1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
3 * New feature: Experimental support for bivalent streams: streams
4 opened with :element-type :default now allow character and binary
6 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
7 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
9 * bug fix: no more highly sporadic "couldn't check whether ~S is
10 readable" when reading a stream and an interrupt hits in the middle
11 of a select system call
12 * compiler better recognizes complex arrays (reported by Eduardo
14 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
15 for variables on SBCL built with linkage-tables. (reported by Luis
17 * various error reporting improvements.
18 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
19 (Thanks to Thiemo Seufer)
20 * bug fix: nested reader invokations work correctly
21 * bug fix: it is possible to have simultaneous references to foreign
22 code and foreign data with the same name.
24 ** added x86-64 support
25 ** incompatible change: the threading api now works with thread
26 objects instead of thread ids
27 ** bug fix: threads are protected from signals and interruption when
28 starting up or going down
29 ** bug fix: a race where an exiting thread could lose its stack to gc
30 ** bug fix: don't halt on infinite error in threads if possible
31 ** fixed numerous gc deadlocks introduced in the pthread merge
32 ** bug fix: fixed thread safety issues in read and print
33 ** bug fix: debugger doesn't hang on session lock if interrupted at
34 an inappropriate moment
35 ** bug fix: run-program is now thread safe(r)
36 ** bug fix: inner with-recursive-lock no longer releases the mutex
37 * fixed some bugs revealed by Paul Dietz' test suite:
38 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
39 EXPECTED-TYPE slots filled.
40 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
42 ** MAKE-CONDITION accepts classes as type-designators.
43 ** COMPILE may never return NIL.
45 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
46 * numerous signal handling fixes to increase stability
47 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
48 * minor incompatible change: we now correctly canonize default
49 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
50 as per the MOP, rather than the historical (INITARG INITFUNCTION
51 INITFORM). (reported by Bruno Haible)
52 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
53 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
55 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
56 pathname is a directory pathname.
57 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
58 * dynamic space size on PPC has been increased to 768Mb. (thanks to
60 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
61 :METACLASS argument in addition to a class name. (reported by
62 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
63 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
64 (reported by Bruno Haible; patch from Gerd Moellmann)
65 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
67 * bug fix: more cleanups to the floating point exception handling on
68 x86-64 (thanks to James Knight)
69 * bug fix: the compiler does not try to constant fold calls of
70 COERCE and BIT-* functions when they return freshly constructed
71 objects. (reported by Nikodemus Siivola and Paul Dietz)
72 * optimization: improved the MIPS versions of generic (in the
73 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
74 * optimization: direct conversion of (unsigned-byte 32) to floats on
76 * optimization: structure instances with raw slots now use less
77 memory, and probably show better memory locality. (thanks to
79 * optimization: DYNAMIC-EXTENT declarations for lists and closures
80 are treated as requests for stack allocation on the x86-64,
81 Alpha-32, PPC and SPARC platforms.
82 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
83 the wrong file descriptor; implementation of SOCKET-OPEN-P.
84 (thanks to Tony Martinez)
86 ** gcing a dead thread can no longer lead to lockups
87 ** threads block signals until they are set up properly
88 ** errno is no longer shared by threads
89 ** interrupt-thread restores the eflags register on x86
90 ** fixed some lockups due to gc/thread interaction
91 * fixed some bugs revealed by Paul Dietz' test suite:
92 ** invalid dotted lists no longer raise a read error when
94 ** signal an error if a symbol that names a declaration is used
95 as the name of a type, or vice versa
96 ** allow using the (declare (typespec var*)) abbreviation for
97 (declare (type typespec var*)) with all type specifiers
98 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
99 ** always return NIL from PPRINT-POP when OBJECT is NIL
100 ** don't signal errors when pretty-printing malformed LABELS,
101 FLET or MACROLET forms
102 ** declarations in a DOLIST body are also in scope for the
104 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
106 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
109 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
110 * fixed cross-compiler leakages that prevented building a 32-bit
111 target with a 64-bit host compiler.
112 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
113 opened with :IF-EXISTS :OVERWRITE.
114 * fixed bug 281: error for an invalid qualifier in a short-form method
115 combination method is not signalled until the faulty method is called.
116 * bug fix: iteration variable type inferrer failed to deal with open
117 intervals. (reported by Alan Shields)
118 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
119 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
120 or more runtime options were provided to the sbcl binary.
121 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
122 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
123 * optimization: there's now a fast-path for fixnum arguments in the
124 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
125 * optimization: the code generated on x86-64 is more compact thanks
126 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
127 * bug fix: floating-point exception handling now partly works on
128 x86-64. (Thanks to James Knight)
129 * improvement to the MIPS backend from Thiemo Seufer: C-style
130 64-bit long long arguments and return values to alien functions
132 * fixed some bugs revealed by Paul Dietz' test suite:
133 ** the type-error signalled from WARN has a filled-in DATUM slot.
134 ** the type-error required when a stream is not associated with
135 a file has the stream as its datum.
136 ** type-errors on single-floats on x86-64 no longer have
137 :INVALID-OBJECT as the datum
138 ** the type-errors signalled for invalid function names now have
139 a correct expected type
140 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
141 for structures defined with a :TYPE. Documentation strings for
142 typed structures are no longer immediately discarded
143 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
144 broadcast streams and synonym streams. FILE-LENGTH now also works
145 on broadcast streams.
147 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
148 * incompatible change: the --noprogrammer option, deprecated since
149 version 0.7.5, has been removed. Please use the equivalent
150 --disable-debugger option instead.
151 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
153 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
154 inoperational for a while now) has been completely removed.
155 * null lexical environments are now printed as #<NULL-LEXENV>,
156 significantly reducing the amount of clutter in typical backtraces.
157 * documentation on weak pointers, finalization, and after GC hooks
158 has been added to the manual.
159 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
160 as other specialized array types, is much faster. SUBSEQ and
161 COPY-SEQ on such arrays have also been sped up.
162 * optimization: EQL is now more efficient when at least other argument
163 is known to be of type (OR FIXNUM (NOT NUMBER)).
164 * fixed bug: the runtime is now less vulnerable to changes in the
165 size of the SBCL object on OS X, and virtual memory is reserved for
166 all spaces that need to be at a fixed address.
167 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
168 * fixed bug: finalizers and after GC hooks that cause consing are now
170 * fixed bug: compiler error messages and summaries are now printed to
171 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
172 * fixed inference of the upper bound of an iteration variable.
173 (reported by Rajat Datta).
174 * fixed bug 211e: calling local functions with duplicated constant
175 keyword argument no longer causes a bogus style warning about an
177 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
178 ftype declaration to be disregarded. (reported by Dave Roberts)
179 * fixed bug 373: caused by erronous compilation of references to alien
180 variables in the runtime on ppc/darwin.
181 * fixed bug 376: CONJUGATE type deriver.
182 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
183 (reported by Baughn on #lisp)
184 * a cross-compiler bug on non-x86 platforms has been identified and
185 fixed. (thanks to Bruno Haible)
186 * improvements to the MIPS runtime code for increased stability.
187 (thanks to Thiemo Seufer)
188 * increased the maximimum compact environment size to allow
189 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
190 * partial workaround for bug 135: don't name the function we're
191 calling for hairy cases of EVAL, so as not to accumulate one environment
192 entry per call to EVAL. (reported by Kevin Reid)
193 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
194 * optimization: type testing for non-vector arrays should be faster.
195 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
197 * bug fix: setting 31st element of a bit vector to zero did not work
199 * bug fix: redefining a class definition which failed due to a
200 previous accessor / function clash now works (but see BUGS entry
201 #380 for more problems in this area). (thanks to Zach Beane)
202 * the long form of DEFINE-METHOD-COMBINATION disables method group
203 checking when given a single method group with pattern *. (thanks
205 * on x86 compiler supports stack allocation of results of simple
206 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
207 * support for building and running on Mac OS X 10.4 "Tiger" has been added
208 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
209 * fixed some bugs related to Unicode integration:
210 ** the restarts for recovering from input and output encoding
211 errors only appear when there is in fact such an error to
213 * fixed some bugs revealed by Paul Dietz' test suite:
214 ** MISC.549 and similar: late transformation of unsafe type
215 assertions into derived types caused unexpected code
217 ** SCALE-FLOAT type deriver is less wrong.
218 ** type derivers for EXP, LOG and similar functions compute result
219 types for complex arguments better.
220 ** (MISC.563) CONJUGATE type deriver works for very restricted
222 ** out-of-line type testers for character strings are available.
223 ** EQUAL compiler transform understands specialness of objects
225 ** accessing double-floats stored on the stack now works on x86-64.
226 ** debugger internals could sometimes create invalid lispobjs,
227 resulting in GC crashes.
228 ** MISC.548: type check weakening can convert required type into
230 ** initialization forms for bindings are not in scope of free special
233 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
234 * incompatible change: thread support for non-NPTL systems has
235 been removed - locking is buggy and unreliable. A threaded
236 SBCL build will now warn at startup and refuse to create
237 new threads, unless futex support is detected in the kernel
238 * incompatible change: the top level REPL now has only an ABORT
239 restart associated with it, not TOPLEVEL and ABORT as it used to.
240 TOP and TOPLEVEL are now available as debugger commands for
241 returning to the top level.
242 * incompatible change: forms evaluated in the REPL now use the
243 global optimization policy.
244 * incompatible change: user- and system-initialization files are
245 no longer processed with LOAD, but by READ and EVAL; hence the
246 global optimization policy, startup package, readtable, etc,
248 * The .fasl file format number has been incremented because of
249 various incompatible changes.
250 * internal entry point details and argument counts no longer appear
251 in backtraces unless explicitly requested by setting
252 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
253 * built-in and standard functions no longer have names like "top
254 level local call to FOO".
255 * fixed bug 32: functions defined in non-null lexical environments
256 now have more legible printed representation
257 * fixed bug 33: functions defined in non-null lexical environemnts
258 are now more amenable to inspection by INSPECT.
259 * workaround for bug 354: XEPs no longer appear in backtraces unless
260 explicitly requested.
261 * fixed bug: receiving the signal which results from attempting to
262 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
263 SIGBUS on other *BSDs) on architectures where the C stack is also
264 the Lisp stack (x86 and x86-64) from foreign code no longer leads
265 to debugger confusion or wild instability. (reported by Cheuksan
267 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
268 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
269 (reported by Lutz Euler)
270 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
271 compiled code. (reported by Paul Dietz)
272 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
273 specialized parameter to a method to something that is not TYPEP
274 the specializer is now possible.
275 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
276 face of package deletion.
277 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
278 pathname. (reported by David Tolpin on c.l.l)
279 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
280 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
281 STRING-STREAM and FILE-STREAM, respectively.
282 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
283 than previously; in addition, its test suite is now run on build.
284 (thanks to Luke Gorrie)
285 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
286 * optimization: added a immediate representation for single-floats
288 * fixed some bugs revealed by Paul Dietz' test suite:
289 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
290 ** attempting to create a package with a colliding nickname causes
291 correctable errors to be signalled.
292 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
293 ** several x86-64 backend bugs related to sign-extension of immediate
296 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
297 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
298 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
300 * fixed loading of multiply forward-referenced layouts.
301 (thanks to Cheuksan Wang)
302 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
303 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
304 (see COMPILE-FILE documentation for details of :PRINT option)
305 * fixed bugs 19 and 317: fixed-format floating point printing is
306 more accurate. This also fixes a bug reported by Adam Warner
307 related to the ~@F format directive.
308 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
310 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
311 dispatch table if the from-char is a dispatch macro character.
312 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
313 by the wordsize no longer ignore the last word. (reported by Lutz
315 * fixed bug in type checking of dynamic-extent variables. (reported
317 * optimization: sequence traversal functions use their freedom to
318 coerce function designators to functions.
319 * optimization: code with many calls to CLOS methods specialized on
320 CLOS classes has had redundant type checks removed.
321 * optimization: added declarations to speed up operations that access
322 the internal character database (for example STRING-UPCASE)
323 * optimization: comparison operations between floats and sufficiently small
324 fixnums no longer create extra rationals
325 * fixed some bugs related to Unicode integration:
326 ** portions of multibyte characters at the end of buffers for
327 character-based file input are correctly transferred to the
328 start of the buffer at the next read.
329 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
330 passing it through to OPEN.
331 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
332 argument given to internal calls to OPEN.
333 ** the built-in comment readers (introduced by character sequences
334 ";" and "#|") are more forgiving to encoding errors; they will
335 STYLE-WARN and then attempt to resync the stream at a character
336 boundary. (thanks to Teemu Kalvas)
337 * fixed some bugs in the x86-64 port:
338 ** Negative short int return values from c-calls are sign-extended
340 ** The stack is aligned to 16-bytes for c-calls, as required by
341 the ABI. (thanks to Cheuksan Wang)
342 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
343 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
344 for enums. (thanks to Vincent Arkesteijn)
345 ** Multiple small optimizations and bugfixes for floating point
347 * fixed some bugs revealed by Paul Dietz' test suite:
348 ** Space, Tab, Linefeed, Return and Page have the invalid
349 secondary constituent character trait.
350 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
352 ** WITH-INPUT-FROM-STRING should only update the index place on
354 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
356 ** Bit-array operations (BIT-AND and similar) worked incorrectly
357 with one-dimensional arrays with fill pointers.
358 ** TYPE-OF failed on a complex with an integer realpart and a
360 ** compiler failure during type inference for the code of form
361 (IF (EQL X (THE ...)) ...) (MISC.535).
363 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
364 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
365 platform. The port supports 61-bit fixnums, large memory spaces
366 and reloading shared object files.
367 * enhancement: saving cores with foreign code loaded is now
368 supported on ppc/Darwin in addition to the previously supported
370 * enhancement: the statistical profiler now walks deeper into the
371 call stack for more meaningful call-graphs and accrued time
372 reports (x86/x86-64 only). It also now reports time spent in
374 * enhancement: it is now possible to trace most individual methods
375 of a generic function in addition to tracing the generic function
377 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
379 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
380 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
381 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
382 produces an error. (thanks to Vincent Arkesteijn)
383 * bug fix: NAMESTRING on pathnames with :WILD components in their
384 directories works correctly. (thanks to Artem V. Andreev)
385 * fixed bug 125: compiler preserves identity of closures. (reported
387 * bug fixed: functions with &REST arguments sometimes failed with
388 "Undefined function" when compiled with (DEBUG 3). (reported by
390 * bug fix: overflow during compiling of setting element of a bit
391 vector with constant index and value. (reported by Timmy Douglas)
392 * build fix: fixed the dependence on *LOAD-PATHNAME* and
393 *COMPILE-FILE-PATHNAME* being absolute pathnames.
394 * on x86 compiler partially supports stack allocation of dynamic-extent
396 * GO and RETURN-FROM do not check the extent of their exit points
397 when compiled with SAFETY 0.
398 * fixed some bugs related to Unicode integration:
399 ** encoding and decoding errors are now much more robustly
400 handled; it should now be possible to recover even from invalid
401 input or output to the terminal. (thanks to Teemu Kalvas)
402 ** provided a first cut at implementing STRING-TO-OCTETS and
403 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
404 ** altered the SB-MD5 contributed module slightly, changing the
405 interface just enough for it to be supportable for builds where
406 lisp characters are not eight bits.
407 * fixed some bugs revealed by Paul Dietz' test suite:
408 ** the FORMATTER-generated functions for ~V[ conditionals require
409 the correct number of arguments.
410 ** READ-FROM-STRING returns the mandated second value when applied
411 to displaced strings.
412 ** the #\Rubout and #\Backspace characters are treated as invalid
413 constituent characters by the tokenizer.
415 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
416 * new feature: reloading changed shared object files with
417 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
418 * new feature: references to foreign variables and functions
419 can now be compiled and loaded before the corresponding shared
420 object file is loaded, as long as the foreign definitions are
421 available at runtime.
422 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
423 supported. (thanks to Dan Debertin)
424 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
425 just keywords. (thanks to Vincent Arkesteijn)
426 * fixed bug #331: structure-class instances corresponding to
427 DEFSTRUCT forms are now created eagerly.
428 * fixed bug #345: backtraces from calls to undefined functions work
429 on x86 as well. Related bug #61 is now also partially fixed on x86
430 (backtraces from throws to unknown catch tags.)
431 * bug fix: lambda-list parsing is now stricter vrt. order and number
432 of lambda-list keywords.
433 * bug fix: as specified by AMOP, an error is signalled if a
434 class-option appears multiple times in a DEFCLASS form. (reported
436 * bug fix: RANDOM can be compiled when the compiler derives the type
437 of its numeric argument as a disjoint set of small integers.
438 (reported by Paul Dietz)
439 * bug fix: starting a core saved with shared objects loaded when
440 those objects are not available no longer causes threaded SBCL to
441 hang. (reported by Sean Ross)
442 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
444 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
445 argument, which must always be NIL. (reported by Kalle Niemitalo)
446 * bug fix: printing 1.0d+23 no longer results in an error.
447 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
448 * bug fix: structure slot setters preserve evaluation order. (thanks
450 * fixed some bugs related to Unicode integration:
451 ** RUN-PROGRAM can allow its child to take input from a Lisp
452 stream. (reported by Stefan Scholl)
453 ** ASDF-INSTALL successfully downloads .tar.gz files in all
454 locales. (reported by Ken Causey)
455 * fixed some bugs revealed by Paul Dietz' test suite:
456 ** INCF, DECF and REMF evaluate their place form as specified in
458 ** FORMATTER expands ~{ ~} iteration directives with V or #
459 parameters correctly.
460 ** FORMATTER deals with the ~@[ ~] conditional directive where the
461 consequent uses no arguments correctly.
462 ** the system has a partial understanding of the (COMPLEX RATIO)
464 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
467 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
468 * new feature: a build-time option (controlled by the :SB-UNICODE
469 keyword feature, enabled by default) for building the system with
470 support for the entire 21-bit character space defined by the
472 * new feature: the system now has rudimentary external-format
473 support; the primary user-visible change at this time is that
474 characters with the high bit set (such as lower-case-e-acute) will
475 print correctly to a terminal in a UTF-8 environment.
476 * minor incompatible change: BASE-CHAR no longer names a class;
477 however, CHARACTER continues to do so, as required by ANSI.
478 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
479 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
480 has been moved to the SB-EXT package (temporarily re-exported from
482 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
483 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
485 * fixed bug #351: better error-handlind and reporting for malformed
487 * fixed bug #350: bignum-printing is now more memory-efficient,
488 allowing printing of very large bignums, eg. (expt 2 10000000).
489 (reported by Bruno Haible)
490 * fixed bug #302: better primitive-type selection for intersection
492 * fixed bug #308: non-graphic characters now all have names, as
493 required. (reported by Bruno Haible)
494 * bug fix: redefining a class with different superclasses now correctly
495 removes it from the direct-subclasses of its previous superclasses.
496 (reported by David Morse)
497 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
498 new value now works. (reported by Bruno Haible)
499 * bug fix: correct canonicalization of multiple non-standard slot
500 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
501 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
502 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
503 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
504 now exists, an signals an error.
505 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
506 during macroexpansion for non-iteration variables as well. (reported
507 by Bruno Haible for CMUCL)
508 * bug fix: Cyclic structures and unprintable objects in compiler
509 messages no longer cause errors. (reported by Bruno Haible)
510 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
511 and READ-FROM-STRING all now return a primary value of NIL if
512 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
513 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
514 (reported by Bruno Haible for CMUCL)
515 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
516 do not designate a condition. (reported by Bruno Haible for
518 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
519 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
520 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
521 permitting resolution in favour of any of the conflicting symbols.
522 (reported by Bruno Haible for CMUCL)
523 * bug fix: EQUAL compiler optimizations is less aggressive on
524 strings which can potentially compare true despite having distinct
525 specialized array element types.
526 * bug fix: unit enumerations can be defined without dividing by
527 zero. (thanks to Vincent Arkesteijn)
528 * FORMAT compile-time argument count checking has been enhanced.
529 (report from Bruno Haible for CMUCL)
530 * a partial workaround for the bug 262: the compiler does not try to
531 inline-expand a local function doing RETURN-FROM from a deleted
532 BLOCK. (thanks to Peter Denno for the bug report and to David
533 Wragg for the simple test case)
534 * fixed some bugs revealed by Paul Dietz' test suite:
535 ** RENAME-PACKAGE allows all package designators as new package
537 ** constraint propagation and lambda variable substitution are
538 more cautious in dealing with partially deleted code.
539 ** compiler sometimes lost reoptimization passes.
540 ** CERROR, given a condition as condition designator, passes its
541 remaining arguments to the continue format control without
543 ** Case-altering FORMAT directives work correctly on non-ASCII
545 ** The REAL type specifier handles bounds outside the floating
546 point ranges without signalling FLOATING-POINT-OVERFLOW.
547 ** Functions with IR1-transformations can create intercomponent
548 references to global functions.
549 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
551 ** FORMAT ~R treats a nil value for its first parameter correctly.
553 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
554 * enhancement: saving cores with foreign code loaded is now
555 supported on x86/NetBSD and sparc/Linux in addition to the previously
557 * bug fix: on some platforms repeated installations caused multiple
558 copies of HTML documentation to be installed -- should not happen
559 any more. (reported by Stefan Scholl)
560 * bug fix: parsing self-recursive alien record types multiple times
561 no longer causes infinite recursion. (reported by Thomas F. Burdick,
562 original patch by Helmut Eller for CMUCL)
563 * bug fix: stack-exhaustion detection works now on NetBSD as well.
564 (thanks to Richard Kreuter)
565 * bug fix: defining classes whose accessors are methods on existing
566 generic functions in other (locked) packages no longer signals
567 bogus package lock violations. (reported by François-René Rideau)
568 * bug fix: special variables as DEFMETHOD parameters no longer have
569 associated bogus type declarations. (reported by David Wragg and
571 * bug fix: read-write consistency on streams of element-type
572 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
573 * bug fix: redefiniton of the only method of a generic function with
574 no DEFGENERIC no longer emits a full WARNING. In addition,
575 redefinition of generic functions with no DEFGENERIC to an
576 incompatible lambda list now signals an error. (thanks to Zach
578 * bug fix: DEFGENERIC now works even when there's a function of the
579 same name in an enclosing lexical environment. (thanks to Zach
581 * fixed compiler failure, caused by instrumenting code during
582 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
583 * optimization: added loop analysis and improved register allocation
584 * fixed some bugs revealed by Paul Dietz' test suite:
585 ** POSITION on displaced vectors with non-zero displacement
586 returns the right answer.
587 ** (SIMPLE-STRING) is a valid type specifier for sequence
589 ** *PRINT-LEVEL* handling for slotless structures is pedantically
591 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
592 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
594 ** The justification version of the FORMAT ~< directive treats
595 non-zero minpad parameter correctly.
597 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
598 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
599 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
600 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
601 the supported interface.
602 * new feature: Single-stepping of code compiled with DEBUG 2 or
603 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
604 * new feature: saving cores with foreign code loaded is now
605 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
606 Timothy Moore's work for CMUCL)
607 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
608 parameters to * instead of NIL if no initform is supplied.
609 (reported by Johan Bockgaard)
610 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
611 correctly when FUNCALL appears as the car of the form. Note:
612 despite this FUNCALL forms are not currently subject to
613 compiler-macro expansion. (port of Raymond Toy's fix for the
614 same from CMUCL, reported by Johan Bockgaard)
615 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
616 (thanks for Teemu Kalvas)
617 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
618 case instead of terminating the iteration (thanks for Julian
619 Squires, Sean Champ and Raymond Toy)
620 * bug fix: incorrect expansion of defgeneric that caused a style
621 warning. (thanks for Zach Beane)
622 * bug fix: slot accessor effective method computation works properly
623 for classes with multiple non-standard applicable methods on
624 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
625 * on x86 compiler supports stack allocation of results of LIST and
626 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
627 CMUCL implementation by Gerd Moellmann)
628 * fixed some bugs revealed by Paul Dietz' test suite:
629 ** FORMAT strings with both the ~<~:;~> form of the justification
630 directive and pretty-printing directives cause an error.
632 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
633 * incompatible change: the internal functions
634 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
635 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
636 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
637 instead of the old functions.
638 * new feature: on platforms where "dladdr" is available foreign
639 function names now appear in backtraces. (based on Helmut Eller's
641 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
642 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
644 * documentation: ASDF manual is now distributed with SBCL.
645 * bug fix: SBCL can now load its contributed modules with REQUIRE
646 even if the system-provided entries have been removed from
647 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
649 * bug fix: backtraces involving undefined functions or assembly
650 routines are more informative. (thanks to Brian Downing)
651 * bug fix: mutually referent alien structures now work correctly.
652 (reported by Rick Taube)
653 * bug fix: structures defined by WITH-ALIEN can be referred to
654 within other definitions in the same WITH-ALIEN.
655 * bug fix: division operators (MOD, TRUNCATE and the like) with
656 constant zero divisors and integer dividends no longer generate
658 * bug fix: provide default methods for INPUT-STREAM-P and
659 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
660 * bug fix: improve the detection and resolution of MOP metacycles
661 (where effective method computation depends on the generic function
662 having its effective method computed). (reported by Bruno Haible)
663 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
664 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
665 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
666 * bug fix: pathnames with relative directory components can be
667 represented relative to default pathnames.
668 * optimization: in taking the GCD of bignums, reduce the two bignums
669 to approximately the same size (using Euclid's algorithm) before
670 applying the more sophisticated binary GCD. (thanks to Juho
672 * optimization: COUNT on bitvectors now operates word-at-a-time.
673 * optimization: ASH with a positive, but not necessarily constant,
674 (leftwards) shift, when in a modular context, is compiled to a
676 * fixed some bugs revealed by Paul Dietz' test suite:
677 ** FORMAT variable parameters ("~V<char>") are defaulted properly
678 if the corresponding argument is NIL.
679 ** FORMAT directives accepting a minpad parameter treat negative
681 ** literal commas as character parameters to FORMAT directives are
683 ** literal spaces directly after ~<Newline> directives within a
684 format-logical-block (~:< ~@:>) do not induce :FILL-style
685 conditional newlines.
686 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
687 as specified: it no longer includes conditional newlines).
688 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
690 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
691 :PER-LINE-PREFIX argument does not evaluate to a string.
693 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
694 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
695 the manual for details; for now, package locks can be disabled by
696 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
697 if no major problems are found then it is likely that they will be
698 compiled in unconditionally.
699 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
700 unsupported operators on all platforms. To load a shared library
701 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
702 object file, link it into a shared library outside of SBCL and
703 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
705 * bug workaround: ROOM T and the GC were not working together
706 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
707 reporting facility aren't preserved. That reporting has been
708 disabled, so now until and unless someone figures out how to make it
709 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
710 * minor incompatible change: as threatened around sbcl-0.8.0, the
711 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
712 The symbols are also exported from SB-PCL for backwards
713 compatibility, but more so than before SB-PCL should be treated as
714 an implementation-internal package.
715 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
717 * the SB-SPROF contrib now works on (most) non-x86 architectures.
718 It is known as of this release not to work on the Alpha, however.
719 * fixed bug #167: errors signalled due to illegal syntax in method
720 bodies are now more legible.
721 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
722 designators and can hence be used with TYPEP.
723 * fixed bug #333: CHECK-TYPE now ensures that the type error
724 signalled, if any, has the right object to be accessed by
725 TYPE-ERROR-DATUM. (reported by Tony Martinez)
726 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
727 5.1.2.3 for multiple-value place subforms. (reported by Kalle
729 * fixed bug #334: programmatic addition of slots using specialized
730 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
731 and :ALLOCATION :CLASS slots.
732 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
733 float exponents. (rereported by Peter Seibel)
734 * fixed a bug: #\Space (and other whitespace characters) are no
735 longer considered to be macro characters in standard syntax by
737 * fixed bug: initialization of condition class metaobjects no longer
738 causes an instance of the condition to be created. (reported by
740 * fixed bug: it is now possible to have slots such that
741 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
742 description is neither :INSTANCE nor :CLASS.
743 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
744 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
745 SLOT-VALUE-USING-CLASS), and no longer causes errors with
746 non-standard SLOT-DEFINITION-ALLOCATION values.
747 * fixed bugs: various race conditions handling exiting threads.
748 CL-PPCRE's thread torture test now passes.
749 * fixed arguably-a-bug: GC time proportional to number of threads in
750 system even when most of them are idle
751 * optimization: improved performance of BIT and SBIT on bit-vectors.
752 * .fasl file incompatibility: The fasl file version number has
753 been incremented because of changes associated with package locks.
755 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
756 * minor incompatible change: the system no longer provides
757 optimization hints (conditions which are TYPEP
758 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
759 mode; these hints are still emitted when higher SPEED optimization
761 * new contrib module: a sampling profiler (profiling by statistical
762 sampling, rather than by instrumenting functions) is available as
763 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
764 * the behaviour of the standard function ED is now customizeable by
765 third parties through a hook variable: see ED's documentation
766 string for information on the protocol.
767 * the compiler no longer emits efficiency notes for (FUNCALL X)
768 when the type of X is uncertain under default optimization
770 * fixed bug 276: mutating a binding of a specialized parameter to a
771 method to something that is not TYPEP the specializer is now
773 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
774 exhibit the required behaviour on the x86 platform. (thanks to
775 Peter van Eynde, Eric Marsden and Bruno Haible)
776 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
777 even for difficult arguments. (reported by Peter Graves)
778 * fixed bug 141a: the backquote printer now descends quoted
780 * fixed another bug in backquote printing: no more destructive
781 modification of the form's list structure. (reported by Brian
783 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
784 move between its address being taken and the call to
785 interrupt_thread, fixing a crashing race condition.
786 * the SB-POSIX contrib implementation has been adjusted so that it
787 no longer exhibits ridiculously poor performance when constructing
788 instances corresponding to C structs.
790 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
791 * minor incompatible change: the sb-grovel contrib now treats C
792 structures as alien (in the sense of SB-ALIEN) objects rather than
793 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
794 has implications for memory management of client code
795 (sb-grovel-returned objects must now be manually managed) and for
796 type safety (alien objects now have full types).
797 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
798 used to control emission of compiler diagnostics, rather than the
799 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
800 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
801 quality should be considered deprecated.
802 * install.sh now installs the user manual as well
803 * (not quite a new documentable feature, but worth considering in
804 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
805 beginnings of a semantically meaningful condition hierarchy is
806 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
808 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
809 designator as the defaults argument.
810 * fixed bug: Displaced arrays whose displaced-to array has become
811 too small now cause ARRAY-DIMENSION to signal an error, providing
812 for safer bounds-checking. (reported by Bruno Haible)
813 * fixed bug: DEFCLASS slot definitions with identical :READER and
814 :WRITER names now signal a reasonable error. (reported by Thomas
816 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
818 * fixed bug: Invalid slot specification errors now print correctly.
819 (thanks to Zach Beane)
820 * fixed bug 320: Shared to local slot value transfers in class
821 redefinitions now happen corrently from superclasses as
822 well. (reported by Bruno Haible)
823 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
825 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
826 improper lists correctly. (reported by Bruno Haible)
827 * fixed bug 313: source-transform for <fun-name> was erroneously
828 applied to a call of a value of a variable with name <fun-name>.
829 (reported by Antonio Menezes Leitao)
830 * fixed bug 307: The obsolete instance protocol ensures that
831 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
832 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
833 remove dead unknown-values globs from the middle of the stack.
834 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
835 easier to resume long computations after using *BREAK-ON-SIGNALS*
836 to diagnose and fix failures. (thanks to Nikodemus Siivola)
837 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
838 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
839 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
840 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
841 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
843 * fixed a (fairly theoretical) bug in string printing: if
844 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
845 does not have array-element-type equal to the most general string
847 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
848 function name. (thanks to Nikodemus Siivola)
849 * fixed bug: compile-time format string checker failed on
850 non-closed ~{. (reported by Thomas F Burdick)
851 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
852 don't warn on legal loop code involving more than one aggregate
853 boolean. (thanks to Nikodemus Siivola)
854 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
855 in keyword position to MAKE-PACKAGE was not being reported as
857 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
858 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
860 * fixed bugs 280 and 312: the checking for multiple definitions in a
861 file is less likely to become confused by uses of inline
863 * fixed bug: the #S reader macro performs the keyword coercion
864 specified for slot names. (reported by Kalle Niemitalo)
865 * fixed bug: lambda lists may contain symbols whose names start with
866 & but are not lambda-list-keywords; their occurrence triggers a
868 * fixed bug 321: define-method-combination argument lambda lists do
869 not cause invalid code to be generated when &optional or &aux
870 variables have default values. (reported by Bruno Haible)
871 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
872 have CLOS classes; this appears also to have fixed reported
873 instability in stack exhaustion detection.
874 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
875 MUFFLE-WARNING when no associated restart is present is now
877 * optimization: rearranged the expansion of various defining macros
878 so that each expands into only one top-level form in a
879 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
881 * optimization: used a previously little-used slot in symbols to
882 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
884 * fixed some bugs revealed by Paul Dietz' test suite:
885 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
887 ** Fixed an optimization bug related to inheritance of initforms
888 from local to shared slots.
889 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
890 ** CAST optimizer forgot to flush argument derived type.
891 ** print/read consistency on floats is now orders of magnitude
892 more likely. (thanks also to Bruno Haible for a similar report
894 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
895 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
896 for objects of type REAL. Make it so.
897 ** SXHASH is specified (infelicitously) to respect similarity,
898 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
899 it so. (thanks to Markus Ziegler)
900 ** on the Alpha, the compiler succeeds in compiling functions
901 returning a known number of arguments greater than 63.
902 ** fixed handling of invalid NIL arguments in keyword position
904 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
906 ** provided a readably-printable representation for RANDOM-STATE
908 ** ensured that pathnames, where they have a namestring, always
909 print using #P"..." syntax.
911 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
912 * Support for the forthcoming 2.0 version of the NetBSD kernel
913 running on x86 hardware has been added. (thanks to Perry
914 E. Metzger most immediately, and others for their past work)
915 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
916 current release version; SBCL's OpenBSD support had been broken
917 since about the time of OpenBSD's switch to ELF binary format.)
918 * [placeholder for DX summary]
919 ** user code with &REST lists declared dynamic-extent, under high
920 speed or space and low safety and debug optimization policy.
921 * The manual has been converted to Texinfo format and the debugger
922 chapter from the cmucl manual has been added.
923 * A facility has been added to extract documentation strings from
924 sbcl and store them as Texinfo-formatted snippets for inclusion in
925 the manual (via Texinfo's @include directive)
926 * bug fix: compiler emitted division in optimized DEREF. (thanks for
927 the test case to Dave Roberts)
928 * bug fix: multidimensional simple arrays loaded from FASLs had fill
929 pointers. (reported by Sean Ross)
930 * bug fix: PROFILE output is printed nicely even for large numerical
931 values. (thanks to Zach Beane)
932 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
933 greater than 32 handle EOF correctly.
934 * bug fix: on X86 an immediate argument of the IMUL instruction is
935 correctly printed in disassembly. (thanks to Lutz Euler)
936 * bug fix: class slots in redefined classes preserve their old
937 values. (thanks to Bruno Haible and Nikodemus Siivola)
938 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
939 succeeds. (reported by Marco Baringer)
940 * fixed some bugs revealed by Paul Dietz' test suite:
941 ** READ-SEQUENCE now works on ECHO-STREAMs.
942 ** RATIONALIZE works more according to its specification. (thanks
945 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
946 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
947 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
948 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
949 should matter to you only if you rebind the printer control
950 variables and then find you want different bindings in the
951 debugger than in the ordinary execution of your program.)
952 * The runtime build system has been tweaked to support building
953 (on SPARC/SunOS) using a C compiler which invokes Sun's own
954 assembler and linker. (thanks to Nikodemus Siivola)
955 * Unbound, undefined, undeclared variables now trigger full
956 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
957 more often programmer error than deliberate exploitation of undefined
959 * optimization: the hash algorithm for strings has changed to one
960 that is less vulnerable to spurious collisions. (thanks to Juho
962 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
963 needless bounds checking. (thanks to Juho Snellman)
964 * optimization: implemented multiplication as a modular
965 (UNSIGNED-BYTE 32) operation on the PPC backend.
966 * fixed some bugs revealed by Paul Dietz' test suite:
967 ** ADJUST-ARRAY now copies the datum in a zero rank array if
969 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
970 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
971 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
973 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
975 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
977 ** ECHO-STREAMs no longer attempt to echo the end of file value to
978 their output stream on EOF from read.
979 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
980 have been read to end-of-file.
981 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
983 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
984 description of determination of which consecutive characters
986 ** Printing the "Space" character with escaping on now yields "#\\ ",
987 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
988 ** Reading floating-point numbers with *READ-BASE* set to a number
989 less than 10 works correctly.
990 ** Reading floating-point numbers with *READ-BASE* set to a number
991 more than 10 works correctly.
992 ** Printing with *PRINT-READABLY* targets the standard readtable, not
993 the readtable currently in effect.
995 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
996 * minor incompatible change: parsing of namestrings on a physical
997 (Unix) host has changed; numbers after the final #\. in a
998 namestring are no longer interpreted as a version field. This is
999 intented to be largely invisible to the user, except that the
1000 meaning of the namestring "*.*.*" has changed: it now refers to a
1001 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1002 should usually be replaced by
1003 (make-pathname :name :wild :type :wild :version :wild)
1004 with the added benefit that this is more likely to be portable.
1005 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1006 option to OPEN now signals an error if the file being opened
1007 exists; this may have an impact on existing code.
1008 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1009 signals received, on the PowerPC platforms (both Linux and
1010 Darwin). (thanks to Pierre Mai for pointing out the location of
1012 * several fixes on OS X: The system now builds and runs cleanly on
1013 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1014 (thanks to Brian Mastenbrook)
1015 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1016 second-resolution: integer multiples of 1/3600 between -24 and 24.
1017 (thanks to Vincent Arkesteijn)
1018 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1019 type when called with 1 argument; PEEK-CHAR checked type of
1020 PEEK-TYPE only after having read first character from a
1021 stream. (reported by Peter Graves)
1022 * bug fix: the garbage collector now has much better locality
1023 behaviour, and in particular no longer treats all memory as being
1024 exhausted when many small objects point to each other in a deeply
1026 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1027 recognized as being TYPEP their class.
1028 * bug fix: the PUSHNEW documentation string has been corrected.
1029 (thanks to Vincent Arkesteijn)
1030 * bug fix: defaulting of the value for the last of an atypically
1031 large number of multiple values being bound was not being
1032 performed correctly on the Alpha or PPC platforms
1033 * optimization: implemented multiplication as a modular
1034 (UNSIGNED-BYTE 32) operation on the x86 backend.
1035 * optimization: SEARCH on simple-base-strings can now be open-coded.
1036 (see also contrib/compiler-extras.lisp for inspiration for
1037 teaching the compiler about the Boyer-Moore algorithm).
1038 * value, returned by MAX (and MIN) called with several EQUALP, but
1039 not EQL, arguments now does not depend on compiler settings.
1040 * fixed some bugs revealed by Paul Dietz' test suite:
1041 ** in stack analysis liveness information is propagated from
1042 non-local entry points.
1043 ** pathwise CAST removing failed when the CAST node did not start
1045 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1046 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1048 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1049 host is already defined.
1050 ** RENAME-FILE works on streams instead of signalling an internal
1052 ** PEEK-CHAR uses the current readtable when determining whether
1053 or not a character is whitespace.
1054 ** MERGE-PATHNAMES handles the case when the pathname does not
1055 specify a name while the default-pathname specifies a version.
1056 ** Pathnames now stand a better chance of respecting print/read
1058 ** Attempting to use standardized file system operators with a
1059 pathname with invalid :DIRECTORY components signals a
1061 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1063 ** DIRECTORY on logical pathnames is more correct.
1064 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1065 signal a TYPE-ERROR if their argument is not a stream.
1066 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1067 designator argument does not designate a stream.
1068 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1069 examining the synonym.
1070 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1072 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1073 element-type larger than ([UN]SIGNED-BYTE 32).
1075 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1076 * When built with the :SB-FUTEX feature, threaded builds now take
1077 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1078 for faster/more reliable mutex and condition variable support.
1079 * Incompatible change (but one you probably shouldn't have been using
1080 anyway): the interface and code for arbitrating between multiple
1081 threads in the same user session has been redesigned.
1082 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1083 error when called without an explicit environment argument.
1084 (thanks to Nikodemus Siivola)
1085 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1086 confused about their position. (thanks to Adam Warner and Gerd
1088 * bug fix: Pretty printing backquoted forms with unquotations in the
1089 argument list position of various code constructs such as LAMBDA
1090 now works correctly. (reported by Paul Dietz)
1091 * bug fix: Pretty printing unquotations no longer loses all
1092 stream position information.
1093 * optimization: performance of string output streams is now less
1094 poor for multiple small sequence writes.
1095 * optimization: performance of CSUBTYPEP in the presence of complex
1096 expressions involving CONS and NOT many times has been improved.
1097 (reported by Paul Dietz)
1098 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1100 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1101 streams for the REPL, for the debugger, and so on. (thanks to
1103 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1104 * fixed some bugs revealed by Paul Dietz' test suite:
1105 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1106 is ignored if it is too small, rather than propagating through
1107 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1108 ** extremely complex negations of CONS types were not being
1109 sufficiently canonicalized, leading to inconsistencies in
1111 ** VALUES tranformer lost derived type.
1113 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1114 * fixed a bootstrapping bug: the build process no longer assumes
1115 that the various BOOLE-related constants have the same value in
1116 host and target lisps. (noted by Paul Dietz' test suite on an
1117 SBCL binary built from CLISP)
1118 * The system can now be dynamically linked on the MIPS platform,
1119 which enables dynamic loading of foreign code from Lisp. (thanks
1120 to Ralf Baechle for discussions on the MIPS ABI)
1121 * The system now records debugging information for its own source
1122 files in a filesystem-position-independent manner, relative to
1123 the "SYS" logical host.
1124 * fixed a compiler bug: MV-LET convertion did not check references
1125 to the "max args" entry point. (reported by Brian Downing)
1126 * tweaked disassembly notes to be less confident about proclaiming
1127 some instruction as an LRA. (thanks to Brian Downing)
1128 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1129 now each have their own history, command character, and other
1130 characteristics. (thanks to David Lichteblau)
1131 * fixed some bugs revealed by Paul Dietz' test suite:
1132 ** compiler failure in compiling LOGAND expressions including a
1134 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1135 shift greater than 32.
1136 ** FLUSH-DEST did not mark blocks for type check regeneration.
1137 ** HANDLER-CASE failed to accept declarations in handler clauses
1138 in some circumstances.
1140 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1141 * New code in contrib/sb-introspect (still probably not entirely
1142 stable yet) provides some support for smart Lisp development
1143 environments like SLIME.
1144 * The conditions signalled for errors occurring when loading .fasl
1145 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1146 in a way which should help ASDF recover gracefully.
1147 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1148 stuff has been cleaned up. If you code contrib/ stuff, this might
1149 affect you, and you can look at contrib/README, contrib/STANDARDS,
1150 and/or the 0.8.4.27 diff to check.
1151 * In full calls the compiler now does not generate checks for declared
1152 argument types for all arguments.
1153 * various threading fixes
1154 ** and some experimental patches which didn't make it into
1155 the main tree for this release, but which are shipped in
1156 contrib/experimental-thread.patch as a possible fix for some
1157 failures (deadlock, spinning...) in GC-intensive multithreaded
1159 * fixed PPC build problem (source code incompatibility of different
1160 library versions): added offsetof() hackery which attempts to divine
1161 where glibc maintainers put uc_mcontext today
1162 * fixed bug 282: compiler does not trust type assertions while passing
1163 arguments to a full call.
1164 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1165 &OPTIONAL VALUES type parameter.
1166 * fix bug 214: algorithm for noting rejected templates is now more
1167 similar to that of template seletion. (also reported by rydis on
1169 * fixed bug 141b: printing backquoted information readably and prettily
1170 inserts a space where necessary.
1171 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1172 (CHARACTER 10) are now reported as errors, rather than propagated
1173 as unknown types. (reported by piso on #lisp)
1174 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1175 with values NIL and :ERROR. (thanks to Milan Zamazal)
1176 * fixed bug 191c: CLOS now does proper keyword argument checking as
1177 described in CLHS 7.6.5 and 7.6.5.1.
1178 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1179 longer raise an error; further, using a list as a for-as-arithmetic
1180 counter now raises a meaningful error.
1181 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1182 now understood by sequence creation functions such as MAKE-SEQUENCE
1184 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1185 a STRING-INPUT-STREAM.
1186 * compiler enhancement: SIGNUM is now better able to derive the type
1188 * type declarations inside WITH-SLOTS are checked. (reported by
1190 * fixed some bugs revealed by Paul Dietz' test suite:
1191 ** incorrect optimization of TRUNCATE for a positive first
1192 argument and negative second.
1193 ** compiler failure in let-convertion during flushing dead code.
1194 ** compiler failure while deriving type of TRUNCATE on an
1195 interval, containing 0.
1196 ** ASH of a negative bignum by a negative bignum count now returns
1198 ** intersection of CONS types now canonicalizes properly, fixing
1199 inconsistencies in SUBTYPEP.
1201 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1202 * incompatible change: The --disable-debugger command line
1203 option now clobbers the debugger at a more fundamental
1204 level, by redefining #'INVOKE-DEBUGGER instead of by
1205 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1206 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1207 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1208 enter the debugger REPL and then suffer recursive errors
1209 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1210 while under the new system, BREAK in --disable-debugger mode
1211 terminates the system just as an unhandled error would.
1212 * fixed compiler performance when processing loops with a step >1;
1213 * bug fix: DOCUMENTATION now retrieves generic function
1214 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1215 support has been systematized, and now supports the methods
1216 specified by ANSI, along with a default method and a method for
1217 slot documentation. (reported by Nathan Froyd)
1218 * bug fix: effective methods associated with a generic function are
1219 no longer cached over a change of that generic function's method
1220 combination. (reported by Andreas Fuchs)
1221 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1222 names a non-existent directory. (thanks to Andreas Fuchs)
1223 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1224 the PPC platform, lead to stack corruption; this has been fixed.
1225 (reported by Rainer Joswig)
1226 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1227 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1228 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1229 arithmetic in general, is now much more reliable on the PPC
1231 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1232 platform now returns the right answer.
1233 * optimization: restored some effective method precomputation in
1234 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1235 precomputation is now tunable.
1236 * optimization: compiler-internal data structure use has been
1237 reviewed, and changes have been made that should improve the
1238 performance of the compiler by about 20%.
1239 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1240 simple-base-strings and simple-bit-vectors is improved.
1241 * optimization: the optimization of 32-bit logical and arithmetic
1242 functions introduced in version 0.8.3 on the x86 has been
1243 implemented on the mips, ppc and sparc platforms; an
1244 implementation of the same facility, but for 64-bit arithmetic,
1245 has been added for the alpha.
1246 * microoptimization: the compiler is better able to make use of the
1247 x86 LEA instruction for multiplication by constants.
1248 * bug fix: in some situations compiler did not report usage of
1249 generic arithmetic in (SPEED 3) policy.
1250 * bug 145b fix: compiler used wrong type specifier while converting
1251 MEMBER-types to numeric.
1252 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1254 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1255 index argument now works properly on the Alpha platform.
1256 * bug fix: floating point exception treatment on the Alpha platform
1258 * bug fix: FILE-POSITION works much better on string input and
1259 output streams. (thanks to Nikodemus Siivola)
1260 * bug fix: many threading/garbage collection symptoms sorted.
1261 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1262 might be pseudo-atomic.
1263 * internal change: Stopping for GC is now done with signals not
1264 ptrace. GC is now done in whichever thread wanted it, instead of
1266 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1268 * bug fix: result form in DO is not contained in the implicit
1270 * incompatible change: ICR structure is changed; the value part of
1271 CONTINUATION is now called LVAR; corresponding functions are
1272 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1273 * added type deriver for ISQRT (thanks to Robert E. Brown).
1274 * bug fix: better support for loading from the command line when an
1275 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1277 * fixed some bugs revealed by Paul Dietz' test suite:
1278 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1279 ** ROUND and FROUND now give the right answer when given very
1280 small float arguments.
1281 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1283 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1284 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1285 ** LCM should return a non-negative integer.
1286 ** PARSE-INTEGER returned the index of a terminator instead of the
1287 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1288 ** PARSE-INTEGER returned an incorrect index being applied to a
1290 ** LCM with two arguments of 0 returns 0 rather than signalling
1292 ** unsigned addition of a 32-bit constant with the high bit set no
1293 longer causes an internal compiler error on the x86.
1294 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1295 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1296 with negative last argument.
1297 ** byte specifiers with zero size and position no longer cause
1298 an error during type derivation.
1299 ** bignum multiplication on the Alpha platform now returns the
1301 * porting: The system now builds on SuSE AMD64, although it still
1302 generates a 32-bit binary.
1303 * .fasl file incompatibility: The fasl file version number has
1304 been incremented (because of the changes to internal compiler
1305 data structures referred to above).
1307 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1308 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1309 more accurately, on the Darwin kernel running on PowerPC hardware.
1310 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1311 * Compiler code deletion notes now signal a condition of type
1312 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1313 an associated MUFFLE-WARNING restart.
1314 * The compiler now performs limited argument count validation of
1315 constant format strings in FORMAT, and where appropriate in ERROR,
1316 CERROR and WARN. (thanks to Gerd Moellmann)
1317 * New ASDF-INSTALL contrib can be used for automatic download and
1318 installation of third-party Lisp code from CCLAN or other sites
1320 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1321 which forces another thread to execute a function supplied by the
1323 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1324 now accept and act upon their :ELEMENT-TYPE keyword argument.
1325 (reported by Martin Atzmueller, Edi Weitz)
1326 * bug fix: FILE-POSITION now accepts position designators up to
1327 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1328 is the greater. (thanks to Patrik Nordebo)
1329 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1330 Kalle Olavi Niemitalo)
1331 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1332 printable. (reported by Eric Marsden)
1333 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1334 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1335 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1336 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1337 (thanks to Stig Sandoe)
1338 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1339 UNREAD-CHAR now correctly includes the unread character in the
1340 target sequence. (thanks to Gerd Moellmann)
1341 * bug fix in threaded builds: the system can now be suspended and
1342 resumed by shell job control with minimal disruption.
1343 * bug fixes in times and timezones >2038AD
1344 * better handling of "where is GNU make?" problem in build scripts
1345 (thanks to Nikodemus Siivola)
1346 * new optimization: inside a named function any reference to a
1347 function with the same name is considered to be a self-reference;
1348 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1349 optimization quality.
1350 * new optimization on x86: logical functions and + now have
1351 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1352 used when the result is truncated to 32 bits.
1353 * VALUES declaration is partially enabled.
1354 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1355 * fixed some bugs revealed by Paul Dietz' test suite:
1356 ** The system now obeys the constraint imposed by
1357 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1358 types form a lattice under type intersection.
1359 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1360 ** ASSOC now ignores NIL elements in an alist.
1361 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1362 and (1+ MOST-POSITIVE-FIXNUM) answers.
1363 ** The addition of a method with invalid qualifiers to a generic
1364 function does not cause an error to be signalled immediately;
1365 a warning is signalled, and the error is generated only on
1366 calling the generic function.
1367 * changed .fasl file version number, in part to add type codes for
1368 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1369 obscure ANSI requirements
1371 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1372 * fixed bug 148: failure to inline-expand a local function left
1373 garbage, confusing the compiler.
1374 * fixed bugs 3cd: structure slot readers perform type check if the
1375 slot can have an invalid value (i.e. it is either not initialized
1376 or can be written with a less specific slot writer).
1377 * bug fix: the compiler now traps array references to elements off
1378 the end of an array; previously, the bounds checking in some
1379 circumstances could go off-by-one.
1380 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1382 * type declarations for array element types now obey the description
1383 on the CLHS page "Declaration TYPE", as per discussions on
1384 sbcl-help around 2003-05-08. This means that a declaration
1385 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1386 declaration, all references to BAR will be asserted or assumed
1387 (with THE, so dependent on compiler policy) to involve objects of
1388 type FOO. Note that no such declaration is implied in
1389 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1390 * declared types of functions from the "Conditions"
1391 chapter. (reported by Paul Dietz)
1392 * bug fix: CERROR accepts a function as its first argument.
1393 * bug fix: NTH an NTHCDR accept a bignum as index
1394 arguments. (reported by Adam Warner)
1395 * optimization: character compare routines now optimize comparing
1396 against a constant character. (reported by Gilbert Baumann)
1397 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1398 argument now works properly on the MIPS platform.
1399 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1400 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1401 declared to have a functional type, does not cause an error
1403 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1404 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1405 * fixed bug: if last continuation of a deleted block has a
1406 destination, this destination should be deleted too. (reported by
1408 * fixed a bug in the bootstrap process: the host compiler's values
1409 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1410 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1411 test case from Patrik Nordebo)
1412 * improved the ability of the disassembler on the PPC platform to
1413 provide helpful disassembly notes.
1414 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1415 the class in more cases than previously.
1416 * bug fix: FILE-POSITION now understands :START and :END for
1417 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1418 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1419 8) data. (thanks to David Lichteblau)
1420 * bug fix: it is possible to add a method to a generic function
1421 without lambda list.
1422 * bug fix: reader failed to signal END-OF-FILE inside an
1423 object representation. (reported by Nikodemus Siivola)
1424 * fixed some bugs revealed by Paul Dietz' test suite:
1425 ** LAST and [N]BUTLAST should accept a bignum.
1426 ** condition slot accessors are methods.
1427 ** (VECTOR NIL) is a subtype of STRING.
1429 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1430 * minor incompatible change: some nonsensical specialized lambda
1431 lists (used in DEFMETHOD) which were previously ignored now signal
1433 * minor incompatible change: the system is now aware of the types of
1434 variables in the COMMON-LISP package, and will signal errors for
1435 most violations of these type constraints (where previously they
1436 were silently accepted).
1437 * minor incompatible change: COMPILE-FILE now uses the freedom
1438 afforded (ANSI 3.2.2.3) to use derived function types for
1439 functions defined in the same file. This also permits the system
1440 to warn on static type mismatches and function
1441 redefinition. (Currently it does not work with high DEBUG level.)
1442 * minor incompatible change: VALUES declaration is disabled.
1443 * When issuing notes, the compiler now signals a condition of type
1444 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1445 restart for use in user handlers. It is expected that the
1446 COMPILER-NOTE condition will eventually become a condition
1447 supertype to a hierarchy of note types, which will then be
1448 handleable in a similar fashion. However, at the moment, no such
1449 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1450 but it's an implementation detail, not a classification for the
1452 * Changes in type checking closed the following bugs:
1453 ** type checking of unused values (192b, 194d, 203);
1454 ** template selection based on unsafe type assertions (192c, 236);
1455 ** type checking in branches (194bc).
1456 * A short form of VALUES type specifier has ANSI meaning (it has
1457 increased the number of situations when SBCL cannot perform type
1459 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1460 keywords or constants is permissible.
1461 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1462 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1463 classes. (thanks to Antonio Martinez)
1464 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1465 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1466 argument to a pretty-print stream. Instead, it leaves any such
1467 operations to DESCRIBE-OBJECT methods.
1468 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1470 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1471 lambda lists. (thanks to David Lichteblau)
1472 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1473 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1474 (thanks to Gerd Moellmann)
1475 * fixed bug 246: increased compilation speed of long
1476 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1478 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1480 * DEFCONSTANT now throws a condition of type
1481 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1482 constant to a non-EQL value; CONTINUE and ABORT restarts
1483 respectively change and preserve the value.
1484 * fixed bug 63: The code walker, part of the implementation of CLOS,
1485 is now better at handling symbol macros.
1486 * bug fix: There is no longer an internal implementation type named
1487 CL:LENGTH. (reported by Raymond Toy)
1488 * bug fix: In macro-like defining macros/special operators the
1489 implicit block does not enclose the lambda list.
1490 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1491 enclosed in parentheses) are not suitable as type specifiers, and
1492 their use properly signals an error now.
1493 * bug fix: An argument count mismatch for a type specifier in code
1494 being compiled no longer causes an unhandled error at compile
1495 time, but signals a compile-time warning.
1496 * fixed simple vector readable printing
1497 * bug fix: DESCRIBE takes more care over whether the class
1498 precedence list slot of a class is bound before accessing it.
1499 (reported by Markus Krummenacker)
1500 * bug fix: FORMATTER can successfully compile pretty-printer format
1501 strings which use variants of the ~* directive inside.
1502 * bug fix: SEARCH now applies its TEST predicate to the elements of
1503 the arguments in the correct order. (thanks to Wolfhard Buss)
1504 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1505 in type inference. (thanks to Robert E. Brown)
1506 * bug fix: Reading in symbols with an explicit package name of ""
1507 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1508 * fixed some bugs revealed by Paul Dietz' test suite:
1509 ** NIL is now allowed as a structure slot name.
1510 ** Arbitrary numbers, not just REALs, are allowed in certain
1511 circumstances in LOOP for-as-arithmetic clauses.
1512 ** Multiple class redefinitions before slot access no longer
1513 causes a type error.
1514 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1515 association between the name and a class.
1516 ** Generic functions with non-standard method-combination and over
1517 five methods all of which return constants no longer return NIL
1518 after the first few invocations. (thanks to Gerd Moellmann)
1519 ** CALL-NEXT-METHOD with no arguments now passes the original
1520 values of the arguments, even in the presence of assignment.
1521 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1523 ** DISASSEMBLE works with closures and funcallable instances.
1524 ** ADD-METHOD now returns the generic function, not the new method.
1525 ** FIND-METHOD signals an error if the lengths of the specializers
1526 is incompatible with the generic function, even if the ERRORP
1528 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1529 which its argument is a member.
1530 ** DEFCLASS only redefines the class named by its class-name
1531 argument if that name is the proper name of the class;
1532 otherwise, it creates a new class.
1533 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1534 of the UNBOUND-SLOT condition to the name of the slot.
1535 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1536 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1537 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1538 SLOT-MAKUNBOUND in the specified fashion.
1540 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1541 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1542 cross-compilation host. As a consequence, we can now bootstrap our
1543 way up to SBCL starting with a bare gcc toolchain and human-readable
1544 source code (first the source to CLISP, then the source to SBCL).
1545 * A contributed module containing a partial implementation of the
1546 simple-streams interface has been included. (thanks to Rudi
1548 * A contributed module implementing the RFC1321 Message Digest
1549 Algorithm, known as MD5, has been included.
1550 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1551 no longer has any effect, as the code controlled by this feature
1552 has been deleted. (As far as we know, no-one has ever built using
1553 this feature, and its semantics were confused in any case).
1554 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1555 work on conditions (as required by the ANSI specification),
1556 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1557 expected behaviour on conditions. Users should note, however,
1558 that such behaviour is not required by the ANSI specification,
1559 and so use of this behaviour may render their code unportable.
1560 * fixed some bugs revealed by Paul Dietz' test suite:
1561 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1563 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1564 ** SXHASH on condition objects no longer returns NIL.
1565 ** :ALLOCATION :CLASS slots are better treated; their values are
1566 updated on class redefinition, and initforms inherited from
1567 superclasses are applied.
1568 ** REMOVE-METHOD returns its generic function argument even when
1569 no method was removed.
1570 ** SHARED-INITIALIZE now initializes the values of the requested
1571 slots, including those with :ALLOCATION :CLASS.
1572 ** ALLOCATE-INSTANCE now works on structure classes defined via
1573 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1575 ** SLOT-EXISTS-P now works on conditions, as well as structures
1577 ** MAKE-LOAD-FORM now has the required methods on
1578 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1579 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1580 keyword, and now implements the SLOT-NAMES argument.
1581 ** methods with &OPTIONAL arguments no longer allow too many
1582 arguments to be passed in the call without error.
1583 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1584 option is consistent with the required arguments of the generic
1585 function lambda list.
1586 * bug fix: REQUIRE accepts a string designator. (Thanks to
1588 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1589 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1590 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1591 * bug fix: The compiler now checks for duplicated variables in macro
1593 * bug fix: SETQ on globals returns the correct value.
1594 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1595 not a COMPILER-ERROR (followed by some other strange error on
1596 choosing the CONTINUE restart).
1597 * bug fix: make.sh and friends are now more consistent in the way that
1598 they look for GNU "make".
1600 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1601 * experimental native threads support (on x86 Linux >=2.4 only).
1602 This is not compiled in by default: you need to add :SB-THREAD to
1603 the target features. See the "Beyond ANSI" chapter of the manual
1605 * fix for longstanding nonANSIism: The old distinction between
1606 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1607 The return value from CL:FIND-CLASS is now a CLOS class, and
1608 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1609 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1610 * An interface to the MetaObject Protocol, as described in Kiczales,
1611 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1612 MIT Press, 1991, is available from the SB-MOP package.
1613 * incompatible change: the SB-PCL package should now be considered
1614 a private implementation detail, and no longer a semi-private MOP
1616 * minor incompatible change: due to rearrangement for threads, the
1617 control stack and binding stack are now allocated at arbitrary
1618 addresses instead of being hardcoded per-port. Users affected by
1619 this probably have to be doing advanced things with shared
1620 libraries, and will know who they are.
1621 * minor incompatible change: Previously, all --eval forms used were
1622 processed with READ before any of them were processed with EVAL.
1623 Now each --eval form is processed with both READ and EVAL before
1624 the next --eval form is processed. (Thus package operations like
1625 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1626 work as the user might reasonably expect.)
1627 * minor incompatible change: *STANDARD-INPUT* is now only an
1628 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1630 * minor incompatible change: Y-OR-N-P is now character-oriented,
1631 not line oriented. Also, YES-OR-NO-P now works without errors.
1632 (thanks to Antonio Martinez)
1633 * sb-aclrepl module improvements: an integrated inspector, added
1634 repl features, and a bug fix to :trace command.
1635 * Known functions, which cannot be open coded by the backend, are
1636 considered to be able to check types of their arguments. (fixing
1637 a bug report by Nathan J. Froyd)
1638 * fixed a bug in computing method discriminating functions: It is
1639 now possible to define methods specialized on classes which have
1640 forward-referenced superclasses. (thanks to Gerd Moellmann)
1641 * fixed evaluation order in optional entries (reported by Gilbert
1643 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1644 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1645 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1646 required-by-AMOP NAME argument, as well as CLASS and
1647 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1648 * fixed bug 20: DEFMETHOD can define methods using names that are
1649 not the proper names of classes to designate class specializers.
1650 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1651 Unix file descriptors, instead of blowing up. (thanks to Antonio
1653 * Garbage collection refactoring: user-visible change is that a
1654 call to the GC function during WITHOUT-GCING will not do garbage
1655 collection until the end of the WITHOUT-GCING. If you were doing
1656 this you were probably losing anyway.
1657 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1658 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1659 TYPEP the latter but not the former.
1660 * The compiler issues a full WARNING for calls to undefined functions
1661 with names from the CL package.
1662 * MAP-INTO for a vector destination is open coded. (reported by
1663 Brian Downing on c.l.l)
1664 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1665 documentation string.
1666 * fixed some bugs revealed by Paul Dietz' test suite:
1667 ** COPY-ALIST now signals an error if its argument is a dotted
1669 ** Condition slots are now accessed more correctly in the presence
1670 of multiple initargs for a given slot.
1671 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1672 exclude restarts of the same name associated with a different
1674 ** DEFCLASS of forward-referenced classes with another
1675 forward-referenced class in the superclasses list no longer
1677 ** Condition slots are now initialized once each, not multiple
1678 times. (thanks to Gerd Moellmann)
1679 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1681 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1682 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1683 arguments contain duplicated elements.
1684 ** RESTART-CASE understands local macros.
1685 ** RESTART-CASE associates exactly its own restarts with a condition.
1686 ** ENDP in safe mode checks its argument to be of type LIST.
1687 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1688 in question is unbound.
1689 ** Optimized MAKE-INSTANCE functions no longer cause internal
1690 assertion failures in the presence of duplicate initargs.
1691 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1692 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1693 propagated even in the ctor optimized implementation of
1695 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1697 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1698 * a better implementation of SXHASH on (simple) bit vectors,
1699 measured both in execution speed and in distribution of results
1700 over the positive fixnums, has been installed. Likewise, a better
1701 implementation of EQUAL for simple bit vectors is now available.
1702 * fixed CEILING optimization for a divisor of form 2^k.
1703 * fixed bug 240 (emitting extra style warnings "using the lexical
1704 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1705 by Antonio Martinez)
1706 * fixed SXHASH, giving different results for NIL depending on type
1707 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1708 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1709 variable. (found by Rolf Wester)
1710 * SBCL does not ignore type declarations for special
1711 variables. (reported by rif on c.l.l 2003-03-05)
1712 * some bug fixes in contrib/sb-aclrepl/
1713 * fixed some bugs revealed by Paul Dietz' test suite:
1714 ** a bug in the CONS type specifier, whereby the CAR and CDR
1715 types got intertwined, has been fixed;
1716 ** the type system is now able to reason about the interaction
1717 between INTEGER and RATIO types more completely;
1718 ** APPEND, [N]REVERSE and NRECONC check that those their
1719 arguments, which must be proper lists, are really so;
1720 ** An array specialized to be unable to hold elements has been
1721 implemented, as required -- yes, really -- by ANSI;
1722 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1723 on malformed property lists;
1725 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1726 * incompatible packaging change: in line with Unix convention,
1727 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1728 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1729 * REQUIRE and PROVIDE are now optionally capable of doing something
1730 useful. See the documentation string for REQUIRE.
1731 * infrastructure for a managed SBCL contrib system: contributed
1732 modules in this release include:
1733 ** the ASDF system definition facility;
1734 ** an interface to the BSD Sockets API;
1735 ** an ACL-like convenience interface to the repl;
1736 (thanks to Kevin Rosenberg)
1737 ** an implementation of ROTATE-BYTE, with efficient implementation
1739 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1740 gives a closer approximation to the right answer than previously.
1741 (thanks to Raymond Toy)
1742 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1743 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1744 optional environment arguments, as required by ANSI.
1745 * fixed bugs in other functions taking environment objects, allowing
1746 calls with an explicit NIL environment argument to be compiled
1748 * fixed bug 228: primary return values from
1749 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1750 COMPILE or FUNCTION.
1751 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1752 :TYPE will no longer signal errors on innocuous objects.
1753 * fixed bug 231b: SETQ is better at respecting type declarations in
1754 the lexical environment.
1755 * fixed a bug in DEFCLASS: classes named by symbols with no or
1756 unprintable packages can now be defined.
1757 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1758 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1759 * fixed compiler failure related to checking types of functions.
1760 (reported by Robert E. Brown)
1761 * the compiler is now much more consistent in its error-checking
1762 treatment of bounding index arguments to sequence functions: in
1763 (SAFETY 3) code, errors will be signalled in almost all cases if
1764 invalid sequence bounding indices are passed to functions defined
1765 by ANSI to operate on sequences.
1766 * fixed a bug in the build procedure: documentation of SBCL-specific
1767 packages is now preserved and available in the final Lisp image.
1768 * lifted FDEFINITION lookup out of loops in the implementation of
1769 many list operations. (thanks to Robert E. Brown)
1770 * fixed a bug in the reader: the #n# reader macro now works for
1771 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1772 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1773 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1774 E. Brown and Tony Martinez respectively)
1775 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1776 SLOT-BOUNDP: in particular, it is now able to optimize them much
1777 better, and is now not vulnerable to having packages renamed.
1778 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1779 always called when a slot is not present in an instance. (thanks
1781 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1782 longer signal an error on unreading a character following EOF on
1783 the previous constituent stream. (thanks to Tony Martinez)
1784 * fixed some bugs revealed by Paul Dietz' test suite:
1785 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1786 not just nonnegative fixnums;
1787 ** the logical bit-array operators such as BIT-AND now accept an
1788 explicit NIL for their "opt-arg" argument (to indicate a
1789 freshly-consed result bit-array);
1790 ** ELT now signals an error on an invalid sequence index in safe
1792 ** the type system is now cleverer about negations of numeric
1793 types, and consequently understands the BIGNUM and RATIO types
1795 ** the type system is now cleverer about the interaction between
1796 INTEGER and RATIO types: while bugs still remain, many more
1797 cases are accurately computed;
1798 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1799 if it is in the last clause;
1800 ** CONSTANTLY now correctly returns a side-effect-free function in
1802 ** DECLARE is no longer treated as a special-operator; in
1803 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1804 * incremented fasl file version number due to the change in the
1805 DEFSTRUCT-SLOT-DESCRIPTION structure.
1807 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1808 * minor incompatible change: code processed by the "interpreter" or
1809 EVAL now has a compilation optimization policy of (DEBUG 2)
1810 (changed from (DEBUG 1)) to improve debuggability of interactive
1811 development, and to allow the use of the debug RETURN command in
1813 * an experimental implementation of the RETURN command for the
1814 debugger has been included. (thanks to Frederik Kuivinen)
1815 * fixed bug 62: constraints were not propagated into a loop.
1816 * fixed bug in embedded calls of SORT (reported and investigated by
1818 * fixed some bugs revealed by Paul Dietz' test suite:
1819 ** printing and reading of arrays with some dimensions having
1820 length 0 (thanks to Gerd Moellmann);
1821 ** BOA constructor with &AUX argument without a default value does
1822 not cause a type error;
1823 ** CONSTANTP now returns true for all self-evaluating objects.
1825 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1826 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1827 accessors that are related by inheritance, as specified in the
1828 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1829 Valtteri Vuorikoski)
1830 * The compiler is now able to inline functions that were defined in
1831 a complex lexical environment (e.g. inside a MACROLET).
1832 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1834 * Support for the upcoming FreeBSD-5.0 release has been included.
1835 (thanks to Dag-Erling Smorgrav)
1836 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1837 effect when it is not in a toplevel context.
1838 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1839 stand a better chance of being correct. (thanks to Gerd
1841 * fixed bug in COERCE, which now signals an error on coercing a
1842 rational to a bounded real type which excludes the expected
1844 * The compiler is now able to derive types more accurately from the
1845 COERCE and COMPILE functions.
1846 * fixed bug 223: functional binding is considered to be constant
1847 only for symbols in the CL package.
1848 * fixed bug 231: SETQ did not check the type of a variable being set
1849 (reported by Robert E. Brown)
1850 * A new optimization for MAKE-INSTANCE has been included, fixing
1851 various bugs (including relating to :ALLOCATION :CLASS slots and
1852 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1854 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1855 ** As required by ANSI, LOOP now disallows anonymous collection
1856 clauses such as COLLECT I in conjunction with aggregate boolean
1857 clauses such as THEREIS (= I 1);
1858 ** LOOP now signals an error when any variable is reused in the
1859 same loop (including the potentially useful construct analogous
1860 to WITH A = 1 WITH A = (1+ A);
1861 ** IT is only a special loop symbol within the first clause of a
1862 conditional loop clause;
1863 ** LOOP with a typed iteration variable over a hashtable now
1864 signals a type error iff it should.
1865 * fixed some other bugs revealed by Paul Dietz' test suite:
1866 ** FILE-STREAM now names the class previously known as FD-STREAM;
1867 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1868 argument) no longer signals an error;
1869 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1870 of the slot symbol, rather than using the current package
1871 ((:CONC-NAME "") continues to intern the slot's name in the
1873 * incremented fasl file version number, because of the incompatible
1874 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1875 of the new implementation of DEFINE-COMPILER-MACRO.
1877 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1878 * Support for building SBCL for MIPS platforms running in
1879 little-endian mode has now been checked in, and basic
1880 functionality on said platforms verified.
1881 * minor incompatible change: PCL now records the pathname of a file
1882 in which methods and the like are defined, rather than its
1884 * minor incompatible change: TRUENAME now considers the truename of
1885 a file naming a directory to be the pathname with :DIRECTORY
1886 component indicating that directory.
1887 * minor incompatible change: a NAMED clause in the extended form of
1888 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1889 reason for the previous behaviour is unclear.
1890 * more systematization and improvement of CLOS and MOP conformance
1891 in PCL (thanks to Gerd Moellman and Pierre Mai):
1892 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1894 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1895 primary methods with no specializers;
1896 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1898 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1899 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1900 FORWARD-REFERENCED-CLASSes; error reporting on
1901 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1903 ** SXHASH on CLOS instances now uses a slot internal to the
1904 instance to return different numbers on distinct instances,
1905 while preserving the same return value through invocations of
1907 ** DEFMETHOD signals errors when methods with longer incongruent
1908 lambda lists are added to generic functions;
1909 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1910 CLASS, as specified in AMOP;
1911 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1912 on the order returned by the primary method for classes of
1913 class STANDARD-CLASS;
1914 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1915 * fixed some bugs shown by Paul Dietz' test suite:
1916 ** DOLIST puts its body in TAGBODY;
1917 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1919 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1920 value producing form;
1921 ** if more variables are given to PROGV than values, extra
1922 variables are bound and made to have no value;
1923 ** NSUBSTITUTE on list arguments gets the right answer with
1925 ** ELT signals an error of type TYPE-ERROR when the index argument
1926 is not a valid sequence index;
1927 ** LOOP signals (at macroexpansion time) an error of type
1928 PROGRAM-ERROR when duplicate variable names are found;
1929 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1930 ** FUNCALL of special-operators now cause an error of type
1932 ** PSETQ now works as required in the presence of side-effecting
1933 symbol-macro places;
1934 ** NCONC accepts any object as its last argument;
1935 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1937 ** loop-for-as-package does not require a package to be explicitely
1939 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1941 * fixed bug 166: compiler preserves "there is a way to go"
1942 invariant when deleting code.
1943 * fixed bug 172: macro lambda lists with required arguments after
1944 &REST arguments now cause an error to be signalled. (thanks to
1946 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1947 bug. (thanks to Matthew Danish)
1948 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1950 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1951 arguments are lexically rebound. (thanks to Gerd Moellmann and
1953 * fixed bug 194: error messages are now more informative when there
1954 is no primary method applicable in a call to a generic
1955 function. (thanks to Gerd Moellmann)
1956 * fixed bug in command line argument checking (thanks to Julian
1958 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1959 (thanks to Matthew Danish)
1960 * incremented fasl file version number, because of the
1961 SXHASH-related changes in the layout of CLOS data structures
1963 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1964 * minor incompatible change: The runtime (the Unix executable named
1965 "sbcl") is now much pickier about the .core files it will load.
1966 Essentially it now requires .core files to descend from the same
1967 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1968 as the runtime does. (The intent is to prevent the crashes which
1969 can occur, and which can even be reported as mysterious failures,
1970 when people patch the sources or change the build parameters
1971 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1972 sbcl and .core files.)
1973 * fixed bug: VALUES-LIST is no longer optimized away.
1974 * fixed bug 142: The FFI conversion of C string values to Lisp
1975 string values no longer conses excessively. (thanks to Nathan
1976 Froyd porting Raymond Toy's fix to CMU CL)
1977 * began to systematize and improve MOP conformance in PCL (thanks to
1978 Nathan Froyd, Gerd Moellman and Pierre Mai):
1979 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1981 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1982 ** FINALIZE-INHERITANCE is now called on class finalization;
1983 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1984 argument precedence order.
1985 * fixed bug 202: The compiler no longer fails on functions whose
1986 derived types contradict their declared type.
1987 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1988 so it can be non-toplevel.
1989 * The fasl file version number has changed (because of the new
1990 implementation of DEFMACRO).
1991 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1992 safe code, that any length requirement by their type-specifier
1993 argument is valid. The exceptions to this are described in bug
1995 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1996 functions now check the types of their inputs as required by ANSI.
1997 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1998 introduced symbol is DECLAREd to be SPECIAL.
1999 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2000 * fixed bug: :COUNT argument to sequence functions may be negative
2001 * fixed bug: body of DO-SYMBOLS may contain declarations
2002 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2003 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2004 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2005 symbol macro only once
2006 * fixed printing of call frame when argument list is unavailable
2007 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2008 * compiler no longer signals WARNING on unknown keyword
2011 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2012 * A beta-quality port to the mips architecture running Linux,
2013 based on the old CMUCL backend, has been made. It has been tested
2014 on a big-endian kernel, and works sufficiently well to be able to
2015 rebuild itself; it has not been tested in little-endian mode.
2016 * fixed an inconsistency between gencgc.c and purify.c which made
2017 dumping/loading .core files unreliable
2018 * fixed bug 120a: The compiler now deals correctly with IFs where
2019 the consequent is the same as the alternative, instead of
2020 misderiving the return type. (thanks to Alexey Dejneka)
2021 * fixed bug 113: Logical pathnames are now dumpable (the logical
2022 host is resolved at load-time, throwing an error if it is not
2024 * fixed bug 174: FORMAT's error message is slightly clearer when a
2025 non-printing character is used in a format directive.
2026 * fixed several bugs in compiler checking of type declarations, i.e.
2027 violations of the Python "declarations are assertions" principle
2028 (thanks to Alexey Dejneka)
2029 * fixed several bugs in PCL's error checking (thanks to Gerd
2031 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2033 * fixed bug in compilation of functions as first class values
2034 (thanks to Antonio Martinez-Shotton)
2035 * The compiler's handling TYPE-ERRORs which it can prove will
2036 inevitably happen at runtime has been cleaned up and corrected
2037 in several ways. (thanks to Alexey Dejneka)
2038 * improved argument type checking for various basic arithmetic
2039 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2040 lavished on them in the past that they can be compiled in many
2041 ways in different special cases
2042 * fixed bug 181: compiler checks validity of user supplied type
2044 * cleaned up code flushing in optimization: Function calls which
2045 should signal errors for safety purposes (e.g. which ANSI says
2046 should signal errors when their arguments are of incorrect type)
2047 are no longer optimized away.
2048 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2049 * incremented fasl file version number, because changes in the
2050 implementation of sequence functions like COERCE caused
2051 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2052 in old inline expansions) to become undefined. (Actually these
2053 changes were later undone, so we might very well be binary
2054 compatible with 0.7.7 after all, but leaving the version number
2055 incremented seemed like the simplest and most conservative
2058 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2059 * An alpha-quality port to the parisc architecture running Linux,
2060 based on the old CMUCL backend, has been made. This, even more so
2061 than the other backends, should be considered still a work in
2062 progress; known problems include that the Linux kernel in 64-bit
2063 mode does not propagate the correct sigcontext structure to
2064 userspace, and consequently SBCL on a parisc64 kernel will not
2066 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2067 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2068 suppressed.) Also now that the compiler is looking at declarations
2069 in the environment, it checks optimization declarations as well,
2070 and suppresses inlining when (> DEBUG SPEED).
2071 * More fixes have been made to treatment of floating point exception
2072 treatment and other Unix signals. In particular, floating point
2073 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2074 * The detection and handling of control stack exhaustion (infinite
2075 or very deeply nested recursion) has changed. Stack exhaustion
2076 detection is now done by write-protecting pages at the OS level
2077 and applies at all optimization settings; when found, a
2078 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2079 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2080 be caught using IGNORE-ERRORS.
2081 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2082 correctly when transforming recursive LABELS functions to LETs.
2083 (thanks to Alexey Dejneka)
2084 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2085 that are names of constants or global variables.
2086 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2087 alien routines with docstrings.
2088 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2089 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2091 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2092 (thanks to Eric Marsden)
2093 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2094 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2095 * Bug fix: The SPARC backend can now compile functions involving
2096 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2097 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2098 to a functions expecting a PATHNAME-DESIGNATOR.
2099 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2100 lambda lists. (thanks to Alexey Dejneka)
2101 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2102 functions so that EXPORTing the name of the class doesn't cause
2103 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2104 to Antonio Martinez for reporting this)
2105 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2106 constant-folded without failing an assertion. (thanks to Einar
2107 Floystad Dorum for reporting this)
2108 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2109 types are now dealt with more correctly by the compiler.
2110 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2111 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2112 behaviour analogous to RENAME-FILE. This puts its behaviour more
2113 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2115 * The fasl file version number has changed again. (because of the
2116 bug fix involving the names of PCL MAKE-INSTANCE functions)
2118 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2119 * bug fix: Floating point exceptions are treated much more
2120 consistently on the x86/Linux and PPC/Linux platforms.
2121 * Array initialization with :INITIAL-ELEMENT is now much faster for
2122 cases when the compiler cannot open code the array creation, but
2123 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2124 array accesses have also seen a speed increase.
2125 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2126 specified. (This was at the root of some bad interactions between
2127 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2129 * bug fix: Internal error arguments for undefined functions are now
2130 computed correctly on the PPC/Linux platform.
2131 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2132 Raymond Toy's patch for CMU CL)
2133 * Support for the Solaris 9 operating environment has been included
2134 (thanks to Daniel Merritt)
2135 * A very ugly but hopefully complete draft of the missing FFI chapter
2136 of the manual has been created by reformatting the corresponding
2137 CMU CL manual chapter into (currently very ugly and incoherent)
2138 DocBook and bringing it up to date for SBCL behavior. Thus, the
2139 manual is now essentially complete, at least by my extreme
2140 once-and-only-once standards, whereby it's acceptable to refer to
2141 the doc strings of SB-EXT functions as the primary documentation.
2142 * The fasl file version number has changed again, due to cleanup of
2143 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2144 is no longer a static symbol.)
2146 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2147 * SBCL now builds with OpenMCL (version 0.12) as the
2148 cross-compilation host; also, more progress has been made toward
2149 bootstrapping under CLISP.
2150 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2152 * bug 158 fixed: The compiler can now deal with integer loop
2153 increments different from 1; fixing this turned out also to fix
2155 * bug 169 fixed: no more bogus warnings about using lexical bindings
2156 despite the presence of perfectly good SPECIAL declarations (thanks
2157 to David Lichteblau)
2158 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2159 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2160 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2162 * bug 180 fixed: Method combination specifications no longer ignore
2163 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2164 * bug fix: Structure type predicate functions now check their argument
2165 count as they should.
2166 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2167 correctly. (thanks to Pierre Mai)
2168 * minor incompatible change: The --noprogrammer option is deprecated
2169 in favor of the new --disable-debugger option, which is very similar.
2170 (The major difference is that it takes effect at a slightly different
2171 time at startup, causing handling of errors in --sysinit and
2172 --userinit files will be affected differently.) The
2173 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2174 been added to allow this functionality to be controlled from ordinary
2175 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2176 maintainers, who might want to run non-interactive scripts to
2177 build SBCL cores which will later be used interactively.)
2178 * minor incompatible change: The LOAD function no longer, when given
2179 a wild pathname to load, loads all files matching that pathname.
2180 Instead, an error of type FILE-ERROR is signalled.
2182 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2183 * bug 147 fixed: The compiler preserves its block link/count
2184 invariants more correctly now so that it doesn't crash. (thanks
2186 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2188 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2189 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2190 no-value-for-this-character as NIL (as specified by ANSI).
2191 * HOST-NAMESTRING on physical pathnames now returns a string that is
2192 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2193 (thanks to Christophe Rhodes)
2194 * The Alpha port handles icache flushing more correctly. (thanks to
2196 * More progress has been made toward bootstrapping under CLISP. (thanks
2197 to Christophe Rhodes)
2198 * The fasl file format has changed again, because dynamic loading
2199 on OpenBSD (which has non-ELF object files) motivated some cleanups
2200 in the way that foreign symbols are transformed and passed around.
2201 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2202 is no longer treated as whitespace by the reader, but instead as
2203 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
2204 |A
\x7fB|, instead of A as it used to.
2206 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2207 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2208 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2209 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2210 this as of 0.7.1.45, but was left out of the previous news section
2211 (thanks to Dan Barlow)
2212 * SBCL now runs on the Solaris operating system on SPARC architectures
2213 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2214 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2215 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2216 * SPARC backend cleanups, allowing builds of cores optimized for V8
2217 and V9 SPARCS, and also emission of code targeted to a particular
2218 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2220 * SBCL is closer to bootstrapping under CLISP, thanks to various
2221 fixes by Christophe Rhodes.
2222 * The fasl file format has changed again, to allow the compiler's
2223 INFO database to support symbol macros.
2224 * The user manual (in doc/) is formatted into HTML more nicely.
2225 (thanks to coreythomas)
2226 * The system is smarter about SUBTYPEP relationships, especially
2227 those involving NOT types (including types such as ATOM which are
2228 represented internally using NOT types). Thus SUBTYPEP is less
2229 likely to return (VALUES NIL NIL) in general, and in particular
2230 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2232 * The fasl file format has changed again, because the internal
2233 representation of types now includes a new slot to support the new
2234 SUBTYPEP-of-NOT-types logic.
2235 * (not a change in the main branch of SBCL, but a related prototype
2236 which can hopefully be merged into the main branch of SBCL in the
2237 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2238 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2239 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2240 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2241 dispatch macro character combinations. (thanks to Alexey Dejneka)
2242 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2243 that superficially look like logical namestrings correctly.
2244 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2246 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2247 default supplied-p) for &optional and &key arguments. (thanks to
2250 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2251 * incompatible change: The compiler is now less aggressive about
2252 tail call optimization, doing it only when (> SPACE DEBUG) or
2253 (> SPEED DEBUG). (This is an incompatible change because there are
2254 programs which relied on the old CMU-CL-style behavior to optimize
2255 away their unbounded recursion which will now die of stack overflow.)
2256 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2257 for non-GENCGC systems has been increased to 20M (since that
2258 seems much closer to the likely performance optimum for modern
2259 systems than the old 4M value was)
2260 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2261 and *DEBUG-PRINT-LEVEL*
2262 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2263 of CMU CL's support for SPARC, and various endianness and other
2264 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2265 * new syntactic sugar for the Unix command line: --load foo.bar is now
2266 an alternate notation for --eval '(load "foo.bar")'.
2268 ** The system now detects stack overflow and handles it gracefully,
2269 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2270 optimization settings. (This is a good thing in general, and
2271 its introduction in this version should be particularly timely
2272 for anyone whose code fails because of suppression of tail
2274 ** The system now hunts for the C variable "environ" in a more
2275 devious way, to avoid segfaults when the C library version differs
2276 between compile time and run time. (thanks to Christophe Rhodes)
2277 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2278 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2279 ** The compiler no longer issues bogus style warnings for undefined
2280 classes in the same source file as the DEFCLASSes which defined
2281 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2283 ** fixes in CONDITION class precedence list for undefined function
2284 errors (thanks to Alexei Dejneka)
2285 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2286 correctly. (thanks to Dan Barlow)
2287 ** portability fixes aiming at bootstrapping under CLISP (thanks
2288 to Dave McDonald and Christophe Rhodes)
2289 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2290 ** fixes in type translation and and type inference (thanks to
2292 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2293 ** various fixes in the new ports (thanks to Dan Barlow)
2294 * several changes related to debugging:
2295 ** suppression of tail recursion, as noted above
2296 ** stack overflow detection, as noted above
2297 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2298 is now the default. (For some time encapsulation has been more
2299 reliable than the breakpoint-based :ENCAPSULATE NIL
2300 implementation, at least on X86 systems; and I just noticed that
2301 encapsulation also seems closer to the spirit of the ANSI
2304 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2306 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2307 up properly again. (There was a packaging bug in 0.7.0 which
2308 left their definitions in SB-SYS::LOAD-FOREIGN and
2309 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2310 vital for most things which interface to C-level interfaces,
2311 like extensions working with sockets or databases or
2312 Perl-compatible regexes or whatever, and the need to fix
2313 this bug is the main reason that 0.7.1 was released so
2315 ** DEFGENERIC is now choosier about the methods it redefines, so that
2316 reLOADing a previously-LOADed file containing DEFGENERICs does
2317 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2318 cycle now works as it should. (thanks to Alexey Dejneka)
2319 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2320 fixed by Christophe Rhodes. (He actually submitted this patch
2321 months ago, and I delayed until after 0.7.0.)
2322 ** Bug 111 (internal compiler confusion about runtime checks on
2323 FUNCTION types) was fixed by Alexey Dejneka.
2324 * Some internal cleanups (getting rid of variables which aren't
2325 needed now that the byte interpreter is gone) caused the fasl
2326 file format number to change again.
2328 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2329 * major incompatible change: The default fasl file extension, i.e. the
2330 default extension for files produced by COMPILE-FILE, has changed
2331 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2333 ** There are many changes in the implementation of the compiler.
2334 SBCL is now essentially a compiler-only implementation of ANSI
2335 Common Lisp. EVAL still "interprets" a few special cases, but
2336 almost all the interesting cases are handled by creating
2337 a LAMBDA expression, calling COMPILE on it, then calling
2338 FUNCALL on the result.
2339 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2340 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2341 Since the code is newer, there might still be some new bugs
2342 (though not as many as before Martin Atzmueller's fixes:-). But
2343 the new code is substantially simpler and clearer, and hopefully
2344 any remaining bugs will be simpler, less fundamental, and more
2345 fixable then the bugs in the old code.
2346 ** The revised compiler is still a little unsteady on its feet.
2348 *** The debugging information it produces (particularly the names
2349 of FUNCTION objects) is sometimes much less useful than what
2350 the old compiler produced.
2351 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2352 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2353 within a MACROLET) has been temporarily weakened.
2354 ** There are new compiler optimizations for various functions:
2355 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2356 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2357 *** the math functions TRUNCATE, FLOOR, and CEILING
2358 *** the function-of-all-trades COERCE
2359 Mostly these should be transparent, but there's one
2360 potentially-annoying problem (bug 117): when the compiler
2361 inline-expands a function and does type analysis on the result,
2362 it can create control paths which have type mismatches, and
2363 when it can't prove that those control paths aren't taken,
2364 it will issue WARNINGs about the type mismatches. This is
2365 a particular problem in practice for the new sequence functions.
2366 It's not clear how this should be fixed, and for now, a
2367 workaround is given in the entry for 117 in the BUGS file.
2368 ** (Because of the interaction between the two previous items --
2369 occasional inlining problems and new inline expansions -- some
2370 of the new sequence function optimizations won't really kick in
2371 completely until debugging information, and then inlining, are
2372 straightened out in some future version.)
2373 * minor incompatible changes:
2374 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2375 DIRECTORY no longer implicitly promotes NIL slots of its
2376 pathname argument to :WILD. In particular, when you ask for the
2377 contents of a directory (which you used to be able to do without
2378 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2379 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2380 ** changes in behavior that ANSI explicitly defines to be
2381 implementation dependent:
2382 *** The new compiler-only implementation still conforms with ANSI,
2383 but acts a little different than before. Besides the obvious
2384 changes in performance tradeoffs (that the cost per form passed
2385 to EVAL has gone up, and the cost per form executed by EVAL
2386 has gone down), the behavior of the system changes a little
2387 because there are no longer any interpreted function objects.
2388 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2389 e.g. doing COMPILE on the output of interactive DEFUN is
2391 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2393 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2394 has changed from (:CL) to NIL.
2395 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2396 e.g. #\Newline and #\Tab, don't have names specified in the
2397 ANSI Common Lisp standard, is now based on their ASCII symbolic
2398 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2399 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2400 are no longer used for output.
2401 ** changes in internal implementation constants:
2402 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2403 4 million. (If your application spends a lot of time GCing and
2404 you have a lot of RAM, you might want to experiment with
2405 increasing it even more.)
2406 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2407 However, almost all old code should still continue to work without
2408 immediate update, as SB-C-CALL is now a (deprecated) nickname
2410 ** Old operator names in the style DEF-FOO are now deprecated in
2411 favor of new corresponding names DEFINE-FOO, for consistency with
2412 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2413 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2414 internal symbols, but a few supported extensions like
2415 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2416 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2417 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2418 etc. as you get deeper into recursive calls to the debugger
2419 command loop, instead of the old "5]", "5]]", "5]]]"
2420 sequence. (I was motivated to do this when squabbles between
2421 ILISP and SBCL left me very deeply nested in the debugger. In the
2422 short term, this change will probably provoke more ILISP/SBCL
2423 squabbles, but hopefully it will be an improvement in the long run.)
2424 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2425 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2426 function, since declaiming return types involving aliens is
2427 (1) annoyingly messy to do by hand and (2) vital to efficient
2428 compilation of code which calls such functions.
2429 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2430 longer reexported by the SB-EXT package. They're solely useful
2431 for alien code, so it seems more logical that you should get
2432 them from the SB-ALIEN package, not in SB-EXT.
2433 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2434 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2435 features. Instead, the code that they used to control is always
2436 built into the system.
2437 * many other bug fixes
2438 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2439 advantage of the new EVAL-WHEN stuff and to clean them up in
2440 general, and they are now more ANSI-compliant in a number of
2441 ways. Martin Atzmueller is responsible for a lot of this.
2442 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2444 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2445 *** correct ERROR type for various file operations
2446 *** some fixes for Lisp streams
2447 *** DEFMETHOD syntax checking
2448 *** changing old weird representation of debug information as
2449 strings (which, among their other deficiencies, don't transform
2450 correctly when you rename packages, and don't change their
2451 print representation when you change things like *PACKAGE*
2452 and *PRINT-LENGTH*) to symbols and lists of symbols
2453 He also made several improvements and fixed several bugs in DESCRIBE.
2454 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2456 *** misbehavior of WRITE-STRING/WRITE-LINE
2457 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2458 and several other LOOP problems as well
2459 *** DIRECTORY when similar filenames are present
2460 *** DEFGENERIC with :METHOD options
2461 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2462 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2463 *** argument ordering in FIND with :TEST option
2464 *** mishandled package designator argument in APROPOS-LIST
2465 *** various problems in the backquote readmacro
2466 *** a bug in APROPOS
2467 *** probably some others that I'm not describing very well here,
2468 since the CVS log documents them by reference to sbcl-devel
2469 messages, and the SourceForge archives aren't working well.:-(
2470 ** Dan Barlow improved the Alpha port (and is making progress on the
2471 PPC port, for those of you who think different).
2472 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2473 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2474 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2475 continues to work on the SPARC port (for those of you in a position
2476 to look down upon our little PC-compatible boxes from a great height).
2477 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2478 and uses that copy, rather than the current dynamic value, when
2479 it's trying to decide whether to truncate output. Thus e.g.
2480 (let ((*print-lines* 50))
2481 (pprint-logical-block (stream nil)
2483 (let ((*print-lines* 8))
2484 (print (aref possiblybigthings i) stream)))))
2485 should now truncate the logical block only at 50 lines, instead of
2486 often truncating it at 8 lines, as it did before.
2487 * The doc/cmucl/ directory, containing old CMU CL documentation
2488 from the time of the fork, is no longer part of the base system.
2489 SourceForge has shut down its anonymous FTP service, and with it
2490 my original plan for distributing the old CMU CL documentation
2491 there. For now, if you need these files you can download an old
2492 SBCL source release and extract them from it.
2493 * The fasl file version number changed again, for dozens of reasons,
2494 some of which are apparent above.
2496 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2497 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2498 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2499 patch, so that DISASSEMBLE output is much nicer.
2500 * The code in the SB-PROFILE package now seems reasonably stable.
2501 I still haven't decided what the final interface should look like
2502 (I'd like PROFILE to interact cleanly with TRACE, since both
2503 facilities use function encapsulation) but if you have a need
2504 for profiling now, you can probably use it successfully with
2505 the current CMU-CL-style interface.
2506 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2507 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2508 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2509 link now returns the dangling link itself, and for similar
2510 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2511 link itself. (In these cases the old code signalled an error and
2512 looped endlessly, respectively.) Thus, DIRECTORY now works even
2513 in the presence of dangling and cyclic symbolic links.
2514 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2515 is now a supported extension again, since the consensus on
2516 sbcl-devel was that it can be useful for ordinary development
2517 work, not just for debugging SBCL itself.
2518 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2519 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2520 that currently-defined functions might be redefined later with
2521 different return types.
2522 * Hash tables can be printed readably, as inspired by CMU CL code
2523 of Eric Marsden and SBCL code of Martin Atzmueller.
2524 * better error handling in CLOS method combination, thanks to
2525 Martin Atzmueller porting Pierre Mai's CMU CL patches
2526 * more overflow fixes for >16Mbyte I/O buffers
2527 * A bug in READ has been fixed, so that now a single Ctrl-D
2528 character suffices to cause end-of-file on character streams.
2529 In particular, now you only need one Ctrl-D at the command
2530 line (not two) to exit SBCL.
2531 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2533 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2534 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2535 does the right thing.
2536 * The implementation of some type tests, especially for CONDITION
2537 types, is now tidier and maybe faster, due to CMU CL code
2538 originally by Douglas Crosher, ported by Martin Atzmueller.
2539 * Some math functions have been fixed, and there are new
2540 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2541 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2542 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2543 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2544 likely to be built into sbcl-0.7.0.)
2545 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2546 CL, ported to SBCL by Martin Atzmueller)
2547 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2548 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2549 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2550 * There's a new slam.sh hack to shorten the edit/compile/debug
2551 cycle for low-level changes to SBCL itself, and a new
2552 :SB-AFTER-XC-CORE target feature to control the generation of
2553 the after-xc.core file needed by slam.sh.
2554 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2555 COMPILE-FILE is no longer supported, so that now every function
2556 gets an entry point, so that block compilation looks a little
2557 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2558 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2559 returns the number of bytes consed since the system started,
2560 rather than the number consed since the first time the function
2561 was called. (The new definition parallels ANSI functions like
2562 CL:GET-INTERNAL-RUN-TIME.)
2563 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2564 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2565 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2566 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2567 * Fasl file version numbers are now independent of the target CPU,
2568 since historically most system changes which required version
2569 number changes have affected all CPUs equally. Similarly,
2570 the byte fasl file version is now equal to the ordinary
2573 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2574 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2575 is no longer recognized. I apologize for this, because it was
2576 listed in SB-EXT as a supported extension, but I found that
2577 its existing behavior was poorly specified, as well as incorrectly
2578 specified, and it looked like too much of a mess to straighten it
2579 out. I have enough on my hands trying to get ANSI stuff to work..
2580 * many patches ported from CMU CL by Martin Atzmueller, with
2581 half a dozen bug fixes in pretty-printing and the debugger, and
2582 half a dozen others elsewhere
2583 * fixed bug 13: Floating point infinities are now supported again.
2584 They might still be a little bit flaky, but thanks to bug reports
2585 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2586 as flaky as they were.
2587 * The --noprogrammer command line option is now supported. (Its
2588 behavior is slightly different in detail from what the old man
2589 page claimed it would do, but it's still appropriate under the
2590 same circumstances that the man page talks about.)
2591 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2592 are now supported, and enabled by default. Thus, the compiler can
2593 handle many floating point and complex operations much less
2594 inefficiently. (Thus e.g. you can implement a complex FFT
2596 * The compiler now detects type mismatches between DECLAIM FTYPE
2597 and DEFUN better, and implements CHECK-TYPE more correctly, and
2598 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2599 thanks to patches from Martin Atzmueller.
2600 * various fixes to make the cross-compiler more portable to
2601 ANSI-conforming-but-different cross-compilation hosts (notably
2602 Lispworks for Windows, following bug reports from Arthur Lemmens)
2603 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2604 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2605 fixed thanks to Pierre Mai's CMU CL patches.
2606 * fixes to keep the system from overflowing internal counters when
2607 it tries to use i/o buffers larger than 16M bytes
2608 * fixed bug 45a: Various internal functions required to support
2609 complex special functions have been merged from CMU CL sources.
2610 (When I was first setting up SBCL, I misunderstood a compile-time
2611 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2612 * improved support for type intersection and union, fixing bug 12
2613 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2614 more obscure bugs as well
2615 * some steps toward byte-compiling non-performance-critical
2616 parts of the system, courtesy of patches from Martin Atzmueller
2617 * Christophe Rhodes has made some debian packages of sbcl at
2618 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2619 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2620 stable, but are nonetheless usable. When he's ready, I'd be happy
2621 to add them to the SourceForge "File Releases" section. (And if
2622 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2623 * new fasl file format version number (because of changes in
2624 internal representation of (OR ..) types to accommodate the new
2625 support for (AND ..) types, among other things)
2627 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2628 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2630 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2632 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2633 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2634 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2635 * DESCRIBE now works on structure objects again.
2636 * Most function call argument type mismatches are now handled as
2637 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2638 know whether the function will be redefined before the call is
2639 executed. (The compiler could flag local calls with full WARNINGs,
2640 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2641 it doesn't keep track of enough information to know whether calls
2642 are local in this sense.)
2643 * Compiler output is now more verbose, with messages truncated
2644 later than before. (There should be some supported way for users
2645 to override the default verbosity, but I haven't decided how to
2646 provide it yet, so this behavior is still controlled by the internal
2647 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2648 src/compiler/ir1util.lisp.)
2649 * Fasl file format version numbers have increased again, because
2650 support for the Gray streams extension changes the layout of the
2651 system's STREAM objects.
2652 * The Gray subclassable streams extension now works, thanks to a
2653 patch from Martin Atzmueller.
2654 * The full LOAD-FOREIGN extension (not just the primitive
2655 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2656 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2657 but like most other programs, it defaults to copying the Unix
2658 environment from the original process instead of starting the
2659 new process in an empty environment.
2660 * Extensions which manipulate the Unix environment now support
2661 an :ENVIRONMENT keyword option which doesn't smash case or
2662 do other bad things. The CMU-CL-style :ENV option is retained
2663 for porting convenience.
2664 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2665 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2667 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2669 * A patch from Martin Atzmueller seems to have solved the SIGINT
2670 problem, and as far as we know, signal-handling now works cleanly.
2671 (If you find any new bugs, please report them!)
2672 * The system no longer defaults Lisp source file names to types
2673 ".l", ".cl", or ".lsp", but only to ".lisp".
2674 * The compiler no longer uses special default file extensions for
2675 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2676 seems to expect a single default extension for all compiled code,
2677 and there's no compelling reason to try to stretch the standard
2678 to allow two different extensions.) Instead, byte-compiled files
2679 default to the same extension as native-compiled files.
2680 * Fasl file format version numbers have increased again, because
2681 a rearrangement of internal implementation packages made some
2682 dumped symbols in old fasl files unreadable in new cores.
2683 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2684 many fewer weird special cases.
2685 * Bug #17 (differing COMPILE-FILE behavior between logical and
2686 physical pathnames) has been fixed, and some related misbehavior too,
2687 thanks to a patch from Martin Atzmueller.
2688 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2689 by Tim Moore, ported to SBCL by Martin Atzmueller.
2690 * Martin Atzmueller fixed several filesystem-related problems,
2691 including bug #36, in part by porting CMU CL patches, which were
2692 written in part by Paul Werkowski.
2693 * More compiler warnings in src/runtime/ are gone, thanks to
2694 more patches from Martin Atzmueller.
2695 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2698 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2700 * DESCRIBE now works on CONDITION objects.
2701 * The debugger now handles errors which arise when trying to print
2702 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2704 * The build system now uses an additional file, customize-target-features.lisp,
2705 to allow local modifications to the target *FEATURES* list. (The point of
2706 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2707 debugging features enabled, without having to worry about propagating it
2708 into everyone's system when I do a "cvs update".) When no
2709 customize-target-features.lisp file exists, the target *FEATURES* list
2710 should be constructed the same way as before.
2711 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2712 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2713 warn about behavior which is undefined under the ANSI spec. Note: This
2714 is specified by ANSI, but it's not very popular with programmers.
2715 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2716 macro in the SBCL sources for an example of a workaround which you
2717 might use to make portable ANSI-standard code which does what you want.
2718 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2719 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2720 when it's used as a non-toplevel form. (This is required in order
2721 to implement the DEFCONSTANT-EQX macro.)
2722 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2723 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2724 funny things, due to bugs in the implementation of EVAL-WHEN
2725 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2726 non-toplevel DEFCONSTANTs are uncommon.)
2727 * The core file version number and fasl file version number have been
2728 incremented, because the old noncompliant DEFCONSTANT behavior involved
2729 calling functions which no longer exist, and because I also took the
2730 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2731 * fixed bug 1 (error handling before read-eval-print loop starts), and
2732 redid debugger restarts and related debugger commands somewhat while
2734 ** The QUIT debugger command is gone, since it did something
2735 rather different than the SB-EXT:QUIT command, and since it never
2736 worked properly outside the main toplevel read/eval/print loop.
2737 Invoking the new TOPLEVEL restart provides the same functionality.
2738 ** The GO debugger command is also gone, since you can just invoke
2739 the CONTINUE restart directly instead.
2740 ** The TOP debugger command is also gone, since it's redundant with the
2741 FRAME 0 command, and since it interfered with abbreviations for the
2743 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2745 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2746 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2747 some porting work by Martin Atzmueller)
2748 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2749 with a boatload of patches which he ported to SBCL. Now that those
2751 ** The system tries to make sure that its low-priority messages
2752 are prefixed by semicolons, to help people who like to use
2753 syntax highlighting in their ILISP buffer. (This patch
2754 was originally due to Raymond Toy.)
2755 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2756 patches originally written by Raymond Toy.
2757 ** The compiler understands coercion between single-value and
2758 multiple-VALUES type expressions better, getting rid of some very
2759 weird behavior, thanks to patches originally by Robert MacLachlan
2760 and Douglas Crosher.
2761 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2762 lambda lists, thanks to a patch originally by Pierre Mai.
2763 ** The system no longer bogusly warns about "abbreviated type
2765 ** The compiler gets less confused by inlining and RETURN-FROM,
2766 thanks to some patches originally by Tim Moore.
2767 ** The system no longer hangs when dumping circular lists to fasl
2768 files, thanks to a patch originally from Douglas Crosher.
2769 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2770 undefined function error.
2771 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2772 in structure slot initforms) for now, documented workaround instead:-|
2773 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2774 * fixed bug 5: added stubs for various Gray stream functions called
2775 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2776 installed, at least appropriate type errors are generated
2777 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2778 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2779 consistently in DEFMETHOD forms.
2780 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2781 it doesn't seem to affect SBCL after all
2782 * The C runtime system now builds with better optimization and many
2783 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2785 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2787 * The system is now under CVS at SourceForge (instead of the
2788 CVS repository on my home machine).
2789 * The new signal handling code has been tweaked to treat register
2790 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2791 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2792 downstream, e.g. in debug-int.lisp, has implicit dependencies
2793 on the unsignedness of integer representation of machine words,
2794 and that caused the system to bomb out with infinite regress
2795 when trying to recover from type errors involving signed values,
2796 e.g. (BUTLAST '(1 2 3) -1).)
2797 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2798 (This was one of the bugs Peter Van Eynde reported back in July.)
2799 * The system now uses code inspired by Colin Walters' O(N)
2800 implementation of MAP (from the cmucl-imp@cons.org mailing
2801 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2802 inline the MAP operation, and there is more than one
2803 sequence argument to the MAP call (so that it can't just
2804 do ETYPECASE once and for all based on the type of the
2805 single sequence argument). (The old non-inline implementation
2806 of the general M-argument sequence-of-length-N case required
2807 O(M*N*N) time when any of the sequence arguments were LISTs.)
2808 * The QUIT :UNIX-CODE keyword argument has been renamed to
2809 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2810 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2812 * Raymond Wiker's patches to port dynamic loading from Linux to
2813 FreeBSD have been added.
2814 * The BUGS file is now more nearly up to date, thanks in large part
2815 to Martin Atzmueller's review of it.
2816 * The debugger now flushes standard output streams before it begins
2817 its output ("debugger invoked" and so forth).
2818 * The core version number and fasl file version number have both
2819 been incremented, because of incompatible changes in the layout
2821 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2822 used to not be called for a saved Lisp image.)
2823 * Martin Atzmueller's version of a patch to fix a compiler crash,
2824 as posted on sbcl-devel 13 September 2000, has been installed.
2825 * Instead of installing Martin Atzmueller's patch for the
2826 compiler transform for SUBSEQ, I deleted the compiler transform,
2827 and transforms for some similar consing operations.
2828 * A bug in signal handling which kept TRACE from working on OpenBSD
2830 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2831 away by constant folding
2832 * The system now defines its address space constants in one place
2833 (in the Lisp sources), and propagates them automatically elsewhere
2834 (through GENESIS and the sbcl.h file). Therefore, patching the
2835 address map is less unnecessarily tedious and error-prone. The
2836 Lisp names of address space constants have also been systematized.
2837 * CVS tags like dollar-Header-dollar have been removed from
2838 the sources, because they have never saved me trouble and
2839 they've been source of trouble working with patches and other
2840 diff-related operations.
2841 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2842 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2844 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2846 * The system has been ported to OpenBSD.
2847 * The system now compiles with a simple "sh make.sh" on the systems
2848 that it's supported on. I.e., now you no longer need to tweak
2849 text in the target-features.lisp-expr and symlinks in src/runtime/
2850 by hand, the make.sh takes care of it for you.
2851 * The system is no longer so grossly inefficient when compiling code
2852 involving vectors implemented as general (not simple) vectors (VECTOR T),
2853 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2854 which dares to use the various sequence functions on non-simple
2855 vectors, takes less of a performance hit.
2856 * There is now a primitive type predicate VECTOR-T-P
2857 to test for the (VECTOR T) type, so that e.g.
2858 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2859 can now be compiled with some semblance of efficiency. (The old code
2860 turned the type declaration into a full call to %TYPEP at runtime!)
2861 * AREF on (VECTOR T) is still not fast, since it's still compiled
2862 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2863 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2864 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2865 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2866 without making full calls to SUBTYPEP at runtime.
2867 (Some analogous efficiency issues for non-simple vectors specialized to
2868 element types other than T, or for non-simple multidimensional arrays,
2869 have not been addressed. They could almost certainly be handled the
2870 same way if anyone is motivated to do so.)
2871 * The changes in array handling break binary compatibility, so
2872 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2873 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2874 instead of (VALUES T T).
2875 * By following the instructions that Dan Barlow posted to sbcl-devel
2876 on 2 July 2000, I was able to enable primitive dynamic object
2877 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2878 functionality is not implemented (since it calls ld to resolve
2879 library references automatically, requiring RUN-PROGRAM for its
2880 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2881 try to resolve library references) is now supported.
2882 * The system now flushes the standard output streams when it terminates,
2883 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2884 them at several other probably-convenient times, e.g. in each pass of
2885 the toplevel read-eval-print loop, and after evaluating a form given
2886 as an "--eval" command-line option. (These changes were motivated by a
2887 discussion of stream flushing issues on cmucl-imp in August 2000.)
2888 * The source transform for TYPEP of array types no longer assumes
2889 that an array whose element type is a not-yet-defined type
2890 is implemented as an array of T, but instead punts, so that the
2891 type will be interpreted at runtime.
2892 * There is now some support for cross-compiling in make.sh: each of
2893 the phases of make.sh has its own script. (This should be transparent
2894 to people doing ordinary, non-cross-compile builds.)
2895 * Since my laptop doesn't have hundreds of megabytes of memory like
2896 my desktop machine, I became more motivated to do some items on
2897 my to-do list in order to reduce the size of the system a little:
2898 ** Arrange for various needed-only-at-cold-init things to be
2899 uninterned after cold init. To support this, those things have
2900 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2901 symbols with such names are now uninterned after cold init).
2902 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2903 fasl files for cold load.
2904 ** Remove the old compiler structure pooling code (which used to
2905 be conditional on the target feature :SB-ALLOC) completely.
2906 ** Redo the representation of some data in cold init to be more compact.
2907 (I also looked into supporting byte compiled code at bootstrap time,
2908 which would probably reduce the size of the system a lot, but that
2909 looked too complicated, so I punted for now.)
2910 * The maximum signal nesting depth in the src/runtime/ support code has
2911 been reduced from 4096 to 256. (I don't know any reason for the very
2912 large old value. If the new smaller value turns out to break something,
2913 I'll probably just bump it back up.)
2914 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2916 * Many, many bugs reported by Peter Van Eynde have been added to
2917 the BUGS list; some have even been fixed.
2918 * While enabling dynamic object file loading, I tried to make the
2919 code easier to understand, renaming various functions and variables
2920 with less ambiguous names, and changing some function calling
2921 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2922 * While trying to figure out how to do the OpenBSD port, I tried to
2923 clean up some of the code in src/runtime/. In particular, I dropped
2924 support for non-POSIX signal handling, added various comments,
2925 tweaked the code to reduce the number of compilation warnings, and
2926 renamed some files to increase consistency.
2927 * To support the new automatic configuration functionality in make.sh,
2928 the source file target-features.lisp-expr has been replaced with the
2929 source file base-target-features.lisp-expr and the machine-generated
2930 file local-target-features.lisp-expr.
2931 * fixed a stupid quoting error in make.sh so that using CMU CL
2932 "lisp -batch" as cross-compilation host works again
2934 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2936 * DESCRIBE no longer tries to call itself recursively to describe
2937 bound/fbound values, so that it no longer fails on symbols which are
2938 bound to themselves (like keywords, T, and NIL).
2939 * DESCRIBE now works on generic functions.
2940 * The printer now prints less-screwed-up representations of closures
2941 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2942 * A private symbol is used instead of the :EMPTY keyword previously
2943 used to mark empty slots in hash tables. Thus
2944 (DEFVAR *HT* (MAKE-HASH-TABLE))
2945 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2946 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2947 now does what ANSI says that it should. (You can still get
2948 similar noncompliant behavior if bang on the hash table
2949 implementation with all the symbols you get back from
2950 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2951 This breaks binary compatibility, since tests for equality to
2952 :EMPTY are wired into things like the macroexpansion of
2953 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2955 * There's now a minimal placeholder implementation for CL:STEP,
2956 as required by ANSI.
2957 * An obscure bug in the interaction of the normal compiler, the byte
2958 compiler, inlining, and structure predicates has been patched
2959 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2960 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2962 * Missing ordinary arguments in a macro call are now detected even
2963 when the macro lambda list contains &KEY or &REST.
2964 * The debugger no longer complains about encountering the top of the
2965 stack when you type "FRAME 0" to explicitly instruct it to go to
2966 the top of the stack. And it now prints the frame you request even
2967 if it's the current frame (instead of saying "You are here.").
2968 * As specified by ANSI, the system now always prints keywords
2969 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2971 * The default initial SIZE of HASH-TABLEs is now smaller.
2972 * Type information from CLOS class dispatch is now propagated
2973 into DEFMETHOD bodies, so that e.g.
2974 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2976 is now basically equivalent to
2977 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2978 (DECLARE (TYPE SINGLE-FLOAT X))
2980 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2981 operation, without having to do run-time type dispatch.
2982 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2983 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2984 * Since I don't seem to be making big file reorganizations very often
2985 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2986 is overflowing my ability to conveniently back them up), I've finally
2987 checked the system into CVS. (The CVS repository is on my home system,
2988 not at SourceForge -- putting it on SourceForge might come later.)
2989 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2990 high-level GC-NOTIFY-FOO functions send their output. (There's
2991 still very little control of where low-level verbose GC functions
2992 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2993 less than it used to -- the GC-NOTIFY-FOO functions are now under
2994 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2995 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2996 in only one place in the source code, and propagates it automatically
2997 everywhere that it's needed. Thus e.g. when I bump the version from
2998 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2999 * The C source files now include boilerplate legalese and documentation
3000 at the head of each file (just as the Lisp source files already did).
3001 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3002 to his page will be replaced with a link to his new CLiki service.
3004 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3006 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3007 * The build process now looks for GNU make under the default name "gmake",
3008 instead of "make" as it used to. If GNU make is not available as "gmake"
3009 on your system, you can change this default behavior by setting the
3010 GNUMAKE environment variable.
3011 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3012 can build without error under CMU CL.
3014 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3016 * There is now a partial SBCL user manual (with some new text and some
3017 text cribbed from the CMU CL manual).
3018 * The beginnings of a profiler have been added (starting with the
3019 CMU CL profiler and simplifying and cleaning up). Eventually the
3020 main interface should be through the TRACE macro, but for now,
3021 it's still accessed through vaguely CMU-CL-style functions and macros
3022 exported from the package SB-PROFILE.
3023 * Some problems left over from porting CMU CL to the new
3024 cross-compilation bootstrap process have been cleaned up:
3025 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3026 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3027 part when getting rid of the compiler *BACKEND* stuff.)
3028 ** The value of *NULL-TYPE* was screwed up, because it was
3029 being initialized before the type system knew the final
3030 definition of the 'NULL type. This screwed up several key
3031 optimizations in the compiler, causing inefficiency in all sorts
3032 of places. (I found it because I wanted to understand why
3033 GET-INTERNAL-RUN-TIME was consing.)
3034 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3035 PROCLAIM FTYPE information. Unfortunately this broke binary
3036 compatibility again, since now the forms output by DEFGENERIC
3037 to refer to functions which didn't exist in 0.6.3.
3038 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3039 can use the new (as of 0.6.3) transform for SOME into MAP into
3041 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3042 values to (INTEGER 0 1000000), so that the time code will no longer
3043 occasionally get blown up by Linux returning 1000000 microseconds
3044 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3045 its output conform to the ANSI spec. (Alas, this makes its output
3046 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3047 * A full call to MAP NIL with a single sequence argument no longer conses.
3048 * fixes to problems pointed out by Martin Atzmueller:
3049 * The manual page no longer talks about multiprocessing as though
3050 it were currently supported.
3051 * The ILISP support patches have been removed from the distribution,
3052 because as of version 5.10.1, ILISP now supports SBCL without us
3053 having to maintain patches.
3054 * added a modified version of Raymond Toy's recent CMU CL patch for
3055 EQUALP comparison of HASH-TABLE
3057 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3059 * The system still can't cross-compile itself with
3060 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3061 entails), but at least it can compile more of itself that way
3062 than it used to be able to, and various buglets which were uncovered
3063 by trying to cross-compile itself that way have now been fixed.
3064 * This release breaks binary compatibility again. This time
3065 at least I've incremented the FASL file format version to 2, so that the
3066 problem can be detected reliably instead of just causing weird errors.
3067 * various new style warnings:
3068 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3069 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3070 ** using the lexical binding of a variable named in the *FOO* style
3071 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3072 as specified by ANSI.
3073 * *RANDOM-STATE* is no longer automatically initialized from
3074 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3075 default behavior of the system is to repeat its behavior every time
3076 it's run. If you'd like to change this behavior, you can always
3077 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3078 old convention there was no comparably easy way to get the system to
3079 repeat its behavior every time it was run.
3080 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3081 FTYPE declarations has been removed, in favor of their ANSI
3082 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3083 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3084 the types of their sequence arguments can be determined at compile time.
3085 This is done through a new open code expansion for MAP which eliminates
3086 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3087 argument types can be determined at compile time.
3088 * The optimizer now transforms COERCE into an identity operation when it
3089 can prove that the coerced object is already of the correct type. (This
3090 can be a win for machine generated code, including the output of other
3091 optimization transforms, such as the MAP transform above.)
3092 * Credit information has been moved from source file headers into CREDITS.
3093 * Source file headers have been made more standard.
3094 * The CASE macro now compiles without complaining even when it has
3097 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3099 * (Note that the way that the PCL macroexpansions were rewritten
3100 to accommodate the change in DEFGENERIC below breaks binary
3101 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3102 not run under sbcl-0.6.2. Once we get out of alpha releases,
3103 i.e. hit release 1.0.0, we'll probably try to maintain binary
3104 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3105 and sbcl-1.4.4. Until then, however, it might be fairly common
3106 for maintenance releases to break binary compatibility.)
3107 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3109 * The reporting of unhandled signals has been changed to print some
3110 explanatory text as well as the report form. (Previously only
3111 the report form was printed.)
3112 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3113 it defines, so that the compiler no longer issues undefined function
3114 warnings for compiled-but-not-yet-loaded generic functions.
3115 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3116 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3117 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3118 section of the ANSI spec are supported.
3119 * The "" nickname for the "KEYWORD" package has been removed.
3120 The reader still handles symbol tokens which begin with a package marker
3121 as keywords, but it doesn't expose its mechanism for doing so in the
3122 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3123 * The system now issues STYLE-WARNINGs for contradictory TYPE
3124 proclamations. (Warnings for contradictory FTYPE proclamations would
3125 be nice too, but those can't be done usefully unless the type system
3126 is made smarter about FUNCTION types.)
3127 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3128 systematized, so that "*target-*.lisp is supposed to exist only on the
3129 target and imply that there's a related file which exists on the
3130 host, and *host-*.lisp is supposed to exist only on the host and imply
3131 that there's a related file which exists on the target. This involves a
3132 lot of renaming. Hopefully the acute confusion caused by the renaming
3133 will be justified by the reduction in chronic confusion..
3134 ** runtime-type.lisp -> early-target-type.lisp
3135 ** target-type.lisp -> late-target-type.lisp
3136 ** early-host-format.lisp -> early-format.lisp
3137 ** late-host-format.lisp -> late-format.lisp
3138 ** host-error.lisp -> misc-error.lisp
3139 ** early-error.lisp -> early-target-error.lisp
3140 ** late-error.lisp -> late-target-error.lisp
3141 ** host-defboot.lisp -> early-defboot.lisp
3142 ** code/misc.lisp -> code/target-misc.lisp
3143 ** code/host-misc.lisp -> code/misc.lisp
3144 ** code/numbers.lisp -> code/target-numbers.lisp
3145 ** code/early-numbers.lisp -> numbers.lisp
3146 ** early-host-type.lisp -> early-type.lisp
3147 ** late-host-type.lisp -> late-type.lisp
3148 ** host-typep.lisp -> typep.lisp
3149 ** load.lisp -> target-load.lisp
3150 ** host-load.lisp -> load.lisp
3151 ** host-disassem.lisp -> disassem.lisp
3152 ** host-insts.lisp -> insts.lisp
3153 ** byte-comp.lisp -> target-byte-comp.lisp
3154 ** host-byte-comp.lisp -> byte-comp.lisp
3155 ** host-signal.lisp -> signal.lisp
3156 ** host-defstruct.lisp -> defstruct.lisp
3157 ** late-target-type.lisp -> deftypes-for-target.lisp
3158 Furthermore, several other previously target-only files foo.lisp (e.g.
3159 hash-table.lisp and random.lisp) have been split into a target-and-host
3160 foo.lisp file and a target-only target-foo.lisp file, with their key type
3161 definitions in the target-and-host part, so that the cross-compiler will
3162 know more about target types.
3163 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3164 gone away. In their place are various *BACKEND-FOO* variables
3165 corresponding to the slots of the old structure.
3166 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3167 into the target SBCL has been fixed.
3168 * The chill.lisp system for loading cold code into a running SBCL
3170 * Support for the CMU CL "scavenger hook" extension has been removed.
3171 (It was undocumented and unused in the CMU CL sources that SBCL was
3172 derived from, and stale in sbcl-0.6.1.)
3173 * Various errors in the cross-compiler type system were detected
3174 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3175 (enabling various consistency checks). Many of them were fixed,
3176 but some hard problems remain, so the compiler is back to
3177 running without *TYPE-SYSTEM-INITIALIZED* for now.
3178 * As part of the cross-compiler type system cleanup, I implemented
3179 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3180 * I have started adding UNCROSS calls throughout the type system
3181 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3182 on cross-compiler input files will be able to go away, and various
3184 * CONSTANTP now returns true for quoted forms (as explicitly required
3187 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3189 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3190 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3191 (MAKE-LIST -1). In the longer term, it ought to become true
3192 that declarations are assertions even at SAFETY 1. For now, it's not
3193 quite true even at SAFETY 3, but it's at least more nearly true..
3194 (Note that this change seems to increases the size of the system by
3195 O(5%) and to decrease the speed of the compiler by 20% or more.)
3196 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3197 for the problem of printing dozens of lines of distracting information
3198 about low-level system machinery as part of the top stack frame
3199 on entry to the debugger when an undefined function was called.
3200 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3201 is not reset to COMMON-LISP-USER.
3202 * Compilation of stuff related to dyncount.lisp has been made conditional
3203 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3204 smaller. The various dyncount-related symbols have been moved into
3205 a new "SB-DYNCOUNT" package.
3206 * tty-inspect.lisp has been renamed to inspect.lisp.
3207 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3208 feature has gone away. (When we eventually port to other flavors of
3209 libc and/or Unix, we'll try to make the differences between flavors
3210 invisible at the user level.)
3211 * Various other *FEATURES* tags, and/or their associated conditionals,
3212 have been removed if obsolescent, or given better documentation, or
3213 sometimes given more-mnemonic names.
3215 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3217 * tidied up "make.sh" script
3218 * tidied up system directory structure
3219 * better "clean.sh" behavior
3220 * added doc/FOR-CMUCL-DEVELOPERS
3221 * many many small tweaks to output format, e.g. removing possibly-confusing
3222 trailing #\. character in DESCRIBE-INSTANCE
3223 * (EQUALP #\A 'A) no longer signals an error.
3224 * new hashing code, including EQUALP hashing
3225 * tidied up Lisp initialization and toplevel
3226 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3227 * command line argument processing
3228 * added POSIX-GETENV function to deal with Unix-ish environment variables
3229 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3230 terminating SBCL on EOF
3231 * non-verbose GC by default
3232 * There is no more "sbcl" shell script; the sbcl file is now the C
3233 runtime executable (just like CMU CL).
3234 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3236 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3237 symbol and package information as Lisp native symbol and package objects
3238 instead of strings naming symbols and strings naming packages. This way,
3239 whenever packages are renamed (as in warm init), debug information is
3240 transformed along with everything else.
3241 * tweaked the optimization policy declarations which control the building
3242 of SBCL itself. Now, among other things, the system no longer saves
3243 source location debugging information. (This helps two problems at once
3244 by reducing SBCL size and by keeping SBCL from trying to look for its
3245 sources -- which may not exist -- when reporting errors.)
3246 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3247 debugging and testing purposes
3248 * cleaned up printing, making the printer call PRINT-OBJECT for
3249 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3250 methods, giving nearly-ANSI behavior
3251 * converted almost all special variables to use *FOO* naming convention
3252 * deleted PARSE-TIME functionality, since it can be done portably
3253 * moved some files out of cold init into warm init
3254 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3256 * regularized formatting of source files
3257 * added an install.sh script
3258 * fixed ridiculous memory usage of cross-compiler by making
3259 compiler/alloc.lisp not try to do pooling unless it can hook
3260 itself into the GC of the cross-compilation host. Now the system
3261 builds nicely on my old laptop.
3262 * added :SB-ALLOC in target-features.lisp-expr
3263 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3264 was not implemented)
3265 * re-did condition handling and note reporting in the compiler. Notes
3266 are no longer handled by signalling conditions. Style warnings
3267 and warnings are handled more correctly and reported in such a way
3268 that it's easy to find one or the other in your output (so that you
3269 can e.g. figure out which of many problems caused COMPILE-FILE to
3271 * changed the severity of several compiler warnings from full WARNING
3272 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3273 compiler note reporting so that it doesn't use the condition system
3274 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3275 COMPILE-FILE command)
3276 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3277 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3278 no longer supported, since their implementation was deeply intertwingled
3279 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3280 * removed bogus "support" for compiler macros named (SETF FOO), and
3281 removed the compiler macro for SETF INFO (but only after making a fool
3282 of myself on the cmucl-imp mailing list by posting a bogus patch for
3283 DEFINE-COMPILER-MACRO..)
3284 * Compiled files containing forms which have side effects on the Lisp
3285 reader (such as DEFPACKAGE forms) are now handled more correctly.
3286 (Compiler queuing of top level lambdas has been suppressed by setting
3287 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3288 * deleted various currently-unused source files, e.g. gengc.lisp. They
3289 may be added back at some point e.g. when porting to other architectures,
3290 but until they are it's distracting to distribute them and to try to
3292 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3293 there were so many of them they're just distractions, and UNCROSS is
3294 known to be able to handle the current sources
3295 * moved PROFILE functionality into TRACE, so that it will be clear
3296 how the wrapping and unwrapping of functions when you profile them
3297 interacts with the wrapping and unwrapping of functions when you
3298 trace them. (Actually, the functionality isn't there yet, but at least
3299 the interface specification is there. Hopefully, the functionality will
3300 arrive with some future maintenance release.)
3301 * removed host-oops.lisp
3302 * changed signature of QUIT function to allow UNIX-CODE argument
3303 * fixed READ-SEQUENCE bug
3304 * tweaked verbose GC output so that it looks more like the progress
3305 output that ANSI specifies for functions like LOAD
3306 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3307 * added <http://sbcl.sourceforge.com> to the banner information printed by