1 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2 * minor incompatible change: we now correctly canonize default
3 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
4 as per the MOP, rather than the historical (INITARG INITFUNCTION
5 INITFORM). (reported by Bruno Haible)
6 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
7 * fixed some lockups due to gc/thread interaction
8 * dynamic space size on PPC has been increased to 768Mb. (thanks to
10 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
11 :METACLASS argument in addition to a class name. (reported by
12 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellman)
13 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
15 * bug fix: more cleanups to the floating point exception handling on
16 x86-64 (thanks to James Knight)
17 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
18 the wrong file descriptor. (thanks to Tony Martinez)
19 * fixed some bugs revealed by Paul Dietz' test suite:
20 ** Invalid dotted lists no longer raise a read error when
22 ** Signal an error if a symbol that names a declaration is used
23 as the name of a type, or vice versa
25 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
26 * fixed cross-compiler leakages that prevented building a 32-bit
27 target with a 64-bit host compiler.
28 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
29 opened with :IF-EXISTS :OVERWRITE.
30 * fixed bug 281: error for an invalid qualifier in a short-form method
31 combination method is not signalled until the faulty method is called.
32 * bug fix: iteration variable type inferrer failed to deal with open
33 intervals. (reported by Alan Shields)
34 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
35 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
36 or more runtime options were provided to the sbcl binary.
37 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
38 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
39 * optimization: there's now a fast-path for fixnum arguments in the
40 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
41 * optimization: the code generated on x86-64 is more compact thanks
42 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
43 * bug fix: floating-point exception handling now partly works on
44 x86-64. (Thanks to James Knight)
45 * improvement to the MIPS backend from Thiemo Seufer: C-style
46 64-bit long long arguments and return values to alien functions
48 * fixed some bugs revealed by Paul Dietz' test suite:
49 ** the type-error signalled from WARN has a filled-in DATUM slot.
50 ** the type-error required when a stream is not associated with
51 a file has the stream as its datum.
52 ** type-errors on single-floats on x86-64 no longer have
53 :INVALID-OBJECT as the datum
54 ** the type-errors signalled for invalid function names now have
55 a correct expected type
56 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
57 for structures defined with a :TYPE. Documentation strings for
58 typed structures are no longer immediately discarded
59 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
60 broadcast streams and synonym streams. FILE-LENGTH now also works
63 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
64 * incompatible change: the --noprogrammer option, deprecated since
65 version 0.7.5, has been removed. Please use the equivalent
66 --disable-debugger option instead.
67 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
69 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
70 inoperational for a while now) has been completely removed.
71 * null lexical environments are now printed as #<NULL-LEXENV>,
72 significantly reducing the amount of clutter in typical backtraces.
73 * documentation on weak pointers, finalization, and after GC hooks
74 has been added to the manual.
75 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
76 as other specialized array types, is much faster. SUBSEQ and
77 COPY-SEQ on such arrays have also been sped up.
78 * optimization: EQL is now more efficient when at least other argument
79 is known to be of type (OR FIXNUM (NOT NUMBER)).
80 * fixed bug: the runtime is now less vulnerable to changes in the
81 size of the SBCL object on OS X, and virtual memory is reserved for
82 all spaces that need to be at a fixed address.
83 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
84 * fixed bug: finalizers and after GC hooks that cause consing are now
86 * fixed bug: compiler error messages and summaries are now printed to
87 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
88 * fixed inference of the upper bound of an iteration variable.
89 (reported by Rajat Datta).
90 * fixed bug 211e: calling local functions with duplicated constant
91 keyword argument no longer causes a bogus style warning about an
93 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
94 ftype declaration to be disregarded. (reported by Dave Roberts)
95 * fixed bug 373: caused by erronous compilation of references to alien
96 variables in the runtime on ppc/darwin.
97 * fixed bug 376: CONJUGATE type deriver.
98 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
99 (reported by Baughn on #lisp)
100 * a cross-compiler bug on non-x86 platforms has been identified and
101 fixed. (thanks to Bruno Haible)
102 * improvements to the MIPS runtime code for increased stability.
103 (thanks to Thiemo Seufer)
104 * increased the maximimum compact environment size to allow
105 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
106 * partial workaround for bug 135: don't name the function we're
107 calling for hairy cases of EVAL, so as not to accumulate one environment
108 entry per call to EVAL. (reported by Kevin Reid)
109 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
110 * optimization: type testing for non-vector arrays should be faster.
111 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
113 * bug fix: setting 31st element of a bit vector to zero did not work
115 * bug fix: redefining a class definition which failed due to a
116 previous accessor / function clash now works (but see BUGS entry
117 #380 for more problems in this area). (thanks to Zach Beane)
118 * the long form of DEFINE-METHOD-COMBINATION disables method group
119 checking when given a single method group with pattern *. (thanks
121 * on x86 compiler supports stack allocation of results of simple
122 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
123 * support for building and running on Mac OS X 10.4 "Tiger" has been added
124 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
125 * fixed some bugs related to Unicode integration:
126 ** the restarts for recovering from input and output encoding
127 errors only appear when there is in fact such an error to
129 * fixed some bugs revealed by Paul Dietz' test suite:
130 ** MISC.549 and similar: late transformation of unsafe type
131 assertions into derived types caused unexpected code
133 ** SCALE-FLOAT type deriver is less wrong.
134 ** type derivers for EXP, LOG and similar functions compute result
135 types for complex arguments better.
136 ** (MISC.563) CONJUGATE type deriver works for very restricted
138 ** out-of-line type testers for character strings are available.
139 ** EQUAL compiler transform understands specialness of objects
141 ** accessing double-floats stored on the stack now works on x86-64.
142 ** debugger internals could sometimes create invalid lispobjs,
143 resulting in GC crashes.
144 ** MISC.548: type check weakening can convert required type into
146 ** initialization forms for bindings are not in scope of free special
149 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
150 * incompatible change: thread support for non-NPTL systems has
151 been removed - locking is buggy and unreliable. A threaded
152 SBCL build will now warn at startup and refuse to create
153 new threads, unless futex support is detected in the kernel
154 * incompatible change: the top level REPL now has only an ABORT
155 restart associated with it, not TOPLEVEL and ABORT as it used to.
156 TOP and TOPLEVEL are now available as debugger commands for
157 returning to the top level.
158 * incompatible change: forms evaluated in the REPL now use the
159 global optimization policy.
160 * incompatible change: user- and system-initialization files are
161 no longer processed with LOAD, but by READ and EVAL; hence the
162 global optimization policy, startup package, readtable, etc,
164 * The .fasl file format number has been incremented because of
165 various incompatible changes.
166 * internal entry point details and argument counts no longer appear
167 in backtraces unless explicitly requested by setting
168 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
169 * built-in and standard functions no longer have names like "top
170 level local call to FOO".
171 * fixed bug 32: functions defined in non-null lexical environments
172 now have more legible printed representation
173 * fixed bug 33: functions defined in non-null lexical environemnts
174 are now more amenable to inspection by INSPECT.
175 * workaround for bug 354: XEPs no longer appear in backtraces unless
176 explicitly requested.
177 * fixed bug: receiving the signal which results from attempting to
178 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
179 SIGBUS on other *BSDs) on architectures where the C stack is also
180 the Lisp stack (x86 and x86-64) from foreign code no longer leads
181 to debugger confusion or wild instability. (reported by Cheuksan
183 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
184 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
185 (reported by Lutz Euler)
186 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
187 compiled code. (reported by Paul Dietz)
188 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
189 specialized parameter to a method to something that is not TYPEP
190 the specializer is now possible.
191 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
192 face of package deletion.
193 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
194 pathname. (reported by David Tolpin on c.l.l)
195 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
196 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
197 STRING-STREAM and FILE-STREAM, respectively.
198 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
199 than previously; in addition, its test suite is now run on build.
200 (thanks to Luke Gorrie)
201 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
202 * optimization: added a immediate representation for single-floats
204 * fixed some bugs revealed by Paul Dietz' test suite:
205 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
206 ** attempting to create a package with a colliding nickname causes
207 correctable errors to be signalled.
208 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
209 ** several x86-64 backend bugs related to sign-extension of immediate
212 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
213 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
214 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
216 * fixed loading of multiply forward-referenced layouts.
217 (thanks to Cheuksan Wang)
218 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
219 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
220 (see COMPILE-FILE documentation for details of :PRINT option)
221 * fixed bugs 19 and 317: fixed-format floating point printing is
222 more accurate. This also fixes a bug reported by Adam Warner
223 related to the ~@F format directive.
224 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
226 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
227 dispatch table if the from-char is a dispatch macro character.
228 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
229 by the wordsize no longer ignore the last word. (reported by Lutz
231 * fixed bug in type checking of dynamic-extent variables. (reported
233 * optimization: sequence traversal functions use their freedom to
234 coerce function designators to functions.
235 * optimization: code with many calls to CLOS methods specialized on
236 CLOS classes has had redundant type checks removed.
237 * optimization: added declarations to speed up operations that access
238 the internal character database (for example STRING-UPCASE)
239 * optimization: comparison operations between floats and sufficiently small
240 fixnums no longer create extra rationals
241 * fixed some bugs related to Unicode integration:
242 ** portions of multibyte characters at the end of buffers for
243 character-based file input are correctly transferred to the
244 start of the buffer at the next read.
245 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
246 passing it through to OPEN.
247 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
248 argument given to internal calls to OPEN.
249 ** the built-in comment readers (introduced by character sequences
250 ";" and "#|") are more forgiving to encoding errors; they will
251 STYLE-WARN and then attempt to resync the stream at a character
252 boundary. (thanks to Teemu Kalvas)
253 * fixed some bugs in the x86-64 port:
254 ** Negative short int return values from c-calls are sign-extended
256 ** The stack is aligned to 16-bytes for c-calls, as required by
257 the ABI. (thanks to Cheuksan Wang)
258 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
259 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
260 for enums. (thanks to Vincent Arkesteijn)
261 ** Multiple small optimizations and bugfixes for floating point
263 * fixed some bugs revealed by Paul Dietz' test suite:
264 ** Space, Tab, Linefeed, Return and Page have the invalid
265 secondary constituent character trait.
266 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
268 ** WITH-INPUT-FROM-STRING should only update the index place on
270 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
272 ** Bit-array operations (BIT-AND and similar) worked incorrectly
273 with one-dimensional arrays with fill pointers.
274 ** TYPE-OF failed on a complex with an integer realpart and a
276 ** compiler failure during type inference for the code of form
277 (IF (EQL X (THE ...)) ...) (MISC.535).
279 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
280 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
281 platform. The port supports 61-bit fixnums, large memory spaces
282 and reloading shared object files.
283 * enhancement: saving cores with foreign code loaded is now
284 supported on ppc/Darwin in addition to the previously supported
286 * enhancement: the statistical profiler now walks deeper into the
287 call stack for more meaningful call-graphs and accrued time
288 reports (x86/x86-64 only). It also now reports time spent in
290 * enhancement: it is now possible to trace most individual methods
291 of a generic function in addition to tracing the generic function
293 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
295 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
296 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
297 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
298 produces an error. (thanks to Vincent Arkesteijn)
299 * bug fix: NAMESTRING on pathnames with :WILD components in their
300 directories works correctly. (thanks to Artem V. Andreev)
301 * fixed bug 125: compiler preserves identity of closures. (reported
303 * bug fixed: functions with &REST arguments sometimes failed with
304 "Undefined function" when compiled with (DEBUG 3). (reported by
306 * bug fix: overflow during compiling of setting element of a bit
307 vector with constant index and value. (reported by Timmy Douglas)
308 * build fix: fixed the dependence on *LOAD-PATHNAME* and
309 *COMPILE-FILE-PATHNAME* being absolute pathnames.
310 * on x86 compiler partially supports stack allocation of dynamic-extent
312 * GO and RETURN-FROM do not check the extent of their exit points
313 when compiled with SAFETY 0.
314 * fixed some bugs related to Unicode integration:
315 ** encoding and decoding errors are now much more robustly
316 handled; it should now be possible to recover even from invalid
317 input or output to the terminal. (thanks to Teemu Kalvas)
318 ** provided a first cut at implementing STRING-TO-OCTETS and
319 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
320 ** altered the SB-MD5 contributed module slightly, changing the
321 interface just enough for it to be supportable for builds where
322 lisp characters are not eight bits.
323 * fixed some bugs revealed by Paul Dietz' test suite:
324 ** the FORMATTER-generated functions for ~V[ conditionals require
325 the correct number of arguments.
326 ** READ-FROM-STRING returns the mandated second value when applied
327 to displaced strings.
328 ** the #\Rubout and #\Backspace characters are treated as invalid
329 constituent characters by the tokenizer.
331 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
332 * new feature: reloading changed shared object files with
333 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
334 * new feature: references to foreign variables and functions
335 can now be compiled and loaded before the corresponding shared
336 object file is loaded, as long as the foreign definitions are
337 available at runtime.
338 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
339 supported. (thanks to Dan Debertin)
340 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
341 just keywords. (thanks to Vincent Arkesteijn)
342 * fixed bug #331: structure-class instances corresponding to
343 DEFSTRUCT forms are now created eagerly.
344 * fixed bug #345: backtraces from calls to undefined functions work
345 on x86 as well. Related bug #61 is now also partially fixed on x86
346 (backtraces from throws to unknown catch tags.)
347 * bug fix: lambda-list parsing is now stricter vrt. order and number
348 of lambda-list keywords.
349 * bug fix: as specified by AMOP, an error is signalled if a
350 class-option appears multiple times in a DEFCLASS form. (reported
352 * bug fix: RANDOM can be compiled when the compiler derives the type
353 of its numeric argument as a disjoint set of small integers.
354 (reported by Paul Dietz)
355 * bug fix: starting a core saved with shared objects loaded when
356 those objects are not available no longer causes threaded SBCL to
357 hang. (reported by Sean Ross)
358 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
360 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
361 argument, which must always be NIL. (reported by Kalle Niemitalo)
362 * bug fix: printing 1.0d+23 no longer results in an error.
363 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
364 * bug fix: structure slot setters preserve evaluation order. (thanks
366 * fixed some bugs related to Unicode integration:
367 ** RUN-PROGRAM can allow its child to take input from a Lisp
368 stream. (reported by Stefan Scholl)
369 ** ASDF-INSTALL successfully downloads .tar.gz files in all
370 locales. (reported by Ken Causey)
371 * fixed some bugs revealed by Paul Dietz' test suite:
372 ** INCF, DECF and REMF evaluate their place form as specified in
374 ** FORMATTER expands ~{ ~} iteration directives with V or #
375 parameters correctly.
376 ** FORMATTER deals with the ~@[ ~] conditional directive where the
377 consequent uses no arguments correctly.
378 ** the system has a partial understanding of the (COMPLEX RATIO)
380 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
383 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
384 * new feature: a build-time option (controlled by the :SB-UNICODE
385 keyword feature, enabled by default) for building the system with
386 support for the entire 21-bit character space defined by the
388 * new feature: the system now has rudimentary external-format
389 support; the primary user-visible change at this time is that
390 characters with the high bit set (such as lower-case-e-acute) will
391 print correctly to a terminal in a UTF-8 environment.
392 * minor incompatible change: BASE-CHAR no longer names a class;
393 however, CHARACTER continues to do so, as required by ANSI.
394 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
395 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
396 has been moved to the SB-EXT package (temporarily re-exported from
398 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
399 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
401 * fixed bug #351: better error-handlind and reporting for malformed
403 * fixed bug #350: bignum-printing is now more memory-efficient,
404 allowing printing of very large bignums, eg. (expt 2 10000000).
405 (reported by Bruno Haible)
406 * fixed bug #302: better primitive-type selection for intersection
408 * fixed bug #308: non-graphic characters now all have names, as
409 required. (reported by Bruno Haible)
410 * bug fix: redefining a class with different superclasses now correctly
411 removes it from the direct-subclasses of its previous superclasses.
412 (reported by David Morse)
413 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
414 new value now works. (reported by Bruno Haible)
415 * bug fix: correct canonicalization of multiple non-standard slot
416 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
417 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
418 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
419 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
420 now exists, an signals an error.
421 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
422 during macroexpansion for non-iteration variables as well. (reported
423 by Bruno Haible for CMUCL)
424 * bug fix: Cyclic structures and unprintable objects in compiler
425 messages no longer cause errors. (reported by Bruno Haible)
426 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
427 and READ-FROM-STRING all now return a primary value of NIL if
428 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
429 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
430 (reported by Bruno Haible for CMUCL)
431 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
432 do not designate a condition. (reported by Bruno Haible for
434 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
435 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
436 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
437 permitting resolution in favour of any of the conflicting symbols.
438 (reported by Bruno Haible for CMUCL)
439 * bug fix: EQUAL compiler optimizations is less aggressive on
440 strings which can potentially compare true despite having distinct
441 specialized array element types.
442 * bug fix: unit enumerations can be defined without dividing by
443 zero. (thanks to Vincent Arkesteijn)
444 * FORMAT compile-time argument count checking has been enhanced.
445 (report from Bruno Haible for CMUCL)
446 * a partial workaround for the bug 262: the compiler does not try to
447 inline-expand a local function doing RETURN-FROM from a deleted
448 BLOCK. (thanks to Peter Denno for the bug report and to David
449 Wragg for the simple test case)
450 * fixed some bugs revealed by Paul Dietz' test suite:
451 ** RENAME-PACKAGE allows all package designators as new package
453 ** constraint propagation and lambda variable substitution are
454 more cautious in dealing with partially deleted code.
455 ** compiler sometimes lost reoptimization passes.
456 ** CERROR, given a condition as condition designator, passes its
457 remaining arguments to the continue format control without
459 ** Case-altering FORMAT directives work correctly on non-ASCII
461 ** The REAL type specifier handles bounds outside the floating
462 point ranges without signalling FLOATING-POINT-OVERFLOW.
463 ** Functions with IR1-transformations can create intercomponent
464 references to global functions.
465 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
467 ** FORMAT ~R treats a nil value for its first parameter correctly.
469 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
470 * enhancement: saving cores with foreign code loaded is now
471 supported on x86/NetBSD and sparc/Linux in addition to the previously
473 * bug fix: on some platforms repeated installations caused multiple
474 copies of HTML documentation to be installed -- should not happen
475 any more. (reported by Stefan Scholl)
476 * bug fix: parsing self-recursive alien record types multiple times
477 no longer causes infinite recursion. (reported by Thomas F. Burdick,
478 original patch by Helmut Eller for CMUCL)
479 * bug fix: stack-exhaustion detection works now on NetBSD as well.
480 (thanks to Richard Kreuter)
481 * bug fix: defining classes whose accessors are methods on existing
482 generic functions in other (locked) packages no longer signals
483 bogus package lock violations. (reported by François-René Rideau)
484 * bug fix: special variables as DEFMETHOD parameters no longer have
485 associated bogus type declarations. (reported by David Wragg and
487 * bug fix: read-write consistency on streams of element-type
488 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
489 * bug fix: redefiniton of the only method of a generic function with
490 no DEFGENERIC no longer emits a full WARNING. In addition,
491 redefinition of generic functions with no DEFGENERIC to an
492 incompatible lambda list now signals an error. (thanks to Zach
494 * bug fix: DEFGENERIC now works even when there's a function of the
495 same name in an enclosing lexical environment. (thanks to Zach
497 * fixed compiler failure, caused by instrumenting code during
498 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
499 * optimization: added loop analysis and improved register allocation
500 * fixed some bugs revealed by Paul Dietz' test suite:
501 ** POSITION on displaced vectors with non-zero displacement
502 returns the right answer.
503 ** (SIMPLE-STRING) is a valid type specifier for sequence
505 ** *PRINT-LEVEL* handling for slotless structures is pedantically
507 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
508 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
510 ** The justification version of the FORMAT ~< directive treats
511 non-zero minpad parameter correctly.
513 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
514 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
515 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
516 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
517 the supported interface.
518 * new feature: Single-stepping of code compiled with DEBUG 2 or
519 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
520 * new feature: saving cores with foreign code loaded is now
521 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
522 Timothy Moore's work for CMUCL)
523 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
524 parameters to * instead of NIL if no initform is supplied.
525 (reported by Johan Bockgaard)
526 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
527 correctly when FUNCALL appears as the car of the form. Note:
528 despite this FUNCALL forms are not currently subject to
529 compiler-macro expansion. (port of Raymond Toy's fix for the
530 same from CMUCL, reported by Johan Bockgaard)
531 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
532 (thanks for Teemu Kalvas)
533 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
534 case instead of terminating the iteration (thanks for Julian
535 Squires, Sean Champ and Raymond Toy)
536 * bug fix: incorrect expansion of defgeneric that caused a style
537 warning. (thanks for Zach Beane)
538 * bug fix: slot accessor effective method computation works properly
539 for classes with multiple non-standard applicable methods on
540 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
541 * on x86 compiler supports stack allocation of results of LIST and
542 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
543 CMUCL implementation by Gerd Moellmann)
544 * fixed some bugs revealed by Paul Dietz' test suite:
545 ** FORMAT strings with both the ~<~:;~> form of the justification
546 directive and pretty-printing directives cause an error.
548 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
549 * incompatible change: the internal functions
550 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
551 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
552 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
553 instead of the old functions.
554 * new feature: on platforms where "dladdr" is available foreign
555 function names now appear in backtraces. (based on Helmut Eller's
557 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
558 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
560 * documentation: ASDF manual is now distributed with SBCL.
561 * bug fix: SBCL can now load its contributed modules with REQUIRE
562 even if the system-provided entries have been removed from
563 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
565 * bug fix: backtraces involving undefined functions or assembly
566 routines are more informative. (thanks to Brian Downing)
567 * bug fix: mutually referent alien structures now work correctly.
568 (reported by Rick Taube)
569 * bug fix: structures defined by WITH-ALIEN can be referred to
570 within other definitions in the same WITH-ALIEN.
571 * bug fix: division operators (MOD, TRUNCATE and the like) with
572 constant zero divisors and integer dividends no longer generate
574 * bug fix: provide default methods for INPUT-STREAM-P and
575 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
576 * bug fix: improve the detection and resolution of MOP metacycles
577 (where effective method computation depends on the generic function
578 having its effective method computed). (reported by Bruno Haible)
579 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
580 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
581 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
582 * bug fix: pathnames with relative directory components can be
583 represented relative to default pathnames.
584 * optimization: in taking the GCD of bignums, reduce the two bignums
585 to approximately the same size (using Euclid's algorithm) before
586 applying the more sophisticated binary GCD. (thanks to Juho
588 * optimization: COUNT on bitvectors now operates word-at-a-time.
589 * optimization: ASH with a positive, but not necessarily constant,
590 (leftwards) shift, when in a modular context, is compiled to a
592 * fixed some bugs revealed by Paul Dietz' test suite:
593 ** FORMAT variable parameters ("~V<char>") are defaulted properly
594 if the corresponding argument is NIL.
595 ** FORMAT directives accepting a minpad parameter treat negative
597 ** literal commas as character parameters to FORMAT directives are
599 ** literal spaces directly after ~<Newline> directives within a
600 format-logical-block (~:< ~@:>) do not induce :FILL-style
601 conditional newlines.
602 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
603 as specified: it no longer includes conditional newlines).
604 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
606 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
607 :PER-LINE-PREFIX argument does not evaluate to a string.
609 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
610 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
611 the manual for details; for now, package locks can be disabled by
612 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
613 if no major problems are found then it is likely that they will be
614 compiled in unconditionally.
615 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
616 unsupported operators on all platforms. To load a shared library
617 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
618 object file, link it into a shared library outside of SBCL and
619 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
621 * bug workaround: ROOM T and the GC were not working together
622 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
623 reporting facility aren't preserved. That reporting has been
624 disabled, so now until and unless someone figures out how to make it
625 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
626 * minor incompatible change: as threatened around sbcl-0.8.0, the
627 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
628 The symbols are also exported from SB-PCL for backwards
629 compatibility, but more so than before SB-PCL should be treated as
630 an implementation-internal package.
631 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
633 * the SB-SPROF contrib now works on (most) non-x86 architectures.
634 It is known as of this release not to work on the Alpha, however.
635 * fixed bug #167: errors signalled due to illegal syntax in method
636 bodies are now more legible.
637 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
638 designators and can hence be used with TYPEP.
639 * fixed bug #333: CHECK-TYPE now ensures that the type error
640 signalled, if any, has the right object to be accessed by
641 TYPE-ERROR-DATUM. (reported by Tony Martinez)
642 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
643 5.1.2.3 for multiple-value place subforms. (reported by Kalle
645 * fixed bug #334: programmatic addition of slots using specialized
646 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
647 and :ALLOCATION :CLASS slots.
648 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
649 float exponents. (rereported by Peter Seibel)
650 * fixed a bug: #\Space (and other whitespace characters) are no
651 longer considered to be macro characters in standard syntax by
653 * fixed bug: initialization of condition class metaobjects no longer
654 causes an instance of the condition to be created. (reported by
656 * fixed bug: it is now possible to have slots such that
657 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
658 description is neither :INSTANCE nor :CLASS.
659 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
660 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
661 SLOT-VALUE-USING-CLASS), and no longer causes errors with
662 non-standard SLOT-DEFINITION-ALLOCATION values.
663 * fixed bugs: various race conditions handling exiting threads.
664 CL-PPCRE's thread torture test now passes.
665 * fixed arguably-a-bug: GC time proportional to number of threads in
666 system even when most of them are idle
667 * optimization: improved performance of BIT and SBIT on bit-vectors.
668 * .fasl file incompatibility: The fasl file version number has
669 been incremented because of changes associated with package locks.
671 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
672 * minor incompatible change: the system no longer provides
673 optimization hints (conditions which are TYPEP
674 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
675 mode; these hints are still emitted when higher SPEED optimization
677 * new contrib module: a sampling profiler (profiling by statistical
678 sampling, rather than by instrumenting functions) is available as
679 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
680 * the behaviour of the standard function ED is now customizeable by
681 third parties through a hook variable: see ED's documentation
682 string for information on the protocol.
683 * the compiler no longer emits efficiency notes for (FUNCALL X)
684 when the type of X is uncertain under default optimization
686 * fixed bug 276: mutating a binding of a specialized parameter to a
687 method to something that is not TYPEP the specializer is now
689 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
690 exhibit the required behaviour on the x86 platform. (thanks to
691 Peter van Eynde, Eric Marsden and Bruno Haible)
692 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
693 even for difficult arguments. (reported by Peter Graves)
694 * fixed bug 141a: the backquote printer now descends quoted
696 * fixed another bug in backquote printing: no more destructive
697 modification of the form's list structure. (reported by Brian
699 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
700 move between its address being taken and the call to
701 interrupt_thread, fixing a crashing race condition.
702 * the SB-POSIX contrib implementation has been adjusted so that it
703 no longer exhibits ridiculously poor performance when constructing
704 instances corresponding to C structs.
706 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
707 * minor incompatible change: the sb-grovel contrib now treats C
708 structures as alien (in the sense of SB-ALIEN) objects rather than
709 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
710 has implications for memory management of client code
711 (sb-grovel-returned objects must now be manually managed) and for
712 type safety (alien objects now have full types).
713 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
714 used to control emission of compiler diagnostics, rather than the
715 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
716 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
717 quality should be considered deprecated.
718 * install.sh now installs the user manual as well
719 * (not quite a new documentable feature, but worth considering in
720 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
721 beginnings of a semantically meaningful condition hierarchy is
722 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
724 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
725 designator as the defaults argument.
726 * fixed bug: Displaced arrays whose displaced-to array has become
727 too small now cause ARRAY-DIMENSION to signal an error, providing
728 for safer bounds-checking. (reported by Bruno Haible)
729 * fixed bug: DEFCLASS slot definitions with identical :READER and
730 :WRITER names now signal a reasonable error. (reported by Thomas
732 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
734 * fixed bug: Invalid slot specification errors now print correctly.
735 (thanks to Zach Beane)
736 * fixed bug 320: Shared to local slot value transfers in class
737 redefinitions now happen corrently from superclasses as
738 well. (reported by Bruno Haible)
739 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
741 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
742 improper lists correctly. (reported by Bruno Haible)
743 * fixed bug 313: source-transform for <fun-name> was erroneously
744 applied to a call of a value of a variable with name <fun-name>.
745 (reported by Antonio Menezes Leitao)
746 * fixed bug 307: The obsolete instance protocol ensures that
747 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
748 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
749 remove dead unknown-values globs from the middle of the stack.
750 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
751 easier to resume long computations after using *BREAK-ON-SIGNALS*
752 to diagnose and fix failures. (thanks to Nikodemus Siivola)
753 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
754 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
755 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
756 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
757 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
759 * fixed a (fairly theoretical) bug in string printing: if
760 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
761 does not have array-element-type equal to the most general string
763 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
764 function name. (thanks to Nikodemus Siivola)
765 * fixed bug: compile-time format string checker failed on
766 non-closed ~{. (reported by Thomas F Burdick)
767 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
768 don't warn on legal loop code involving more than one aggregate
769 boolean. (thanks to Nikodemus Siivola)
770 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
771 in keyword position to MAKE-PACKAGE was not being reported as
773 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
774 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
776 * fixed bugs 280 and 312: the checking for multiple definitions in a
777 file is less likely to become confused by uses of inline
779 * fixed bug: the #S reader macro performs the keyword coercion
780 specified for slot names. (reported by Kalle Niemitalo)
781 * fixed bug: lambda lists may contain symbols whose names start with
782 & but are not lambda-list-keywords; their occurrence triggers a
784 * fixed bug 321: define-method-combination argument lambda lists do
785 not cause invalid code to be generated when &optional or &aux
786 variables have default values. (reported by Bruno Haible)
787 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
788 have CLOS classes; this appears also to have fixed reported
789 instability in stack exhaustion detection.
790 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
791 MUFFLE-WARNING when no associated restart is present is now
793 * optimization: rearranged the expansion of various defining macros
794 so that each expands into only one top-level form in a
795 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
797 * optimization: used a previously little-used slot in symbols to
798 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
800 * fixed some bugs revealed by Paul Dietz' test suite:
801 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
803 ** Fixed an optimization bug related to inheritance of initforms
804 from local to shared slots.
805 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
806 ** CAST optimizer forgot to flush argument derived type.
807 ** print/read consistency on floats is now orders of magnitude
808 more likely. (thanks also to Bruno Haible for a similar report
810 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
811 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
812 for objects of type REAL. Make it so.
813 ** SXHASH is specified (infelicitously) to respect similarity,
814 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
815 it so. (thanks to Markus Ziegler)
816 ** on the Alpha, the compiler succeeds in compiling functions
817 returning a known number of arguments greater than 63.
818 ** fixed handling of invalid NIL arguments in keyword position
820 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
822 ** provided a readably-printable representation for RANDOM-STATE
824 ** ensured that pathnames, where they have a namestring, always
825 print using #P"..." syntax.
827 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
828 * Support for the forthcoming 2.0 version of the NetBSD kernel
829 running on x86 hardware has been added. (thanks to Perry
830 E. Metzger most immediately, and others for their past work)
831 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
832 current release version; SBCL's OpenBSD support had been broken
833 since about the time of OpenBSD's switch to ELF binary format.)
834 * [placeholder for DX summary]
835 ** user code with &REST lists declared dynamic-extent, under high
836 speed or space and low safety and debug optimization policy.
837 * The manual has been converted to Texinfo format and the debugger
838 chapter from the cmucl manual has been added.
839 * A facility has been added to extract documentation strings from
840 sbcl and store them as Texinfo-formatted snippets for inclusion in
841 the manual (via Texinfo's @include directive)
842 * bug fix: compiler emitted division in optimized DEREF. (thanks for
843 the test case to Dave Roberts)
844 * bug fix: multidimensional simple arrays loaded from FASLs had fill
845 pointers. (reported by Sean Ross)
846 * bug fix: PROFILE output is printed nicely even for large numerical
847 values. (thanks to Zach Beane)
848 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
849 greater than 32 handle EOF correctly.
850 * bug fix: on X86 an immediate argument of the IMUL instruction is
851 correctly printed in disassembly. (thanks to Lutz Euler)
852 * bug fix: class slots in redefined classes preserve their old
853 values. (thanks to Bruno Haible and Nikodemus Siivola)
854 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
855 succeeds. (reported by Marco Baringer)
856 * fixed some bugs revealed by Paul Dietz' test suite:
857 ** READ-SEQUENCE now works on ECHO-STREAMs.
858 ** RATIONALIZE works more according to its specification. (thanks
861 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
862 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
863 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
864 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
865 should matter to you only if you rebind the printer control
866 variables and then find you want different bindings in the
867 debugger than in the ordinary execution of your program.)
868 * The runtime build system has been tweaked to support building
869 (on SPARC/SunOS) using a C compiler which invokes Sun's own
870 assembler and linker. (thanks to Nikodemus Siivola)
871 * Unbound, undefined, undeclared variables now trigger full
872 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
873 more often programmer error than deliberate exploitation of undefined
875 * optimization: the hash algorithm for strings has changed to one
876 that is less vulnerable to spurious collisions. (thanks to Juho
878 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
879 needless bounds checking. (thanks to Juho Snellman)
880 * optimization: implemented multiplication as a modular
881 (UNSIGNED-BYTE 32) operation on the PPC backend.
882 * fixed some bugs revealed by Paul Dietz' test suite:
883 ** ADJUST-ARRAY now copies the datum in a zero rank array if
885 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
886 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
887 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
889 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
891 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
893 ** ECHO-STREAMs no longer attempt to echo the end of file value to
894 their output stream on EOF from read.
895 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
896 have been read to end-of-file.
897 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
899 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
900 description of determination of which consecutive characters
902 ** Printing the "Space" character with escaping on now yields "#\\ ",
903 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
904 ** Reading floating-point numbers with *READ-BASE* set to a number
905 less than 10 works correctly.
906 ** Reading floating-point numbers with *READ-BASE* set to a number
907 more than 10 works correctly.
908 ** Printing with *PRINT-READABLY* targets the standard readtable, not
909 the readtable currently in effect.
911 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
912 * minor incompatible change: parsing of namestrings on a physical
913 (Unix) host has changed; numbers after the final #\. in a
914 namestring are no longer interpreted as a version field. This is
915 intented to be largely invisible to the user, except that the
916 meaning of the namestring "*.*.*" has changed: it now refers to a
917 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
918 should usually be replaced by
919 (make-pathname :name :wild :type :wild :version :wild)
920 with the added benefit that this is more likely to be portable.
921 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
922 option to OPEN now signals an error if the file being opened
923 exists; this may have an impact on existing code.
924 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
925 signals received, on the PowerPC platforms (both Linux and
926 Darwin). (thanks to Pierre Mai for pointing out the location of
928 * several fixes on OS X: The system now builds and runs cleanly on
929 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
930 (thanks to Brian Mastenbrook)
931 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
932 second-resolution: integer multiples of 1/3600 between -24 and 24.
933 (thanks to Vincent Arkesteijn)
934 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
935 type when called with 1 argument; PEEK-CHAR checked type of
936 PEEK-TYPE only after having read first character from a
937 stream. (reported by Peter Graves)
938 * bug fix: the garbage collector now has much better locality
939 behaviour, and in particular no longer treats all memory as being
940 exhausted when many small objects point to each other in a deeply
942 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
943 recognized as being TYPEP their class.
944 * bug fix: the PUSHNEW documentation string has been corrected.
945 (thanks to Vincent Arkesteijn)
946 * bug fix: defaulting of the value for the last of an atypically
947 large number of multiple values being bound was not being
948 performed correctly on the Alpha or PPC platforms
949 * optimization: implemented multiplication as a modular
950 (UNSIGNED-BYTE 32) operation on the x86 backend.
951 * optimization: SEARCH on simple-base-strings can now be open-coded.
952 (see also contrib/compiler-extras.lisp for inspiration for
953 teaching the compiler about the Boyer-Moore algorithm).
954 * value, returned by MAX (and MIN) called with several EQUALP, but
955 not EQL, arguments now does not depend on compiler settings.
956 * fixed some bugs revealed by Paul Dietz' test suite:
957 ** in stack analysis liveness information is propagated from
958 non-local entry points.
959 ** pathwise CAST removing failed when the CAST node did not start
961 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
962 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
964 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
965 host is already defined.
966 ** RENAME-FILE works on streams instead of signalling an internal
968 ** PEEK-CHAR uses the current readtable when determining whether
969 or not a character is whitespace.
970 ** MERGE-PATHNAMES handles the case when the pathname does not
971 specify a name while the default-pathname specifies a version.
972 ** Pathnames now stand a better chance of respecting print/read
974 ** Attempting to use standardized file system operators with a
975 pathname with invalid :DIRECTORY components signals a
977 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
979 ** DIRECTORY on logical pathnames is more correct.
980 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
981 signal a TYPE-ERROR if their argument is not a stream.
982 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
983 designator argument does not designate a stream.
984 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
985 examining the synonym.
986 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
988 ** OPEN and WITH-OPEN-STREAM allow opening streams with
989 element-type larger than ([UN]SIGNED-BYTE 32).
991 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
992 * When built with the :SB-FUTEX feature, threaded builds now take
993 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
994 for faster/more reliable mutex and condition variable support.
995 * Incompatible change (but one you probably shouldn't have been using
996 anyway): the interface and code for arbitrating between multiple
997 threads in the same user session has been redesigned.
998 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
999 error when called without an explicit environment argument.
1000 (thanks to Nikodemus Siivola)
1001 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1002 confused about their position. (thanks to Adam Warner and Gerd
1004 * bug fix: Pretty printing backquoted forms with unquotations in the
1005 argument list position of various code constructs such as LAMBDA
1006 now works correctly. (reported by Paul Dietz)
1007 * bug fix: Pretty printing unquotations no longer loses all
1008 stream position information.
1009 * optimization: performance of string output streams is now less
1010 poor for multiple small sequence writes.
1011 * optimization: performance of CSUBTYPEP in the presence of complex
1012 expressions involving CONS and NOT many times has been improved.
1013 (reported by Paul Dietz)
1014 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1016 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1017 streams for the REPL, for the debugger, and so on. (thanks to
1019 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1020 * fixed some bugs revealed by Paul Dietz' test suite:
1021 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1022 is ignored if it is too small, rather than propagating through
1023 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1024 ** extremely complex negations of CONS types were not being
1025 sufficiently canonicalized, leading to inconsistencies in
1027 ** VALUES tranformer lost derived type.
1029 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1030 * fixed a bootstrapping bug: the build process no longer assumes
1031 that the various BOOLE-related constants have the same value in
1032 host and target lisps. (noted by Paul Dietz' test suite on an
1033 SBCL binary built from CLISP)
1034 * The system can now be dynamically linked on the MIPS platform,
1035 which enables dynamic loading of foreign code from Lisp. (thanks
1036 to Ralf Baechle for discussions on the MIPS ABI)
1037 * The system now records debugging information for its own source
1038 files in a filesystem-position-independent manner, relative to
1039 the "SYS" logical host.
1040 * fixed a compiler bug: MV-LET convertion did not check references
1041 to the "max args" entry point. (reported by Brian Downing)
1042 * tweaked disassembly notes to be less confident about proclaiming
1043 some instruction as an LRA. (thanks to Brian Downing)
1044 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1045 now each have their own history, command character, and other
1046 characteristics. (thanks to David Lichteblau)
1047 * fixed some bugs revealed by Paul Dietz' test suite:
1048 ** compiler failure in compiling LOGAND expressions including a
1050 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1051 shift greater than 32.
1052 ** FLUSH-DEST did not mark blocks for type check regeneration.
1053 ** HANDLER-CASE failed to accept declarations in handler clauses
1054 in some circumstances.
1056 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1057 * New code in contrib/sb-introspect (still probably not entirely
1058 stable yet) provides some support for smart Lisp development
1059 environments like SLIME.
1060 * The conditions signalled for errors occurring when loading .fasl
1061 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1062 in a way which should help ASDF recover gracefully.
1063 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1064 stuff has been cleaned up. If you code contrib/ stuff, this might
1065 affect you, and you can look at contrib/README, contrib/STANDARDS,
1066 and/or the 0.8.4.27 diff to check.
1067 * In full calls the compiler now does not generate checks for declared
1068 argument types for all arguments.
1069 * various threading fixes
1070 ** and some experimental patches which didn't make it into
1071 the main tree for this release, but which are shipped in
1072 contrib/experimental-thread.patch as a possible fix for some
1073 failures (deadlock, spinning...) in GC-intensive multithreaded
1075 * fixed PPC build problem (source code incompatibility of different
1076 library versions): added offsetof() hackery which attempts to divine
1077 where glibc maintainers put uc_mcontext today
1078 * fixed bug 282: compiler does not trust type assertions while passing
1079 arguments to a full call.
1080 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1081 &OPTIONAL VALUES type parameter.
1082 * fix bug 214: algorithm for noting rejected templates is now more
1083 similar to that of template seletion. (also reported by rydis on
1085 * fixed bug 141b: printing backquoted information readably and prettily
1086 inserts a space where necessary.
1087 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1088 (CHARACTER 10) are now reported as errors, rather than propagated
1089 as unknown types. (reported by piso on #lisp)
1090 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1091 with values NIL and :ERROR. (thanks to Milan Zamazal)
1092 * fixed bug 191c: CLOS now does proper keyword argument checking as
1093 described in CLHS 7.6.5 and 7.6.5.1.
1094 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1095 longer raise an error; further, using a list as a for-as-arithmetic
1096 counter now raises a meaningful error.
1097 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1098 now understood by sequence creation functions such as MAKE-SEQUENCE
1100 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1101 a STRING-INPUT-STREAM.
1102 * compiler enhancement: SIGNUM is now better able to derive the type
1104 * type declarations inside WITH-SLOTS are checked. (reported by
1106 * fixed some bugs revealed by Paul Dietz' test suite:
1107 ** incorrect optimization of TRUNCATE for a positive first
1108 argument and negative second.
1109 ** compiler failure in let-convertion during flushing dead code.
1110 ** compiler failure while deriving type of TRUNCATE on an
1111 interval, containing 0.
1112 ** ASH of a negative bignum by a negative bignum count now returns
1114 ** intersection of CONS types now canonicalizes properly, fixing
1115 inconsistencies in SUBTYPEP.
1117 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1118 * incompatible change: The --disable-debugger command line
1119 option now clobbers the debugger at a more fundamental
1120 level, by redefining #'INVOKE-DEBUGGER instead of by
1121 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1122 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1123 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1124 enter the debugger REPL and then suffer recursive errors
1125 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1126 while under the new system, BREAK in --disable-debugger mode
1127 terminates the system just as an unhandled error would.
1128 * fixed compiler performance when processing loops with a step >1;
1129 * bug fix: DOCUMENTATION now retrieves generic function
1130 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1131 support has been systematized, and now supports the methods
1132 specified by ANSI, along with a default method and a method for
1133 slot documentation. (reported by Nathan Froyd)
1134 * bug fix: effective methods associated with a generic function are
1135 no longer cached over a change of that generic function's method
1136 combination. (reported by Andreas Fuchs)
1137 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1138 names a non-existent directory. (thanks to Andreas Fuchs)
1139 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1140 the PPC platform, lead to stack corruption; this has been fixed.
1141 (reported by Rainer Joswig)
1142 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1143 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1144 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1145 arithmetic in general, is now much more reliable on the PPC
1147 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1148 platform now returns the right answer.
1149 * optimization: restored some effective method precomputation in
1150 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1151 precomputation is now tunable.
1152 * optimization: compiler-internal data structure use has been
1153 reviewed, and changes have been made that should improve the
1154 performance of the compiler by about 20%.
1155 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1156 simple-base-strings and simple-bit-vectors is improved.
1157 * optimization: the optimization of 32-bit logical and arithmetic
1158 functions introduced in version 0.8.3 on the x86 has been
1159 implemented on the mips, ppc and sparc platforms; an
1160 implementation of the same facility, but for 64-bit arithmetic,
1161 has been added for the alpha.
1162 * microoptimization: the compiler is better able to make use of the
1163 x86 LEA instruction for multiplication by constants.
1164 * bug fix: in some situations compiler did not report usage of
1165 generic arithmetic in (SPEED 3) policy.
1166 * bug 145b fix: compiler used wrong type specifier while converting
1167 MEMBER-types to numeric.
1168 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1170 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1171 index argument now works properly on the Alpha platform.
1172 * bug fix: floating point exception treatment on the Alpha platform
1174 * bug fix: FILE-POSITION works much better on string input and
1175 output streams. (thanks to Nikodemus Siivola)
1176 * bug fix: many threading/garbage collection symptoms sorted.
1177 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1178 might be pseudo-atomic.
1179 * internal change: Stopping for GC is now done with signals not
1180 ptrace. GC is now done in whichever thread wanted it, instead of
1182 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1184 * bug fix: result form in DO is not contained in the implicit
1186 * incompatible change: ICR structure is changed; the value part of
1187 CONTINUATION is now called LVAR; corresponding functions are
1188 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1189 * added type deriver for ISQRT (thanks to Robert E. Brown).
1190 * bug fix: better support for loading from the command line when an
1191 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1193 * fixed some bugs revealed by Paul Dietz' test suite:
1194 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1195 ** ROUND and FROUND now give the right answer when given very
1196 small float arguments.
1197 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1199 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1200 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1201 ** LCM should return a non-negative integer.
1202 ** PARSE-INTEGER returned the index of a terminator instead of the
1203 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1204 ** PARSE-INTEGER returned an incorrect index being applied to a
1206 ** LCM with two arguments of 0 returns 0 rather than signalling
1208 ** unsigned addition of a 32-bit constant with the high bit set no
1209 longer causes an internal compiler error on the x86.
1210 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1211 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1212 with negative last argument.
1213 ** byte specifiers with zero size and position no longer cause
1214 an error during type derivation.
1215 ** bignum multiplication on the Alpha platform now returns the
1217 * porting: The system now builds on SuSE AMD64, although it still
1218 generates a 32-bit binary.
1219 * .fasl file incompatibility: The fasl file version number has
1220 been incremented (because of the changes to internal compiler
1221 data structures referred to above).
1223 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1224 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1225 more accurately, on the Darwin kernel running on PowerPC hardware.
1226 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1227 * Compiler code deletion notes now signal a condition of type
1228 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1229 an associated MUFFLE-WARNING restart.
1230 * The compiler now performs limited argument count validation of
1231 constant format strings in FORMAT, and where appropriate in ERROR,
1232 CERROR and WARN. (thanks to Gerd Moellmann)
1233 * New ASDF-INSTALL contrib can be used for automatic download and
1234 installation of third-party Lisp code from CCLAN or other sites
1236 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1237 which forces another thread to execute a function supplied by the
1239 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1240 now accept and act upon their :ELEMENT-TYPE keyword argument.
1241 (reported by Martin Atzmueller, Edi Weitz)
1242 * bug fix: FILE-POSITION now accepts position designators up to
1243 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1244 is the greater. (thanks to Patrik Nordebo)
1245 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1246 Kalle Olavi Niemitalo)
1247 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1248 printable. (reported by Eric Marsden)
1249 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1250 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1251 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1252 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1253 (thanks to Stig Sandoe)
1254 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1255 UNREAD-CHAR now correctly includes the unread character in the
1256 target sequence. (thanks to Gerd Moellmann)
1257 * bug fix in threaded builds: the system can now be suspended and
1258 resumed by shell job control with minimal disruption.
1259 * bug fixes in times and timezones >2038AD
1260 * better handling of "where is GNU make?" problem in build scripts
1261 (thanks to Nikodemus Siivola)
1262 * new optimization: inside a named function any reference to a
1263 function with the same name is considered to be a self-reference;
1264 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1265 optimization quality.
1266 * new optimization on x86: logical functions and + now have
1267 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1268 used when the result is truncated to 32 bits.
1269 * VALUES declaration is partially enabled.
1270 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1271 * fixed some bugs revealed by Paul Dietz' test suite:
1272 ** The system now obeys the constraint imposed by
1273 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1274 types form a lattice under type intersection.
1275 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1276 ** ASSOC now ignores NIL elements in an alist.
1277 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1278 and (1+ MOST-POSITIVE-FIXNUM) answers.
1279 ** The addition of a method with invalid qualifiers to a generic
1280 function does not cause an error to be signalled immediately;
1281 a warning is signalled, and the error is generated only on
1282 calling the generic function.
1283 * changed .fasl file version number, in part to add type codes for
1284 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1285 obscure ANSI requirements
1287 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1288 * fixed bug 148: failure to inline-expand a local function left
1289 garbage, confusing the compiler.
1290 * fixed bugs 3cd: structure slot readers perform type check if the
1291 slot can have an invalid value (i.e. it is either not initialized
1292 or can be written with a less specific slot writer).
1293 * bug fix: the compiler now traps array references to elements off
1294 the end of an array; previously, the bounds checking in some
1295 circumstances could go off-by-one.
1296 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1298 * type declarations for array element types now obey the description
1299 on the CLHS page "Declaration TYPE", as per discussions on
1300 sbcl-help around 2003-05-08. This means that a declaration
1301 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1302 declaration, all references to BAR will be asserted or assumed
1303 (with THE, so dependent on compiler policy) to involve objects of
1304 type FOO. Note that no such declaration is implied in
1305 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1306 * declared types of functions from the "Conditions"
1307 chapter. (reported by Paul Dietz)
1308 * bug fix: CERROR accepts a function as its first argument.
1309 * bug fix: NTH an NTHCDR accept a bignum as index
1310 arguments. (reported by Adam Warner)
1311 * optimization: character compare routines now optimize comparing
1312 against a constant character. (reported by Gilbert Baumann)
1313 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1314 argument now works properly on the MIPS platform.
1315 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1316 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1317 declared to have a functional type, does not cause an error
1319 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1320 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1321 * fixed bug: if last continuation of a deleted block has a
1322 destination, this destination should be deleted too. (reported by
1324 * fixed a bug in the bootstrap process: the host compiler's values
1325 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1326 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1327 test case from Patrik Nordebo)
1328 * improved the ability of the disassembler on the PPC platform to
1329 provide helpful disassembly notes.
1330 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1331 the class in more cases than previously.
1332 * bug fix: FILE-POSITION now understands :START and :END for
1333 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1334 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1335 8) data. (thanks to David Lichteblau)
1336 * bug fix: it is possible to add a method to a generic function
1337 without lambda list.
1338 * bug fix: reader failed to signal END-OF-FILE inside an
1339 object representation. (reported by Nikodemus Siivola)
1340 * fixed some bugs revealed by Paul Dietz' test suite:
1341 ** LAST and [N]BUTLAST should accept a bignum.
1342 ** condition slot accessors are methods.
1343 ** (VECTOR NIL) is a subtype of STRING.
1345 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1346 * minor incompatible change: some nonsensical specialized lambda
1347 lists (used in DEFMETHOD) which were previously ignored now signal
1349 * minor incompatible change: the system is now aware of the types of
1350 variables in the COMMON-LISP package, and will signal errors for
1351 most violations of these type constraints (where previously they
1352 were silently accepted).
1353 * minor incompatible change: COMPILE-FILE now uses the freedom
1354 afforded (ANSI 3.2.2.3) to use derived function types for
1355 functions defined in the same file. This also permits the system
1356 to warn on static type mismatches and function
1357 redefinition. (Currently it does not work with high DEBUG level.)
1358 * minor incompatible change: VALUES declaration is disabled.
1359 * When issuing notes, the compiler now signals a condition of type
1360 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1361 restart for use in user handlers. It is expected that the
1362 COMPILER-NOTE condition will eventually become a condition
1363 supertype to a hierarchy of note types, which will then be
1364 handleable in a similar fashion. However, at the moment, no such
1365 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1366 but it's an implementation detail, not a classification for the
1368 * Changes in type checking closed the following bugs:
1369 ** type checking of unused values (192b, 194d, 203);
1370 ** template selection based on unsafe type assertions (192c, 236);
1371 ** type checking in branches (194bc).
1372 * A short form of VALUES type specifier has ANSI meaning (it has
1373 increased the number of situations when SBCL cannot perform type
1375 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1376 keywords or constants is permissible.
1377 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1378 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1379 classes. (thanks to Antonio Martinez)
1380 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1381 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1382 argument to a pretty-print stream. Instead, it leaves any such
1383 operations to DESCRIBE-OBJECT methods.
1384 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1386 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1387 lambda lists. (thanks to David Lichteblau)
1388 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1389 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1390 (thanks to Gerd Moellmann)
1391 * fixed bug 246: increased compilation speed of long
1392 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1394 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1396 * DEFCONSTANT now throws a condition of type
1397 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1398 constant to a non-EQL value; CONTINUE and ABORT restarts
1399 respectively change and preserve the value.
1400 * fixed bug 63: The code walker, part of the implementation of CLOS,
1401 is now better at handling symbol macros.
1402 * bug fix: There is no longer an internal implementation type named
1403 CL:LENGTH. (reported by Raymond Toy)
1404 * bug fix: In macro-like defining macros/special operators the
1405 implicit block does not enclose the lambda list.
1406 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1407 enclosed in parentheses) are not suitable as type specifiers, and
1408 their use properly signals an error now.
1409 * bug fix: An argument count mismatch for a type specifier in code
1410 being compiled no longer causes an unhandled error at compile
1411 time, but signals a compile-time warning.
1412 * fixed simple vector readable printing
1413 * bug fix: DESCRIBE takes more care over whether the class
1414 precedence list slot of a class is bound before accessing it.
1415 (reported by Markus Krummenacker)
1416 * bug fix: FORMATTER can successfully compile pretty-printer format
1417 strings which use variants of the ~* directive inside.
1418 * bug fix: SEARCH now applies its TEST predicate to the elements of
1419 the arguments in the correct order. (thanks to Wolfhard Buss)
1420 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1421 in type inference. (thanks to Robert E. Brown)
1422 * bug fix: Reading in symbols with an explicit package name of ""
1423 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1424 * fixed some bugs revealed by Paul Dietz' test suite:
1425 ** NIL is now allowed as a structure slot name.
1426 ** Arbitrary numbers, not just REALs, are allowed in certain
1427 circumstances in LOOP for-as-arithmetic clauses.
1428 ** Multiple class redefinitions before slot access no longer
1429 causes a type error.
1430 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1431 association between the name and a class.
1432 ** Generic functions with non-standard method-combination and over
1433 five methods all of which return constants no longer return NIL
1434 after the first few invocations. (thanks to Gerd Moellmann)
1435 ** CALL-NEXT-METHOD with no arguments now passes the original
1436 values of the arguments, even in the presence of assignment.
1437 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1439 ** DISASSEMBLE works with closures and funcallable instances.
1440 ** ADD-METHOD now returns the generic function, not the new method.
1441 ** FIND-METHOD signals an error if the lengths of the specializers
1442 is incompatible with the generic function, even if the ERRORP
1444 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1445 which its argument is a member.
1446 ** DEFCLASS only redefines the class named by its class-name
1447 argument if that name is the proper name of the class;
1448 otherwise, it creates a new class.
1449 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1450 of the UNBOUND-SLOT condition to the name of the slot.
1451 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1452 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1453 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1454 SLOT-MAKUNBOUND in the specified fashion.
1456 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1457 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1458 cross-compilation host. As a consequence, we can now bootstrap our
1459 way up to SBCL starting with a bare gcc toolchain and human-readable
1460 source code (first the source to CLISP, then the source to SBCL).
1461 * A contributed module containing a partial implementation of the
1462 simple-streams interface has been included. (thanks to Rudi
1464 * A contributed module implementing the RFC1321 Message Digest
1465 Algorithm, known as MD5, has been included.
1466 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1467 no longer has any effect, as the code controlled by this feature
1468 has been deleted. (As far as we know, no-one has ever built using
1469 this feature, and its semantics were confused in any case).
1470 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1471 work on conditions (as required by the ANSI specification),
1472 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1473 expected behaviour on conditions. Users should note, however,
1474 that such behaviour is not required by the ANSI specification,
1475 and so use of this behaviour may render their code unportable.
1476 * fixed some bugs revealed by Paul Dietz' test suite:
1477 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1479 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1480 ** SXHASH on condition objects no longer returns NIL.
1481 ** :ALLOCATION :CLASS slots are better treated; their values are
1482 updated on class redefinition, and initforms inherited from
1483 superclasses are applied.
1484 ** REMOVE-METHOD returns its generic function argument even when
1485 no method was removed.
1486 ** SHARED-INITIALIZE now initializes the values of the requested
1487 slots, including those with :ALLOCATION :CLASS.
1488 ** ALLOCATE-INSTANCE now works on structure classes defined via
1489 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1491 ** SLOT-EXISTS-P now works on conditions, as well as structures
1493 ** MAKE-LOAD-FORM now has the required methods on
1494 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1495 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1496 keyword, and now implements the SLOT-NAMES argument.
1497 ** methods with &OPTIONAL arguments no longer allow too many
1498 arguments to be passed in the call without error.
1499 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1500 option is consistent with the required arguments of the generic
1501 function lambda list.
1502 * bug fix: REQUIRE accepts a string designator. (Thanks to
1504 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1505 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1506 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1507 * bug fix: The compiler now checks for duplicated variables in macro
1509 * bug fix: SETQ on globals returns the correct value.
1510 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1511 not a COMPILER-ERROR (followed by some other strange error on
1512 choosing the CONTINUE restart).
1513 * bug fix: make.sh and friends are now more consistent in the way that
1514 they look for GNU "make".
1516 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1517 * experimental native threads support (on x86 Linux >=2.4 only).
1518 This is not compiled in by default: you need to add :SB-THREAD to
1519 the target features. See the "Beyond ANSI" chapter of the manual
1521 * fix for longstanding nonANSIism: The old distinction between
1522 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1523 The return value from CL:FIND-CLASS is now a CLOS class, and
1524 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1525 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1526 * An interface to the MetaObject Protocol, as described in Kiczales,
1527 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1528 MIT Press, 1991, is available from the SB-MOP package.
1529 * incompatible change: the SB-PCL package should now be considered
1530 a private implementation detail, and no longer a semi-private MOP
1532 * minor incompatible change: due to rearrangement for threads, the
1533 control stack and binding stack are now allocated at arbitrary
1534 addresses instead of being hardcoded per-port. Users affected by
1535 this probably have to be doing advanced things with shared
1536 libraries, and will know who they are.
1537 * minor incompatible change: Previously, all --eval forms used were
1538 processed with READ before any of them were processed with EVAL.
1539 Now each --eval form is processed with both READ and EVAL before
1540 the next --eval form is processed. (Thus package operations like
1541 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1542 work as the user might reasonably expect.)
1543 * minor incompatible change: *STANDARD-INPUT* is now only an
1544 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1546 * minor incompatible change: Y-OR-N-P is now character-oriented,
1547 not line oriented. Also, YES-OR-NO-P now works without errors.
1548 (thanks to Antonio Martinez)
1549 * sb-aclrepl module improvements: an integrated inspector, added
1550 repl features, and a bug fix to :trace command.
1551 * Known functions, which cannot be open coded by the backend, are
1552 considered to be able to check types of their arguments. (fixing
1553 a bug report by Nathan J. Froyd)
1554 * fixed a bug in computing method discriminating functions: It is
1555 now possible to define methods specialized on classes which have
1556 forward-referenced superclasses. (thanks to Gerd Moellmann)
1557 * fixed evaluation order in optional entries (reported by Gilbert
1559 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1560 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1561 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1562 required-by-AMOP NAME argument, as well as CLASS and
1563 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1564 * fixed bug 20: DEFMETHOD can define methods using names that are
1565 not the proper names of classes to designate class specializers.
1566 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1567 Unix file descriptors, instead of blowing up. (thanks to Antonio
1569 * Garbage collection refactoring: user-visible change is that a
1570 call to the GC function during WITHOUT-GCING will not do garbage
1571 collection until the end of the WITHOUT-GCING. If you were doing
1572 this you were probably losing anyway.
1573 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1574 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1575 TYPEP the latter but not the former.
1576 * The compiler issues a full WARNING for calls to undefined functions
1577 with names from the CL package.
1578 * MAP-INTO for a vector destination is open coded. (reported by
1579 Brian Downing on c.l.l)
1580 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1581 documentation string.
1582 * fixed some bugs revealed by Paul Dietz' test suite:
1583 ** COPY-ALIST now signals an error if its argument is a dotted
1585 ** Condition slots are now accessed more correctly in the presence
1586 of multiple initargs for a given slot.
1587 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1588 exclude restarts of the same name associated with a different
1590 ** DEFCLASS of forward-referenced classes with another
1591 forward-referenced class in the superclasses list no longer
1593 ** Condition slots are now initialized once each, not multiple
1594 times. (thanks to Gerd Moellmann)
1595 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1597 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1598 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1599 arguments contain duplicated elements.
1600 ** RESTART-CASE understands local macros.
1601 ** RESTART-CASE associates exactly its own restarts with a condition.
1602 ** ENDP in safe mode checks its argument to be of type LIST.
1603 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1604 in question is unbound.
1605 ** Optimized MAKE-INSTANCE functions no longer cause internal
1606 assertion failures in the presence of duplicate initargs.
1607 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1608 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1609 propagated even in the ctor optimized implementation of
1611 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1613 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1614 * a better implementation of SXHASH on (simple) bit vectors,
1615 measured both in execution speed and in distribution of results
1616 over the positive fixnums, has been installed. Likewise, a better
1617 implementation of EQUAL for simple bit vectors is now available.
1618 * fixed CEILING optimization for a divisor of form 2^k.
1619 * fixed bug 240 (emitting extra style warnings "using the lexical
1620 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1621 by Antonio Martinez)
1622 * fixed SXHASH, giving different results for NIL depending on type
1623 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1624 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1625 variable. (found by Rolf Wester)
1626 * SBCL does not ignore type declarations for special
1627 variables. (reported by rif on c.l.l 2003-03-05)
1628 * some bug fixes in contrib/sb-aclrepl/
1629 * fixed some bugs revealed by Paul Dietz' test suite:
1630 ** a bug in the CONS type specifier, whereby the CAR and CDR
1631 types got intertwined, has been fixed;
1632 ** the type system is now able to reason about the interaction
1633 between INTEGER and RATIO types more completely;
1634 ** APPEND, [N]REVERSE and NRECONC check that those their
1635 arguments, which must be proper lists, are really so;
1636 ** An array specialized to be unable to hold elements has been
1637 implemented, as required -- yes, really -- by ANSI;
1638 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1639 on malformed property lists;
1641 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1642 * incompatible packaging change: in line with Unix convention,
1643 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1644 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1645 * REQUIRE and PROVIDE are now optionally capable of doing something
1646 useful. See the documentation string for REQUIRE.
1647 * infrastructure for a managed SBCL contrib system: contributed
1648 modules in this release include:
1649 ** the ASDF system definition facility;
1650 ** an interface to the BSD Sockets API;
1651 ** an ACL-like convenience interface to the repl;
1652 (thanks to Kevin Rosenberg)
1653 ** an implementation of ROTATE-BYTE, with efficient implementation
1655 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1656 gives a closer approximation to the right answer than previously.
1657 (thanks to Raymond Toy)
1658 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1659 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1660 optional environment arguments, as required by ANSI.
1661 * fixed bugs in other functions taking environment objects, allowing
1662 calls with an explicit NIL environment argument to be compiled
1664 * fixed bug 228: primary return values from
1665 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1666 COMPILE or FUNCTION.
1667 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1668 :TYPE will no longer signal errors on innocuous objects.
1669 * fixed bug 231b: SETQ is better at respecting type declarations in
1670 the lexical environment.
1671 * fixed a bug in DEFCLASS: classes named by symbols with no or
1672 unprintable packages can now be defined.
1673 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1674 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1675 * fixed compiler failure related to checking types of functions.
1676 (reported by Robert E. Brown)
1677 * the compiler is now much more consistent in its error-checking
1678 treatment of bounding index arguments to sequence functions: in
1679 (SAFETY 3) code, errors will be signalled in almost all cases if
1680 invalid sequence bounding indices are passed to functions defined
1681 by ANSI to operate on sequences.
1682 * fixed a bug in the build procedure: documentation of SBCL-specific
1683 packages is now preserved and available in the final Lisp image.
1684 * lifted FDEFINITION lookup out of loops in the implementation of
1685 many list operations. (thanks to Robert E. Brown)
1686 * fixed a bug in the reader: the #n# reader macro now works for
1687 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1688 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1689 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1690 E. Brown and Tony Martinez respectively)
1691 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1692 SLOT-BOUNDP: in particular, it is now able to optimize them much
1693 better, and is now not vulnerable to having packages renamed.
1694 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1695 always called when a slot is not present in an instance. (thanks
1697 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1698 longer signal an error on unreading a character following EOF on
1699 the previous constituent stream. (thanks to Tony Martinez)
1700 * fixed some bugs revealed by Paul Dietz' test suite:
1701 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1702 not just nonnegative fixnums;
1703 ** the logical bit-array operators such as BIT-AND now accept an
1704 explicit NIL for their "opt-arg" argument (to indicate a
1705 freshly-consed result bit-array);
1706 ** ELT now signals an error on an invalid sequence index in safe
1708 ** the type system is now cleverer about negations of numeric
1709 types, and consequently understands the BIGNUM and RATIO types
1711 ** the type system is now cleverer about the interaction between
1712 INTEGER and RATIO types: while bugs still remain, many more
1713 cases are accurately computed;
1714 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1715 if it is in the last clause;
1716 ** CONSTANTLY now correctly returns a side-effect-free function in
1718 ** DECLARE is no longer treated as a special-operator; in
1719 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1720 * incremented fasl file version number due to the change in the
1721 DEFSTRUCT-SLOT-DESCRIPTION structure.
1723 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1724 * minor incompatible change: code processed by the "interpreter" or
1725 EVAL now has a compilation optimization policy of (DEBUG 2)
1726 (changed from (DEBUG 1)) to improve debuggability of interactive
1727 development, and to allow the use of the debug RETURN command in
1729 * an experimental implementation of the RETURN command for the
1730 debugger has been included. (thanks to Frederik Kuivinen)
1731 * fixed bug 62: constraints were not propagated into a loop.
1732 * fixed bug in embedded calls of SORT (reported and investigated by
1734 * fixed some bugs revealed by Paul Dietz' test suite:
1735 ** printing and reading of arrays with some dimensions having
1736 length 0 (thanks to Gerd Moellmann);
1737 ** BOA constructor with &AUX argument without a default value does
1738 not cause a type error;
1739 ** CONSTANTP now returns true for all self-evaluating objects.
1741 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1742 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1743 accessors that are related by inheritance, as specified in the
1744 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1745 Valtteri Vuorikoski)
1746 * The compiler is now able to inline functions that were defined in
1747 a complex lexical environment (e.g. inside a MACROLET).
1748 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1750 * Support for the upcoming FreeBSD-5.0 release has been included.
1751 (thanks to Dag-Erling Smorgrav)
1752 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1753 effect when it is not in a toplevel context.
1754 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1755 stand a better chance of being correct. (thanks to Gerd
1757 * fixed bug in COERCE, which now signals an error on coercing a
1758 rational to a bounded real type which excludes the expected
1760 * The compiler is now able to derive types more accurately from the
1761 COERCE and COMPILE functions.
1762 * fixed bug 223: functional binding is considered to be constant
1763 only for symbols in the CL package.
1764 * fixed bug 231: SETQ did not check the type of a variable being set
1765 (reported by Robert E. Brown)
1766 * A new optimization for MAKE-INSTANCE has been included, fixing
1767 various bugs (including relating to :ALLOCATION :CLASS slots and
1768 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1770 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1771 ** As required by ANSI, LOOP now disallows anonymous collection
1772 clauses such as COLLECT I in conjunction with aggregate boolean
1773 clauses such as THEREIS (= I 1);
1774 ** LOOP now signals an error when any variable is reused in the
1775 same loop (including the potentially useful construct analogous
1776 to WITH A = 1 WITH A = (1+ A);
1777 ** IT is only a special loop symbol within the first clause of a
1778 conditional loop clause;
1779 ** LOOP with a typed iteration variable over a hashtable now
1780 signals a type error iff it should.
1781 * fixed some other bugs revealed by Paul Dietz' test suite:
1782 ** FILE-STREAM now names the class previously known as FD-STREAM;
1783 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1784 argument) no longer signals an error;
1785 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1786 of the slot symbol, rather than using the current package
1787 ((:CONC-NAME "") continues to intern the slot's name in the
1789 * incremented fasl file version number, because of the incompatible
1790 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1791 of the new implementation of DEFINE-COMPILER-MACRO.
1793 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1794 * Support for building SBCL for MIPS platforms running in
1795 little-endian mode has now been checked in, and basic
1796 functionality on said platforms verified.
1797 * minor incompatible change: PCL now records the pathname of a file
1798 in which methods and the like are defined, rather than its
1800 * minor incompatible change: TRUENAME now considers the truename of
1801 a file naming a directory to be the pathname with :DIRECTORY
1802 component indicating that directory.
1803 * minor incompatible change: a NAMED clause in the extended form of
1804 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1805 reason for the previous behaviour is unclear.
1806 * more systematization and improvement of CLOS and MOP conformance
1807 in PCL (thanks to Gerd Moellman and Pierre Mai):
1808 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1810 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1811 primary methods with no specializers;
1812 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1814 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1815 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1816 FORWARD-REFERENCED-CLASSes; error reporting on
1817 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1819 ** SXHASH on CLOS instances now uses a slot internal to the
1820 instance to return different numbers on distinct instances,
1821 while preserving the same return value through invocations of
1823 ** DEFMETHOD signals errors when methods with longer incongruent
1824 lambda lists are added to generic functions;
1825 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1826 CLASS, as specified in AMOP;
1827 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1828 on the order returned by the primary method for classes of
1829 class STANDARD-CLASS;
1830 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1831 * fixed some bugs shown by Paul Dietz' test suite:
1832 ** DOLIST puts its body in TAGBODY;
1833 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1835 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1836 value producing form;
1837 ** if more variables are given to PROGV than values, extra
1838 variables are bound and made to have no value;
1839 ** NSUBSTITUTE on list arguments gets the right answer with
1841 ** ELT signals an error of type TYPE-ERROR when the index argument
1842 is not a valid sequence index;
1843 ** LOOP signals (at macroexpansion time) an error of type
1844 PROGRAM-ERROR when duplicate variable names are found;
1845 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1846 ** FUNCALL of special-operators now cause an error of type
1848 ** PSETQ now works as required in the presence of side-effecting
1849 symbol-macro places;
1850 ** NCONC accepts any object as its last argument;
1851 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1853 ** loop-for-as-package does not require a package to be explicitely
1855 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1857 * fixed bug 166: compiler preserves "there is a way to go"
1858 invariant when deleting code.
1859 * fixed bug 172: macro lambda lists with required arguments after
1860 &REST arguments now cause an error to be signalled. (thanks to
1862 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1863 bug. (thanks to Matthew Danish)
1864 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1866 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1867 arguments are lexically rebound. (thanks to Gerd Moellmann and
1869 * fixed bug 194: error messages are now more informative when there
1870 is no primary method applicable in a call to a generic
1871 function. (thanks to Gerd Moellmann)
1872 * fixed bug in command line argument checking (thanks to Julian
1874 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1875 (thanks to Matthew Danish)
1876 * incremented fasl file version number, because of the
1877 SXHASH-related changes in the layout of CLOS data structures
1879 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1880 * minor incompatible change: The runtime (the Unix executable named
1881 "sbcl") is now much pickier about the .core files it will load.
1882 Essentially it now requires .core files to descend from the same
1883 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1884 as the runtime does. (The intent is to prevent the crashes which
1885 can occur, and which can even be reported as mysterious failures,
1886 when people patch the sources or change the build parameters
1887 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1888 sbcl and .core files.)
1889 * fixed bug: VALUES-LIST is no longer optimized away.
1890 * fixed bug 142: The FFI conversion of C string values to Lisp
1891 string values no longer conses excessively. (thanks to Nathan
1892 Froyd porting Raymond Toy's fix to CMU CL)
1893 * began to systematize and improve MOP conformance in PCL (thanks to
1894 Nathan Froyd, Gerd Moellman and Pierre Mai):
1895 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1897 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1898 ** FINALIZE-INHERITANCE is now called on class finalization;
1899 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1900 argument precedence order.
1901 * fixed bug 202: The compiler no longer fails on functions whose
1902 derived types contradict their declared type.
1903 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1904 so it can be non-toplevel.
1905 * The fasl file version number has changed (because of the new
1906 implementation of DEFMACRO).
1907 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1908 safe code, that any length requirement by their type-specifier
1909 argument is valid. The exceptions to this are described in bug
1911 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1912 functions now check the types of their inputs as required by ANSI.
1913 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1914 introduced symbol is DECLAREd to be SPECIAL.
1915 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1916 * fixed bug: :COUNT argument to sequence functions may be negative
1917 * fixed bug: body of DO-SYMBOLS may contain declarations
1918 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1919 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1920 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1921 symbol macro only once
1922 * fixed printing of call frame when argument list is unavailable
1923 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1924 * compiler no longer signals WARNING on unknown keyword
1927 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1928 * A beta-quality port to the mips architecture running Linux,
1929 based on the old CMUCL backend, has been made. It has been tested
1930 on a big-endian kernel, and works sufficiently well to be able to
1931 rebuild itself; it has not been tested in little-endian mode.
1932 * fixed an inconsistency between gencgc.c and purify.c which made
1933 dumping/loading .core files unreliable
1934 * fixed bug 120a: The compiler now deals correctly with IFs where
1935 the consequent is the same as the alternative, instead of
1936 misderiving the return type. (thanks to Alexey Dejneka)
1937 * fixed bug 113: Logical pathnames are now dumpable (the logical
1938 host is resolved at load-time, throwing an error if it is not
1940 * fixed bug 174: FORMAT's error message is slightly clearer when a
1941 non-printing character is used in a format directive.
1942 * fixed several bugs in compiler checking of type declarations, i.e.
1943 violations of the Python "declarations are assertions" principle
1944 (thanks to Alexey Dejneka)
1945 * fixed several bugs in PCL's error checking (thanks to Gerd
1947 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1949 * fixed bug in compilation of functions as first class values
1950 (thanks to Antonio Martinez-Shotton)
1951 * The compiler's handling TYPE-ERRORs which it can prove will
1952 inevitably happen at runtime has been cleaned up and corrected
1953 in several ways. (thanks to Alexey Dejneka)
1954 * improved argument type checking for various basic arithmetic
1955 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1956 lavished on them in the past that they can be compiled in many
1957 ways in different special cases
1958 * fixed bug 181: compiler checks validity of user supplied type
1960 * cleaned up code flushing in optimization: Function calls which
1961 should signal errors for safety purposes (e.g. which ANSI says
1962 should signal errors when their arguments are of incorrect type)
1963 are no longer optimized away.
1964 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1965 * incremented fasl file version number, because changes in the
1966 implementation of sequence functions like COERCE caused
1967 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1968 in old inline expansions) to become undefined. (Actually these
1969 changes were later undone, so we might very well be binary
1970 compatible with 0.7.7 after all, but leaving the version number
1971 incremented seemed like the simplest and most conservative
1974 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1975 * An alpha-quality port to the parisc architecture running Linux,
1976 based on the old CMUCL backend, has been made. This, even more so
1977 than the other backends, should be considered still a work in
1978 progress; known problems include that the Linux kernel in 64-bit
1979 mode does not propagate the correct sigcontext structure to
1980 userspace, and consequently SBCL on a parisc64 kernel will not
1982 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1983 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1984 suppressed.) Also now that the compiler is looking at declarations
1985 in the environment, it checks optimization declarations as well,
1986 and suppresses inlining when (> DEBUG SPEED).
1987 * More fixes have been made to treatment of floating point exception
1988 treatment and other Unix signals. In particular, floating point
1989 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1990 * The detection and handling of control stack exhaustion (infinite
1991 or very deeply nested recursion) has changed. Stack exhaustion
1992 detection is now done by write-protecting pages at the OS level
1993 and applies at all optimization settings; when found, a
1994 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1995 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1996 be caught using IGNORE-ERRORS.
1997 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1998 correctly when transforming recursive LABELS functions to LETs.
1999 (thanks to Alexey Dejneka)
2000 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2001 that are names of constants or global variables.
2002 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2003 alien routines with docstrings.
2004 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2005 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2007 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2008 (thanks to Eric Marsden)
2009 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2010 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2011 * Bug fix: The SPARC backend can now compile functions involving
2012 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2013 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2014 to a functions expecting a PATHNAME-DESIGNATOR.
2015 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2016 lambda lists. (thanks to Alexey Dejneka)
2017 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2018 functions so that EXPORTing the name of the class doesn't cause
2019 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2020 to Antonio Martinez for reporting this)
2021 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2022 constant-folded without failing an assertion. (thanks to Einar
2023 Floystad Dorum for reporting this)
2024 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2025 types are now dealt with more correctly by the compiler.
2026 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2027 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2028 behaviour analogous to RENAME-FILE. This puts its behaviour more
2029 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2031 * The fasl file version number has changed again. (because of the
2032 bug fix involving the names of PCL MAKE-INSTANCE functions)
2034 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2035 * bug fix: Floating point exceptions are treated much more
2036 consistently on the x86/Linux and PPC/Linux platforms.
2037 * Array initialization with :INITIAL-ELEMENT is now much faster for
2038 cases when the compiler cannot open code the array creation, but
2039 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2040 array accesses have also seen a speed increase.
2041 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2042 specified. (This was at the root of some bad interactions between
2043 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2045 * bug fix: Internal error arguments for undefined functions are now
2046 computed correctly on the PPC/Linux platform.
2047 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2048 Raymond Toy's patch for CMU CL)
2049 * Support for the Solaris 9 operating environment has been included
2050 (thanks to Daniel Merritt)
2051 * A very ugly but hopefully complete draft of the missing FFI chapter
2052 of the manual has been created by reformatting the corresponding
2053 CMU CL manual chapter into (currently very ugly and incoherent)
2054 DocBook and bringing it up to date for SBCL behavior. Thus, the
2055 manual is now essentially complete, at least by my extreme
2056 once-and-only-once standards, whereby it's acceptable to refer to
2057 the doc strings of SB-EXT functions as the primary documentation.
2058 * The fasl file version number has changed again, due to cleanup of
2059 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2060 is no longer a static symbol.)
2062 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2063 * SBCL now builds with OpenMCL (version 0.12) as the
2064 cross-compilation host; also, more progress has been made toward
2065 bootstrapping under CLISP.
2066 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2068 * bug 158 fixed: The compiler can now deal with integer loop
2069 increments different from 1; fixing this turned out also to fix
2071 * bug 169 fixed: no more bogus warnings about using lexical bindings
2072 despite the presence of perfectly good SPECIAL declarations (thanks
2073 to David Lichteblau)
2074 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2075 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2076 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2078 * bug 180 fixed: Method combination specifications no longer ignore
2079 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2080 * bug fix: Structure type predicate functions now check their argument
2081 count as they should.
2082 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2083 correctly. (thanks to Pierre Mai)
2084 * minor incompatible change: The --noprogrammer option is deprecated
2085 in favor of the new --disable-debugger option, which is very similar.
2086 (The major difference is that it takes effect at a slightly different
2087 time at startup, causing handling of errors in --sysinit and
2088 --userinit files will be affected differently.) The
2089 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2090 been added to allow this functionality to be controlled from ordinary
2091 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2092 maintainers, who might want to run non-interactive scripts to
2093 build SBCL cores which will later be used interactively.)
2094 * minor incompatible change: The LOAD function no longer, when given
2095 a wild pathname to load, loads all files matching that pathname.
2096 Instead, an error of type FILE-ERROR is signalled.
2098 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2099 * bug 147 fixed: The compiler preserves its block link/count
2100 invariants more correctly now so that it doesn't crash. (thanks
2102 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2104 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2105 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2106 no-value-for-this-character as NIL (as specified by ANSI).
2107 * HOST-NAMESTRING on physical pathnames now returns a string that is
2108 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2109 (thanks to Christophe Rhodes)
2110 * The Alpha port handles icache flushing more correctly. (thanks to
2112 * More progress has been made toward bootstrapping under CLISP. (thanks
2113 to Christophe Rhodes)
2114 * The fasl file format has changed again, because dynamic loading
2115 on OpenBSD (which has non-ELF object files) motivated some cleanups
2116 in the way that foreign symbols are transformed and passed around.
2117 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2118 is no longer treated as whitespace by the reader, but instead as
2119 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2120 |A
\7fB|, instead of A as it used to.
2122 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2123 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2124 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2125 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2126 this as of 0.7.1.45, but was left out of the previous news section
2127 (thanks to Dan Barlow)
2128 * SBCL now runs on the Solaris operating system on SPARC architectures
2129 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2130 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2131 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2132 * SPARC backend cleanups, allowing builds of cores optimized for V8
2133 and V9 SPARCS, and also emission of code targeted to a particular
2134 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2136 * SBCL is closer to bootstrapping under CLISP, thanks to various
2137 fixes by Christophe Rhodes.
2138 * The fasl file format has changed again, to allow the compiler's
2139 INFO database to support symbol macros.
2140 * The user manual (in doc/) is formatted into HTML more nicely.
2141 (thanks to coreythomas)
2142 * The system is smarter about SUBTYPEP relationships, especially
2143 those involving NOT types (including types such as ATOM which are
2144 represented internally using NOT types). Thus SUBTYPEP is less
2145 likely to return (VALUES NIL NIL) in general, and in particular
2146 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2148 * The fasl file format has changed again, because the internal
2149 representation of types now includes a new slot to support the new
2150 SUBTYPEP-of-NOT-types logic.
2151 * (not a change in the main branch of SBCL, but a related prototype
2152 which can hopefully be merged into the main branch of SBCL in the
2153 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2154 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2155 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2156 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2157 dispatch macro character combinations. (thanks to Alexey Dejneka)
2158 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2159 that superficially look like logical namestrings correctly.
2160 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2162 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2163 default supplied-p) for &optional and &key arguments. (thanks to
2166 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2167 * incompatible change: The compiler is now less aggressive about
2168 tail call optimization, doing it only when (> SPACE DEBUG) or
2169 (> SPEED DEBUG). (This is an incompatible change because there are
2170 programs which relied on the old CMU-CL-style behavior to optimize
2171 away their unbounded recursion which will now die of stack overflow.)
2172 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2173 for non-GENCGC systems has been increased to 20M (since that
2174 seems much closer to the likely performance optimum for modern
2175 systems than the old 4M value was)
2176 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2177 and *DEBUG-PRINT-LEVEL*
2178 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2179 of CMU CL's support for SPARC, and various endianness and other
2180 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2181 * new syntactic sugar for the Unix command line: --load foo.bar is now
2182 an alternate notation for --eval '(load "foo.bar")'.
2184 ** The system now detects stack overflow and handles it gracefully,
2185 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2186 optimization settings. (This is a good thing in general, and
2187 its introduction in this version should be particularly timely
2188 for anyone whose code fails because of suppression of tail
2190 ** The system now hunts for the C variable "environ" in a more
2191 devious way, to avoid segfaults when the C library version differs
2192 between compile time and run time. (thanks to Christophe Rhodes)
2193 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2194 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2195 ** The compiler no longer issues bogus style warnings for undefined
2196 classes in the same source file as the DEFCLASSes which defined
2197 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2199 ** fixes in CONDITION class precedence list for undefined function
2200 errors (thanks to Alexei Dejneka)
2201 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2202 correctly. (thanks to Dan Barlow)
2203 ** portability fixes aiming at bootstrapping under CLISP (thanks
2204 to Dave McDonald and Christophe Rhodes)
2205 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2206 ** fixes in type translation and and type inference (thanks to
2208 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2209 ** various fixes in the new ports (thanks to Dan Barlow)
2210 * several changes related to debugging:
2211 ** suppression of tail recursion, as noted above
2212 ** stack overflow detection, as noted above
2213 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2214 is now the default. (For some time encapsulation has been more
2215 reliable than the breakpoint-based :ENCAPSULATE NIL
2216 implementation, at least on X86 systems; and I just noticed that
2217 encapsulation also seems closer to the spirit of the ANSI
2220 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2222 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2223 up properly again. (There was a packaging bug in 0.7.0 which
2224 left their definitions in SB-SYS::LOAD-FOREIGN and
2225 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2226 vital for most things which interface to C-level interfaces,
2227 like extensions working with sockets or databases or
2228 Perl-compatible regexes or whatever, and the need to fix
2229 this bug is the main reason that 0.7.1 was released so
2231 ** DEFGENERIC is now choosier about the methods it redefines, so that
2232 reLOADing a previously-LOADed file containing DEFGENERICs does
2233 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2234 cycle now works as it should. (thanks to Alexey Dejneka)
2235 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2236 fixed by Christophe Rhodes. (He actually submitted this patch
2237 months ago, and I delayed until after 0.7.0.)
2238 ** Bug 111 (internal compiler confusion about runtime checks on
2239 FUNCTION types) was fixed by Alexey Dejneka.
2240 * Some internal cleanups (getting rid of variables which aren't
2241 needed now that the byte interpreter is gone) caused the fasl
2242 file format number to change again.
2244 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2245 * major incompatible change: The default fasl file extension, i.e. the
2246 default extension for files produced by COMPILE-FILE, has changed
2247 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2249 ** There are many changes in the implementation of the compiler.
2250 SBCL is now essentially a compiler-only implementation of ANSI
2251 Common Lisp. EVAL still "interprets" a few special cases, but
2252 almost all the interesting cases are handled by creating
2253 a LAMBDA expression, calling COMPILE on it, then calling
2254 FUNCALL on the result.
2255 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2256 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2257 Since the code is newer, there might still be some new bugs
2258 (though not as many as before Martin Atzmueller's fixes:-). But
2259 the new code is substantially simpler and clearer, and hopefully
2260 any remaining bugs will be simpler, less fundamental, and more
2261 fixable then the bugs in the old code.
2262 ** The revised compiler is still a little unsteady on its feet.
2264 *** The debugging information it produces (particularly the names
2265 of FUNCTION objects) is sometimes much less useful than what
2266 the old compiler produced.
2267 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2268 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2269 within a MACROLET) has been temporarily weakened.
2270 ** There are new compiler optimizations for various functions:
2271 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2272 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2273 *** the math functions TRUNCATE, FLOOR, and CEILING
2274 *** the function-of-all-trades COERCE
2275 Mostly these should be transparent, but there's one
2276 potentially-annoying problem (bug 117): when the compiler
2277 inline-expands a function and does type analysis on the result,
2278 it can create control paths which have type mismatches, and
2279 when it can't prove that those control paths aren't taken,
2280 it will issue WARNINGs about the type mismatches. This is
2281 a particular problem in practice for the new sequence functions.
2282 It's not clear how this should be fixed, and for now, a
2283 workaround is given in the entry for 117 in the BUGS file.
2284 ** (Because of the interaction between the two previous items --
2285 occasional inlining problems and new inline expansions -- some
2286 of the new sequence function optimizations won't really kick in
2287 completely until debugging information, and then inlining, are
2288 straightened out in some future version.)
2289 * minor incompatible changes:
2290 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2291 DIRECTORY no longer implicitly promotes NIL slots of its
2292 pathname argument to :WILD. In particular, when you ask for the
2293 contents of a directory (which you used to be able to do without
2294 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2295 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2296 ** changes in behavior that ANSI explicitly defines to be
2297 implementation dependent:
2298 *** The new compiler-only implementation still conforms with ANSI,
2299 but acts a little different than before. Besides the obvious
2300 changes in performance tradeoffs (that the cost per form passed
2301 to EVAL has gone up, and the cost per form executed by EVAL
2302 has gone down), the behavior of the system changes a little
2303 because there are no longer any interpreted function objects.
2304 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2305 e.g. doing COMPILE on the output of interactive DEFUN is
2307 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2309 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2310 has changed from (:CL) to NIL.
2311 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2312 e.g. #\Newline and #\Tab, don't have names specified in the
2313 ANSI Common Lisp standard, is now based on their ASCII symbolic
2314 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2315 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2316 are no longer used for output.
2317 ** changes in internal implementation constants:
2318 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2319 4 million. (If your application spends a lot of time GCing and
2320 you have a lot of RAM, you might want to experiment with
2321 increasing it even more.)
2322 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2323 However, almost all old code should still continue to work without
2324 immediate update, as SB-C-CALL is now a (deprecated) nickname
2326 ** Old operator names in the style DEF-FOO are now deprecated in
2327 favor of new corresponding names DEFINE-FOO, for consistency with
2328 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2329 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2330 internal symbols, but a few supported extensions like
2331 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2332 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2333 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2334 etc. as you get deeper into recursive calls to the debugger
2335 command loop, instead of the old "5]", "5]]", "5]]]"
2336 sequence. (I was motivated to do this when squabbles between
2337 ILISP and SBCL left me very deeply nested in the debugger. In the
2338 short term, this change will probably provoke more ILISP/SBCL
2339 squabbles, but hopefully it will be an improvement in the long run.)
2340 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2341 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2342 function, since declaiming return types involving aliens is
2343 (1) annoyingly messy to do by hand and (2) vital to efficient
2344 compilation of code which calls such functions.
2345 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2346 longer reexported by the SB-EXT package. They're solely useful
2347 for alien code, so it seems more logical that you should get
2348 them from the SB-ALIEN package, not in SB-EXT.
2349 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2350 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2351 features. Instead, the code that they used to control is always
2352 built into the system.
2353 * many other bug fixes
2354 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2355 advantage of the new EVAL-WHEN stuff and to clean them up in
2356 general, and they are now more ANSI-compliant in a number of
2357 ways. Martin Atzmueller is responsible for a lot of this.
2358 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2360 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2361 *** correct ERROR type for various file operations
2362 *** some fixes for Lisp streams
2363 *** DEFMETHOD syntax checking
2364 *** changing old weird representation of debug information as
2365 strings (which, among their other deficiencies, don't transform
2366 correctly when you rename packages, and don't change their
2367 print representation when you change things like *PACKAGE*
2368 and *PRINT-LENGTH*) to symbols and lists of symbols
2369 He also made several improvements and fixed several bugs in DESCRIBE.
2370 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2372 *** misbehavior of WRITE-STRING/WRITE-LINE
2373 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2374 and several other LOOP problems as well
2375 *** DIRECTORY when similar filenames are present
2376 *** DEFGENERIC with :METHOD options
2377 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2378 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2379 *** argument ordering in FIND with :TEST option
2380 *** mishandled package designator argument in APROPOS-LIST
2381 *** various problems in the backquote readmacro
2382 *** a bug in APROPOS
2383 *** probably some others that I'm not describing very well here,
2384 since the CVS log documents them by reference to sbcl-devel
2385 messages, and the SourceForge archives aren't working well.:-(
2386 ** Dan Barlow improved the Alpha port (and is making progress on the
2387 PPC port, for those of you who think different).
2388 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2389 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2390 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2391 continues to work on the SPARC port (for those of you in a position
2392 to look down upon our little PC-compatible boxes from a great height).
2393 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2394 and uses that copy, rather than the current dynamic value, when
2395 it's trying to decide whether to truncate output. Thus e.g.
2396 (let ((*print-lines* 50))
2397 (pprint-logical-block (stream nil)
2399 (let ((*print-lines* 8))
2400 (print (aref possiblybigthings i) stream)))))
2401 should now truncate the logical block only at 50 lines, instead of
2402 often truncating it at 8 lines, as it did before.
2403 * The doc/cmucl/ directory, containing old CMU CL documentation
2404 from the time of the fork, is no longer part of the base system.
2405 SourceForge has shut down its anonymous FTP service, and with it
2406 my original plan for distributing the old CMU CL documentation
2407 there. For now, if you need these files you can download an old
2408 SBCL source release and extract them from it.
2409 * The fasl file version number changed again, for dozens of reasons,
2410 some of which are apparent above.
2412 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2413 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2414 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2415 patch, so that DISASSEMBLE output is much nicer.
2416 * The code in the SB-PROFILE package now seems reasonably stable.
2417 I still haven't decided what the final interface should look like
2418 (I'd like PROFILE to interact cleanly with TRACE, since both
2419 facilities use function encapsulation) but if you have a need
2420 for profiling now, you can probably use it successfully with
2421 the current CMU-CL-style interface.
2422 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2423 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2424 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2425 link now returns the dangling link itself, and for similar
2426 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2427 link itself. (In these cases the old code signalled an error and
2428 looped endlessly, respectively.) Thus, DIRECTORY now works even
2429 in the presence of dangling and cyclic symbolic links.
2430 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2431 is now a supported extension again, since the consensus on
2432 sbcl-devel was that it can be useful for ordinary development
2433 work, not just for debugging SBCL itself.
2434 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2435 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2436 that currently-defined functions might be redefined later with
2437 different return types.
2438 * Hash tables can be printed readably, as inspired by CMU CL code
2439 of Eric Marsden and SBCL code of Martin Atzmueller.
2440 * better error handling in CLOS method combination, thanks to
2441 Martin Atzmueller porting Pierre Mai's CMU CL patches
2442 * more overflow fixes for >16Mbyte I/O buffers
2443 * A bug in READ has been fixed, so that now a single Ctrl-D
2444 character suffices to cause end-of-file on character streams.
2445 In particular, now you only need one Ctrl-D at the command
2446 line (not two) to exit SBCL.
2447 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2449 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2450 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2451 does the right thing.
2452 * The implementation of some type tests, especially for CONDITION
2453 types, is now tidier and maybe faster, due to CMU CL code
2454 originally by Douglas Crosher, ported by Martin Atzmueller.
2455 * Some math functions have been fixed, and there are new
2456 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2457 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2458 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2459 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2460 likely to be built into sbcl-0.7.0.)
2461 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2462 CL, ported to SBCL by Martin Atzmueller)
2463 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2464 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2465 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2466 * There's a new slam.sh hack to shorten the edit/compile/debug
2467 cycle for low-level changes to SBCL itself, and a new
2468 :SB-AFTER-XC-CORE target feature to control the generation of
2469 the after-xc.core file needed by slam.sh.
2470 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2471 COMPILE-FILE is no longer supported, so that now every function
2472 gets an entry point, so that block compilation looks a little
2473 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2474 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2475 returns the number of bytes consed since the system started,
2476 rather than the number consed since the first time the function
2477 was called. (The new definition parallels ANSI functions like
2478 CL:GET-INTERNAL-RUN-TIME.)
2479 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2480 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2481 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2482 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2483 * Fasl file version numbers are now independent of the target CPU,
2484 since historically most system changes which required version
2485 number changes have affected all CPUs equally. Similarly,
2486 the byte fasl file version is now equal to the ordinary
2489 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2490 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2491 is no longer recognized. I apologize for this, because it was
2492 listed in SB-EXT as a supported extension, but I found that
2493 its existing behavior was poorly specified, as well as incorrectly
2494 specified, and it looked like too much of a mess to straighten it
2495 out. I have enough on my hands trying to get ANSI stuff to work..
2496 * many patches ported from CMU CL by Martin Atzmueller, with
2497 half a dozen bug fixes in pretty-printing and the debugger, and
2498 half a dozen others elsewhere
2499 * fixed bug 13: Floating point infinities are now supported again.
2500 They might still be a little bit flaky, but thanks to bug reports
2501 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2502 as flaky as they were.
2503 * The --noprogrammer command line option is now supported. (Its
2504 behavior is slightly different in detail from what the old man
2505 page claimed it would do, but it's still appropriate under the
2506 same circumstances that the man page talks about.)
2507 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2508 are now supported, and enabled by default. Thus, the compiler can
2509 handle many floating point and complex operations much less
2510 inefficiently. (Thus e.g. you can implement a complex FFT
2512 * The compiler now detects type mismatches between DECLAIM FTYPE
2513 and DEFUN better, and implements CHECK-TYPE more correctly, and
2514 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2515 thanks to patches from Martin Atzmueller.
2516 * various fixes to make the cross-compiler more portable to
2517 ANSI-conforming-but-different cross-compilation hosts (notably
2518 Lispworks for Windows, following bug reports from Arthur Lemmens)
2519 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2520 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2521 fixed thanks to Pierre Mai's CMU CL patches.
2522 * fixes to keep the system from overflowing internal counters when
2523 it tries to use i/o buffers larger than 16M bytes
2524 * fixed bug 45a: Various internal functions required to support
2525 complex special functions have been merged from CMU CL sources.
2526 (When I was first setting up SBCL, I misunderstood a compile-time
2527 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2528 * improved support for type intersection and union, fixing bug 12
2529 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2530 more obscure bugs as well
2531 * some steps toward byte-compiling non-performance-critical
2532 parts of the system, courtesy of patches from Martin Atzmueller
2533 * Christophe Rhodes has made some debian packages of sbcl at
2534 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2535 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2536 stable, but are nonetheless usable. When he's ready, I'd be happy
2537 to add them to the SourceForge "File Releases" section. (And if
2538 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2539 * new fasl file format version number (because of changes in
2540 internal representation of (OR ..) types to accommodate the new
2541 support for (AND ..) types, among other things)
2543 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2544 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2546 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2548 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2549 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2550 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2551 * DESCRIBE now works on structure objects again.
2552 * Most function call argument type mismatches are now handled as
2553 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2554 know whether the function will be redefined before the call is
2555 executed. (The compiler could flag local calls with full WARNINGs,
2556 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2557 it doesn't keep track of enough information to know whether calls
2558 are local in this sense.)
2559 * Compiler output is now more verbose, with messages truncated
2560 later than before. (There should be some supported way for users
2561 to override the default verbosity, but I haven't decided how to
2562 provide it yet, so this behavior is still controlled by the internal
2563 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2564 src/compiler/ir1util.lisp.)
2565 * Fasl file format version numbers have increased again, because
2566 support for the Gray streams extension changes the layout of the
2567 system's STREAM objects.
2568 * The Gray subclassable streams extension now works, thanks to a
2569 patch from Martin Atzmueller.
2570 * The full LOAD-FOREIGN extension (not just the primitive
2571 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2572 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2573 but like most other programs, it defaults to copying the Unix
2574 environment from the original process instead of starting the
2575 new process in an empty environment.
2576 * Extensions which manipulate the Unix environment now support
2577 an :ENVIRONMENT keyword option which doesn't smash case or
2578 do other bad things. The CMU-CL-style :ENV option is retained
2579 for porting convenience.
2580 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2581 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2583 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2585 * A patch from Martin Atzmueller seems to have solved the SIGINT
2586 problem, and as far as we know, signal-handling now works cleanly.
2587 (If you find any new bugs, please report them!)
2588 * The system no longer defaults Lisp source file names to types
2589 ".l", ".cl", or ".lsp", but only to ".lisp".
2590 * The compiler no longer uses special default file extensions for
2591 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2592 seems to expect a single default extension for all compiled code,
2593 and there's no compelling reason to try to stretch the standard
2594 to allow two different extensions.) Instead, byte-compiled files
2595 default to the same extension as native-compiled files.
2596 * Fasl file format version numbers have increased again, because
2597 a rearrangement of internal implementation packages made some
2598 dumped symbols in old fasl files unreadable in new cores.
2599 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2600 many fewer weird special cases.
2601 * Bug #17 (differing COMPILE-FILE behavior between logical and
2602 physical pathnames) has been fixed, and some related misbehavior too,
2603 thanks to a patch from Martin Atzmueller.
2604 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2605 by Tim Moore, ported to SBCL by Martin Atzmueller.
2606 * Martin Atzmueller fixed several filesystem-related problems,
2607 including bug #36, in part by porting CMU CL patches, which were
2608 written in part by Paul Werkowski.
2609 * More compiler warnings in src/runtime/ are gone, thanks to
2610 more patches from Martin Atzmueller.
2611 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2614 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2616 * DESCRIBE now works on CONDITION objects.
2617 * The debugger now handles errors which arise when trying to print
2618 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2620 * The build system now uses an additional file, customize-target-features.lisp,
2621 to allow local modifications to the target *FEATURES* list. (The point of
2622 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2623 debugging features enabled, without having to worry about propagating it
2624 into everyone's system when I do a "cvs update".) When no
2625 customize-target-features.lisp file exists, the target *FEATURES* list
2626 should be constructed the same way as before.
2627 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2628 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2629 warn about behavior which is undefined under the ANSI spec. Note: This
2630 is specified by ANSI, but it's not very popular with programmers.
2631 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2632 macro in the SBCL sources for an example of a workaround which you
2633 might use to make portable ANSI-standard code which does what you want.
2634 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2635 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2636 when it's used as a non-toplevel form. (This is required in order
2637 to implement the DEFCONSTANT-EQX macro.)
2638 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2639 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2640 funny things, due to bugs in the implementation of EVAL-WHEN
2641 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2642 non-toplevel DEFCONSTANTs are uncommon.)
2643 * The core file version number and fasl file version number have been
2644 incremented, because the old noncompliant DEFCONSTANT behavior involved
2645 calling functions which no longer exist, and because I also took the
2646 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2647 * fixed bug 1 (error handling before read-eval-print loop starts), and
2648 redid debugger restarts and related debugger commands somewhat while
2650 ** The QUIT debugger command is gone, since it did something
2651 rather different than the SB-EXT:QUIT command, and since it never
2652 worked properly outside the main toplevel read/eval/print loop.
2653 Invoking the new TOPLEVEL restart provides the same functionality.
2654 ** The GO debugger command is also gone, since you can just invoke
2655 the CONTINUE restart directly instead.
2656 ** The TOP debugger command is also gone, since it's redundant with the
2657 FRAME 0 command, and since it interfered with abbreviations for the
2659 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2661 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2662 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2663 some porting work by Martin Atzmueller)
2664 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2665 with a boatload of patches which he ported to SBCL. Now that those
2667 ** The system tries to make sure that its low-priority messages
2668 are prefixed by semicolons, to help people who like to use
2669 syntax highlighting in their ILISP buffer. (This patch
2670 was originally due to Raymond Toy.)
2671 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2672 patches originally written by Raymond Toy.
2673 ** The compiler understands coercion between single-value and
2674 multiple-VALUES type expressions better, getting rid of some very
2675 weird behavior, thanks to patches originally by Robert MacLachlan
2676 and Douglas Crosher.
2677 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2678 lambda lists, thanks to a patch originally by Pierre Mai.
2679 ** The system no longer bogusly warns about "abbreviated type
2681 ** The compiler gets less confused by inlining and RETURN-FROM,
2682 thanks to some patches originally by Tim Moore.
2683 ** The system no longer hangs when dumping circular lists to fasl
2684 files, thanks to a patch originally from Douglas Crosher.
2685 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2686 undefined function error.
2687 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2688 in structure slot initforms) for now, documented workaround instead:-|
2689 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2690 * fixed bug 5: added stubs for various Gray stream functions called
2691 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2692 installed, at least appropriate type errors are generated
2693 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2694 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2695 consistently in DEFMETHOD forms.
2696 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2697 it doesn't seem to affect SBCL after all
2698 * The C runtime system now builds with better optimization and many
2699 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2701 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2703 * The system is now under CVS at SourceForge (instead of the
2704 CVS repository on my home machine).
2705 * The new signal handling code has been tweaked to treat register
2706 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2707 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2708 downstream, e.g. in debug-int.lisp, has implicit dependencies
2709 on the unsignedness of integer representation of machine words,
2710 and that caused the system to bomb out with infinite regress
2711 when trying to recover from type errors involving signed values,
2712 e.g. (BUTLAST '(1 2 3) -1).)
2713 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2714 (This was one of the bugs Peter Van Eynde reported back in July.)
2715 * The system now uses code inspired by Colin Walters' O(N)
2716 implementation of MAP (from the cmucl-imp@cons.org mailing
2717 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2718 inline the MAP operation, and there is more than one
2719 sequence argument to the MAP call (so that it can't just
2720 do ETYPECASE once and for all based on the type of the
2721 single sequence argument). (The old non-inline implementation
2722 of the general M-argument sequence-of-length-N case required
2723 O(M*N*N) time when any of the sequence arguments were LISTs.)
2724 * The QUIT :UNIX-CODE keyword argument has been renamed to
2725 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2726 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2728 * Raymond Wiker's patches to port dynamic loading from Linux to
2729 FreeBSD have been added.
2730 * The BUGS file is now more nearly up to date, thanks in large part
2731 to Martin Atzmueller's review of it.
2732 * The debugger now flushes standard output streams before it begins
2733 its output ("debugger invoked" and so forth).
2734 * The core version number and fasl file version number have both
2735 been incremented, because of incompatible changes in the layout
2737 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2738 used to not be called for a saved Lisp image.)
2739 * Martin Atzmueller's version of a patch to fix a compiler crash,
2740 as posted on sbcl-devel 13 September 2000, has been installed.
2741 * Instead of installing Martin Atzmueller's patch for the
2742 compiler transform for SUBSEQ, I deleted the compiler transform,
2743 and transforms for some similar consing operations.
2744 * A bug in signal handling which kept TRACE from working on OpenBSD
2746 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2747 away by constant folding
2748 * The system now defines its address space constants in one place
2749 (in the Lisp sources), and propagates them automatically elsewhere
2750 (through GENESIS and the sbcl.h file). Therefore, patching the
2751 address map is less unnecessarily tedious and error-prone. The
2752 Lisp names of address space constants have also been systematized.
2753 * CVS tags like dollar-Header-dollar have been removed from
2754 the sources, because they have never saved me trouble and
2755 they've been source of trouble working with patches and other
2756 diff-related operations.
2757 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2758 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2760 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2762 * The system has been ported to OpenBSD.
2763 * The system now compiles with a simple "sh make.sh" on the systems
2764 that it's supported on. I.e., now you no longer need to tweak
2765 text in the target-features.lisp-expr and symlinks in src/runtime/
2766 by hand, the make.sh takes care of it for you.
2767 * The system is no longer so grossly inefficient when compiling code
2768 involving vectors implemented as general (not simple) vectors (VECTOR T),
2769 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2770 which dares to use the various sequence functions on non-simple
2771 vectors, takes less of a performance hit.
2772 * There is now a primitive type predicate VECTOR-T-P
2773 to test for the (VECTOR T) type, so that e.g.
2774 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2775 can now be compiled with some semblance of efficiency. (The old code
2776 turned the type declaration into a full call to %TYPEP at runtime!)
2777 * AREF on (VECTOR T) is still not fast, since it's still compiled
2778 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2779 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2780 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2781 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2782 without making full calls to SUBTYPEP at runtime.
2783 (Some analogous efficiency issues for non-simple vectors specialized to
2784 element types other than T, or for non-simple multidimensional arrays,
2785 have not been addressed. They could almost certainly be handled the
2786 same way if anyone is motivated to do so.)
2787 * The changes in array handling break binary compatibility, so
2788 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2789 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2790 instead of (VALUES T T).
2791 * By following the instructions that Dan Barlow posted to sbcl-devel
2792 on 2 July 2000, I was able to enable primitive dynamic object
2793 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2794 functionality is not implemented (since it calls ld to resolve
2795 library references automatically, requiring RUN-PROGRAM for its
2796 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2797 try to resolve library references) is now supported.
2798 * The system now flushes the standard output streams when it terminates,
2799 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2800 them at several other probably-convenient times, e.g. in each pass of
2801 the toplevel read-eval-print loop, and after evaluating a form given
2802 as an "--eval" command-line option. (These changes were motivated by a
2803 discussion of stream flushing issues on cmucl-imp in August 2000.)
2804 * The source transform for TYPEP of array types no longer assumes
2805 that an array whose element type is a not-yet-defined type
2806 is implemented as an array of T, but instead punts, so that the
2807 type will be interpreted at runtime.
2808 * There is now some support for cross-compiling in make.sh: each of
2809 the phases of make.sh has its own script. (This should be transparent
2810 to people doing ordinary, non-cross-compile builds.)
2811 * Since my laptop doesn't have hundreds of megabytes of memory like
2812 my desktop machine, I became more motivated to do some items on
2813 my to-do list in order to reduce the size of the system a little:
2814 ** Arrange for various needed-only-at-cold-init things to be
2815 uninterned after cold init. To support this, those things have
2816 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2817 symbols with such names are now uninterned after cold init).
2818 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2819 fasl files for cold load.
2820 ** Remove the old compiler structure pooling code (which used to
2821 be conditional on the target feature :SB-ALLOC) completely.
2822 ** Redo the representation of some data in cold init to be more compact.
2823 (I also looked into supporting byte compiled code at bootstrap time,
2824 which would probably reduce the size of the system a lot, but that
2825 looked too complicated, so I punted for now.)
2826 * The maximum signal nesting depth in the src/runtime/ support code has
2827 been reduced from 4096 to 256. (I don't know any reason for the very
2828 large old value. If the new smaller value turns out to break something,
2829 I'll probably just bump it back up.)
2830 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2832 * Many, many bugs reported by Peter Van Eynde have been added to
2833 the BUGS list; some have even been fixed.
2834 * While enabling dynamic object file loading, I tried to make the
2835 code easier to understand, renaming various functions and variables
2836 with less ambiguous names, and changing some function calling
2837 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2838 * While trying to figure out how to do the OpenBSD port, I tried to
2839 clean up some of the code in src/runtime/. In particular, I dropped
2840 support for non-POSIX signal handling, added various comments,
2841 tweaked the code to reduce the number of compilation warnings, and
2842 renamed some files to increase consistency.
2843 * To support the new automatic configuration functionality in make.sh,
2844 the source file target-features.lisp-expr has been replaced with the
2845 source file base-target-features.lisp-expr and the machine-generated
2846 file local-target-features.lisp-expr.
2847 * fixed a stupid quoting error in make.sh so that using CMU CL
2848 "lisp -batch" as cross-compilation host works again
2850 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2852 * DESCRIBE no longer tries to call itself recursively to describe
2853 bound/fbound values, so that it no longer fails on symbols which are
2854 bound to themselves (like keywords, T, and NIL).
2855 * DESCRIBE now works on generic functions.
2856 * The printer now prints less-screwed-up representations of closures
2857 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2858 * A private symbol is used instead of the :EMPTY keyword previously
2859 used to mark empty slots in hash tables. Thus
2860 (DEFVAR *HT* (MAKE-HASH-TABLE))
2861 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2862 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2863 now does what ANSI says that it should. (You can still get
2864 similar noncompliant behavior if bang on the hash table
2865 implementation with all the symbols you get back from
2866 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2867 This breaks binary compatibility, since tests for equality to
2868 :EMPTY are wired into things like the macroexpansion of
2869 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2871 * There's now a minimal placeholder implementation for CL:STEP,
2872 as required by ANSI.
2873 * An obscure bug in the interaction of the normal compiler, the byte
2874 compiler, inlining, and structure predicates has been patched
2875 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2876 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2878 * Missing ordinary arguments in a macro call are now detected even
2879 when the macro lambda list contains &KEY or &REST.
2880 * The debugger no longer complains about encountering the top of the
2881 stack when you type "FRAME 0" to explicitly instruct it to go to
2882 the top of the stack. And it now prints the frame you request even
2883 if it's the current frame (instead of saying "You are here.").
2884 * As specified by ANSI, the system now always prints keywords
2885 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2887 * The default initial SIZE of HASH-TABLEs is now smaller.
2888 * Type information from CLOS class dispatch is now propagated
2889 into DEFMETHOD bodies, so that e.g.
2890 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2892 is now basically equivalent to
2893 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2894 (DECLARE (TYPE SINGLE-FLOAT X))
2896 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2897 operation, without having to do run-time type dispatch.
2898 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2899 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2900 * Since I don't seem to be making big file reorganizations very often
2901 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2902 is overflowing my ability to conveniently back them up), I've finally
2903 checked the system into CVS. (The CVS repository is on my home system,
2904 not at SourceForge -- putting it on SourceForge might come later.)
2905 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2906 high-level GC-NOTIFY-FOO functions send their output. (There's
2907 still very little control of where low-level verbose GC functions
2908 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2909 less than it used to -- the GC-NOTIFY-FOO functions are now under
2910 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2911 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2912 in only one place in the source code, and propagates it automatically
2913 everywhere that it's needed. Thus e.g. when I bump the version from
2914 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2915 * The C source files now include boilerplate legalese and documentation
2916 at the head of each file (just as the Lisp source files already did).
2917 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2918 to his page will be replaced with a link to his new CLiki service.
2920 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2922 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2923 * The build process now looks for GNU make under the default name "gmake",
2924 instead of "make" as it used to. If GNU make is not available as "gmake"
2925 on your system, you can change this default behavior by setting the
2926 GNUMAKE environment variable.
2927 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2928 can build without error under CMU CL.
2930 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2932 * There is now a partial SBCL user manual (with some new text and some
2933 text cribbed from the CMU CL manual).
2934 * The beginnings of a profiler have been added (starting with the
2935 CMU CL profiler and simplifying and cleaning up). Eventually the
2936 main interface should be through the TRACE macro, but for now,
2937 it's still accessed through vaguely CMU-CL-style functions and macros
2938 exported from the package SB-PROFILE.
2939 * Some problems left over from porting CMU CL to the new
2940 cross-compilation bootstrap process have been cleaned up:
2941 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2942 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2943 part when getting rid of the compiler *BACKEND* stuff.)
2944 ** The value of *NULL-TYPE* was screwed up, because it was
2945 being initialized before the type system knew the final
2946 definition of the 'NULL type. This screwed up several key
2947 optimizations in the compiler, causing inefficiency in all sorts
2948 of places. (I found it because I wanted to understand why
2949 GET-INTERNAL-RUN-TIME was consing.)
2950 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2951 PROCLAIM FTYPE information. Unfortunately this broke binary
2952 compatibility again, since now the forms output by DEFGENERIC
2953 to refer to functions which didn't exist in 0.6.3.
2954 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2955 can use the new (as of 0.6.3) transform for SOME into MAP into
2957 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2958 values to (INTEGER 0 1000000), so that the time code will no longer
2959 occasionally get blown up by Linux returning 1000000 microseconds
2960 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2961 its output conform to the ANSI spec. (Alas, this makes its output
2962 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2963 * A full call to MAP NIL with a single sequence argument no longer conses.
2964 * fixes to problems pointed out by Martin Atzmueller:
2965 * The manual page no longer talks about multiprocessing as though
2966 it were currently supported.
2967 * The ILISP support patches have been removed from the distribution,
2968 because as of version 5.10.1, ILISP now supports SBCL without us
2969 having to maintain patches.
2970 * added a modified version of Raymond Toy's recent CMU CL patch for
2971 EQUALP comparison of HASH-TABLE
2973 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2975 * The system still can't cross-compile itself with
2976 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2977 entails), but at least it can compile more of itself that way
2978 than it used to be able to, and various buglets which were uncovered
2979 by trying to cross-compile itself that way have now been fixed.
2980 * This release breaks binary compatibility again. This time
2981 at least I've incremented the FASL file format version to 2, so that the
2982 problem can be detected reliably instead of just causing weird errors.
2983 * various new style warnings:
2984 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2985 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2986 ** using the lexical binding of a variable named in the *FOO* style
2987 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2988 as specified by ANSI.
2989 * *RANDOM-STATE* is no longer automatically initialized from
2990 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2991 default behavior of the system is to repeat its behavior every time
2992 it's run. If you'd like to change this behavior, you can always
2993 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2994 old convention there was no comparably easy way to get the system to
2995 repeat its behavior every time it was run.
2996 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2997 FTYPE declarations has been removed, in favor of their ANSI
2998 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2999 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3000 the types of their sequence arguments can be determined at compile time.
3001 This is done through a new open code expansion for MAP which eliminates
3002 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3003 argument types can be determined at compile time.
3004 * The optimizer now transforms COERCE into an identity operation when it
3005 can prove that the coerced object is already of the correct type. (This
3006 can be a win for machine generated code, including the output of other
3007 optimization transforms, such as the MAP transform above.)
3008 * Credit information has been moved from source file headers into CREDITS.
3009 * Source file headers have been made more standard.
3010 * The CASE macro now compiles without complaining even when it has
3013 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3015 * (Note that the way that the PCL macroexpansions were rewritten
3016 to accommodate the change in DEFGENERIC below breaks binary
3017 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3018 not run under sbcl-0.6.2. Once we get out of alpha releases,
3019 i.e. hit release 1.0.0, we'll probably try to maintain binary
3020 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3021 and sbcl-1.4.4. Until then, however, it might be fairly common
3022 for maintenance releases to break binary compatibility.)
3023 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3025 * The reporting of unhandled signals has been changed to print some
3026 explanatory text as well as the report form. (Previously only
3027 the report form was printed.)
3028 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3029 it defines, so that the compiler no longer issues undefined function
3030 warnings for compiled-but-not-yet-loaded generic functions.
3031 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3032 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3033 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3034 section of the ANSI spec are supported.
3035 * The "" nickname for the "KEYWORD" package has been removed.
3036 The reader still handles symbol tokens which begin with a package marker
3037 as keywords, but it doesn't expose its mechanism for doing so in the
3038 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3039 * The system now issues STYLE-WARNINGs for contradictory TYPE
3040 proclamations. (Warnings for contradictory FTYPE proclamations would
3041 be nice too, but those can't be done usefully unless the type system
3042 is made smarter about FUNCTION types.)
3043 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3044 systematized, so that "*target-*.lisp is supposed to exist only on the
3045 target and imply that there's a related file which exists on the
3046 host, and *host-*.lisp is supposed to exist only on the host and imply
3047 that there's a related file which exists on the target. This involves a
3048 lot of renaming. Hopefully the acute confusion caused by the renaming
3049 will be justified by the reduction in chronic confusion..
3050 ** runtime-type.lisp -> early-target-type.lisp
3051 ** target-type.lisp -> late-target-type.lisp
3052 ** early-host-format.lisp -> early-format.lisp
3053 ** late-host-format.lisp -> late-format.lisp
3054 ** host-error.lisp -> misc-error.lisp
3055 ** early-error.lisp -> early-target-error.lisp
3056 ** late-error.lisp -> late-target-error.lisp
3057 ** host-defboot.lisp -> early-defboot.lisp
3058 ** code/misc.lisp -> code/target-misc.lisp
3059 ** code/host-misc.lisp -> code/misc.lisp
3060 ** code/numbers.lisp -> code/target-numbers.lisp
3061 ** code/early-numbers.lisp -> numbers.lisp
3062 ** early-host-type.lisp -> early-type.lisp
3063 ** late-host-type.lisp -> late-type.lisp
3064 ** host-typep.lisp -> typep.lisp
3065 ** load.lisp -> target-load.lisp
3066 ** host-load.lisp -> load.lisp
3067 ** host-disassem.lisp -> disassem.lisp
3068 ** host-insts.lisp -> insts.lisp
3069 ** byte-comp.lisp -> target-byte-comp.lisp
3070 ** host-byte-comp.lisp -> byte-comp.lisp
3071 ** host-signal.lisp -> signal.lisp
3072 ** host-defstruct.lisp -> defstruct.lisp
3073 ** late-target-type.lisp -> deftypes-for-target.lisp
3074 Furthermore, several other previously target-only files foo.lisp (e.g.
3075 hash-table.lisp and random.lisp) have been split into a target-and-host
3076 foo.lisp file and a target-only target-foo.lisp file, with their key type
3077 definitions in the target-and-host part, so that the cross-compiler will
3078 know more about target types.
3079 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3080 gone away. In their place are various *BACKEND-FOO* variables
3081 corresponding to the slots of the old structure.
3082 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3083 into the target SBCL has been fixed.
3084 * The chill.lisp system for loading cold code into a running SBCL
3086 * Support for the CMU CL "scavenger hook" extension has been removed.
3087 (It was undocumented and unused in the CMU CL sources that SBCL was
3088 derived from, and stale in sbcl-0.6.1.)
3089 * Various errors in the cross-compiler type system were detected
3090 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3091 (enabling various consistency checks). Many of them were fixed,
3092 but some hard problems remain, so the compiler is back to
3093 running without *TYPE-SYSTEM-INITIALIZED* for now.
3094 * As part of the cross-compiler type system cleanup, I implemented
3095 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3096 * I have started adding UNCROSS calls throughout the type system
3097 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3098 on cross-compiler input files will be able to go away, and various
3100 * CONSTANTP now returns true for quoted forms (as explicitly required
3103 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3105 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3106 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3107 (MAKE-LIST -1). In the longer term, it ought to become true
3108 that declarations are assertions even at SAFETY 1. For now, it's not
3109 quite true even at SAFETY 3, but it's at least more nearly true..
3110 (Note that this change seems to increases the size of the system by
3111 O(5%) and to decrease the speed of the compiler by 20% or more.)
3112 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3113 for the problem of printing dozens of lines of distracting information
3114 about low-level system machinery as part of the top stack frame
3115 on entry to the debugger when an undefined function was called.
3116 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3117 is not reset to COMMON-LISP-USER.
3118 * Compilation of stuff related to dyncount.lisp has been made conditional
3119 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3120 smaller. The various dyncount-related symbols have been moved into
3121 a new "SB-DYNCOUNT" package.
3122 * tty-inspect.lisp has been renamed to inspect.lisp.
3123 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3124 feature has gone away. (When we eventually port to other flavors of
3125 libc and/or Unix, we'll try to make the differences between flavors
3126 invisible at the user level.)
3127 * Various other *FEATURES* tags, and/or their associated conditionals,
3128 have been removed if obsolescent, or given better documentation, or
3129 sometimes given more-mnemonic names.
3131 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3133 * tidied up "make.sh" script
3134 * tidied up system directory structure
3135 * better "clean.sh" behavior
3136 * added doc/FOR-CMUCL-DEVELOPERS
3137 * many many small tweaks to output format, e.g. removing possibly-confusing
3138 trailing #\. character in DESCRIBE-INSTANCE
3139 * (EQUALP #\A 'A) no longer signals an error.
3140 * new hashing code, including EQUALP hashing
3141 * tidied up Lisp initialization and toplevel
3142 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3143 * command line argument processing
3144 * added POSIX-GETENV function to deal with Unix-ish environment variables
3145 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3146 terminating SBCL on EOF
3147 * non-verbose GC by default
3148 * There is no more "sbcl" shell script; the sbcl file is now the C
3149 runtime executable (just like CMU CL).
3150 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3152 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3153 symbol and package information as Lisp native symbol and package objects
3154 instead of strings naming symbols and strings naming packages. This way,
3155 whenever packages are renamed (as in warm init), debug information is
3156 transformed along with everything else.
3157 * tweaked the optimization policy declarations which control the building
3158 of SBCL itself. Now, among other things, the system no longer saves
3159 source location debugging information. (This helps two problems at once
3160 by reducing SBCL size and by keeping SBCL from trying to look for its
3161 sources -- which may not exist -- when reporting errors.)
3162 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3163 debugging and testing purposes
3164 * cleaned up printing, making the printer call PRINT-OBJECT for
3165 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3166 methods, giving nearly-ANSI behavior
3167 * converted almost all special variables to use *FOO* naming convention
3168 * deleted PARSE-TIME functionality, since it can be done portably
3169 * moved some files out of cold init into warm init
3170 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3172 * regularized formatting of source files
3173 * added an install.sh script
3174 * fixed ridiculous memory usage of cross-compiler by making
3175 compiler/alloc.lisp not try to do pooling unless it can hook
3176 itself into the GC of the cross-compilation host. Now the system
3177 builds nicely on my old laptop.
3178 * added :SB-ALLOC in target-features.lisp-expr
3179 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3180 was not implemented)
3181 * re-did condition handling and note reporting in the compiler. Notes
3182 are no longer handled by signalling conditions. Style warnings
3183 and warnings are handled more correctly and reported in such a way
3184 that it's easy to find one or the other in your output (so that you
3185 can e.g. figure out which of many problems caused COMPILE-FILE to
3187 * changed the severity of several compiler warnings from full WARNING
3188 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3189 compiler note reporting so that it doesn't use the condition system
3190 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3191 COMPILE-FILE command)
3192 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3193 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3194 no longer supported, since their implementation was deeply intertwingled
3195 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3196 * removed bogus "support" for compiler macros named (SETF FOO), and
3197 removed the compiler macro for SETF INFO (but only after making a fool
3198 of myself on the cmucl-imp mailing list by posting a bogus patch for
3199 DEFINE-COMPILER-MACRO..)
3200 * Compiled files containing forms which have side effects on the Lisp
3201 reader (such as DEFPACKAGE forms) are now handled more correctly.
3202 (Compiler queuing of top level lambdas has been suppressed by setting
3203 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3204 * deleted various currently-unused source files, e.g. gengc.lisp. They
3205 may be added back at some point e.g. when porting to other architectures,
3206 but until they are it's distracting to distribute them and to try to
3208 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3209 there were so many of them they're just distractions, and UNCROSS is
3210 known to be able to handle the current sources
3211 * moved PROFILE functionality into TRACE, so that it will be clear
3212 how the wrapping and unwrapping of functions when you profile them
3213 interacts with the wrapping and unwrapping of functions when you
3214 trace them. (Actually, the functionality isn't there yet, but at least
3215 the interface specification is there. Hopefully, the functionality will
3216 arrive with some future maintenance release.)
3217 * removed host-oops.lisp
3218 * changed signature of QUIT function to allow UNIX-CODE argument
3219 * fixed READ-SEQUENCE bug
3220 * tweaked verbose GC output so that it looks more like the progress
3221 output that ANSI specifies for functions like LOAD
3222 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3223 * added <http://sbcl.sourceforge.com> to the banner information printed by