do garbage
1626 collection until the end of the WITHOUT-GCING. If you were doing
1627 this you were probably losing anyway.
1628 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1629 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1630 TYPEP the latter but not the former.
1631 * The compiler issues a full WARNING for calls to undefined functions
1632 with names from the CL package.
1633 * MAP-INTO for a vector destination is open coded. (reported by
1634 Brian Downing on c.l.l)
1635 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1636 documentation string.
1637 * fixed some bugs revealed by Paul Dietz' test suite:
1638 ** COPY-ALIST now signals an error if its argument is a dotted
1640 ** Condition slots are now accessed more correctly in the presence
1641 of multiple initargs for a given slot.
1642 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1643 exclude restarts of the same name associated with a different
1645 ** DEFCLASS of forward-referenced classes with another
1646 forward-referenced class in the superclasses list no longer
1648 ** Condition slots are now initialized once each, not multiple
1649 times. (thanks to Gerd Moellmann)
1650 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1652 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1653 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1654 arguments contain duplicated elements.
1655 ** RESTART-CASE understands local macros.
1656 ** RESTART-CASE associates exactly its own restarts with a condition.
1657 ** ENDP in safe mode checks its argument to be of type LIST.
1658 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1659 in question is unbound.
1660 ** Optimized MAKE-INSTANCE functions no longer cause internal
1661 assertion failures in the presence of duplicate initargs.
1662 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1663 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1664 propagated even in the ctor optimized implementation of
1666 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1668 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1669 * a better implementation of SXHASH on (simple) bit vectors,
1670 measured both in execution speed and in distribution of results
1671 over the positive fixnums, has been installed. Likewise, a better
1672 implementation of EQUAL for simple bit vectors is now available.
1673 * fixed CEILING optimization for a divisor of form 2^k.
1674 * fixed bug 240 (emitting extra style warnings "using the lexical
1675 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1676 by Antonio Martinez)
1677 * fixed SXHASH, giving different results for NIL depending on type
1678 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1679 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1680 variable. (found by Rolf Wester)
1681 * SBCL does not ignore type declarations for special
1682 variables. (reported by rif on c.l.l 2003-03-05)
1683 * some bug fixes in contrib/sb-aclrepl/
1684 * fixed some bugs revealed by Paul Dietz' test suite:
1685 ** a bug in the CONS type specifier, whereby the CAR and CDR
1686 types got intertwined, has been fixed;
1687 ** the type system is now able to reason about the interaction
1688 between INTEGER and RATIO types more completely;
1689 ** APPEND, [N]REVERSE and NRECONC check that those their
1690 arguments, which must be proper lists, are really so;
1691 ** An array specialized to be unable to hold elements has been
1692 implemented, as required -- yes, really -- by ANSI;
1693 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1694 on malformed property lists;
1696 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1697 * incompatible packaging change: in line with Unix convention,
1698 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1699 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1700 * REQUIRE and PROVIDE are now optionally capable of doing something
1701 useful. See the documentation string for REQUIRE.
1702 * infrastructure for a managed SBCL contrib system: contributed
1703 modules in this release include:
1704 ** the ASDF system definition facility;
1705 ** an interface to the BSD Sockets API;
1706 ** an ACL-like convenience interface to the repl;
1707 (thanks to Kevin Rosenberg)
1708 ** an implementation of ROTATE-BYTE, with efficient implementation
1710 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1711 gives a closer approximation to the right answer than previously.
1712 (thanks to Raymond Toy)
1713 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1714 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1715 optional environment arguments, as required by ANSI.
1716 * fixed bugs in other functions taking environment objects, allowing
1717 calls with an explicit NIL environment argument to be compiled
1719 * fixed bug 228: primary return values from
1720 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1721 COMPILE or FUNCTION.
1722 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1723 :TYPE will no longer signal errors on innocuous objects.
1724 * fixed bug 231b: SETQ is better at respecting type declarations in
1725 the lexical environment.
1726 * fixed a bug in DEFCLASS: classes named by symbols with no or
1727 unprintable packages can now be defined.
1728 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1729 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1730 * fixed compiler failure related to checking types of functions.
1731 (reported by Robert E. Brown)
1732 * the compiler is now much more consistent in its error-checking
1733 treatment of bounding index arguments to sequence functions: in
1734 (SAFETY 3) code, errors will be signalled in almost all cases if
1735 invalid sequence bounding indices are passed to functions defined
1736 by ANSI to operate on sequences.
1737 * fixed a bug in the build procedure: documentation of SBCL-specific
1738 packages is now preserved and available in the final Lisp image.
1739 * lifted FDEFINITION lookup out of loops in the implementation of
1740 many list operations. (thanks to Robert E. Brown)
1741 * fixed a bug in the reader: the #n# reader macro now works for
1742 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1743 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1744 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1745 E. Brown and Tony Martinez respectively)
1746 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1747 SLOT-BOUNDP: in particular, it is now able to optimize them much
1748 better, and is now not vulnerable to having packages renamed.
1749 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1750 always called when a slot is not present in an instance. (thanks
1752 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1753 longer signal an error on unreading a character following EOF on
1754 the previous constituent stream. (thanks to Tony Martinez)
1755 * fixed some bugs revealed by Paul Dietz' test suite:
1756 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1757 not just nonnegative fixnums;
1758 ** the logical bit-array operators such as BIT-AND now accept an
1759 explicit NIL for their "opt-arg" argument (to indicate a
1760 freshly-consed result bit-array);
1761 ** ELT now signals an error on an invalid sequence index in safe
1763 ** the type system is now cleverer about negations of numeric
1764 types, and consequently understands the BIGNUM and RATIO types
1766 ** the type system is now cleverer about the interaction between
1767 INTEGER and RATIO types: while bugs still remain, many more
1768 cases are accurately computed;
1769 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1770 if it is in the last clause;
1771 ** CONSTANTLY now correctly returns a side-effect-free function in
1773 ** DECLARE is no longer treated as a special-operator; in
1774 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1775 * incremented fasl file version number due to the change in the
1776 DEFSTRUCT-SLOT-DESCRIPTION structure.
1778 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1779 * minor incompatible change: code processed by the "interpreter" or
1780 EVAL now has a compilation optimization policy of (DEBUG 2)
1781 (changed from (DEBUG 1)) to improve debuggability of interactive
1782 development, and to allow the use of the debug RETURN command in
1784 * an experimental implementation of the RETURN command for the
1785 debugger has been included. (thanks to Frederik Kuivinen)
1786 * fixed bug 62: constraints were not propagated into a loop.
1787 * fixed bug in embedded calls of SORT (reported and investigated by
1789 * fixed some bugs revealed by Paul Dietz' test suite:
1790 ** printing and reading of arrays with some dimensions having
1791 length 0 (thanks to Gerd Moellmann);
1792 ** BOA constructor with &AUX argument without a default value does
1793 not cause a type error;
1794 ** CONSTANTP now returns true for all self-evaluating objects.
1796 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1797 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1798 accessors that are related by inheritance, as specified in the
1799 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1800 Valtteri Vuorikoski)
1801 * The compiler is now able to inline functions that were defined in
1802 a complex lexical environment (e.g. inside a MACROLET).
1803 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1805 * Support for the upcoming FreeBSD-5.0 release has been included.
1806 (thanks to Dag-Erling Smorgrav)
1807 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1808 effect when it is not in a toplevel context.
1809 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1810 stand a better chance of being correct. (thanks to Gerd
1812 * fixed bug in COERCE, which now signals an error on coercing a
1813 rational to a bounded real type which excludes the expected
1815 * The compiler is now able to derive types more accurately from the
1816 COERCE and COMPILE functions.
1817 * fixed bug 223: functional binding is considered to be constant
1818 only for symbols in the CL package.
1819 * fixed bug 231: SETQ did not check the type of a variable being set
1820 (reported by Robert E. Brown)
1821 * A new optimization for MAKE-INSTANCE has been included, fixing
1822 various bugs (including relating to :ALLOCATION :CLASS slots and
1823 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1825 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1826 ** As required by ANSI, LOOP now disallows anonymous collection
1827 clauses such as COLLECT I in conjunction with aggregate boolean
1828 clauses such as THEREIS (= I 1);
1829 ** LOOP now signals an error when any variable is reused in the
1830 same loop (including the potentially useful construct analogous
1831 to WITH A = 1 WITH A = (1+ A);
1832 ** IT is only a special loop symbol within the first clause of a
1833 conditional loop clause;
1834 ** LOOP with a typed iteration variable over a hashtable now
1835 signals a type error iff it should.
1836 * fixed some other bugs revealed by Paul Dietz' test suite:
1837 ** FILE-STREAM now names the class previously known as FD-STREAM;
1838 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1839 argument) no longer signals an error;
1840 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1841 of the slot symbol, rather than using the current package
1842 ((:CONC-NAME "") continues to intern the slot's name in the
1844 * incremented fasl file version number, because of the incompatible
1845 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1846 of the new implementation of DEFINE-COMPILER-MACRO.
1848 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1849 * Support for building SBCL for MIPS platforms running in
1850 little-endian mode has now been checked in, and basic
1851 functionality on said platforms verified.
1852 * minor incompatible change: PCL now records the pathname of a file
1853 in which methods and the like are defined, rather than its
1855 * minor incompatible change: TRUENAME now considers the truename of
1856 a file naming a directory to be the pathname with :DIRECTORY
1857 component indicating that directory.
1858 * minor incompatible change: a NAMED clause in the extended form of
1859 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1860 reason for the previous behaviour is unclear.
1861 * more systematization and improvement of CLOS and MOP conformance
1862 in PCL (thanks to Gerd Moellman and Pierre Mai):
1863 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1865 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1866 primary methods with no specializers;
1867 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1869 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1870 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1871 FORWARD-REFERENCED-CLASSes; error reporting on
1872 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1874 ** SXHASH on CLOS instances now uses a slot internal to the
1875 instance to return different numbers on distinct instances,
1876 while preserving the same return value through invocations of
1878 ** DEFMETHOD signals errors when methods with longer incongruent
1879 lambda lists are added to generic functions;
1880 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1881 CLASS, as specified in AMOP;
1882 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1883 on the order returned by the primary method for classes of
1884 class STANDARD-CLASS;
1885 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1886 * fixed some bugs shown by Paul Dietz' test suite:
1887 ** DOLIST puts its body in TAGBODY;
1888 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1890 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1891 value producing form;
1892 ** if more variables are given to PROGV than values, extra
1893 variables are bound and made to have no value;
1894 ** NSUBSTITUTE on list arguments gets the right answer with
1896 ** ELT signals an error of type TYPE-ERROR when the index argument
1897 is not a valid sequence index;
1898 ** LOOP signals (at macroexpansion time) an error of type
1899 PROGRAM-ERROR when duplicate variable names are found;
1900 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1901 ** FUNCALL of special-operators now cause an error of type
1903 ** PSETQ now works as required in the presence of side-effecting
1904 symbol-macro places;
1905 ** NCONC accepts any object as its last argument;
1906 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1908 ** loop-for-as-package does not require a package to be explicitely
1910 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1912 * fixed bug 166: compiler preserves "there is a way to go"
1913 invariant when deleting code.
1914 * fixed bug 172: macro lambda lists with required arguments after
1915 &REST arguments now cause an error to be signalled. (thanks to
1917 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1918 bug. (thanks to Matthew Danish)
1919 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1921 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1922 arguments are lexically rebound. (thanks to Gerd Moellmann and
1924 * fixed bug 194: error messages are now more informative when there
1925 is no primary method applicable in a call to a generic
1926 function. (thanks to Gerd Moellmann)
1927 * fixed bug in command line argument checking (thanks to Julian
1929 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1930 (thanks to Matthew Danish)
1931 * incremented fasl file version number, because of the
1932 SXHASH-related changes in the layout of CLOS data structures
1934 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1935 * minor incompatible change: The runtime (the Unix executable named
1936 "sbcl") is now much pickier about the .core files it will load.
1937 Essentially it now requires .core files to descend from the same
1938 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1939 as the runtime does. (The intent is to prevent the crashes which
1940 can occur, and which can even be reported as mysterious failures,
1941 when people patch the sources or change the build parameters
1942 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1943 sbcl and .core files.)
1944 * fixed bug: VALUES-LIST is no longer optimized away.
1945 * fixed bug 142: The FFI conversion of C string values to Lisp
1946 string values no longer conses excessively. (thanks to Nathan
1947 Froyd porting Raymond Toy's fix to CMU CL)
1948 * began to systematize and improve MOP conformance in PCL (thanks to
1949 Nathan Froyd, Gerd Moellman and Pierre Mai):
1950 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1952 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1953 ** FINALIZE-INHERITANCE is now called on class finalization;
1954 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1955 argument precedence order.
1956 * fixed bug 202: The compiler no longer fails on functions whose
1957 derived types contradict their declared type.
1958 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1959 so it can be non-toplevel.
1960 * The fasl file version number has changed (because of the new
1961 implementation of DEFMACRO).
1962 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1963 safe code, that any length requirement by their type-specifier
1964 argument is valid. The exceptions to this are described in bug
1966 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1967 functions now check the types of their inputs as required by ANSI.
1968 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1969 introduced symbol is DECLAREd to be SPECIAL.
1970 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1971 * fixed bug: :COUNT argument to sequence functions may be negative
1972 * fixed bug: body of DO-SYMBOLS may contain declarations
1973 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1974 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1975 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1976 symbol macro only once
1977 * fixed printing of call frame when argument list is unavailable
1978 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1979 * compiler no longer signals WARNING on unknown keyword
1982 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1983 * A beta-quality port to the mips architecture running Linux,
1984 based on the old CMUCL backend, has been made. It has been tested
1985 on a big-endian kernel, and works sufficiently well to be able to
1986 rebuild itself; it has not been tested in little-endian mode.
1987 * fixed an inconsistency between gencgc.c and purify.c which made
1988 dumping/loading .core files unreliable
1989 * fixed bug 120a: The compiler now deals correctly with IFs where
1990 the consequent is the same as the alternative, instead of
1991 misderiving the return type. (thanks to Alexey Dejneka)
1992 * fixed bug 113: Logical pathnames are now dumpable (the logical
1993 host is resolved at load-time, throwing an error if it is not
1995 * fixed bug 174: FORMAT's error message is slightly clearer when a
1996 non-printing character is used in a format directive.
1997 * fixed several bugs in compiler checking of type declarations, i.e.
1998 violations of the Python "declarations are assertions" principle
1999 (thanks to Alexey Dejneka)
2000 * fixed several bugs in PCL's error checking (thanks to Gerd
2002 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2004 * fixed bug in compilation of functions as first class values
2005 (thanks to Antonio Martinez-Shotton)
2006 * The compiler's handling TYPE-ERRORs which it can prove will
2007 inevitably happen at runtime has been cleaned up and corrected
2008 in several ways. (thanks to Alexey Dejneka)
2009 * improved argument type checking for various basic arithmetic
2010 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2011 lavished on them in the past that they can be compiled in many
2012 ways in different special cases
2013 * fixed bug 181: compiler checks validity of user supplied type
2015 * cleaned up code flushing in optimization: Function calls which
2016 should signal errors for safety purposes (e.g. which ANSI says
2017 should signal errors when their arguments are of incorrect type)
2018 are no longer optimized away.
2019 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2020 * incremented fasl file version number, because changes in the
2021 implementation of sequence functions like COERCE caused
2022 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2023 in old inline expansions) to become undefined. (Actually these
2024 changes were later undone, so we might very well be binary
2025 compatible with 0.7.7 after all, but leaving the version number
2026 incremented seemed like the simplest and most conservative
2029 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2030 * An alpha-quality port to the parisc architecture running Linux,
2031 based on the old CMUCL backend, has been made. This, even more so
2032 than the other backends, should be considered still a work in
2033 progress; known problems include that the Linux kernel in 64-bit
2034 mode does not propagate the correct sigcontext structure to
2035 userspace, and consequently SBCL on a parisc64 kernel will not
2037 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2038 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2039 suppressed.) Also now that the compiler is looking at declarations
2040 in the environment, it checks optimization declarations as well,
2041 and suppresses inlining when (> DEBUG SPEED).
2042 * More fixes have been made to treatment of floating point exception
2043 treatment and other Unix signals. In particular, floating point
2044 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2045 * The detection and handling of control stack exhaustion (infinite
2046 or very deeply nested recursion) has changed. Stack exhaustion
2047 detection is now done by write-protecting pages at the OS level
2048 and applies at all optimization settings; when found, a
2049 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2050 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2051 be caught using IGNORE-ERRORS.
2052 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2053 correctly when transforming recursive LABELS functions to LETs.
2054 (thanks to Alexey Dejneka)
2055 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2056 that are names of constants or global variables.
2057 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2058 alien routines with docstrings.
2059 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2060 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2062 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2063 (thanks to Eric Marsden)
2064 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2065 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2066 * Bug fix: The SPARC backend can now compile functions involving
2067 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2068 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2069 to a functions expecting a PATHNAME-DESIGNATOR.
2070 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2071 lambda lists. (thanks to Alexey Dejneka)
2072 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2073 functions so that EXPORTing the name of the class doesn't cause
2074 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2075 to Antonio Martinez for reporting this)
2076 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2077 constant-folded without failing an assertion. (thanks to Einar
2078 Floystad Dorum for reporting this)
2079 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2080 types are now dealt with more correctly by the compiler.
2081 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2082 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2083 behaviour analogous to RENAME-FILE. This puts its behaviour more
2084 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2086 * The fasl file version number has changed again. (because of the
2087 bug fix involving the names of PCL MAKE-INSTANCE functions)
2089 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2090 * bug fix: Floating point exceptions are treated much more
2091 consistently on the x86/Linux and PPC/Linux platforms.
2092 * Array initialization with :INITIAL-ELEMENT is now much faster for
2093 cases when the compiler cannot open code the array creation, but
2094 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2095 array accesses have also seen a speed increase.
2096 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2097 specified. (This was at the root of some bad interactions between
2098 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2100 * bug fix: Internal error arguments for undefined functions are now
2101 computed correctly on the PPC/Linux platform.
2102 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2103 Raymond Toy's patch for CMU CL)
2104 * Support for the Solaris 9 operating environment has been included
2105 (thanks to Daniel Merritt)
2106 * A very ugly but hopefully complete draft of the missing FFI chapter
2107 of the manual has been created by reformatting the corresponding
2108 CMU CL manual chapter into (currently very ugly and incoherent)
2109 DocBook and bringing it up to date for SBCL behavior. Thus, the
2110 manual is now essentially complete, at least by my extreme
2111 once-and-only-once standards, whereby it's acceptable to refer to
2112 the doc strings of SB-EXT functions as the primary documentation.
2113 * The fasl file version number has changed again, due to cleanup of
2114 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2115 is no longer a static symbol.)
2117 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2118 * SBCL now builds with OpenMCL (version 0.12) as the
2119 cross-compilation host; also, more progress has been made toward
2120 bootstrapping under CLISP.
2121 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2123 * bug 158 fixed: The compiler can now deal with integer loop
2124 increments different from 1; fixing this turned out also to fix
2126 * bug 169 fixed: no more bogus warnings about using lexical bindings
2127 despite the presence of perfectly good SPECIAL declarations (thanks
2128 to David Lichteblau)
2129 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2130 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2131 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2133 * bug 180 fixed: Method combination specifications no longer ignore
2134 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2135 * bug fix: Structure type predicate functions now check their argument
2136 count as they should.
2137 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2138 correctly. (thanks to Pierre Mai)
2139 * minor incompatible change: The --noprogrammer option is deprecated
2140 in favor of the new --disable-debugger option, which is very similar.
2141 (The major difference is that it takes effect at a slightly different
2142 time at startup, causing handling of errors in --sysinit and
2143 --userinit files will be affected differently.) The
2144 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2145 been added to allow this functionality to be controlled from ordinary
2146 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2147 maintainers, who might want to run non-interactive scripts to
2148 build SBCL cores which will later be used interactively.)
2149 * minor incompatible change: The LOAD function no longer, when given
2150 a wild pathname to load, loads all files matching that pathname.
2151 Instead, an error of type FILE-ERROR is signalled.
2153 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2154 * bug 147 fixed: The compiler preserves its block link/count
2155 invariants more correctly now so that it doesn't crash. (thanks
2157 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2159 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2160 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2161 no-value-for-this-character as NIL (as specified by ANSI).
2162 * HOST-NAMESTRING on physical pathnames now returns a string that is
2163 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2164 (thanks to Christophe Rhodes)
2165 * The Alpha port handles icache flushing more correctly. (thanks to
2167 * More progress has been made toward bootstrapping under CLISP. (thanks
2168 to Christophe Rhodes)
2169 * The fasl file format has changed again, because dynamic loading
2170 on OpenBSD (which has non-ELF object files) motivated some cleanups
2171 in the way that foreign symbols are transformed and passed around.
2172 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2173 is no longer treated as whitespace by the reader, but instead as
2174 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2175 |A
\7fB|, instead of A as it used to.
2177 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2178 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2179 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2180 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2181 this as of 0.7.1.45, but was left out of the previous news section
2182 (thanks to Dan Barlow)
2183 * SBCL now runs on the Solaris operating system on SPARC architectures
2184 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2185 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2186 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2187 * SPARC backend cleanups, allowing builds of cores optimized for V8
2188 and V9 SPARCS, and also emission of code targeted to a particular
2189 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2191 * SBCL is closer to bootstrapping under CLISP, thanks to various
2192 fixes by Christophe Rhodes.
2193 * The fasl file format has changed again, to allow the compiler's
2194 INFO database to support symbol macros.
2195 * The user manual (in doc/) is formatted into HTML more nicely.
2196 (thanks to coreythomas)
2197 * The system is smarter about SUBTYPEP relationships, especially
2198 those involving NOT types (including types such as ATOM which are
2199 represented internally using NOT types). Thus SUBTYPEP is less
2200 likely to return (VALUES NIL NIL) in general, and in particular
2201 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2203 * The fasl file format has changed again, because the internal
2204 representation of types now includes a new slot to support the new
2205 SUBTYPEP-of-NOT-types logic.
2206 * (not a change in the main branch of SBCL, but a related prototype
2207 which can hopefully be merged into the main branch of SBCL in the
2208 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2209 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2210 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2211 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2212 dispatch macro character combinations. (thanks to Alexey Dejneka)
2213 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2214 that superficially look like logical namestrings correctly.
2215 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2217 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2218 default supplied-p) for &optional and &key arguments. (thanks to
2221 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2222 * incompatible change: The compiler is now less aggressive about
2223 tail call optimization, doing it only when (> SPACE DEBUG) or
2224 (> SPEED DEBUG). (This is an incompatible change because there are
2225 programs which relied on the old CMU-CL-style behavior to optimize
2226 away their unbounded recursion which will now die of stack overflow.)
2227 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2228 for non-GENCGC systems has been increased to 20M (since that
2229 seems much closer to the likely performance optimum for modern
2230 systems than the old 4M value was)
2231 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2232 and *DEBUG-PRINT-LEVEL*
2233 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2234 of CMU CL's support for SPARC, and various endianness and other
2235 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2236 * new syntactic sugar for the Unix command line: --load foo.bar is now
2237 an alternate notation for --eval '(load "foo.bar")'.
2239 ** The system now detects stack overflow and handles it gracefully,
2240 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2241 optimization settings. (This is a good thing in general, and
2242 its introduction in this version should be particularly timely
2243 for anyone whose code fails because of suppression of tail
2245 ** The system now hunts for the C variable "environ" in a more
2246 devious way, to avoid segfaults when the C library version differs
2247 between compile time and run time. (thanks to Christophe Rhodes)
2248 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2249 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2250 ** The compiler no longer issues bogus style warnings for undefined
2251 classes in the same source file as the DEFCLASSes which defined
2252 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2254 ** fixes in CONDITION class precedence list for undefined function
2255 errors (thanks to Alexei Dejneka)
2256 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2257 correctly. (thanks to Dan Barlow)
2258 ** portability fixes aiming at bootstrapping under CLISP (thanks
2259 to Dave McDonald and Christophe Rhodes)
2260 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2261 ** fixes in type translation and and type inference (thanks to
2263 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2264 ** various fixes in the new ports (thanks to Dan Barlow)
2265 * several changes related to debugging:
2266 ** suppression of tail recursion, as noted above
2267 ** stack overflow detection, as noted above
2268 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2269 is now the default. (For some time encapsulation has been more
2270 reliable than the breakpoint-based :ENCAPSULATE NIL
2271 implementation, at least on X86 systems; and I just noticed that
2272 encapsulation also seems closer to the spirit of the ANSI
2275 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2277 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2278 up properly again. (There was a packaging bug in 0.7.0 which
2279 left their definitions in SB-SYS::LOAD-FOREIGN and
2280 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2281 vital for most things which interface to C-level interfaces,
2282 like extensions working with sockets or databases or
2283 Perl-compatible regexes or whatever, and the need to fix
2284 this bug is the main reason that 0.7.1 was released so
2286 ** DEFGENERIC is now choosier about the methods it redefines, so that
2287 reLOADing a previously-LOADed file containing DEFGENERICs does
2288 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2289 cycle now works as it should. (thanks to Alexey Dejneka)
2290 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2291 fixed by Christophe Rhodes. (He actually submitted this patch
2292 months ago, and I delayed until after 0.7.0.)
2293 ** Bug 111 (internal compiler confusion about runtime checks on
2294 FUNCTION types) was fixed by Alexey Dejneka.
2295 * Some internal cleanups (getting rid of variables which aren't
2296 needed now that the byte interpreter is gone) caused the fasl
2297 file format number to change again.
2299 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2300 * major incompatible change: The default fasl file extension, i.e. the
2301 default extension for files produced by COMPILE-FILE, has changed
2302 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2304 ** There are many changes in the implementation of the compiler.
2305 SBCL is now essentially a compiler-only implementation of ANSI
2306 Common Lisp. EVAL still "interprets" a few special cases, but
2307 almost all the interesting cases are handled by creating
2308 a LAMBDA expression, calling COMPILE on it, then calling
2309 FUNCALL on the result.
2310 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2311 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2312 Since the code is newer, there might still be some new bugs
2313 (though not as many as before Martin Atzmueller's fixes:-). But
2314 the new code is substantially simpler and clearer, and hopefully
2315 any remaining bugs will be simpler, less fundamental, and more
2316 fixable then the bugs in the old code.
2317 ** The revised compiler is still a little unsteady on its feet.
2319 *** The debugging information it produces (particularly the names
2320 of FUNCTION objects) is sometimes much less useful than what
2321 the old compiler produced.
2322 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2323 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2324 within a MACROLET) has been temporarily weakened.
2325 ** There are new compiler optimizations for various functions:
2326 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2327 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2328 *** the math functions TRUNCATE, FLOOR, and CEILING
2329 *** the function-of-all-trades COERCE
2330 Mostly these should be transparent, but there's one
2331 potentially-annoying problem (bug 117): when the compiler
2332 inline-expands a function and does type analysis on the result,
2333 it can create control paths which have type mismatches, and
2334 when it can't prove that those control paths aren't taken,
2335 it will issue WARNINGs about the type mismatches. This is
2336 a particular problem in practice for the new sequence functions.
2337 It's not clear how this should be fixed, and for now, a
2338 workaround is given in the entry for 117 in the BUGS file.
2339 ** (Because of the interaction between the two previous items --
2340 occasional inlining problems and new inline expansions -- some
2341 of the new sequence function optimizations won't really kick in
2342 completely until debugging information, and then inlining, are
2343 straightened out in some future version.)
2344 * minor incompatible changes:
2345 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2346 DIRECTORY no longer implicitly promotes NIL slots of its
2347 pathname argument to :WILD. In particular, when you ask for the
2348 contents of a directory (which you used to be able to do without
2349 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2350 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2351 ** changes in behavior that ANSI explicitly defines to be
2352 implementation dependent:
2353 *** The new compiler-only implementation still conforms with ANSI,
2354 but acts a little different than before. Besides the obvious
2355 changes in performance tradeoffs (that the cost per form passed
2356 to EVAL has gone up, and the cost per form executed by EVAL
2357 has gone down), the behavior of the system changes a little
2358 because there are no longer any interpreted function objects.
2359 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2360 e.g. doing COMPILE on the output of interactive DEFUN is
2362 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2364 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2365 has changed from (:CL) to NIL.
2366 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2367 e.g. #\Newline and #\Tab, don't have names specified in the
2368 ANSI Common Lisp standard, is now based on their ASCII symbolic
2369 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2370 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2371 are no longer used for output.
2372 ** changes in internal implementation constants:
2373 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2374 4 million. (If your application spends a lot of time GCing and
2375 you have a lot of RAM, you might want to experiment with
2376 increasing it even more.)
2377 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2378 However, almost all old code should still continue to work without
2379 immediate update, as SB-C-CALL is now a (deprecated) nickname
2381 ** Old operator names in the style DEF-FOO are now deprecated in
2382 favor of new corresponding names DEFINE-FOO, for consistency with
2383 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2384 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2385 internal symbols, but a few supported extensions like
2386 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2387 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2388 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2389 etc. as you get deeper into recursive calls to the debugger
2390 command loop, instead of the old "5]", "5]]", "5]]]"
2391 sequence. (I was motivated to do this when squabbles between
2392 ILISP and SBCL left me very deeply nested in the debugger. In the
2393 short term, this change will probably provoke more ILISP/SBCL
2394 squabbles, but hopefully it will be an improvement in the long run.)
2395 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2396 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2397 function, since declaiming return types involving aliens is
2398 (1) annoyingly messy to do by hand and (2) vital to efficient
2399 compilation of code which calls such functions.
2400 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2401 longer reexported by the SB-EXT package. They're solely useful
2402 for alien code, so it seems more logical that you should get
2403 them from the SB-ALIEN package, not in SB-EXT.
2404 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2405 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2406 features. Instead, the code that they used to control is always
2407 built into the system.
2408 * many other bug fixes
2409 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2410 advantage of the new EVAL-WHEN stuff and to clean them up in
2411 general, and they are now more ANSI-compliant in a number of
2412 ways. Martin Atzmueller is responsible for a lot of this.
2413 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2415 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2416 *** correct ERROR type for various file operations
2417 *** some fixes for Lisp streams
2418 *** DEFMETHOD syntax checking
2419 *** changing old weird representation of debug information as
2420 strings (which, among their other deficiencies, don't transform
2421 correctly when you rename packages, and don't change their
2422 print representation when you change things like *PACKAGE*
2423 and *PRINT-LENGTH*) to symbols and lists of symbols
2424 He also made several improvements and fixed several bugs in DESCRIBE.
2425 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2427 *** misbehavior of WRITE-STRING/WRITE-LINE
2428 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2429 and several other LOOP problems as well
2430 *** DIRECTORY when similar filenames are present
2431 *** DEFGENERIC with :METHOD options
2432 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2433 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2434 *** argument ordering in FIND with :TEST option
2435 *** mishandled package designator argument in APROPOS-LIST
2436 *** various problems in the backquote readmacro
2437 *** a bug in APROPOS
2438 *** probably some others that I'm not describing very well here,
2439 since the CVS log documents them by reference to sbcl-devel
2440 messages, and the SourceForge archives aren't working well.:-(
2441 ** Dan Barlow improved the Alpha port (and is making progress on the
2442 PPC port, for those of you who think different).
2443 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2444 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2445 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2446 continues to work on the SPARC port (for those of you in a position
2447 to look down upon our little PC-compatible boxes from a great height).
2448 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2449 and uses that copy, rather than the current dynamic value, when
2450 it's trying to decide whether to truncate output. Thus e.g.
2451 (let ((*print-lines* 50))
2452 (pprint-logical-block (stream nil)
2454 (let ((*print-lines* 8))
2455 (print (aref possiblybigthings i) stream)))))
2456 should now truncate the logical block only at 50 lines, instead of
2457 often truncating it at 8 lines, as it did before.
2458 * The doc/cmucl/ directory, containing old CMU CL documentation
2459 from the time of the fork, is no longer part of the base system.
2460 SourceForge has shut down its anonymous FTP service, and with it
2461 my original plan for distributing the old CMU CL documentation
2462 there. For now, if you need these files you can download an old
2463 SBCL source release and extract them from it.
2464 * The fasl file version number changed again, for dozens of reasons,
2465 some of which are apparent above.
2467 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2468 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2469 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2470 patch, so that DISASSEMBLE output is much nicer.
2471 * The code in the SB-PROFILE package now seems reasonably stable.
2472 I still haven't decided what the final interface should look like
2473 (I'd like PROFILE to interact cleanly with TRACE, since both
2474 facilities use function encapsulation) but if you have a need
2475 for profiling now, you can probably use it successfully with
2476 the current CMU-CL-style interface.
2477 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2478 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2479 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2480 link now returns the dangling link itself, and for similar
2481 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2482 link itself. (In these cases the old code signalled an error and
2483 looped endlessly, respectively.) Thus, DIRECTORY now works even
2484 in the presence of dangling and cyclic symbolic links.
2485 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2486 is now a supported extension again, since the consensus on
2487 sbcl-devel was that it can be useful for ordinary development
2488 work, not just for debugging SBCL itself.
2489 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2490 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2491 that currently-defined functions might be redefined later with
2492 different return types.
2493 * Hash tables can be printed readably, as inspired by CMU CL code
2494 of Eric Marsden and SBCL code of Martin Atzmueller.
2495 * better error handling in CLOS method combination, thanks to
2496 Martin Atzmueller porting Pierre Mai's CMU CL patches
2497 * more overflow fixes for >16Mbyte I/O buffers
2498 * A bug in READ has been fixed, so that now a single Ctrl-D
2499 character suffices to cause end-of-file on character streams.
2500 In particular, now you only need one Ctrl-D at the command
2501 line (not two) to exit SBCL.
2502 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2504 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2505 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2506 does the right thing.
2507 * The implementation of some type tests, especially for CONDITION
2508 types, is now tidier and maybe faster, due to CMU CL code
2509 originally by Douglas Crosher, ported by Martin Atzmueller.
2510 * Some math functions have been fixed, and there are new
2511 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2512 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2513 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2514 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2515 likely to be built into sbcl-0.7.0.)
2516 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2517 CL, ported to SBCL by Martin Atzmueller)
2518 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2519 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2520 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2521 * There's a new slam.sh hack to shorten the edit/compile/debug
2522 cycle for low-level changes to SBCL itself, and a new
2523 :SB-AFTER-XC-CORE target feature to control the generation of
2524 the after-xc.core file needed by slam.sh.
2525 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2526 COMPILE-FILE is no longer supported, so that now every function
2527 gets an entry point, so that block compilation looks a little
2528 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2529 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2530 returns the number of bytes consed since the system started,
2531 rather than the number consed since the first time the function
2532 was called. (The new definition parallels ANSI functions like
2533 CL:GET-INTERNAL-RUN-TIME.)
2534 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2535 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2536 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2537 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2538 * Fasl file version numbers are now independent of the target CPU,
2539 since historically most system changes which required version
2540 number changes have affected all CPUs equally. Similarly,
2541 the byte fasl file version is now equal to the ordinary
2544 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2545 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2546 is no longer recognized. I apologize for this, because it was
2547 listed in SB-EXT as a supported extension, but I found that
2548 its existing behavior was poorly specified, as well as incorrectly
2549 specified, and it looked like too much of a mess to straighten it
2550 out. I have enough on my hands trying to get ANSI stuff to work..
2551 * many patches ported from CMU CL by Martin Atzmueller, with
2552 half a dozen bug fixes in pretty-printing and the debugger, and
2553 half a dozen others elsewhere
2554 * fixed bug 13: Floating point infinities are now supported again.
2555 They might still be a little bit flaky, but thanks to bug reports
2556 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2557 as flaky as they were.
2558 * The --noprogrammer command line option is now supported. (Its
2559 behavior is slightly different in detail from what the old man
2560 page claimed it would do, but it's still appropriate under the
2561 same circumstances that the man page talks about.)
2562 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2563 are now supported, and enabled by default. Thus, the compiler can
2564 handle many floating point and complex operations much less
2565 inefficiently. (Thus e.g. you can implement a complex FFT
2567 * The compiler now detects type mismatches between DECLAIM FTYPE
2568 and DEFUN better, and implements CHECK-TYPE more correctly, and
2569 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2570 thanks to patches from Martin Atzmueller.
2571 * various fixes to make the cross-compiler more portable to
2572 ANSI-conforming-but-different cross-compilation hosts (notably
2573 Lispworks for Windows, following bug reports from Arthur Lemmens)
2574 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2575 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2576 fixed thanks to Pierre Mai's CMU CL patches.
2577 * fixes to keep the system from overflowing internal counters when
2578 it tries to use i/o buffers larger than 16M bytes
2579 * fixed bug 45a: Various internal functions required to support
2580 complex special functions have been merged from CMU CL sources.
2581 (When I was first setting up SBCL, I misunderstood a compile-time
2582 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2583 * improved support for type intersection and union, fixing bug 12
2584 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2585 more obscure bugs as well
2586 * some steps toward byte-compiling non-performance-critical
2587 parts of the system, courtesy of patches from Martin Atzmueller
2588 * Christophe Rhodes has made some debian packages of sbcl at
2589 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2590 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2591 stable, but are nonetheless usable. When he's ready, I'd be happy
2592 to add them to the SourceForge "File Releases" section. (And if
2593 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2594 * new fasl file format version number (because of changes in
2595 internal representation of (OR ..) types to accommodate the new
2596 support for (AND ..) types, among other things)
2598 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2599 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2601 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2603 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2604 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2605 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2606 * DESCRIBE now works on structure objects again.
2607 * Most function call argument type mismatches are now handled as
2608 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2609 know whether the function will be redefined before the call is
2610 executed. (The compiler could flag local calls with full WARNINGs,
2611 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2612 it doesn't keep track of enough information to know whether calls
2613 are local in this sense.)
2614 * Compiler output is now more verbose, with messages truncated
2615 later than before. (There should be some supported way for users
2616 to override the default verbosity, but I haven't decided how to
2617 provide it yet, so this behavior is still controlled by the internal
2618 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2619 src/compiler/ir1util.lisp.)
2620 * Fasl file format version numbers have increased again, because
2621 support for the Gray streams extension changes the layout of the
2622 system's STREAM objects.
2623 * The Gray subclassable streams extension now works, thanks to a
2624 patch from Martin Atzmueller.
2625 * The full LOAD-FOREIGN extension (not just the primitive
2626 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2627 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2628 but like most other programs, it defaults to copying the Unix
2629 environment from the original process instead of starting the
2630 new process in an empty environment.
2631 * Extensions which manipulate the Unix environment now support
2632 an :ENVIRONMENT keyword option which doesn't smash case or
2633 do other bad things. The CMU-CL-style :ENV option is retained
2634 for porting convenience.
2635 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2636 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2638 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2640 * A patch from Martin Atzmueller seems to have solved the SIGINT
2641 problem, and as far as we know, signal-handling now works cleanly.
2642 (If you find any new bugs, please report them!)
2643 * The system no longer defaults Lisp source file names to types
2644 ".l", ".cl", or ".lsp", but only to ".lisp".
2645 * The compiler no longer uses special default file extensions for
2646 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2647 seems to expect a single default extension for all compiled code,
2648 and there's no compelling reason to try to stretch the standard
2649 to allow two different extensions.) Instead, byte-compiled files
2650 default to the same extension as native-compiled files.
2651 * Fasl file format version numbers have increased again, because
2652 a rearrangement of internal implementation packages made some
2653 dumped symbols in old fasl files unreadable in new cores.
2654 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2655 many fewer weird special cases.
2656 * Bug #17 (differing COMPILE-FILE behavior between logical and
2657 physical pathnames) has been fixed, and some related misbehavior too,
2658 thanks to a patch from Martin Atzmueller.
2659 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2660 by Tim Moore, ported to SBCL by Martin Atzmueller.
2661 * Martin Atzmueller fixed several filesystem-related problems,
2662 including bug #36, in part by porting CMU CL patches, which were
2663 written in part by Paul Werkowski.
2664 * More compiler warnings in src/runtime/ are gone, thanks to
2665 more patches from Martin Atzmueller.
2666 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2669 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2671 * DESCRIBE now works on CONDITION objects.
2672 * The debugger now handles errors which arise when trying to print
2673 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2675 * The build system now uses an additional file, customize-target-features.lisp,
2676 to allow local modifications to the target *FEATURES* list. (The point of
2677 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2678 debugging features enabled, without having to worry about propagating it
2679 into everyone's system when I do a "cvs update".) When no
2680 customize-target-features.lisp file exists, the target *FEATURES* list
2681 should be constructed the same way as before.
2682 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2683 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2684 warn about behavior which is undefined under the ANSI spec. Note: This
2685 is specified by ANSI, but it's not very popular with programmers.
2686 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2687 macro in the SBCL sources for an example of a workaround which you
2688 might use to make portable ANSI-standard code which does what you want.
2689 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2690 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2691 when it's used as a non-toplevel form. (This is required in order
2692 to implement the DEFCONSTANT-EQX macro.)
2693 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2694 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2695 funny things, due to bugs in the implementation of EVAL-WHEN
2696 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2697 non-toplevel DEFCONSTANTs are uncommon.)
2698 * The core file version number and fasl file version number have been
2699 incremented, because the old noncompliant DEFCONSTANT behavior involved
2700 calling functions which no longer exist, and because I also took the
2701 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2702 * fixed bug 1 (error handling before read-eval-print loop starts), and
2703 redid debugger restarts and related debugger commands somewhat while
2705 ** The QUIT debugger command is gone, since it did something
2706 rather different than the SB-EXT:QUIT command, and since it never
2707 worked properly outside the main toplevel read/eval/print loop.
2708 Invoking the new TOPLEVEL restart provides the same functionality.
2709 ** The GO debugger command is also gone, since you can just invoke
2710 the CONTINUE restart directly instead.
2711 ** The TOP debugger command is also gone, since it's redundant with the
2712 FRAME 0 command, and since it interfered with abbreviations for the
2714 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2716 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2717 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2718 some porting work by Martin Atzmueller)
2719 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2720 with a boatload of patches which he ported to SBCL. Now that those
2722 ** The system tries to make sure that its low-priority messages
2723 are prefixed by semicolons, to help people who like to use
2724 syntax highlighting in their ILISP buffer. (This patch
2725 was originally due to Raymond Toy.)
2726 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2727 patches originally written by Raymond Toy.
2728 ** The compiler understands coercion between single-value and
2729 multiple-VALUES type expressions better, getting rid of some very
2730 weird behavior, thanks to patches originally by Robert MacLachlan
2731 and Douglas Crosher.
2732 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2733 lambda lists, thanks to a patch originally by Pierre Mai.
2734 ** The system no longer bogusly warns about "abbreviated type
2736 ** The compiler gets less confused by inlining and RETURN-FROM,
2737 thanks to some patches originally by Tim Moore.
2738 ** The system no longer hangs when dumping circular lists to fasl
2739 files, thanks to a patch originally from Douglas Crosher.
2740 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2741 undefined function error.
2742 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2743 in structure slot initforms) for now, documented workaround instead:-|
2744 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2745 * fixed bug 5: added stubs for various Gray stream functions called
2746 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2747 installed, at least appropriate type errors are generated
2748 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2749 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2750 consistently in DEFMETHOD forms.
2751 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2752 it doesn't seem to affect SBCL after all
2753 * The C runtime system now builds with better optimization and many
2754 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2756 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2758 * The system is now under CVS at SourceForge (instead of the
2759 CVS repository on my home machine).
2760 * The new signal handling code has been tweaked to treat register
2761 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2762 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2763 downstream, e.g. in debug-int.lisp, has implicit dependencies
2764 on the unsignedness of integer representation of machine words,
2765 and that caused the system to bomb out with infinite regress
2766 when trying to recover from type errors involving signed values,
2767 e.g. (BUTLAST '(1 2 3) -1).)
2768 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2769 (This was one of the bugs Peter Van Eynde reported back in July.)
2770 * The system now uses code inspired by Colin Walters' O(N)
2771 implementation of MAP (from the cmucl-imp@cons.org mailing
2772 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2773 inline the MAP operation, and there is more than one
2774 sequence argument to the MAP call (so that it can't just
2775 do ETYPECASE once and for all based on the type of the
2776 single sequence argument). (The old non-inline implementation
2777 of the general M-argument sequence-of-length-N case required
2778 O(M*N*N) time when any of the sequence arguments were LISTs.)
2779 * The QUIT :UNIX-CODE keyword argument has been renamed to
2780 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2781 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2783 * Raymond Wiker's patches to port dynamic loading from Linux to
2784 FreeBSD have been added.
2785 * The BUGS file is now more nearly up to date, thanks in large part
2786 to Martin Atzmueller's review of it.
2787 * The debugger now flushes standard output streams before it begins
2788 its output ("debugger invoked" and so forth).
2789 * The core version number and fasl file version number have both
2790 been incremented, because of incompatible changes in the layout
2792 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2793 used to not be called for a saved Lisp image.)
2794 * Martin Atzmueller's version of a patch to fix a compiler crash,
2795 as posted on sbcl-devel 13 September 2000, has been installed.
2796 * Instead of installing Martin Atzmueller's patch for the
2797 compiler transform for SUBSEQ, I deleted the compiler transform,
2798 and transforms for some similar consing operations.
2799 * A bug in signal handling which kept TRACE from working on OpenBSD
2801 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2802 away by constant folding
2803 * The system now defines its address space constants in one place
2804 (in the Lisp sources), and propagates them automatically elsewhere
2805 (through GENESIS and the sbcl.h file). Therefore, patching the
2806 address map is less unnecessarily tedious and error-prone. The
2807 Lisp names of address space constants have also been systematized.
2808 * CVS tags like dollar-Header-dollar have been removed from
2809 the sources, because they have never saved me trouble and
2810 they've been source of trouble working with patches and other
2811 diff-related operations.
2812 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2813 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2815 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2817 * The system has been ported to OpenBSD.
2818 * The system now compiles with a simple "sh make.sh" on the systems
2819 that it's supported on. I.e., now you no longer need to tweak
2820 text in the target-features.lisp-expr and symlinks in src/runtime/
2821 by hand, the make.sh takes care of it for you.
2822 * The system is no longer so grossly inefficient when compiling code
2823 involving vectors implemented as general (not simple) vectors (VECTOR T),
2824 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2825 which dares to use the various sequence functions on non-simple
2826 vectors, takes less of a performance hit.
2827 * There is now a primitive type predicate VECTOR-T-P
2828 to test for the (VECTOR T) type, so that e.g.
2829 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2830 can now be compiled with some semblance of efficiency. (The old code
2831 turned the type declaration into a full call to %TYPEP at runtime!)
2832 * AREF on (VECTOR T) is still not fast, since it's still compiled
2833 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2834 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2835 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2836 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2837 without making full calls to SUBTYPEP at runtime.
2838 (Some analogous efficiency issues for non-simple vectors specialized to
2839 element types other than T, or for non-simple multidimensional arrays,
2840 have not been addressed. They could almost certainly be handled the
2841 same way if anyone is motivated to do so.)
2842 * The changes in array handling break binary compatibility, so
2843 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2844 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2845 instead of (VALUES T T).
2846 * By following the instructions that Dan Barlow posted to sbcl-devel
2847 on 2 July 2000, I was able to enable primitive dynamic object
2848 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2849 functionality is not implemented (since it calls ld to resolve
2850 library references automatically, requiring RUN-PROGRAM for its
2851 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2852 try to resolve library references) is now supported.
2853 * The system now flushes the standard output streams when it terminates,
2854 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2855 them at several other probably-convenient times, e.g. in each pass of
2856 the toplevel read-eval-print loop, and after evaluating a form given
2857 as an "--eval" command-line option. (These changes were motivated by a
2858 discussion of stream flushing issues on cmucl-imp in August 2000.)
2859 * The source transform for TYPEP of array types no longer assumes
2860 that an array whose element type is a not-yet-defined type
2861 is implemented as an array of T, but instead punts, so that the
2862 type will be interpreted at runtime.
2863 * There is now some support for cross-compiling in make.sh: each of
2864 the phases of make.sh has its own script. (This should be transparent
2865 to people doing ordinary, non-cross-compile builds.)
2866 * Since my laptop doesn't have hundreds of megabytes of memory like
2867 my desktop machine, I became more motivated to do some items on
2868 my to-do list in order to reduce the size of the system a little:
2869 ** Arrange for various needed-only-at-cold-init things to be
2870 uninterned after cold init. To support this, those things have
2871 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2872 symbols with such names are now uninterned after cold init).
2873 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2874 fasl files for cold load.
2875 ** Remove the old compiler structure pooling code (which used to
2876 be conditional on the target feature :SB-ALLOC) completely.
2877 ** Redo the representation of some data in cold init to be more compact.
2878 (I also looked into supporting byte compiled code at bootstrap time,
2879 which would probably reduce the size of the system a lot, but that
2880 looked too complicated, so I punted for now.)
2881 * The maximum signal nesting depth in the src/runtime/ support code has
2882 been reduced from 4096 to 256. (I don't know any reason for the very
2883 large old value. If the new smaller value turns out to break something,
2884 I'll probably just bump it back up.)
2885 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2887 * Many, many bugs reported by Peter Van Eynde have been added to
2888 the BUGS list; some have even been fixed.
2889 * While enabling dynamic object file loading, I tried to make the
2890 code easier to understand, renaming various functions and variables
2891 with less ambiguous names, and changing some function calling
2892 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2893 * While trying to figure out how to do the OpenBSD port, I tried to
2894 clean up some of the code in src/runtime/. In particular, I dropped
2895 support for non-POSIX signal handling, added various comments,
2896 tweaked the code to reduce the number of compilation warnings, and
2897 renamed some files to increase consistency.
2898 * To support the new automatic configuration functionality in make.sh,
2899 the source file target-features.lisp-expr has been replaced with the
2900 source file base-target-features.lisp-expr and the machine-generated
2901 file local-target-features.lisp-expr.
2902 * fixed a stupid quoting error in make.sh so that using CMU CL
2903 "lisp -batch" as cross-compilation host works again
2905 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2907 * DESCRIBE no longer tries to call itself recursively to describe
2908 bound/fbound values, so that it no longer fails on symbols which are
2909 bound to themselves (like keywords, T, and NIL).
2910 * DESCRIBE now works on generic functions.
2911 * The printer now prints less-screwed-up representations of closures
2912 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2913 * A private symbol is used instead of the :EMPTY keyword previously
2914 used to mark empty slots in hash tables. Thus
2915 (DEFVAR *HT* (MAKE-HASH-TABLE))
2916 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2917 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2918 now does what ANSI says that it should. (You can still get
2919 similar noncompliant behavior if bang on the hash table
2920 implementation with all the symbols you get back from
2921 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2922 This breaks binary compatibility, since tests for equality to
2923 :EMPTY are wired into things like the macroexpansion of
2924 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2926 * There's now a minimal placeholder implementation for CL:STEP,
2927 as required by ANSI.
2928 * An obscure bug in the interaction of the normal compiler, the byte
2929 compiler, inlining, and structure predicates has been patched
2930 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2931 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2933 * Missing ordinary arguments in a macro call are now detected even
2934 when the macro lambda list contains &KEY or &REST.
2935 * The debugger no longer complains about encountering the top of the
2936 stack when you type "FRAME 0" to explicitly instruct it to go to
2937 the top of the stack. And it now prints the frame you request even
2938 if it's the current frame (instead of saying "You are here.").
2939 * As specified by ANSI, the system now always prints keywords
2940 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2942 * The default initial SIZE of HASH-TABLEs is now smaller.
2943 * Type information from CLOS class dispatch is now propagated
2944 into DEFMETHOD bodies, so that e.g.
2945 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2947 is now basically equivalent to
2948 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2949 (DECLARE (TYPE SINGLE-FLOAT X))
2951 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2952 operation, without having to do run-time type dispatch.
2953 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2954 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2955 * Since I don't seem to be making big file reorganizations very often
2956 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2957 is overflowing my ability to conveniently back them up), I've finally
2958 checked the system into CVS. (The CVS repository is on my home system,
2959 not at SourceForge -- putting it on SourceForge might come later.)
2960 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2961 high-level GC-NOTIFY-FOO functions send their output. (There's
2962 still very little control of where low-level verbose GC functions
2963 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2964 less than it used to -- the GC-NOTIFY-FOO functions are now under
2965 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2966 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2967 in only one place in the source code, and propagates it automatically
2968 everywhere that it's needed. Thus e.g. when I bump the version from
2969 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2970 * The C source files now include boilerplate legalese and documentation
2971 at the head of each file (just as the Lisp source files already did).
2972 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2973 to his page will be replaced with a link to his new CLiki service.
2975 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2977 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2978 * The build process now looks for GNU make under the default name "gmake",
2979 instead of "make" as it used to. If GNU make is not available as "gmake"
2980 on your system, you can change this default behavior by setting the
2981 GNUMAKE environment variable.
2982 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2983 can build without error under CMU CL.
2985 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2987 * There is now a partial SBCL user manual (with some new text and some
2988 text cribbed from the CMU CL manual).
2989 * The beginnings of a profiler have been added (starting with the
2990 CMU CL profiler and simplifying and cleaning up). Eventually the
2991 main interface should be through the TRACE macro, but for now,
2992 it's still accessed through vaguely CMU-CL-style functions and macros
2993 exported from the package SB-PROFILE.
2994 * Some problems left over from porting CMU CL to the new
2995 cross-compilation bootstrap process have been cleaned up:
2996 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2997 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2998 part when getting rid of the compiler *BACKEND* stuff.)
2999 ** The value of *NULL-TYPE* was screwed up, because it was
3000 being initialized before the type system knew the final
3001 definition of the 'NULL type. This screwed up several key
3002 optimizations in the compiler, causing inefficiency in all sorts
3003 of places. (I found it because I wanted to understand why
3004 GET-INTERNAL-RUN-TIME was consing.)
3005 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3006 PROCLAIM FTYPE information. Unfortunately this broke binary
3007 compatibility again, since now the forms output by DEFGENERIC
3008 to refer to functions which didn't exist in 0.6.3.
3009 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3010 can use the new (as of 0.6.3) transform for SOME into MAP into
3012 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3013 values to (INTEGER 0 1000000), so that the time code will no longer
3014 occasionally get blown up by Linux returning 1000000 microseconds
3015 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3016 its output conform to the ANSI spec. (Alas, this makes its output
3017 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3018 * A full call to MAP NIL with a single sequence argument no longer conses.
3019 * fixes to problems pointed out by Martin Atzmueller:
3020 * The manual page no longer talks about multiprocessing as though
3021 it were currently supported.
3022 * The ILISP support patches have been removed from the distribution,
3023 because as of version 5.10.1, ILISP now supports SBCL without us
3024 having to maintain patches.
3025 * added a modified version of Raymond Toy's recent CMU CL patch for
3026 EQUALP comparison of HASH-TABLE
3028 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3030 * The system still can't cross-compile itself with
3031 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3032 entails), but at least it can compile more of itself that way
3033 than it used to be able to, and various buglets which were uncovered
3034 by trying to cross-compile itself that way have now been fixed.
3035 * This release breaks binary compatibility again. This time
3036 at least I've incremented the FASL file format version to 2, so that the
3037 problem can be detected reliably instead of just causing weird errors.
3038 * various new style warnings:
3039 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3040 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3041 ** using the lexical binding of a variable named in the *FOO* style
3042 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3043 as specified by ANSI.
3044 * *RANDOM-STATE* is no longer automatically initialized from
3045 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3046 default behavior of the system is to repeat its behavior every time
3047 it's run. If you'd like to change this behavior, you can always
3048 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3049 old convention there was no comparably easy way to get the system to
3050 repeat its behavior every time it was run.
3051 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3052 FTYPE declarations has been removed, in favor of their ANSI
3053 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3054 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3055 the types of their sequence arguments can be determined at compile time.
3056 This is done through a new open code expansion for MAP which eliminates
3057 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3058 argument types can be determined at compile time.
3059 * The optimizer now transforms COERCE into an identity operation when it
3060 can prove that the coerced object is already of the correct type. (This
3061 can be a win for machine generated code, including the output of other
3062 optimization transforms, such as the MAP transform above.)
3063 * Credit information has been moved from source file headers into CREDITS.
3064 * Source file headers have been made more standard.
3065 * The CASE macro now compiles without complaining even when it has
3068 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3070 * (Note that the way that the PCL macroexpansions were rewritten
3071 to accommodate the change in DEFGENERIC below breaks binary
3072 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3073 not run under sbcl-0.6.2. Once we get out of alpha releases,
3074 i.e. hit release 1.0.0, we'll probably try to maintain binary
3075 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3076 and sbcl-1.4.4. Until then, however, it might be fairly common
3077 for maintenance releases to break binary compatibility.)
3078 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3080 * The reporting of unhandled signals has been changed to print some
3081 explanatory text as well as the report form. (Previously only
3082 the report form was printed.)
3083 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3084 it defines, so that the compiler no longer issues undefined function
3085 warnings for compiled-but-not-yet-loaded generic functions.
3086 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3087 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3088 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3089 section of the ANSI spec are supported.
3090 * The "" nickname for the "KEYWORD" package has been removed.
3091 The reader still handles symbol tokens which begin with a package marker
3092 as keywords, but it doesn't expose its mechanism for doing so in the
3093 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3094 * The system now issues STYLE-WARNINGs for contradictory TYPE
3095 proclamations. (Warnings for contradictory FTYPE proclamations would
3096 be nice too, but those can't be done usefully unless the type system
3097 is made smarter about FUNCTION types.)
3098 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3099 systematized, so that "*target-*.lisp is supposed to exist only on the
3100 target and imply that there's a related file which exists on the
3101 host, and *host-*.lisp is supposed to exist only on the host and imply
3102 that there's a related file which exists on the target. This involves a
3103 lot of renaming. Hopefully the acute confusion caused by the renaming
3104 will be justified by the reduction in chronic confusion..
3105 ** runtime-type.lisp -> early-target-type.lisp
3106 ** target-type.lisp -> late-target-type.lisp
3107 ** early-host-format.lisp -> early-format.lisp
3108 ** late-host-format.lisp -> late-format.lisp
3109 ** host-error.lisp -> misc-error.lisp
3110 ** early-error.lisp -> early-target-error.lisp
3111 ** late-error.lisp -> late-target-error.lisp
3112 ** host-defboot.lisp -> early-defboot.lisp
3113 ** code/misc.lisp -> code/target-misc.lisp
3114 ** code/host-misc.lisp -> code/misc.lisp
3115 ** code/numbers.lisp -> code/target-numbers.lisp
3116 ** code/early-numbers.lisp -> numbers.lisp
3117 ** early-host-type.lisp -> early-type.lisp
3118 ** late-host-type.lisp -> late-type.lisp
3119 ** host-typep.lisp -> typep.lisp
3120 ** load.lisp -> target-load.lisp
3121 ** host-load.lisp -> load.lisp
3122 ** host-disassem.lisp -> disassem.lisp
3123 ** host-insts.lisp -> insts.lisp
3124 ** byte-comp.lisp -> target-byte-comp.lisp
3125 ** host-byte-comp.lisp -> byte-comp.lisp
3126 ** host-signal.lisp -> signal.lisp
3127 ** host-defstruct.lisp -> defstruct.lisp
3128 ** late-target-type.lisp -> deftypes-for-target.lisp
3129 Furthermore, several other previously target-only files foo.lisp (e.g.
3130 hash-table.lisp and random.lisp) have been split into a target-and-host
3131 foo.lisp file and a target-only target-foo.lisp file, with their key type
3132 definitions in the target-and-host part, so that the cross-compiler will
3133 know more about target types.
3134 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3135 gone away. In their place are various *BACKEND-FOO* variables
3136 corresponding to the slots of the old structure.
3137 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3138 into the target SBCL has been fixed.
3139 * The chill.lisp system for loading cold code into a running SBCL
3141 * Support for the CMU CL "scavenger hook" extension has been removed.
3142 (It was undocumented and unused in the CMU CL sources that SBCL was
3143 derived from, and stale in sbcl-0.6.1.)
3144 * Various errors in the cross-compiler type system were detected
3145 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3146 (enabling various consistency checks). Many of them were fixed,
3147 but some hard problems remain, so the compiler is back to
3148 running without *TYPE-SYSTEM-INITIALIZED* for now.
3149 * As part of the cross-compiler type system cleanup, I implemented
3150 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3151 * I have started adding UNCROSS calls throughout the type system
3152 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3153 on cross-compiler input files will be able to go away, and various
3155 * CONSTANTP now returns true for quoted forms (as explicitly required
3158 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3160 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3161 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3162 (MAKE-LIST -1). In the longer term, it ought to become true
3163 that declarations are assertions even at SAFETY 1. For now, it's not
3164 quite true even at SAFETY 3, but it's at least more nearly true..
3165 (Note that this change seems to increases the size of the system by
3166 O(5%) and to decrease the speed of the compiler by 20% or more.)
3167 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3168 for the problem of printing dozens of lines of distracting information
3169 about low-level system machinery as part of the top stack frame
3170 on entry to the debugger when an undefined function was called.
3171 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3172 is not reset to COMMON-LISP-USER.
3173 * Compilation of stuff related to dyncount.lisp has been made conditional
3174 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3175 smaller. The various dyncount-related symbols have been moved into
3176 a new "SB-DYNCOUNT" package.
3177 * tty-inspect.lisp has been renamed to inspect.lisp.
3178 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3179 feature has gone away. (When we eventually port to other flavors of
3180 libc and/or Unix, we'll try to make the differences between flavors
3181 invisible at the user level.)
3182 * Various other *FEATURES* tags, and/or their associated conditionals,
3183 have been removed if obsolescent, or given better documentation, or
3184 sometimes given more-mnemonic names.
3186 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3188 * tidied up "make.sh" script
3189 * tidied up system directory structure
3190 * better "clean.sh" behavior
3191 * added doc/FOR-CMUCL-DEVELOPERS
3192 * many many small tweaks to output format, e.g. removing possibly-confusing
3193 trailing #\. character in DESCRIBE-INSTANCE
3194 * (EQUALP #\A 'A) no longer signals an error.
3195 * new hashing code, including EQUALP hashing
3196 * tidied up Lisp initialization and toplevel
3197 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3198 * command line argument processing
3199 * added POSIX-GETENV function to deal with Unix-ish environment variables
3200 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3201 terminating SBCL on EOF
3202 * non-verbose GC by default
3203 * There is no more "sbcl" shell script; the sbcl file is now the C
3204 runtime executable (just like CMU CL).
3205 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3207 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3208 symbol and package information as Lisp native symbol and package objects
3209 instead of strings naming symbols and strings naming packages. This way,
3210 whenever packages are renamed (as in warm init), debug information is
3211 transformed along with everything else.
3212 * tweaked the optimization policy declarations which control the building
3213 of SBCL itself. Now, among other things, the system no longer saves
3214 source location debugging information. (This helps two problems at once
3215 by reducing SBCL size and by keeping SBCL from trying to look for its
3216 sources -- which may not exist -- when reporting errors.)
3217 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3218 debugging and testing purposes
3219 * cleaned up printing, making the printer call PRINT-OBJECT for
3220 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3221 methods, giving nearly-ANSI behavior
3222 * converted almost all special variables to use *FOO* naming convention
3223 * deleted PARSE-TIME functionality, since it can be done portably
3224 * moved some files out of cold init into warm init
3225 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3227 * regularized formatting of source files
3228 * added an install.sh script
3229 * fixed ridiculous memory usage of cross-compiler by making
3230 compiler/alloc.lisp not try to do pooling unless it can hook
3231 itself into the GC of the cross-compilation host. Now the system
3232 builds nicely on my old laptop.
3233 * added :SB-ALLOC in target-features.lisp-expr
3234 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3235 was not implemented)
3236 * re-did condition handling and note reporting in the compiler. Notes
3237 are no longer handled by signalling conditions. Style warnings
3238 and warnings are handled more correctly and reported in such a way
3239 that it's easy to find one or the other in your output (so that you
3240 can e.g. figure out which of many problems caused COMPILE-FILE to
3242 * changed the severity of several compiler warnings from full WARNING
3243 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3244 compiler note reporting so that it doesn't use the condition system
3245 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3246 COMPILE-FILE command)
3247 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3248 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3249 no longer supported, since their implementation was deeply intertwingled
3250 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3251 * removed bogus "support" for compiler macros named (SETF FOO), and
3252 removed the compiler macro for SETF INFO (but only after making a fool
3253 of myself on the cmucl-imp mailing list by posting a bogus patch for
3254 DEFINE-COMPILER-MACRO..)
3255 * Compiled files containing forms which have side effects on the Lisp
3256 reader (such as DEFPACKAGE forms) are now handled more correctly.
3257 (Compiler queuing of top level lambdas has been suppressed by setting
3258 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3259 * deleted various currently-unused source files, e.g. gengc.lisp. They
3260 may be added back at some point e.g. when porting to other architectures,
3261 but until they are it's distracting to distribute them and to try to
3263 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3264 there were so many of them they're just distractions, and UNCROSS is
3265 known to be able to handle the current sources
3266 * moved PROFILE functionality into TRACE, so that it will be clear
3267 how the wrapping and unwrapping of functions when you profile them
3268 interacts with the wrapping and unwrapping of functions when you
3269 trace them. (Actually, the functionality isn't there yet, but at least
3270 the interface specification is there. Hopefully, the functionality will
3271 arrive with some future maintenance release.)
3272 * removed host-oops.lisp
3273 * changed signature of QUIT function to allow UNIX-CODE argument
3274 * fixed READ-SEQUENCE bug
3275 * tweaked verbose GC output so that it looks more like the progress
3276 output that ANSI specifies for functions like LOAD
3277 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3278 * added <http://sbcl.sourceforge.com> to the banner information printed by