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 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
7 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
9 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
10 * fixed some lockups due to gc/thread interaction
11 * dynamic space size on PPC has been increased to 768Mb. (thanks to
13 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
14 :METACLASS argument in addition to a class name. (reported by
15 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellman)
16 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
18 * bug fix: more cleanups to the floating point exception handling on
19 x86-64 (thanks to James Knight)
20 * optimization: improved the MIPS versions of generic (in the
21 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
22 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
23 the wrong file descriptor; implementation of SOCKET-OPEN-P.
24 (thanks to Tony Martinez)
25 * fixed some bugs revealed by Paul Dietz' test suite:
26 ** Invalid dotted lists no longer raise a read error when
28 ** Signal an error if a symbol that names a declaration is used
29 as the name of a type, or vice versa
30 ** Allow using the (declare (typespec var*)) abbreviation for
31 (declare (type typespec var*)) with all type specifiers.
33 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
34 * fixed cross-compiler leakages that prevented building a 32-bit
35 target with a 64-bit host compiler.
36 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
37 opened with :IF-EXISTS :OVERWRITE.
38 * fixed bug 281: error for an invalid qualifier in a short-form method
39 combination method is not signalled until the faulty method is called.
40 * bug fix: iteration variable type inferrer failed to deal with open
41 intervals. (reported by Alan Shields)
42 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
43 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
44 or more runtime options were provided to the sbcl binary.
45 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
46 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
47 * optimization: there's now a fast-path for fixnum arguments in the
48 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
49 * optimization: the code generated on x86-64 is more compact thanks
50 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
51 * bug fix: floating-point exception handling now partly works on
52 x86-64. (Thanks to James Knight)
53 * improvement to the MIPS backend from Thiemo Seufer: C-style
54 64-bit long long arguments and return values to alien functions
56 * fixed some bugs revealed by Paul Dietz' test suite:
57 ** the type-error signalled from WARN has a filled-in DATUM slot.
58 ** the type-error required when a stream is not associated with
59 a file has the stream as its datum.
60 ** type-errors on single-floats on x86-64 no longer have
61 :INVALID-OBJECT as the datum
62 ** the type-errors signalled for invalid function names now have
63 a correct expected type
64 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
65 for structures defined with a :TYPE. Documentation strings for
66 typed structures are no longer immediately discarded
67 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
68 broadcast streams and synonym streams. FILE-LENGTH now also works
71 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
72 * incompatible change: the --noprogrammer option, deprecated since
73 version 0.7.5, has been removed. Please use the equivalent
74 --disable-debugger option instead.
75 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
77 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
78 inoperational for a while now) has been completely removed.
79 * null lexical environments are now printed as #<NULL-LEXENV>,
80 significantly reducing the amount of clutter in typical backtraces.
81 * documentation on weak pointers, finalization, and after GC hooks
82 has been added to the manual.
83 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
84 as other specialized array types, is much faster. SUBSEQ and
85 COPY-SEQ on such arrays have also been sped up.
86 * optimization: EQL is now more efficient when at least other argument
87 is known to be of type (OR FIXNUM (NOT NUMBER)).
88 * fixed bug: the runtime is now less vulnerable to changes in the
89 size of the SBCL object on OS X, and virtual memory is reserved for
90 all spaces that need to be at a fixed address.
91 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
92 * fixed bug: finalizers and after GC hooks that cause consing are now
94 * fixed bug: compiler error messages and summaries are now printed to
95 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
96 * fixed inference of the upper bound of an iteration variable.
97 (reported by Rajat Datta).
98 * fixed bug 211e: calling local functions with duplicated constant
99 keyword argument no longer causes a bogus style warning about an
101 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
102 ftype declaration to be disregarded. (reported by Dave Roberts)
103 * fixed bug 373: caused by erronous compilation of references to alien
104 variables in the runtime on ppc/darwin.
105 * fixed bug 376: CONJUGATE type deriver.
106 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
107 (reported by Baughn on #lisp)
108 * a cross-compiler bug on non-x86 platforms has been identified and
109 fixed. (thanks to Bruno Haible)
110 * improvements to the MIPS runtime code for increased stability.
111 (thanks to Thiemo Seufer)
112 * increased the maximimum compact environment size to allow
113 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
114 * partial workaround for bug 135: don't name the function we're
115 calling for hairy cases of EVAL, so as not to accumulate one environment
116 entry per call to EVAL. (reported by Kevin Reid)
117 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
118 * optimization: type testing for non-vector arrays should be faster.
119 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
121 * bug fix: setting 31st element of a bit vector to zero did not work
123 * bug fix: redefining a class definition which failed due to a
124 previous accessor / function clash now works (but see BUGS entry
125 #380 for more problems in this area). (thanks to Zach Beane)
126 * the long form of DEFINE-METHOD-COMBINATION disables method group
127 checking when given a single method group with pattern *. (thanks
129 * on x86 compiler supports stack allocation of results of simple
130 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
131 * support for building and running on Mac OS X 10.4 "Tiger" has been added
132 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
133 * fixed some bugs related to Unicode integration:
134 ** the restarts for recovering from input and output encoding
135 errors only appear when there is in fact such an error to
137 * fixed some bugs revealed by Paul Dietz' test suite:
138 ** MISC.549 and similar: late transformation of unsafe type
139 assertions into derived types caused unexpected code
141 ** SCALE-FLOAT type deriver is less wrong.
142 ** type derivers for EXP, LOG and similar functions compute result
143 types for complex arguments better.
144 ** (MISC.563) CONJUGATE type deriver works for very restricted
146 ** out-of-line type testers for character strings are available.
147 ** EQUAL compiler transform understands specialness of objects
149 ** accessing double-floats stored on the stack now works on x86-64.
150 ** debugger internals could sometimes create invalid lispobjs,
151 resulting in GC crashes.
152 ** MISC.548: type check weakening can convert required type into
154 ** initialization forms for bindings are not in scope of free special
157 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
158 * incompatible change: thread support for non-NPTL systems has
159 been removed - locking is buggy and unreliable. A threaded
160 SBCL build will now warn at startup and refuse to create
161 new threads, unless futex support is detected in the kernel
162 * incompatible change: the top level REPL now has only an ABORT
163 restart associated with it, not TOPLEVEL and ABORT as it used to.
164 TOP and TOPLEVEL are now available as debugger commands for
165 returning to the top level.
166 * incompatible change: forms evaluated in the REPL now use the
167 global optimization policy.
168 * incompatible change: user- and system-initialization files are
169 no longer processed with LOAD, but by READ and EVAL; hence the
170 global optimization policy, startup package, readtable, etc,
172 * The .fasl file format number has been incremented because of
173 various incompatible changes.
174 * internal entry point details and argument counts no longer appear
175 in backtraces unless explicitly requested by setting
176 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
177 * built-in and standard functions no longer have names like "top
178 level local call to FOO".
179 * fixed bug 32: functions defined in non-null lexical environments
180 now have more legible printed representation
181 * fixed bug 33: functions defined in non-null lexical environemnts
182 are now more amenable to inspection by INSPECT.
183 * workaround for bug 354: XEPs no longer appear in backtraces unless
184 explicitly requested.
185 * fixed bug: receiving the signal which results from attempting to
186 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
187 SIGBUS on other *BSDs) on architectures where the C stack is also
188 the Lisp stack (x86 and x86-64) from foreign code no longer leads
189 to debugger confusion or wild instability. (reported by Cheuksan
191 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
192 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
193 (reported by Lutz Euler)
194 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
195 compiled code. (reported by Paul Dietz)
196 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
197 specialized parameter to a method to something that is not TYPEP
198 the specializer is now possible.
199 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
200 face of package deletion.
201 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
202 pathname. (reported by David Tolpin on c.l.l)
203 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
204 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
205 STRING-STREAM and FILE-STREAM, respectively.
206 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
207 than previously; in addition, its test suite is now run on build.
208 (thanks to Luke Gorrie)
209 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
210 * optimization: added a immediate representation for single-floats
212 * fixed some bugs revealed by Paul Dietz' test suite:
213 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
214 ** attempting to create a package with a colliding nickname causes
215 correctable errors to be signalled.
216 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
217 ** several x86-64 backend bugs related to sign-extension of immediate
220 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
221 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
222 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
224 * fixed loading of multiply forward-referenced layouts.
225 (thanks to Cheuksan Wang)
226 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
227 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
228 (see COMPILE-FILE documentation for details of :PRINT option)
229 * fixed bugs 19 and 317: fixed-format floating point printing is
230 more accurate. This also fixes a bug reported by Adam Warner
231 related to the ~@F format directive.
232 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
234 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
235 dispatch table if the from-char is a dispatch macro character.
236 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
237 by the wordsize no longer ignore the last word. (reported by Lutz
239 * fixed bug in type checking of dynamic-extent variables. (reported
241 * optimization: sequence traversal functions use their freedom to
242 coerce function designators to functions.
243 * optimization: code with many calls to CLOS methods specialized on
244 CLOS classes has had redundant type checks removed.
245 * optimization: added declarations to speed up operations that access
246 the internal character database (for example STRING-UPCASE)
247 * optimization: comparison operations between floats and sufficiently small
248 fixnums no longer create extra rationals
249 * fixed some bugs related to Unicode integration:
250 ** portions of multibyte characters at the end of buffers for
251 character-based file input are correctly transferred to the
252 start of the buffer at the next read.
253 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
254 passing it through to OPEN.
255 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
256 argument given to internal calls to OPEN.
257 ** the built-in comment readers (introduced by character sequences
258 ";" and "#|") are more forgiving to encoding errors; they will
259 STYLE-WARN and then attempt to resync the stream at a character
260 boundary. (thanks to Teemu Kalvas)
261 * fixed some bugs in the x86-64 port:
262 ** Negative short int return values from c-calls are sign-extended
264 ** The stack is aligned to 16-bytes for c-calls, as required by
265 the ABI. (thanks to Cheuksan Wang)
266 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
267 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
268 for enums. (thanks to Vincent Arkesteijn)
269 ** Multiple small optimizations and bugfixes for floating point
271 * fixed some bugs revealed by Paul Dietz' test suite:
272 ** Space, Tab, Linefeed, Return and Page have the invalid
273 secondary constituent character trait.
274 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
276 ** WITH-INPUT-FROM-STRING should only update the index place on
278 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
280 ** Bit-array operations (BIT-AND and similar) worked incorrectly
281 with one-dimensional arrays with fill pointers.
282 ** TYPE-OF failed on a complex with an integer realpart and a
284 ** compiler failure during type inference for the code of form
285 (IF (EQL X (THE ...)) ...) (MISC.535).
287 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
288 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
289 platform. The port supports 61-bit fixnums, large memory spaces
290 and reloading shared object files.
291 * enhancement: saving cores with foreign code loaded is now
292 supported on ppc/Darwin in addition to the previously supported
294 * enhancement: the statistical profiler now walks deeper into the
295 call stack for more meaningful call-graphs and accrued time
296 reports (x86/x86-64 only). It also now reports time spent in
298 * enhancement: it is now possible to trace most individual methods
299 of a generic function in addition to tracing the generic function
301 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
303 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
304 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
305 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
306 produces an error. (thanks to Vincent Arkesteijn)
307 * bug fix: NAMESTRING on pathnames with :WILD components in their
308 directories works correctly. (thanks to Artem V. Andreev)
309 * fixed bug 125: compiler preserves identity of closures. (reported
311 * bug fixed: functions with &REST arguments sometimes failed with
312 "Undefined function" when compiled with (DEBUG 3). (reported by
314 * bug fix: overflow during compiling of setting element of a bit
315 vector with constant index and value. (reported by Timmy Douglas)
316 * build fix: fixed the dependence on *LOAD-PATHNAME* and
317 *COMPILE-FILE-PATHNAME* being absolute pathnames.
318 * on x86 compiler partially supports stack allocation of dynamic-extent
320 * GO and RETURN-FROM do not check the extent of their exit points
321 when compiled with SAFETY 0.
322 * fixed some bugs related to Unicode integration:
323 ** encoding and decoding errors are now much more robustly
324 handled; it should now be possible to recover even from invalid
325 input or output to the terminal. (thanks to Teemu Kalvas)
326 ** provided a first cut at implementing STRING-TO-OCTETS and
327 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
328 ** altered the SB-MD5 contributed module slightly, changing the
329 interface just enough for it to be supportable for builds where
330 lisp characters are not eight bits.
331 * fixed some bugs revealed by Paul Dietz' test suite:
332 ** the FORMATTER-generated functions for ~V[ conditionals require
333 the correct number of arguments.
334 ** READ-FROM-STRING returns the mandated second value when applied
335 to displaced strings.
336 ** the #\Rubout and #\Backspace characters are treated as invalid
337 constituent characters by the tokenizer.
339 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
340 * new feature: reloading changed shared object files with
341 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
342 * new feature: references to foreign variables and functions
343 can now be compiled and loaded before the corresponding shared
344 object file is loaded, as long as the foreign definitions are
345 available at runtime.
346 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
347 supported. (thanks to Dan Debertin)
348 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
349 just keywords. (thanks to Vincent Arkesteijn)
350 * fixed bug #331: structure-class instances corresponding to
351 DEFSTRUCT forms are now created eagerly.
352 * fixed bug #345: backtraces from calls to undefined functions work
353 on x86 as well. Related bug #61 is now also partially fixed on x86
354 (backtraces from throws to unknown catch tags.)
355 * bug fix: lambda-list parsing is now stricter vrt. order and number
356 of lambda-list keywords.
357 * bug fix: as specified by AMOP, an error is signalled if a
358 class-option appears multiple times in a DEFCLASS form. (reported
360 * bug fix: RANDOM can be compiled when the compiler derives the type
361 of its numeric argument as a disjoint set of small integers.
362 (reported by Paul Dietz)
363 * bug fix: starting a core saved with shared objects loaded when
364 those objects are not available no longer causes threaded SBCL to
365 hang. (reported by Sean Ross)
366 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
368 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
369 argument, which must always be NIL. (reported by Kalle Niemitalo)
370 * bug fix: printing 1.0d+23 no longer results in an error.
371 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
372 * bug fix: structure slot setters preserve evaluation order. (thanks
374 * fixed some bugs related to Unicode integration:
375 ** RUN-PROGRAM can allow its child to take input from a Lisp
376 stream. (reported by Stefan Scholl)
377 ** ASDF-INSTALL successfully downloads .tar.gz files in all
378 locales. (reported by Ken Causey)
379 * fixed some bugs revealed by Paul Dietz' test suite:
380 ** INCF, DECF and REMF evaluate their place form as specified in
382 ** FORMATTER expands ~{ ~} iteration directives with V or #
383 parameters correctly.
384 ** FORMATTER deals with the ~@[ ~] conditional directive where the
385 consequent uses no arguments correctly.
386 ** the system has a partial understanding of the (COMPLEX RATIO)
388 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
391 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
392 * new feature: a build-time option (controlled by the :SB-UNICODE
393 keyword feature, enabled by default) for building the system with
394 support for the entire 21-bit character space defined by the
396 * new feature: the system now has rudimentary external-format
397 support; the primary user-visible change at this time is that
398 characters with the high bit set (such as lower-case-e-acute) will
399 print correctly to a terminal in a UTF-8 environment.
400 * minor incompatible change: BASE-CHAR no longer names a class;
401 however, CHARACTER continues to do so, as required by ANSI.
402 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
403 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
404 has been moved to the SB-EXT package (temporarily re-exported from
406 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
407 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
409 * fixed bug #351: better error-handlind and reporting for malformed
411 * fixed bug #350: bignum-printing is now more memory-efficient,
412 allowing printing of very large bignums, eg. (expt 2 10000000).
413 (reported by Bruno Haible)
414 * fixed bug #302: better primitive-type selection for intersection
416 * fixed bug #308: non-graphic characters now all have names, as
417 required. (reported by Bruno Haible)
418 * bug fix: redefining a class with different superclasses now correctly
419 removes it from the direct-subclasses of its previous superclasses.
420 (reported by David Morse)
421 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
422 new value now works. (reported by Bruno Haible)
423 * bug fix: correct canonicalization of multiple non-standard slot
424 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
425 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
426 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
427 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
428 now exists, an signals an error.
429 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
430 during macroexpansion for non-iteration variables as well. (reported
431 by Bruno Haible for CMUCL)
432 * bug fix: Cyclic structures and unprintable objects in compiler
433 messages no longer cause errors. (reported by Bruno Haible)
434 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
435 and READ-FROM-STRING all now return a primary value of NIL if
436 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
437 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
438 (reported by Bruno Haible for CMUCL)
439 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
440 do not designate a condition. (reported by Bruno Haible for
442 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
443 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
444 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
445 permitting resolution in favour of any of the conflicting symbols.
446 (reported by Bruno Haible for CMUCL)
447 * bug fix: EQUAL compiler optimizations is less aggressive on
448 strings which can potentially compare true despite having distinct
449 specialized array element types.
450 * bug fix: unit enumerations can be defined without dividing by
451 zero. (thanks to Vincent Arkesteijn)
452 * FORMAT compile-time argument count checking has been enhanced.
453 (report from Bruno Haible for CMUCL)
454 * a partial workaround for the bug 262: the compiler does not try to
455 inline-expand a local function doing RETURN-FROM from a deleted
456 BLOCK. (thanks to Peter Denno for the bug report and to David
457 Wragg for the simple test case)
458 * fixed some bugs revealed by Paul Dietz' test suite:
459 ** RENAME-PACKAGE allows all package designators as new package
461 ** constraint propagation and lambda variable substitution are
462 more cautious in dealing with partially deleted code.
463 ** compiler sometimes lost reoptimization passes.
464 ** CERROR, given a condition as condition designator, passes its
465 remaining arguments to the continue format control without
467 ** Case-altering FORMAT directives work correctly on non-ASCII
469 ** The REAL type specifier handles bounds outside the floating
470 point ranges without signalling FLOATING-POINT-OVERFLOW.
471 ** Functions with IR1-transformations can create intercomponent
472 references to global functions.
473 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
475 ** FORMAT ~R treats a nil value for its first parameter correctly.
477 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
478 * enhancement: saving cores with foreign code loaded is now
479 supported on x86/NetBSD and sparc/Linux in addition to the previously
481 * bug fix: on some platforms repeated installations caused multiple
482 copies of HTML documentation to be installed -- should not happen
483 any more. (reported by Stefan Scholl)
484 * bug fix: parsing self-recursive alien record types multiple times
485 no longer causes infinite recursion. (reported by Thomas F. Burdick,
486 original patch by Helmut Eller for CMUCL)
487 * bug fix: stack-exhaustion detection works now on NetBSD as well.
488 (thanks to Richard Kreuter)
489 * bug fix: defining classes whose accessors are methods on existing
490 generic functions in other (locked) packages no longer signals
491 bogus package lock violations. (reported by François-René Rideau)
492 * bug fix: special variables as DEFMETHOD parameters no longer have
493 associated bogus type declarations. (reported by David Wragg and
495 * bug fix: read-write consistency on streams of element-type
496 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
497 * bug fix: redefiniton of the only method of a generic function with
498 no DEFGENERIC no longer emits a full WARNING. In addition,
499 redefinition of generic functions with no DEFGENERIC to an
500 incompatible lambda list now signals an error. (thanks to Zach
502 * bug fix: DEFGENERIC now works even when there's a function of the
503 same name in an enclosing lexical environment. (thanks to Zach
505 * fixed compiler failure, caused by instrumenting code during
506 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
507 * optimization: added loop analysis and improved register allocation
508 * fixed some bugs revealed by Paul Dietz' test suite:
509 ** POSITION on displaced vectors with non-zero displacement
510 returns the right answer.
511 ** (SIMPLE-STRING) is a valid type specifier for sequence
513 ** *PRINT-LEVEL* handling for slotless structures is pedantically
515 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
516 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
518 ** The justification version of the FORMAT ~< directive treats
519 non-zero minpad parameter correctly.
521 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
522 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
523 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
524 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
525 the supported interface.
526 * new feature: Single-stepping of code compiled with DEBUG 2 or
527 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
528 * new feature: saving cores with foreign code loaded is now
529 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
530 Timothy Moore's work for CMUCL)
531 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
532 parameters to * instead of NIL if no initform is supplied.
533 (reported by Johan Bockgaard)
534 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
535 correctly when FUNCALL appears as the car of the form. Note:
536 despite this FUNCALL forms are not currently subject to
537 compiler-macro expansion. (port of Raymond Toy's fix for the
538 same from CMUCL, reported by Johan Bockgaard)
539 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
540 (thanks for Teemu Kalvas)
541 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
542 case instead of terminating the iteration (thanks for Julian
543 Squires, Sean Champ and Raymond Toy)
544 * bug fix: incorrect expansion of defgeneric that caused a style
545 warning. (thanks for Zach Beane)
546 * bug fix: slot accessor effective method computation works properly
547 for classes with multiple non-standard applicable methods on
548 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
549 * on x86 compiler supports stack allocation of results of LIST and
550 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
551 CMUCL implementation by Gerd Moellmann)
552 * fixed some bugs revealed by Paul Dietz' test suite:
553 ** FORMAT strings with both the ~<~:;~> form of the justification
554 directive and pretty-printing directives cause an error.
556 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
557 * incompatible change: the internal functions
558 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
559 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
560 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
561 instead of the old functions.
562 * new feature: on platforms where "dladdr" is available foreign
563 function names now appear in backtraces. (based on Helmut Eller's
565 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
566 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
568 * documentation: ASDF manual is now distributed with SBCL.
569 * bug fix: SBCL can now load its contributed modules with REQUIRE
570 even if the system-provided entries have been removed from
571 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
573 * bug fix: backtraces involving undefined functions or assembly
574 routines are more informative. (thanks to Brian Downing)
575 * bug fix: mutually referent alien structures now work correctly.
576 (reported by Rick Taube)
577 * bug fix: structures defined by WITH-ALIEN can be referred to
578 within other definitions in the same WITH-ALIEN.
579 * bug fix: division operators (MOD, TRUNCATE and the like) with
580 constant zero divisors and integer dividends no longer generate
582 * bug fix: provide default methods for INPUT-STREAM-P and
583 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
584 * bug fix: improve the detection and resolution of MOP metacycles
585 (where effective method computation depends on the generic function
586 having its effective method computed). (reported by Bruno Haible)
587 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
588 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
589 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
590 * bug fix: pathnames with relative directory components can be
591 represented relative to default pathnames.
592 * optimization: in taking the GCD of bignums, reduce the two bignums
593 to approximately the same size (using Euclid's algorithm) before
594 applying the more sophisticated binary GCD. (thanks to Juho
596 * optimization: COUNT on bitvectors now operates word-at-a-time.
597 * optimization: ASH with a positive, but not necessarily constant,
598 (leftwards) shift, when in a modular context, is compiled to a
600 * fixed some bugs revealed by Paul Dietz' test suite:
601 ** FORMAT variable parameters ("~V<char>") are defaulted properly
602 if the corresponding argument is NIL.
603 ** FORMAT directives accepting a minpad parameter treat negative
605 ** literal commas as character parameters to FORMAT directives are
607 ** literal spaces directly after ~<Newline> directives within a
608 format-logical-block (~:< ~@:>) do not induce :FILL-style
609 conditional newlines.
610 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
611 as specified: it no longer includes conditional newlines).
612 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
614 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
615 :PER-LINE-PREFIX argument does not evaluate to a string.
617 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
618 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
619 the manual for details; for now, package locks can be disabled by
620 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
621 if no major problems are found then it is likely that they will be
622 compiled in unconditionally.
623 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
624 unsupported operators on all platforms. To load a shared library
625 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
626 object file, link it into a shared library outside of SBCL and
627 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
629 * bug workaround: ROOM T and the GC were not working together
630 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
631 reporting facility aren't preserved. That reporting has been
632 disabled, so now until and unless someone figures out how to make it
633 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
634 * minor incompatible change: as threatened around sbcl-0.8.0, the
635 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
636 The symbols are also exported from SB-PCL for backwards
637 compatibility, but more so than before SB-PCL should be treated as
638 an implementation-internal package.
639 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
641 * the SB-SPROF contrib now works on (most) non-x86 architectures.
642 It is known as of this release not to work on the Alpha, however.
643 * fixed bug #167: errors signalled due to illegal syntax in method
644 bodies are now more legible.
645 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
646 designators and can hence be used with TYPEP.
647 * fixed bug #333: CHECK-TYPE now ensures that the type error
648 signalled, if any, has the right object to be accessed by
649 TYPE-ERROR-DATUM. (reported by Tony Martinez)
650 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
651 5.1.2.3 for multiple-value place subforms. (reported by Kalle
653 * fixed bug #334: programmatic addition of slots using specialized
654 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
655 and :ALLOCATION :CLASS slots.
656 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
657 float exponents. (rereported by Peter Seibel)
658 * fixed a bug: #\Space (and other whitespace characters) are no
659 longer considered to be macro characters in standard syntax by
661 * fixed bug: initialization of condition class metaobjects no longer
662 causes an instance of the condition to be created. (reported by
664 * fixed bug: it is now possible to have slots such that
665 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
666 description is neither :INSTANCE nor :CLASS.
667 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
668 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
669 SLOT-VALUE-USING-CLASS), and no longer causes errors with
670 non-standard SLOT-DEFINITION-ALLOCATION values.
671 * fixed bugs: various race conditions handling exiting threads.
672 CL-PPCRE's thread torture test now passes.
673 * fixed arguably-a-bug: GC time proportional to number of threads in
674 system even when most of them are idle
675 * optimization: improved performance of BIT and SBIT on bit-vectors.
676 * .fasl file incompatibility: The fasl file version number has
677 been incremented because of changes associated with package locks.
679 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
680 * minor incompatible change: the system no longer provides
681 optimization hints (conditions which are TYPEP
682 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
683 mode; these hints are still emitted when higher SPEED optimization
685 * new contrib module: a sampling profiler (profiling by statistical
686 sampling, rather than by instrumenting functions) is available as
687 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
688 * the behaviour of the standard function ED is now customizeable by
689 third parties through a hook variable: see ED's documentation
690 string for information on the protocol.
691 * the compiler no longer emits efficiency notes for (FUNCALL X)
692 when the type of X is uncertain under default optimization
694 * fixed bug 276: mutating a binding of a specialized parameter to a
695 method to something that is not TYPEP the specializer is now
697 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
698 exhibit the required behaviour on the x86 platform. (thanks to
699 Peter van Eynde, Eric Marsden and Bruno Haible)
700 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
701 even for difficult arguments. (reported by Peter Graves)
702 * fixed bug 141a: the backquote printer now descends quoted
704 * fixed another bug in backquote printing: no more destructive
705 modification of the form's list structure. (reported by Brian
707 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
708 move between its address being taken and the call to
709 interrupt_thread, fixing a crashing race condition.
710 * the SB-POSIX contrib implementation has been adjusted so that it
711 no longer exhibits ridiculously poor performance when constructing
712 instances corresponding to C structs.
714 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
715 * minor incompatible change: the sb-grovel contrib now treats C
716 structures as alien (in the sense of SB-ALIEN) objects rather than
717 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
718 has implications for memory management of client code
719 (sb-grovel-returned objects must now be manually managed) and for
720 type safety (alien objects now have full types).
721 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
722 used to control emission of compiler diagnostics, rather than the
723 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
724 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
725 quality should be considered deprecated.
726 * install.sh now installs the user manual as well
727 * (not quite a new documentable feature, but worth considering in
728 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
729 beginnings of a semantically meaningful condition hierarchy is
730 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
732 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
733 designator as the defaults argument.
734 * fixed bug: Displaced arrays whose displaced-to array has become
735 too small now cause ARRAY-DIMENSION to signal an error, providing
736 for safer bounds-checking. (reported by Bruno Haible)
737 * fixed bug: DEFCLASS slot definitions with identical :READER and
738 :WRITER names now signal a reasonable error. (reported by Thomas
740 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
742 * fixed bug: Invalid slot specification errors now print correctly.
743 (thanks to Zach Beane)
744 * fixed bug 320: Shared to local slot value transfers in class
745 redefinitions now happen corrently from superclasses as
746 well. (reported by Bruno Haible)
747 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
749 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
750 improper lists correctly. (reported by Bruno Haible)
751 * fixed bug 313: source-transform for <fun-name> was erroneously
752 applied to a call of a value of a variable with name <fun-name>.
753 (reported by Antonio Menezes Leitao)
754 * fixed bug 307: The obsolete instance protocol ensures that
755 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
756 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
757 remove dead unknown-values globs from the middle of the stack.
758 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
759 easier to resume long computations after using *BREAK-ON-SIGNALS*
760 to diagnose and fix failures. (thanks to Nikodemus Siivola)
761 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
762 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
763 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
764 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
765 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
767 * fixed a (fairly theoretical) bug in string printing: if
768 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
769 does not have array-element-type equal to the most general string
771 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
772 function name. (thanks to Nikodemus Siivola)
773 * fixed bug: compile-time format string checker failed on
774 non-closed ~{. (reported by Thomas F Burdick)
775 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
776 don't warn on legal loop code involving more than one aggregate
777 boolean. (thanks to Nikodemus Siivola)
778 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
779 in keyword position to MAKE-PACKAGE was not being reported as
781 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
782 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
784 * fixed bugs 280 and 312: the checking for multiple definitions in a
785 file is less likely to become confused by uses of inline
787 * fixed bug: the #S reader macro performs the keyword coercion
788 specified for slot names. (reported by Kalle Niemitalo)
789 * fixed bug: lambda lists may contain symbols whose names start with
790 & but are not lambda-list-keywords; their occurrence triggers a
792 * fixed bug 321: define-method-combination argument lambda lists do
793 not cause invalid code to be generated when &optional or &aux
794 variables have default values. (reported by Bruno Haible)
795 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
796 have CLOS classes; this appears also to have fixed reported
797 instability in stack exhaustion detection.
798 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
799 MUFFLE-WARNING when no associated restart is present is now
801 * optimization: rearranged the expansion of various defining macros
802 so that each expands into only one top-level form in a
803 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
805 * optimization: used a previously little-used slot in symbols to
806 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
808 * fixed some bugs revealed by Paul Dietz' test suite:
809 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
811 ** Fixed an optimization bug related to inheritance of initforms
812 from local to shared slots.
813 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
814 ** CAST optimizer forgot to flush argument derived type.
815 ** print/read consistency on floats is now orders of magnitude
816 more likely. (thanks also to Bruno Haible for a similar report
818 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
819 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
820 for objects of type REAL. Make it so.
821 ** SXHASH is specified (infelicitously) to respect similarity,
822 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
823 it so. (thanks to Markus Ziegler)
824 ** on the Alpha, the compiler succeeds in compiling functions
825 returning a known number of arguments greater than 63.
826 ** fixed handling of invalid NIL arguments in keyword position
828 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
830 ** provided a readably-printable representation for RANDOM-STATE
832 ** ensured that pathnames, where they have a namestring, always
833 print using #P"..." syntax.
835 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
836 * Support for the forthcoming 2.0 version of the NetBSD kernel
837 running on x86 hardware has been added. (thanks to Perry
838 E. Metzger most immediately, and others for their past work)
839 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
840 current release version; SBCL's OpenBSD support had been broken
841 since about the time of OpenBSD's switch to ELF binary format.)
842 * [placeholder for DX summary]
843 ** user code with &REST lists declared dynamic-extent, under high
844 speed or space and low safety and debug optimization policy.
845 * The manual has been converted to Texinfo format and the debugger
846 chapter from the cmucl manual has been added.
847 * A facility has been added to extract documentation strings from
848 sbcl and store them as Texinfo-formatted snippets for inclusion in
849 the manual (via Texinfo's @include directive)
850 * bug fix: compiler emitted division in optimized DEREF. (thanks for
851 the test case to Dave Roberts)
852 * bug fix: multidimensional simple arrays loaded from FASLs had fill
853 pointers. (reported by Sean Ross)
854 * bug fix: PROFILE output is printed nicely even for large numerical
855 values. (thanks to Zach Beane)
856 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
857 greater than 32 handle EOF correctly.
858 * bug fix: on X86 an immediate argument of the IMUL instruction is
859 correctly printed in disassembly. (thanks to Lutz Euler)
860 * bug fix: class slots in redefined classes preserve their old
861 values. (thanks to Bruno Haible and Nikodemus Siivola)
862 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
863 succeeds. (reported by Marco Baringer)
864 * fixed some bugs revealed by Paul Dietz' test suite:
865 ** READ-SEQUENCE now works on ECHO-STREAMs.
866 ** RATIONALIZE works more according to its specification. (thanks
869 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
870 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
871 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
872 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
873 should matter to you only if you rebind the printer control
874 variables and then find you want different bindings in the
875 debugger than in the ordinary execution of your program.)
876 * The runtime build system has been tweaked to support building
877 (on SPARC/SunOS) using a C compiler which invokes Sun's own
878 assembler and linker. (thanks to Nikodemus Siivola)
879 * Unbound, undefined, undeclared variables now trigger full
880 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
881 more often programmer error than deliberate exploitation of undefined
883 * optimization: the hash algorithm for strings has changed to one
884 that is less vulnerable to spurious collisions. (thanks to Juho
886 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
887 needless bounds checking. (thanks to Juho Snellman)
888 * optimization: implemented multiplication as a modular
889 (UNSIGNED-BYTE 32) operation on the PPC backend.
890 * fixed some bugs revealed by Paul Dietz' test suite:
891 ** ADJUST-ARRAY now copies the datum in a zero rank array if
893 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
894 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
895 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
897 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
899 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
901 ** ECHO-STREAMs no longer attempt to echo the end of file value to
902 their output stream on EOF from read.
903 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
904 have been read to end-of-file.
905 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
907 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
908 description of determination of which consecutive characters
910 ** Printing the "Space" character with escaping on now yields "#\\ ",
911 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
912 ** Reading floating-point numbers with *READ-BASE* set to a number
913 less than 10 works correctly.
914 ** Reading floating-point numbers with *READ-BASE* set to a number
915 more than 10 works correctly.
916 ** Printing with *PRINT-READABLY* targets the standard readtable, not
917 the readtable currently in effect.
919 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
920 * minor incompatible change: parsing of namestrings on a physical
921 (Unix) host has changed; numbers after the final #\. in a
922 namestring are no longer interpreted as a version field. This is
923 intented to be largely invisible to the user, except that the
924 meaning of the namestring "*.*.*" has changed: it now refers to a
925 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
926 should usually be replaced by
927 (make-pathname :name :wild :type :wild :version :wild)
928 with the added benefit that this is more likely to be portable.
929 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
930 option to OPEN now signals an error if the file being opened
931 exists; this may have an impact on existing code.
932 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
933 signals received, on the PowerPC platforms (both Linux and
934 Darwin). (thanks to Pierre Mai for pointing out the location of
936 * several fixes on OS X: The system now builds and runs cleanly on
937 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
938 (thanks to Brian Mastenbrook)
939 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
940 second-resolution: integer multiples of 1/3600 between -24 and 24.
941 (thanks to Vincent Arkesteijn)
942 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
943 type when called with 1 argument; PEEK-CHAR checked type of
944 PEEK-TYPE only after having read first character from a
945 stream. (reported by Peter Graves)
946 * bug fix: the garbage collector now has much better locality
947 behaviour, and in particular no longer treats all memory as being
948 exhausted when many small objects point to each other in a deeply
950 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
951 recognized as being TYPEP their class.
952 * bug fix: the PUSHNEW documentation string has been corrected.
953 (thanks to Vincent Arkesteijn)
954 * bug fix: defaulting of the value for the last of an atypically
955 large number of multiple values being bound was not being
956 performed correctly on the Alpha or PPC platforms
957 * optimization: implemented multiplication as a modular
958 (UNSIGNED-BYTE 32) operation on the x86 backend.
959 * optimization: SEARCH on simple-base-strings can now be open-coded.
960 (see also contrib/compiler-extras.lisp for inspiration for
961 teaching the compiler about the Boyer-Moore algorithm).
962 * value, returned by MAX (and MIN) called with several EQUALP, but
963 not EQL, arguments now does not depend on compiler settings.
964 * fixed some bugs revealed by Paul Dietz' test suite:
965 ** in stack analysis liveness information is propagated from
966 non-local entry points.
967 ** pathwise CAST removing failed when the CAST node did not start
969 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
970 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
972 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
973 host is already defined.
974 ** RENAME-FILE works on streams instead of signalling an internal
976 ** PEEK-CHAR uses the current readtable when determining whether
977 or not a character is whitespace.
978 ** MERGE-PATHNAMES handles the case when the pathname does not
979 specify a name while the default-pathname specifies a version.
980 ** Pathnames now stand a better chance of respecting print/read
982 ** Attempting to use standardized file system operators with a
983 pathname with invalid :DIRECTORY components signals a
985 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
987 ** DIRECTORY on logical pathnames is more correct.
988 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
989 signal a TYPE-ERROR if their argument is not a stream.
990 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
991 designator argument does not designate a stream.
992 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
993 examining the synonym.
994 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
996 ** OPEN and WITH-OPEN-STREAM allow opening streams with
997 element-type larger than ([UN]SIGNED-BYTE 32).
999 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1000 * When built with the :SB-FUTEX feature, threaded builds now take
1001 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1002 for faster/more reliable mutex and condition variable support.
1003 * Incompatible change (but one you probably shouldn't have been using
1004 anyway): the interface and code for arbitrating between multiple
1005 threads in the same user session has been redesigned.
1006 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1007 error when called without an explicit environment argument.
1008 (thanks to Nikodemus Siivola)
1009 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1010 confused about their position. (thanks to Adam Warner and Gerd
1012 * bug fix: Pretty printing backquoted forms with unquotations in the
1013 argument list position of various code constructs such as LAMBDA
1014 now works correctly. (reported by Paul Dietz)
1015 * bug fix: Pretty printing unquotations no longer loses all
1016 stream position information.
1017 * optimization: performance of string output streams is now less
1018 poor for multiple small sequence writes.
1019 * optimization: performance of CSUBTYPEP in the presence of complex
1020 expressions involving CONS and NOT many times has been improved.
1021 (reported by Paul Dietz)
1022 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1024 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1025 streams for the REPL, for the debugger, and so on. (thanks to
1027 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1028 * fixed some bugs revealed by Paul Dietz' test suite:
1029 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1030 is ignored if it is too small, rather than propagating through
1031 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1032 ** extremely complex negations of CONS types were not being
1033 sufficiently canonicalized, leading to inconsistencies in
1035 ** VALUES tranformer lost derived type.
1037 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1038 * fixed a bootstrapping bug: the build process no longer assumes
1039 that the various BOOLE-related constants have the same value in
1040 host and target lisps. (noted by Paul Dietz' test suite on an
1041 SBCL binary built from CLISP)
1042 * The system can now be dynamically linked on the MIPS platform,
1043 which enables dynamic loading of foreign code from Lisp. (thanks
1044 to Ralf Baechle for discussions on the MIPS ABI)
1045 * The system now records debugging information for its own source
1046 files in a filesystem-position-independent manner, relative to
1047 the "SYS" logical host.
1048 * fixed a compiler bug: MV-LET convertion did not check references
1049 to the "max args" entry point. (reported by Brian Downing)
1050 * tweaked disassembly notes to be less confident about proclaiming
1051 some instruction as an LRA. (thanks to Brian Downing)
1052 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1053 now each have their own history, command character, and other
1054 characteristics. (thanks to David Lichteblau)
1055 * fixed some bugs revealed by Paul Dietz' test suite:
1056 ** compiler failure in compiling LOGAND expressions including a
1058 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1059 shift greater than 32.
1060 ** FLUSH-DEST did not mark blocks for type check regeneration.
1061 ** HANDLER-CASE failed to accept declarations in handler clauses
1062 in some circumstances.
1064 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1065 * New code in contrib/sb-introspect (still probably not entirely
1066 stable yet) provides some support for smart Lisp development
1067 environments like SLIME.
1068 * The conditions signalled for errors occurring when loading .fasl
1069 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1070 in a way which should help ASDF recover gracefully.
1071 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1072 stuff has been cleaned up. If you code contrib/ stuff, this might
1073 affect you, and you can look at contrib/README, contrib/STANDARDS,
1074 and/or the 0.8.4.27 diff to check.
1075 * In full calls the compiler now does not generate checks for declared
1076 argument types for all arguments.
1077 * various threading fixes
1078 ** and some experimental patches which didn't make it into
1079 the main tree for this release, but which are shipped in
1080 contrib/experimental-thread.patch as a possible fix for some
1081 failures (deadlock, spinning...) in GC-intensive multithreaded
1083 * fixed PPC build problem (source code incompatibility of different
1084 library versions): added offsetof() hackery which attempts to divine
1085 where glibc maintainers put uc_mcontext today
1086 * fixed bug 282: compiler does not trust type assertions while passing
1087 arguments to a full call.
1088 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1089 &OPTIONAL VALUES type parameter.
1090 * fix bug 214: algorithm for noting rejected templates is now more
1091 similar to that of template seletion. (also reported by rydis on
1093 * fixed bug 141b: printing backquoted information readably and prettily
1094 inserts a space where necessary.
1095 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1096 (CHARACTER 10) are now reported as errors, rather than propagated
1097 as unknown types. (reported by piso on #lisp)
1098 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1099 with values NIL and :ERROR. (thanks to Milan Zamazal)
1100 * fixed bug 191c: CLOS now does proper keyword argument checking as
1101 described in CLHS 7.6.5 and 7.6.5.1.
1102 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1103 longer raise an error; further, using a list as a for-as-arithmetic
1104 counter now raises a meaningful error.
1105 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1106 now understood by sequence creation functions such as MAKE-SEQUENCE
1108 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1109 a STRING-INPUT-STREAM.
1110 * compiler enhancement: SIGNUM is now better able to derive the type
1112 * type declarations inside WITH-SLOTS are checked. (reported by
1114 * fixed some bugs revealed by Paul Dietz' test suite:
1115 ** incorrect optimization of TRUNCATE for a positive first
1116 argument and negative second.
1117 ** compiler failure in let-convertion during flushing dead code.
1118 ** compiler failure while deriving type of TRUNCATE on an
1119 interval, containing 0.
1120 ** ASH of a negative bignum by a negative bignum count now returns
1122 ** intersection of CONS types now canonicalizes properly, fixing
1123 inconsistencies in SUBTYPEP.
1125 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1126 * incompatible change: The --disable-debugger command line
1127 option now clobbers the debugger at a more fundamental
1128 level, by redefining #'INVOKE-DEBUGGER instead of by
1129 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1130 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1131 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1132 enter the debugger REPL and then suffer recursive errors
1133 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1134 while under the new system, BREAK in --disable-debugger mode
1135 terminates the system just as an unhandled error would.
1136 * fixed compiler performance when processing loops with a step >1;
1137 * bug fix: DOCUMENTATION now retrieves generic function
1138 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1139 support has been systematized, and now supports the methods
1140 specified by ANSI, along with a default method and a method for
1141 slot documentation. (reported by Nathan Froyd)
1142 * bug fix: effective methods associated with a generic function are
1143 no longer cached over a change of that generic function's method
1144 combination. (reported by Andreas Fuchs)
1145 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1146 names a non-existent directory. (thanks to Andreas Fuchs)
1147 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1148 the PPC platform, lead to stack corruption; this has been fixed.
1149 (reported by Rainer Joswig)
1150 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1151 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1152 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1153 arithmetic in general, is now much more reliable on the PPC
1155 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1156 platform now returns the right answer.
1157 * optimization: restored some effective method precomputation in
1158 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1159 precomputation is now tunable.
1160 * optimization: compiler-internal data structure use has been
1161 reviewed, and changes have been made that should improve the
1162 performance of the compiler by about 20%.
1163 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1164 simple-base-strings and simple-bit-vectors is improved.
1165 * optimization: the optimization of 32-bit logical and arithmetic
1166 functions introduced in version 0.8.3 on the x86 has been
1167 implemented on the mips, ppc and sparc platforms; an
1168 implementation of the same facility, but for 64-bit arithmetic,
1169 has been added for the alpha.
1170 * microoptimization: the compiler is better able to make use of the
1171 x86 LEA instruction for multiplication by constants.
1172 * bug fix: in some situations compiler did not report usage of
1173 generic arithmetic in (SPEED 3) policy.
1174 * bug 145b fix: compiler used wrong type specifier while converting
1175 MEMBER-types to numeric.
1176 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1178 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1179 index argument now works properly on the Alpha platform.
1180 * bug fix: floating point exception treatment on the Alpha platform
1182 * bug fix: FILE-POSITION works much better on string input and
1183 output streams. (thanks to Nikodemus Siivola)
1184 * bug fix: many threading/garbage collection symptoms sorted.
1185 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1186 might be pseudo-atomic.
1187 * internal change: Stopping for GC is now done with signals not
1188 ptrace. GC is now done in whichever thread wanted it, instead of
1190 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1192 * bug fix: result form in DO is not contained in the implicit
1194 * incompatible change: ICR structure is changed; the value part of
1195 CONTINUATION is now called LVAR; corresponding functions are
1196 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1197 * added type deriver for ISQRT (thanks to Robert E. Brown).
1198 * bug fix: better support for loading from the command line when an
1199 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1201 * fixed some bugs revealed by Paul Dietz' test suite:
1202 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1203 ** ROUND and FROUND now give the right answer when given very
1204 small float arguments.
1205 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1207 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1208 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1209 ** LCM should return a non-negative integer.
1210 ** PARSE-INTEGER returned the index of a terminator instead of the
1211 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1212 ** PARSE-INTEGER returned an incorrect index being applied to a
1214 ** LCM with two arguments of 0 returns 0 rather than signalling
1216 ** unsigned addition of a 32-bit constant with the high bit set no
1217 longer causes an internal compiler error on the x86.
1218 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1219 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1220 with negative last argument.
1221 ** byte specifiers with zero size and position no longer cause
1222 an error during type derivation.
1223 ** bignum multiplication on the Alpha platform now returns the
1225 * porting: The system now builds on SuSE AMD64, although it still
1226 generates a 32-bit binary.
1227 * .fasl file incompatibility: The fasl file version number has
1228 been incremented (because of the changes to internal compiler
1229 data structures referred to above).
1231 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1232 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1233 more accurately, on the Darwin kernel running on PowerPC hardware.
1234 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1235 * Compiler code deletion notes now signal a condition of type
1236 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1237 an associated MUFFLE-WARNING restart.
1238 * The compiler now performs limited argument count validation of
1239 constant format strings in FORMAT, and where appropriate in ERROR,
1240 CERROR and WARN. (thanks to Gerd Moellmann)
1241 * New ASDF-INSTALL contrib can be used for automatic download and
1242 installation of third-party Lisp code from CCLAN or other sites
1244 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1245 which forces another thread to execute a function supplied by the
1247 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1248 now accept and act upon their :ELEMENT-TYPE keyword argument.
1249 (reported by Martin Atzmueller, Edi Weitz)
1250 * bug fix: FILE-POSITION now accepts position designators up to
1251 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1252 is the greater. (thanks to Patrik Nordebo)
1253 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1254 Kalle Olavi Niemitalo)
1255 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1256 printable. (reported by Eric Marsden)
1257 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1258 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1259 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1260 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1261 (thanks to Stig Sandoe)
1262 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1263 UNREAD-CHAR now correctly includes the unread character in the
1264 target sequence. (thanks to Gerd Moellmann)
1265 * bug fix in threaded builds: the system can now be suspended and
1266 resumed by shell job control with minimal disruption.
1267 * bug fixes in times and timezones >2038AD
1268 * better handling of "where is GNU make?" problem in build scripts
1269 (thanks to Nikodemus Siivola)
1270 * new optimization: inside a named function any reference to a
1271 function with the same name is considered to be a self-reference;
1272 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1273 optimization quality.
1274 * new optimization on x86: logical functions and + now have
1275 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1276 used when the result is truncated to 32 bits.
1277 * VALUES declaration is partially enabled.
1278 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1279 * fixed some bugs revealed by Paul Dietz' test suite:
1280 ** The system now obeys the constraint imposed by
1281 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1282 types form a lattice under type intersection.
1283 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1284 ** ASSOC now ignores NIL elements in an alist.
1285 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1286 and (1+ MOST-POSITIVE-FIXNUM) answers.
1287 ** The addition of a method with invalid qualifiers to a generic
1288 function does not cause an error to be signalled immediately;
1289 a warning is signalled, and the error is generated only on
1290 calling the generic function.
1291 * changed .fasl file version number, in part to add type codes for
1292 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1293 obscure ANSI requirements
1295 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1296 * fixed bug 148: failure to inline-expand a local function left
1297 garbage, confusing the compiler.
1298 * fixed bugs 3cd: structure slot readers perform type check if the
1299 slot can have an invalid value (i.e. it is either not initialized
1300 or can be written with a less specific slot writer).
1301 * bug fix: the compiler now traps array references to elements off
1302 the end of an array; previously, the bounds checking in some
1303 circumstances could go off-by-one.
1304 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1306 * type declarations for array element types now obey the description
1307 on the CLHS page "Declaration TYPE", as per discussions on
1308 sbcl-help around 2003-05-08. This means that a declaration
1309 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1310 declaration, all references to BAR will be asserted or assumed
1311 (with THE, so dependent on compiler policy) to involve objects of
1312 type FOO. Note that no such declaration is implied in
1313 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1314 * declared types of functions from the "Conditions"
1315 chapter. (reported by Paul Dietz)
1316 * bug fix: CERROR accepts a function as its first argument.
1317 * bug fix: NTH an NTHCDR accept a bignum as index
1318 arguments. (reported by Adam Warner)
1319 * optimization: character compare routines now optimize comparing
1320 against a constant character. (reported by Gilbert Baumann)
1321 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1322 argument now works properly on the MIPS platform.
1323 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1324 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1325 declared to have a functional type, does not cause an error
1327 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1328 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1329 * fixed bug: if last continuation of a deleted block has a
1330 destination, this destination should be deleted too. (reported by
1332 * fixed a bug in the bootstrap process: the host compiler's values
1333 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1334 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1335 test case from Patrik Nordebo)
1336 * improved the ability of the disassembler on the PPC platform to
1337 provide helpful disassembly notes.
1338 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1339 the class in more cases than previously.
1340 * bug fix: FILE-POSITION now understands :START and :END for
1341 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1342 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1343 8) data. (thanks to David Lichteblau)
1344 * bug fix: it is possible to add a method to a generic function
1345 without lambda list.
1346 * bug fix: reader failed to signal END-OF-FILE inside an
1347 object representation. (reported by Nikodemus Siivola)
1348 * fixed some bugs revealed by Paul Dietz' test suite:
1349 ** LAST and [N]BUTLAST should accept a bignum.
1350 ** condition slot accessors are methods.
1351 ** (VECTOR NIL) is a subtype of STRING.
1353 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1354 * minor incompatible change: some nonsensical specialized lambda
1355 lists (used in DEFMETHOD) which were previously ignored now signal
1357 * minor incompatible change: the system is now aware of the types of
1358 variables in the COMMON-LISP package, and will signal errors for
1359 most violations of these type constraints (where previously they
1360 were silently accepted).
1361 * minor incompatible change: COMPILE-FILE now uses the freedom
1362 afforded (ANSI 3.2.2.3) to use derived function types for
1363 functions defined in the same file. This also permits the system
1364 to warn on static type mismatches and function
1365 redefinition. (Currently it does not work with high DEBUG level.)
1366 * minor incompatible change: VALUES declaration is disabled.
1367 * When issuing notes, the compiler now signals a condition of type
1368 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1369 restart for use in user handlers. It is expected that the
1370 COMPILER-NOTE condition will eventually become a condition
1371 supertype to a hierarchy of note types, which will then be
1372 handleable in a similar fashion. However, at the moment, no such
1373 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1374 but it's an implementation detail, not a classification for the
1376 * Changes in type checking closed the following bugs:
1377 ** type checking of unused values (192b, 194d, 203);
1378 ** template selection based on unsafe type assertions (192c, 236);
1379 ** type checking in branches (194bc).
1380 * A short form of VALUES type specifier has ANSI meaning (it has
1381 increased the number of situations when SBCL cannot perform type
1383 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1384 keywords or constants is permissible.
1385 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1386 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1387 classes. (thanks to Antonio Martinez)
1388 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1389 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1390 argument to a pretty-print stream. Instead, it leaves any such
1391 operations to DESCRIBE-OBJECT methods.
1392 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1394 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1395 lambda lists. (thanks to David Lichteblau)
1396 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1397 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1398 (thanks to Gerd Moellmann)
1399 * fixed bug 246: increased compilation speed of long
1400 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1402 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1404 * DEFCONSTANT now throws a condition of type
1405 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1406 constant to a non-EQL value; CONTINUE and ABORT restarts
1407 respectively change and preserve the value.
1408 * fixed bug 63: The code walker, part of the implementation of CLOS,
1409 is now better at handling symbol macros.
1410 * bug fix: There is no longer an internal implementation type named
1411 CL:LENGTH. (reported by Raymond Toy)
1412 * bug fix: In macro-like defining macros/special operators the
1413 implicit block does not enclose the lambda list.
1414 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1415 enclosed in parentheses) are not suitable as type specifiers, and
1416 their use properly signals an error now.
1417 * bug fix: An argument count mismatch for a type specifier in code
1418 being compiled no longer causes an unhandled error at compile
1419 time, but signals a compile-time warning.
1420 * fixed simple vector readable printing
1421 * bug fix: DESCRIBE takes more care over whether the class
1422 precedence list slot of a class is bound before accessing it.
1423 (reported by Markus Krummenacker)
1424 * bug fix: FORMATTER can successfully compile pretty-printer format
1425 strings which use variants of the ~* directive inside.
1426 * bug fix: SEARCH now applies its TEST predicate to the elements of
1427 the arguments in the correct order. (thanks to Wolfhard Buss)
1428 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1429 in type inference. (thanks to Robert E. Brown)
1430 * bug fix: Reading in symbols with an explicit package name of ""
1431 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1432 * fixed some bugs revealed by Paul Dietz' test suite:
1433 ** NIL is now allowed as a structure slot name.
1434 ** Arbitrary numbers, not just REALs, are allowed in certain
1435 circumstances in LOOP for-as-arithmetic clauses.
1436 ** Multiple class redefinitions before slot access no longer
1437 causes a type error.
1438 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1439 association between the name and a class.
1440 ** Generic functions with non-standard method-combination and over
1441 five methods all of which return constants no longer return NIL
1442 after the first few invocations. (thanks to Gerd Moellmann)
1443 ** CALL-NEXT-METHOD with no arguments now passes the original
1444 values of the arguments, even in the presence of assignment.
1445 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1447 ** DISASSEMBLE works with closures and funcallable instances.
1448 ** ADD-METHOD now returns the generic function, not the new method.
1449 ** FIND-METHOD signals an error if the lengths of the specializers
1450 is incompatible with the generic function, even if the ERRORP
1452 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1453 which its argument is a member.
1454 ** DEFCLASS only redefines the class named by its class-name
1455 argument if that name is the proper name of the class;
1456 otherwise, it creates a new class.
1457 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1458 of the UNBOUND-SLOT condition to the name of the slot.
1459 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1460 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1461 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1462 SLOT-MAKUNBOUND in the specified fashion.
1464 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1465 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1466 cross-compilation host. As a consequence, we can now bootstrap our
1467 way up to SBCL starting with a bare gcc toolchain and human-readable
1468 source code (first the source to CLISP, then the source to SBCL).
1469 * A contributed module containing a partial implementation of the
1470 simple-streams interface has been included. (thanks to Rudi
1472 * A contributed module implementing the RFC1321 Message Digest
1473 Algorithm, known as MD5, has been included.
1474 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1475 no longer has any effect, as the code controlled by this feature
1476 has been deleted. (As far as we know, no-one has ever built using
1477 this feature, and its semantics were confused in any case).
1478 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1479 work on conditions (as required by the ANSI specification),
1480 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1481 expected behaviour on conditions. Users should note, however,
1482 that such behaviour is not required by the ANSI specification,
1483 and so use of this behaviour may render their code unportable.
1484 * fixed some bugs revealed by Paul Dietz' test suite:
1485 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1487 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1488 ** SXHASH on condition objects no longer returns NIL.
1489 ** :ALLOCATION :CLASS slots are better treated; their values are
1490 updated on class redefinition, and initforms inherited from
1491 superclasses are applied.
1492 ** REMOVE-METHOD returns its generic function argument even when
1493 no method was removed.
1494 ** SHARED-INITIALIZE now initializes the values of the requested
1495 slots, including those with :ALLOCATION :CLASS.
1496 ** ALLOCATE-INSTANCE now works on structure classes defined via
1497 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1499 ** SLOT-EXISTS-P now works on conditions, as well as structures
1501 ** MAKE-LOAD-FORM now has the required methods on
1502 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1503 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1504 keyword, and now implements the SLOT-NAMES argument.
1505 ** methods with &OPTIONAL arguments no longer allow too many
1506 arguments to be passed in the call without error.
1507 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1508 option is consistent with the required arguments of the generic
1509 function lambda list.
1510 * bug fix: REQUIRE accepts a string designator. (Thanks to
1512 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1513 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1514 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1515 * bug fix: The compiler now checks for duplicated variables in macro
1517 * bug fix: SETQ on globals returns the correct value.
1518 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1519 not a COMPILER-ERROR (followed by some other strange error on
1520 choosing the CONTINUE restart).
1521 * bug fix: make.sh and friends are now more consistent in the way that
1522 they look for GNU "make".
1524 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1525 * experimental native threads support (on x86 Linux >=2.4 only).
1526 This is not compiled in by default: you need to add :SB-THREAD to
1527 the target features. See the "Beyond ANSI" chapter of the manual
1529 * fix for longstanding nonANSIism: The old distinction between
1530 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1531 The return value from CL:FIND-CLASS is now a CLOS class, and
1532 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1533 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1534 * An interface to the MetaObject Protocol, as described in Kiczales,
1535 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1536 MIT Press, 1991, is available from the SB-MOP package.
1537 * incompatible change: the SB-PCL package should now be considered
1538 a private implementation detail, and no longer a semi-private MOP
1540 * minor incompatible change: due to rearrangement for threads, the
1541 control stack and binding stack are now allocated at arbitrary
1542 addresses instead of being hardcoded per-port. Users affected by
1543 this probably have to be doing advanced things with shared
1544 libraries, and will know who they are.
1545 * minor incompatible change: Previously, all --eval forms used were
1546 processed with READ before any of them were processed with EVAL.
1547 Now each --eval form is processed with both READ and EVAL before
1548 the next --eval form is processed. (Thus package operations like
1549 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1550 work as the user might reasonably expect.)
1551 * minor incompatible change: *STANDARD-INPUT* is now only an
1552 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1554 * minor incompatible change: Y-OR-N-P is now character-oriented,
1555 not line oriented. Also, YES-OR-NO-P now works without errors.
1556 (thanks to Antonio Martinez)
1557 * sb-aclrepl module improvements: an integrated inspector, added
1558 repl features, and a bug fix to :trace command.
1559 * Known functions, which cannot be open coded by the backend, are
1560 considered to be able to check types of their arguments. (fixing
1561 a bug report by Nathan J. Froyd)
1562 * fixed a bug in computing method discriminating functions: It is
1563 now possible to define methods specialized on classes which have
1564 forward-referenced superclasses. (thanks to Gerd Moellmann)
1565 * fixed evaluation order in optional entries (reported by Gilbert
1567 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1568 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1569 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1570 required-by-AMOP NAME argument, as well as CLASS and
1571 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1572 * fixed bug 20: DEFMETHOD can define methods using names that are
1573 not the proper names of classes to designate class specializers.
1574 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1575 Unix file descriptors, instead of blowing up. (thanks to Antonio
1577 * Garbage collection refactoring: user-visible change is that a
1578 call to the GC function during WITHOUT-GCING will not do garbage
1579 collection until the end of the WITHOUT-GCING. If you were doing
1580 this you were probably losing anyway.
1581 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1582 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1583 TYPEP the latter but not the former.
1584 * The compiler issues a full WARNING for calls to undefined functions
1585 with names from the CL package.
1586 * MAP-INTO for a vector destination is open coded. (reported by
1587 Brian Downing on c.l.l)
1588 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1589 documentation string.
1590 * fixed some bugs revealed by Paul Dietz' test suite:
1591 ** COPY-ALIST now signals an error if its argument is a dotted
1593 ** Condition slots are now accessed more correctly in the presence
1594 of multiple initargs for a given slot.
1595 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1596 exclude restarts of the same name associated with a different
1598 ** DEFCLASS of forward-referenced classes with another
1599 forward-referenced class in the superclasses list no longer
1601 ** Condition slots are now initialized once each, not multiple
1602 times. (thanks to Gerd Moellmann)
1603 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1605 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1606 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1607 arguments contain duplicated elements.
1608 ** RESTART-CASE understands local macros.
1609 ** RESTART-CASE associates exactly its own restarts with a condition.
1610 ** ENDP in safe mode checks its argument to be of type LIST.
1611 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1612 in question is unbound.
1613 ** Optimized MAKE-INSTANCE functions no longer cause internal
1614 assertion failures in the presence of duplicate initargs.
1615 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1616 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1617 propagated even in the ctor optimized implementation of
1619 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1621 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1622 * a better implementation of SXHASH on (simple) bit vectors,
1623 measured both in execution speed and in distribution of results
1624 over the positive fixnums, has been installed. Likewise, a better
1625 implementation of EQUAL for simple bit vectors is now available.
1626 * fixed CEILING optimization for a divisor of form 2^k.
1627 * fixed bug 240 (emitting extra style warnings "using the lexical
1628 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1629 by Antonio Martinez)
1630 * fixed SXHASH, giving different results for NIL depending on type
1631 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1632 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1633 variable. (found by Rolf Wester)
1634 * SBCL does not ignore type declarations for special
1635 variables. (reported by rif on c.l.l 2003-03-05)
1636 * some bug fixes in contrib/sb-aclrepl/
1637 * fixed some bugs revealed by Paul Dietz' test suite:
1638 ** a bug in the CONS type specifier, whereby the CAR and CDR
1639 types got intertwined, has been fixed;
1640 ** the type system is now able to reason about the interaction
1641 between INTEGER and RATIO types more completely;
1642 ** APPEND, [N]REVERSE and NRECONC check that those their
1643 arguments, which must be proper lists, are really so;
1644 ** An array specialized to be unable to hold elements has been
1645 implemented, as required -- yes, really -- by ANSI;
1646 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1647 on malformed property lists;
1649 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1650 * incompatible packaging change: in line with Unix convention,
1651 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1652 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1653 * REQUIRE and PROVIDE are now optionally capable of doing something
1654 useful. See the documentation string for REQUIRE.
1655 * infrastructure for a managed SBCL contrib system: contributed
1656 modules in this release include:
1657 ** the ASDF system definition facility;
1658 ** an interface to the BSD Sockets API;
1659 ** an ACL-like convenience interface to the repl;
1660 (thanks to Kevin Rosenberg)
1661 ** an implementation of ROTATE-BYTE, with efficient implementation
1663 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1664 gives a closer approximation to the right answer than previously.
1665 (thanks to Raymond Toy)
1666 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1667 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1668 optional environment arguments, as required by ANSI.
1669 * fixed bugs in other functions taking environment objects, allowing
1670 calls with an explicit NIL environment argument to be compiled
1672 * fixed bug 228: primary return values from
1673 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1674 COMPILE or FUNCTION.
1675 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1676 :TYPE will no longer signal errors on innocuous objects.
1677 * fixed bug 231b: SETQ is better at respecting type declarations in
1678 the lexical environment.
1679 * fixed a bug in DEFCLASS: classes named by symbols with no or
1680 unprintable packages can now be defined.
1681 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1682 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1683 * fixed compiler failure related to checking types of functions.
1684 (reported by Robert E. Brown)
1685 * the compiler is now much more consistent in its error-checking
1686 treatment of bounding index arguments to sequence functions: in
1687 (SAFETY 3) code, errors will be signalled in almost all cases if
1688 invalid sequence bounding indices are passed to functions defined
1689 by ANSI to operate on sequences.
1690 * fixed a bug in the build procedure: documentation of SBCL-specific
1691 packages is now preserved and available in the final Lisp image.
1692 * lifted FDEFINITION lookup out of loops in the implementation of
1693 many list operations. (thanks to Robert E. Brown)
1694 * fixed a bug in the reader: the #n# reader macro now works for
1695 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1696 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1697 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1698 E. Brown and Tony Martinez respectively)
1699 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1700 SLOT-BOUNDP: in particular, it is now able to optimize them much
1701 better, and is now not vulnerable to having packages renamed.
1702 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1703 always called when a slot is not present in an instance. (thanks
1705 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1706 longer signal an error on unreading a character following EOF on
1707 the previous constituent stream. (thanks to Tony Martinez)
1708 * fixed some bugs revealed by Paul Dietz' test suite:
1709 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1710 not just nonnegative fixnums;
1711 ** the logical bit-array operators such as BIT-AND now accept an
1712 explicit NIL for their "opt-arg" argument (to indicate a
1713 freshly-consed result bit-array);
1714 ** ELT now signals an error on an invalid sequence index in safe
1716 ** the type system is now cleverer about negations of numeric
1717 types, and consequently understands the BIGNUM and RATIO types
1719 ** the type system is now cleverer about the interaction between
1720 INTEGER and RATIO types: while bugs still remain, many more
1721 cases are accurately computed;
1722 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1723 if it is in the last clause;
1724 ** CONSTANTLY now correctly returns a side-effect-free function in
1726 ** DECLARE is no longer treated as a special-operator; in
1727 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1728 * incremented fasl file version number due to the change in the
1729 DEFSTRUCT-SLOT-DESCRIPTION structure.
1731 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1732 * minor incompatible change: code processed by the "interpreter" or
1733 EVAL now has a compilation optimization policy of (DEBUG 2)
1734 (changed from (DEBUG 1)) to improve debuggability of interactive
1735 development, and to allow the use of the debug RETURN command in
1737 * an experimental implementation of the RETURN command for the
1738 debugger has been included. (thanks to Frederik Kuivinen)
1739 * fixed bug 62: constraints were not propagated into a loop.
1740 * fixed bug in embedded calls of SORT (reported and investigated by
1742 * fixed some bugs revealed by Paul Dietz' test suite:
1743 ** printing and reading of arrays with some dimensions having
1744 length 0 (thanks to Gerd Moellmann);
1745 ** BOA constructor with &AUX argument without a default value does
1746 not cause a type error;
1747 ** CONSTANTP now returns true for all self-evaluating objects.
1749 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1750 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1751 accessors that are related by inheritance, as specified in the
1752 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1753 Valtteri Vuorikoski)
1754 * The compiler is now able to inline functions that were defined in
1755 a complex lexical environment (e.g. inside a MACROLET).
1756 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1758 * Support for the upcoming FreeBSD-5.0 release has been included.
1759 (thanks to Dag-Erling Smorgrav)
1760 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1761 effect when it is not in a toplevel context.
1762 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1763 stand a better chance of being correct. (thanks to Gerd
1765 * fixed bug in COERCE, which now signals an error on coercing a
1766 rational to a bounded real type which excludes the expected
1768 * The compiler is now able to derive types more accurately from the
1769 COERCE and COMPILE functions.
1770 * fixed bug 223: functional binding is considered to be constant
1771 only for symbols in the CL package.
1772 * fixed bug 231: SETQ did not check the type of a variable being set
1773 (reported by Robert E. Brown)
1774 * A new optimization for MAKE-INSTANCE has been included, fixing
1775 various bugs (including relating to :ALLOCATION :CLASS slots and
1776 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1778 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1779 ** As required by ANSI, LOOP now disallows anonymous collection
1780 clauses such as COLLECT I in conjunction with aggregate boolean
1781 clauses such as THEREIS (= I 1);
1782 ** LOOP now signals an error when any variable is reused in the
1783 same loop (including the potentially useful construct analogous
1784 to WITH A = 1 WITH A = (1+ A);
1785 ** IT is only a special loop symbol within the first clause of a
1786 conditional loop clause;
1787 ** LOOP with a typed iteration variable over a hashtable now
1788 signals a type error iff it should.
1789 * fixed some other bugs revealed by Paul Dietz' test suite:
1790 ** FILE-STREAM now names the class previously known as FD-STREAM;
1791 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1792 argument) no longer signals an error;
1793 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1794 of the slot symbol, rather than using the current package
1795 ((:CONC-NAME "") continues to intern the slot's name in the
1797 * incremented fasl file version number, because of the incompatible
1798 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1799 of the new implementation of DEFINE-COMPILER-MACRO.
1801 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1802 * Support for building SBCL for MIPS platforms running in
1803 little-endian mode has now been checked in, and basic
1804 functionality on said platforms verified.
1805 * minor incompatible change: PCL now records the pathname of a file
1806 in which methods and the like are defined, rather than its
1808 * minor incompatible change: TRUENAME now considers the truename of
1809 a file naming a directory to be the pathname with :DIRECTORY
1810 component indicating that directory.
1811 * minor incompatible change: a NAMED clause in the extended form of
1812 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1813 reason for the previous behaviour is unclear.
1814 * more systematization and improvement of CLOS and MOP conformance
1815 in PCL (thanks to Gerd Moellman and Pierre Mai):
1816 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1818 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1819 primary methods with no specializers;
1820 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1822 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1823 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1824 FORWARD-REFERENCED-CLASSes; error reporting on
1825 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1827 ** SXHASH on CLOS instances now uses a slot internal to the
1828 instance to return different numbers on distinct instances,
1829 while preserving the same return value through invocations of
1831 ** DEFMETHOD signals errors when methods with longer incongruent
1832 lambda lists are added to generic functions;
1833 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1834 CLASS, as specified in AMOP;
1835 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1836 on the order returned by the primary method for classes of
1837 class STANDARD-CLASS;
1838 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1839 * fixed some bugs shown by Paul Dietz' test suite:
1840 ** DOLIST puts its body in TAGBODY;
1841 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1843 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1844 value producing form;
1845 ** if more variables are given to PROGV than values, extra
1846 variables are bound and made to have no value;
1847 ** NSUBSTITUTE on list arguments gets the right answer with
1849 ** ELT signals an error of type TYPE-ERROR when the index argument
1850 is not a valid sequence index;
1851 ** LOOP signals (at macroexpansion time) an error of type
1852 PROGRAM-ERROR when duplicate variable names are found;
1853 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1854 ** FUNCALL of special-operators now cause an error of type
1856 ** PSETQ now works as required in the presence of side-effecting
1857 symbol-macro places;
1858 ** NCONC accepts any object as its last argument;
1859 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1861 ** loop-for-as-package does not require a package to be explicitely
1863 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1865 * fixed bug 166: compiler preserves "there is a way to go"
1866 invariant when deleting code.
1867 * fixed bug 172: macro lambda lists with required arguments after
1868 &REST arguments now cause an error to be signalled. (thanks to
1870 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1871 bug. (thanks to Matthew Danish)
1872 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1874 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1875 arguments are lexically rebound. (thanks to Gerd Moellmann and
1877 * fixed bug 194: error messages are now more informative when there
1878 is no primary method applicable in a call to a generic
1879 function. (thanks to Gerd Moellmann)
1880 * fixed bug in command line argument checking (thanks to Julian
1882 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1883 (thanks to Matthew Danish)
1884 * incremented fasl file version number, because of the
1885 SXHASH-related changes in the layout of CLOS data structures
1887 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1888 * minor incompatible change: The runtime (the Unix executable named
1889 "sbcl") is now much pickier about the .core files it will load.
1890 Essentially it now requires .core files to descend from the same
1891 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1892 as the runtime does. (The intent is to prevent the crashes which
1893 can occur, and which can even be reported as mysterious failures,
1894 when people patch the sources or change the build parameters
1895 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1896 sbcl and .core files.)
1897 * fixed bug: VALUES-LIST is no longer optimized away.
1898 * fixed bug 142: The FFI conversion of C string values to Lisp
1899 string values no longer conses excessively. (thanks to Nathan
1900 Froyd porting Raymond Toy's fix to CMU CL)
1901 * began to systematize and improve MOP conformance in PCL (thanks to
1902 Nathan Froyd, Gerd Moellman and Pierre Mai):
1903 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1905 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1906 ** FINALIZE-INHERITANCE is now called on class finalization;
1907 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1908 argument precedence order.
1909 * fixed bug 202: The compiler no longer fails on functions whose
1910 derived types contradict their declared type.
1911 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1912 so it can be non-toplevel.
1913 * The fasl file version number has changed (because of the new
1914 implementation of DEFMACRO).
1915 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1916 safe code, that any length requirement by their type-specifier
1917 argument is valid. The exceptions to this are described in bug
1919 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1920 functions now check the types of their inputs as required by ANSI.
1921 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1922 introduced symbol is DECLAREd to be SPECIAL.
1923 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1924 * fixed bug: :COUNT argument to sequence functions may be negative
1925 * fixed bug: body of DO-SYMBOLS may contain declarations
1926 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1927 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1928 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1929 symbol macro only once
1930 * fixed printing of call frame when argument list is unavailable
1931 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1932 * compiler no longer signals WARNING on unknown keyword
1935 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1936 * A beta-quality port to the mips architecture running Linux,
1937 based on the old CMUCL backend, has been made. It has been tested
1938 on a big-endian kernel, and works sufficiently well to be able to
1939 rebuild itself; it has not been tested in little-endian mode.
1940 * fixed an inconsistency between gencgc.c and purify.c which made
1941 dumping/loading .core files unreliable
1942 * fixed bug 120a: The compiler now deals correctly with IFs where
1943 the consequent is the same as the alternative, instead of
1944 misderiving the return type. (thanks to Alexey Dejneka)
1945 * fixed bug 113: Logical pathnames are now dumpable (the logical
1946 host is resolved at load-time, throwing an error if it is not
1948 * fixed bug 174: FORMAT's error message is slightly clearer when a
1949 non-printing character is used in a format directive.
1950 * fixed several bugs in compiler checking of type declarations, i.e.
1951 violations of the Python "declarations are assertions" principle
1952 (thanks to Alexey Dejneka)
1953 * fixed several bugs in PCL's error checking (thanks to Gerd
1955 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1957 * fixed bug in compilation of functions as first class values
1958 (thanks to Antonio Martinez-Shotton)
1959 * The compiler's handling TYPE-ERRORs which it can prove will
1960 inevitably happen at runtime has been cleaned up and corrected
1961 in several ways. (thanks to Alexey Dejneka)
1962 * improved argument type checking for various basic arithmetic
1963 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1964 lavished on them in the past that they can be compiled in many
1965 ways in different special cases
1966 * fixed bug 181: compiler checks validity of user supplied type
1968 * cleaned up code flushing in optimization: Function calls which
1969 should signal errors for safety purposes (e.g. which ANSI says
1970 should signal errors when their arguments are of incorrect type)
1971 are no longer optimized away.
1972 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1973 * incremented fasl file version number, because changes in the
1974 implementation of sequence functions like COERCE caused
1975 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1976 in old inline expansions) to become undefined. (Actually these
1977 changes were later undone, so we might very well be binary
1978 compatible with 0.7.7 after all, but leaving the version number
1979 incremented seemed like the simplest and most conservative
1982 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1983 * An alpha-quality port to the parisc architecture running Linux,
1984 based on the old CMUCL backend, has been made. This, even more so
1985 than the other backends, should be considered still a work in
1986 progress; known problems include that the Linux kernel in 64-bit
1987 mode does not propagate the correct sigcontext structure to
1988 userspace, and consequently SBCL on a parisc64 kernel will not
1990 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1991 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1992 suppressed.) Also now that the compiler is looking at declarations
1993 in the environment, it checks optimization declarations as well,
1994 and suppresses inlining when (> DEBUG SPEED).
1995 * More fixes have been made to treatment of floating point exception
1996 treatment and other Unix signals. In particular, floating point
1997 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1998 * The detection and handling of control stack exhaustion (infinite
1999 or very deeply nested recursion) has changed. Stack exhaustion
2000 detection is now done by write-protecting pages at the OS level
2001 and applies at all optimization settings; when found, a
2002 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2003 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2004 be caught using IGNORE-ERRORS.
2005 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2006 correctly when transforming recursive LABELS functions to LETs.
2007 (thanks to Alexey Dejneka)
2008 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2009 that are names of constants or global variables.
2010 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2011 alien routines with docstrings.
2012 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2013 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2015 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2016 (thanks to Eric Marsden)
2017 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2018 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2019 * Bug fix: The SPARC backend can now compile functions involving
2020 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2021 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2022 to a functions expecting a PATHNAME-DESIGNATOR.
2023 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2024 lambda lists. (thanks to Alexey Dejneka)
2025 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2026 functions so that EXPORTing the name of the class doesn't cause
2027 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2028 to Antonio Martinez for reporting this)
2029 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2030 constant-folded without failing an assertion. (thanks to Einar
2031 Floystad Dorum for reporting this)
2032 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2033 types are now dealt with more correctly by the compiler.
2034 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2035 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2036 behaviour analogous to RENAME-FILE. This puts its behaviour more
2037 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2039 * The fasl file version number has changed again. (because of the
2040 bug fix involving the names of PCL MAKE-INSTANCE functions)
2042 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2043 * bug fix: Floating point exceptions are treated much more
2044 consistently on the x86/Linux and PPC/Linux platforms.
2045 * Array initialization with :INITIAL-ELEMENT is now much faster for
2046 cases when the compiler cannot open code the array creation, but
2047 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2048 array accesses have also seen a speed increase.
2049 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2050 specified. (This was at the root of some bad interactions between
2051 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2053 * bug fix: Internal error arguments for undefined functions are now
2054 computed correctly on the PPC/Linux platform.
2055 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2056 Raymond Toy's patch for CMU CL)
2057 * Support for the Solaris 9 operating environment has been included
2058 (thanks to Daniel Merritt)
2059 * A very ugly but hopefully complete draft of the missing FFI chapter
2060 of the manual has been created by reformatting the corresponding
2061 CMU CL manual chapter into (currently very ugly and incoherent)
2062 DocBook and bringing it up to date for SBCL behavior. Thus, the
2063 manual is now essentially complete, at least by my extreme
2064 once-and-only-once standards, whereby it's acceptable to refer to
2065 the doc strings of SB-EXT functions as the primary documentation.
2066 * The fasl file version number has changed again, due to cleanup of
2067 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2068 is no longer a static symbol.)
2070 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2071 * SBCL now builds with OpenMCL (version 0.12) as the
2072 cross-compilation host; also, more progress has been made toward
2073 bootstrapping under CLISP.
2074 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2076 * bug 158 fixed: The compiler can now deal with integer loop
2077 increments different from 1; fixing this turned out also to fix
2079 * bug 169 fixed: no more bogus warnings about using lexical bindings
2080 despite the presence of perfectly good SPECIAL declarations (thanks
2081 to David Lichteblau)
2082 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2083 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2084 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2086 * bug 180 fixed: Method combination specifications no longer ignore
2087 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2088 * bug fix: Structure type predicate functions now check their argument
2089 count as they should.
2090 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2091 correctly. (thanks to Pierre Mai)
2092 * minor incompatible change: The --noprogrammer option is deprecated
2093 in favor of the new --disable-debugger option, which is very similar.
2094 (The major difference is that it takes effect at a slightly different
2095 time at startup, causing handling of errors in --sysinit and
2096 --userinit files will be affected differently.) The
2097 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2098 been added to allow this functionality to be controlled from ordinary
2099 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2100 maintainers, who might want to run non-interactive scripts to
2101 build SBCL cores which will later be used interactively.)
2102 * minor incompatible change: The LOAD function no longer, when given
2103 a wild pathname to load, loads all files matching that pathname.
2104 Instead, an error of type FILE-ERROR is signalled.
2106 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2107 * bug 147 fixed: The compiler preserves its block link/count
2108 invariants more correctly now so that it doesn't crash. (thanks
2110 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2112 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2113 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2114 no-value-for-this-character as NIL (as specified by ANSI).
2115 * HOST-NAMESTRING on physical pathnames now returns a string that is
2116 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2117 (thanks to Christophe Rhodes)
2118 * The Alpha port handles icache flushing more correctly. (thanks to
2120 * More progress has been made toward bootstrapping under CLISP. (thanks
2121 to Christophe Rhodes)
2122 * The fasl file format has changed again, because dynamic loading
2123 on OpenBSD (which has non-ELF object files) motivated some cleanups
2124 in the way that foreign symbols are transformed and passed around.
2125 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2126 is no longer treated as whitespace by the reader, but instead as
2127 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2128 |A
\7fB|, instead of A as it used to.
2130 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2131 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2132 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2133 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2134 this as of 0.7.1.45, but was left out of the previous news section
2135 (thanks to Dan Barlow)
2136 * SBCL now runs on the Solaris operating system on SPARC architectures
2137 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2138 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2139 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2140 * SPARC backend cleanups, allowing builds of cores optimized for V8
2141 and V9 SPARCS, and also emission of code targeted to a particular
2142 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2144 * SBCL is closer to bootstrapping under CLISP, thanks to various
2145 fixes by Christophe Rhodes.
2146 * The fasl file format has changed again, to allow the compiler's
2147 INFO database to support symbol macros.
2148 * The user manual (in doc/) is formatted into HTML more nicely.
2149 (thanks to coreythomas)
2150 * The system is smarter about SUBTYPEP relationships, especially
2151 those involving NOT types (including types such as ATOM which are
2152 represented internally using NOT types). Thus SUBTYPEP is less
2153 likely to return (VALUES NIL NIL) in general, and in particular
2154 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2156 * The fasl file format has changed again, because the internal
2157 representation of types now includes a new slot to support the new
2158 SUBTYPEP-of-NOT-types logic.
2159 * (not a change in the main branch of SBCL, but a related prototype
2160 which can hopefully be merged into the main branch of SBCL in the
2161 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2162 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2163 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2164 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2165 dispatch macro character combinations. (thanks to Alexey Dejneka)
2166 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2167 that superficially look like logical namestrings correctly.
2168 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2170 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2171 default supplied-p) for &optional and &key arguments. (thanks to
2174 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2175 * incompatible change: The compiler is now less aggressive about
2176 tail call optimization, doing it only when (> SPACE DEBUG) or
2177 (> SPEED DEBUG). (This is an incompatible change because there are
2178 programs which relied on the old CMU-CL-style behavior to optimize
2179 away their unbounded recursion which will now die of stack overflow.)
2180 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2181 for non-GENCGC systems has been increased to 20M (since that
2182 seems much closer to the likely performance optimum for modern
2183 systems than the old 4M value was)
2184 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2185 and *DEBUG-PRINT-LEVEL*
2186 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2187 of CMU CL's support for SPARC, and various endianness and other
2188 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2189 * new syntactic sugar for the Unix command line: --load foo.bar is now
2190 an alternate notation for --eval '(load "foo.bar")'.
2192 ** The system now detects stack overflow and handles it gracefully,
2193 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2194 optimization settings. (This is a good thing in general, and
2195 its introduction in this version should be particularly timely
2196 for anyone whose code fails because of suppression of tail
2198 ** The system now hunts for the C variable "environ" in a more
2199 devious way, to avoid segfaults when the C library version differs
2200 between compile time and run time. (thanks to Christophe Rhodes)
2201 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2202 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2203 ** The compiler no longer issues bogus style warnings for undefined
2204 classes in the same source file as the DEFCLASSes which defined
2205 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2207 ** fixes in CONDITION class precedence list for undefined function
2208 errors (thanks to Alexei Dejneka)
2209 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2210 correctly. (thanks to Dan Barlow)
2211 ** portability fixes aiming at bootstrapping under CLISP (thanks
2212 to Dave McDonald and Christophe Rhodes)
2213 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2214 ** fixes in type translation and and type inference (thanks to
2216 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2217 ** various fixes in the new ports (thanks to Dan Barlow)
2218 * several changes related to debugging:
2219 ** suppression of tail recursion, as noted above
2220 ** stack overflow detection, as noted above
2221 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2222 is now the default. (For some time encapsulation has been more
2223 reliable than the breakpoint-based :ENCAPSULATE NIL
2224 implementation, at least on X86 systems; and I just noticed that
2225 encapsulation also seems closer to the spirit of the ANSI
2228 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2230 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2231 up properly again. (There was a packaging bug in 0.7.0 which
2232 left their definitions in SB-SYS::LOAD-FOREIGN and
2233 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2234 vital for most things which interface to C-level interfaces,
2235 like extensions working with sockets or databases or
2236 Perl-compatible regexes or whatever, and the need to fix
2237 this bug is the main reason that 0.7.1 was released so
2239 ** DEFGENERIC is now choosier about the methods it redefines, so that
2240 reLOADing a previously-LOADed file containing DEFGENERICs does
2241 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2242 cycle now works as it should. (thanks to Alexey Dejneka)
2243 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2244 fixed by Christophe Rhodes. (He actually submitted this patch
2245 months ago, and I delayed until after 0.7.0.)
2246 ** Bug 111 (internal compiler confusion about runtime checks on
2247 FUNCTION types) was fixed by Alexey Dejneka.
2248 * Some internal cleanups (getting rid of variables which aren't
2249 needed now that the byte interpreter is gone) caused the fasl
2250 file format number to change again.
2252 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2253 * major incompatible change: The default fasl file extension, i.e. the
2254 default extension for files produced by COMPILE-FILE, has changed
2255 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2257 ** There are many changes in the implementation of the compiler.
2258 SBCL is now essentially a compiler-only implementation of ANSI
2259 Common Lisp. EVAL still "interprets" a few special cases, but
2260 almost all the interesting cases are handled by creating
2261 a LAMBDA expression, calling COMPILE on it, then calling
2262 FUNCALL on the result.
2263 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2264 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2265 Since the code is newer, there might still be some new bugs
2266 (though not as many as before Martin Atzmueller's fixes:-). But
2267 the new code is substantially simpler and clearer, and hopefully
2268 any remaining bugs will be simpler, less fundamental, and more
2269 fixable then the bugs in the old code.
2270 ** The revised compiler is still a little unsteady on its feet.
2272 *** The debugging information it produces (particularly the names
2273 of FUNCTION objects) is sometimes much less useful than what
2274 the old compiler produced.
2275 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2276 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2277 within a MACROLET) has been temporarily weakened.
2278 ** There are new compiler optimizations for various functions:
2279 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2280 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2281 *** the math functions TRUNCATE, FLOOR, and CEILING
2282 *** the function-of-all-trades COERCE
2283 Mostly these should be transparent, but there's one
2284 potentially-annoying problem (bug 117): when the compiler
2285 inline-expands a function and does type analysis on the result,
2286 it can create control paths which have type mismatches, and
2287 when it can't prove that those control paths aren't taken,
2288 it will issue WARNINGs about the type mismatches. This is
2289 a particular problem in practice for the new sequence functions.
2290 It's not clear how this should be fixed, and for now, a
2291 workaround is given in the entry for 117 in the BUGS file.
2292 ** (Because of the interaction between the two previous items --
2293 occasional inlining problems and new inline expansions -- some
2294 of the new sequence function optimizations won't really kick in
2295 completely until debugging information, and then inlining, are
2296 straightened out in some future version.)
2297 * minor incompatible changes:
2298 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2299 DIRECTORY no longer implicitly promotes NIL slots of its
2300 pathname argument to :WILD. In particular, when you ask for the
2301 contents of a directory (which you used to be able to do without
2302 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2303 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2304 ** changes in behavior that ANSI explicitly defines to be
2305 implementation dependent:
2306 *** The new compiler-only implementation still conforms with ANSI,
2307 but acts a little different than before. Besides the obvious
2308 changes in performance tradeoffs (that the cost per form passed
2309 to EVAL has gone up, and the cost per form executed by EVAL
2310 has gone down), the behavior of the system changes a little
2311 because there are no longer any interpreted function objects.
2312 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2313 e.g. doing COMPILE on the output of interactive DEFUN is
2315 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2317 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2318 has changed from (:CL) to NIL.
2319 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2320 e.g. #\Newline and #\Tab, don't have names specified in the
2321 ANSI Common Lisp standard, is now based on their ASCII symbolic
2322 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2323 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2324 are no longer used for output.
2325 ** changes in internal implementation constants:
2326 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2327 4 million. (If your application spends a lot of time GCing and
2328 you have a lot of RAM, you might want to experiment with
2329 increasing it even more.)
2330 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2331 However, almost all old code should still continue to work without
2332 immediate update, as SB-C-CALL is now a (deprecated) nickname
2334 ** Old operator names in the style DEF-FOO are now deprecated in
2335 favor of new corresponding names DEFINE-FOO, for consistency with
2336 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2337 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2338 internal symbols, but a few supported extensions like
2339 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2340 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2341 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2342 etc. as you get deeper into recursive calls to the debugger
2343 command loop, instead of the old "5]", "5]]", "5]]]"
2344 sequence. (I was motivated to do this when squabbles between
2345 ILISP and SBCL left me very deeply nested in the debugger. In the
2346 short term, this change will probably provoke more ILISP/SBCL
2347 squabbles, but hopefully it will be an improvement in the long run.)
2348 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2349 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2350 function, since declaiming return types involving aliens is
2351 (1) annoyingly messy to do by hand and (2) vital to efficient
2352 compilation of code which calls such functions.
2353 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2354 longer reexported by the SB-EXT package. They're solely useful
2355 for alien code, so it seems more logical that you should get
2356 them from the SB-ALIEN package, not in SB-EXT.
2357 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2358 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2359 features. Instead, the code that they used to control is always
2360 built into the system.
2361 * many other bug fixes
2362 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2363 advantage of the new EVAL-WHEN stuff and to clean them up in
2364 general, and they are now more ANSI-compliant in a number of
2365 ways. Martin Atzmueller is responsible for a lot of this.
2366 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2368 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2369 *** correct ERROR type for various file operations
2370 *** some fixes for Lisp streams
2371 *** DEFMETHOD syntax checking
2372 *** changing old weird representation of debug information as
2373 strings (which, among their other deficiencies, don't transform
2374 correctly when you rename packages, and don't change their
2375 print representation when you change things like *PACKAGE*
2376 and *PRINT-LENGTH*) to symbols and lists of symbols
2377 He also made several improvements and fixed several bugs in DESCRIBE.
2378 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2380 *** misbehavior of WRITE-STRING/WRITE-LINE
2381 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2382 and several other LOOP problems as well
2383 *** DIRECTORY when similar filenames are present
2384 *** DEFGENERIC with :METHOD options
2385 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2386 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2387 *** argument ordering in FIND with :TEST option
2388 *** mishandled package designator argument in APROPOS-LIST
2389 *** various problems in the backquote readmacro
2390 *** a bug in APROPOS
2391 *** probably some others that I'm not describing very well here,
2392 since the CVS log documents them by reference to sbcl-devel
2393 messages, and the SourceForge archives aren't working well.:-(
2394 ** Dan Barlow improved the Alpha port (and is making progress on the
2395 PPC port, for those of you who think different).
2396 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2397 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2398 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2399 continues to work on the SPARC port (for those of you in a position
2400 to look down upon our little PC-compatible boxes from a great height).
2401 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2402 and uses that copy, rather than the current dynamic value, when
2403 it's trying to decide whether to truncate output. Thus e.g.
2404 (let ((*print-lines* 50))
2405 (pprint-logical-block (stream nil)
2407 (let ((*print-lines* 8))
2408 (print (aref possiblybigthings i) stream)))))
2409 should now truncate the logical block only at 50 lines, instead of
2410 often truncating it at 8 lines, as it did before.
2411 * The doc/cmucl/ directory, containing old CMU CL documentation
2412 from the time of the fork, is no longer part of the base system.
2413 SourceForge has shut down its anonymous FTP service, and with it
2414 my original plan for distributing the old CMU CL documentation
2415 there. For now, if you need these files you can download an old
2416 SBCL source release and extract them from it.
2417 * The fasl file version number changed again, for dozens of reasons,
2418 some of which are apparent above.
2420 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2421 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2422 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2423 patch, so that DISASSEMBLE output is much nicer.
2424 * The code in the SB-PROFILE package now seems reasonably stable.
2425 I still haven't decided what the final interface should look like
2426 (I'd like PROFILE to interact cleanly with TRACE, since both
2427 facilities use function encapsulation) but if you have a need
2428 for profiling now, you can probably use it successfully with
2429 the current CMU-CL-style interface.
2430 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2431 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2432 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2433 link now returns the dangling link itself, and for similar
2434 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2435 link itself. (In these cases the old code signalled an error and
2436 looped endlessly, respectively.) Thus, DIRECTORY now works even
2437 in the presence of dangling and cyclic symbolic links.
2438 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2439 is now a supported extension again, since the consensus on
2440 sbcl-devel was that it can be useful for ordinary development
2441 work, not just for debugging SBCL itself.
2442 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2443 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2444 that currently-defined functions might be redefined later with
2445 different return types.
2446 * Hash tables can be printed readably, as inspired by CMU CL code
2447 of Eric Marsden and SBCL code of Martin Atzmueller.
2448 * better error handling in CLOS method combination, thanks to
2449 Martin Atzmueller porting Pierre Mai's CMU CL patches
2450 * more overflow fixes for >16Mbyte I/O buffers
2451 * A bug in READ has been fixed, so that now a single Ctrl-D
2452 character suffices to cause end-of-file on character streams.
2453 In particular, now you only need one Ctrl-D at the command
2454 line (not two) to exit SBCL.
2455 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2457 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2458 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2459 does the right thing.
2460 * The implementation of some type tests, especially for CONDITION
2461 types, is now tidier and maybe faster, due to CMU CL code
2462 originally by Douglas Crosher, ported by Martin Atzmueller.
2463 * Some math functions have been fixed, and there are new
2464 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2465 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2466 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2467 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2468 likely to be built into sbcl-0.7.0.)
2469 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2470 CL, ported to SBCL by Martin Atzmueller)
2471 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2472 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2473 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2474 * There's a new slam.sh hack to shorten the edit/compile/debug
2475 cycle for low-level changes to SBCL itself, and a new
2476 :SB-AFTER-XC-CORE target feature to control the generation of
2477 the after-xc.core file needed by slam.sh.
2478 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2479 COMPILE-FILE is no longer supported, so that now every function
2480 gets an entry point, so that block compilation looks a little
2481 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2482 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2483 returns the number of bytes consed since the system started,
2484 rather than the number consed since the first time the function
2485 was called. (The new definition parallels ANSI functions like
2486 CL:GET-INTERNAL-RUN-TIME.)
2487 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2488 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2489 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2490 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2491 * Fasl file version numbers are now independent of the target CPU,
2492 since historically most system changes which required version
2493 number changes have affected all CPUs equally. Similarly,
2494 the byte fasl file version is now equal to the ordinary
2497 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2498 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2499 is no longer recognized. I apologize for this, because it was
2500 listed in SB-EXT as a supported extension, but I found that
2501 its existing behavior was poorly specified, as well as incorrectly
2502 specified, and it looked like too much of a mess to straighten it
2503 out. I have enough on my hands trying to get ANSI stuff to work..
2504 * many patches ported from CMU CL by Martin Atzmueller, with
2505 half a dozen bug fixes in pretty-printing and the debugger, and
2506 half a dozen others elsewhere
2507 * fixed bug 13: Floating point infinities are now supported again.
2508 They might still be a little bit flaky, but thanks to bug reports
2509 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2510 as flaky as they were.
2511 * The --noprogrammer command line option is now supported. (Its
2512 behavior is slightly different in detail from what the old man
2513 page claimed it would do, but it's still appropriate under the
2514 same circumstances that the man page talks about.)
2515 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2516 are now supported, and enabled by default. Thus, the compiler can
2517 handle many floating point and complex operations much less
2518 inefficiently. (Thus e.g. you can implement a complex FFT
2520 * The compiler now detects type mismatches between DECLAIM FTYPE
2521 and DEFUN better, and implements CHECK-TYPE more correctly, and
2522 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2523 thanks to patches from Martin Atzmueller.
2524 * various fixes to make the cross-compiler more portable to
2525 ANSI-conforming-but-different cross-compilation hosts (notably
2526 Lispworks for Windows, following bug reports from Arthur Lemmens)
2527 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2528 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2529 fixed thanks to Pierre Mai's CMU CL patches.
2530 * fixes to keep the system from overflowing internal counters when
2531 it tries to use i/o buffers larger than 16M bytes
2532 * fixed bug 45a: Various internal functions required to support
2533 complex special functions have been merged from CMU CL sources.
2534 (When I was first setting up SBCL, I misunderstood a compile-time
2535 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2536 * improved support for type intersection and union, fixing bug 12
2537 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2538 more obscure bugs as well
2539 * some steps toward byte-compiling non-performance-critical
2540 parts of the system, courtesy of patches from Martin Atzmueller
2541 * Christophe Rhodes has made some debian packages of sbcl at
2542 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2543 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2544 stable, but are nonetheless usable. When he's ready, I'd be happy
2545 to add them to the SourceForge "File Releases" section. (And if
2546 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2547 * new fasl file format version number (because of changes in
2548 internal representation of (OR ..) types to accommodate the new
2549 support for (AND ..) types, among other things)
2551 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2552 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2554 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2556 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2557 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2558 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2559 * DESCRIBE now works on structure objects again.
2560 * Most function call argument type mismatches are now handled as
2561 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2562 know whether the function will be redefined before the call is
2563 executed. (The compiler could flag local calls with full WARNINGs,
2564 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2565 it doesn't keep track of enough information to know whether calls
2566 are local in this sense.)
2567 * Compiler output is now more verbose, with messages truncated
2568 later than before. (There should be some supported way for users
2569 to override the default verbosity, but I haven't decided how to
2570 provide it yet, so this behavior is still controlled by the internal
2571 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2572 src/compiler/ir1util.lisp.)
2573 * Fasl file format version numbers have increased again, because
2574 support for the Gray streams extension changes the layout of the
2575 system's STREAM objects.
2576 * The Gray subclassable streams extension now works, thanks to a
2577 patch from Martin Atzmueller.
2578 * The full LOAD-FOREIGN extension (not just the primitive
2579 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2580 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2581 but like most other programs, it defaults to copying the Unix
2582 environment from the original process instead of starting the
2583 new process in an empty environment.
2584 * Extensions which manipulate the Unix environment now support
2585 an :ENVIRONMENT keyword option which doesn't smash case or
2586 do other bad things. The CMU-CL-style :ENV option is retained
2587 for porting convenience.
2588 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2589 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2591 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2593 * A patch from Martin Atzmueller seems to have solved the SIGINT
2594 problem, and as far as we know, signal-handling now works cleanly.
2595 (If you find any new bugs, please report them!)
2596 * The system no longer defaults Lisp source file names to types
2597 ".l", ".cl", or ".lsp", but only to ".lisp".
2598 * The compiler no longer uses special default file extensions for
2599 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2600 seems to expect a single default extension for all compiled code,
2601 and there's no compelling reason to try to stretch the standard
2602 to allow two different extensions.) Instead, byte-compiled files
2603 default to the same extension as native-compiled files.
2604 * Fasl file format version numbers have increased again, because
2605 a rearrangement of internal implementation packages made some
2606 dumped symbols in old fasl files unreadable in new cores.
2607 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2608 many fewer weird special cases.
2609 * Bug #17 (differing COMPILE-FILE behavior between logical and
2610 physical pathnames) has been fixed, and some related misbehavior too,
2611 thanks to a patch from Martin Atzmueller.
2612 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2613 by Tim Moore, ported to SBCL by Martin Atzmueller.
2614 * Martin Atzmueller fixed several filesystem-related problems,
2615 including bug #36, in part by porting CMU CL patches, which were
2616 written in part by Paul Werkowski.
2617 * More compiler warnings in src/runtime/ are gone, thanks to
2618 more patches from Martin Atzmueller.
2619 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2622 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2624 * DESCRIBE now works on CONDITION objects.
2625 * The debugger now handles errors which arise when trying to print
2626 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2628 * The build system now uses an additional file, customize-target-features.lisp,
2629 to allow local modifications to the target *FEATURES* list. (The point of
2630 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2631 debugging features enabled, without having to worry about propagating it
2632 into everyone's system when I do a "cvs update".) When no
2633 customize-target-features.lisp file exists, the target *FEATURES* list
2634 should be constructed the same way as before.
2635 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2636 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2637 warn about behavior which is undefined under the ANSI spec. Note: This
2638 is specified by ANSI, but it's not very popular with programmers.
2639 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2640 macro in the SBCL sources for an example of a workaround which you
2641 might use to make portable ANSI-standard code which does what you want.
2642 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2643 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2644 when it's used as a non-toplevel form. (This is required in order
2645 to implement the DEFCONSTANT-EQX macro.)
2646 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2647 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2648 funny things, due to bugs in the implementation of EVAL-WHEN
2649 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2650 non-toplevel DEFCONSTANTs are uncommon.)
2651 * The core file version number and fasl file version number have been
2652 incremented, because the old noncompliant DEFCONSTANT behavior involved
2653 calling functions which no longer exist, and because I also took the
2654 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2655 * fixed bug 1 (error handling before read-eval-print loop starts), and
2656 redid debugger restarts and related debugger commands somewhat while
2658 ** The QUIT debugger command is gone, since it did something
2659 rather different than the SB-EXT:QUIT command, and since it never
2660 worked properly outside the main toplevel read/eval/print loop.
2661 Invoking the new TOPLEVEL restart provides the same functionality.
2662 ** The GO debugger command is also gone, since you can just invoke
2663 the CONTINUE restart directly instead.
2664 ** The TOP debugger command is also gone, since it's redundant with the
2665 FRAME 0 command, and since it interfered with abbreviations for the
2667 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2669 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2670 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2671 some porting work by Martin Atzmueller)
2672 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2673 with a boatload of patches which he ported to SBCL. Now that those
2675 ** The system tries to make sure that its low-priority messages
2676 are prefixed by semicolons, to help people who like to use
2677 syntax highlighting in their ILISP buffer. (This patch
2678 was originally due to Raymond Toy.)
2679 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2680 patches originally written by Raymond Toy.
2681 ** The compiler understands coercion between single-value and
2682 multiple-VALUES type expressions better, getting rid of some very
2683 weird behavior, thanks to patches originally by Robert MacLachlan
2684 and Douglas Crosher.
2685 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2686 lambda lists, thanks to a patch originally by Pierre Mai.
2687 ** The system no longer bogusly warns about "abbreviated type
2689 ** The compiler gets less confused by inlining and RETURN-FROM,
2690 thanks to some patches originally by Tim Moore.
2691 ** The system no longer hangs when dumping circular lists to fasl
2692 files, thanks to a patch originally from Douglas Crosher.
2693 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2694 undefined function error.
2695 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2696 in structure slot initforms) for now, documented workaround instead:-|
2697 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2698 * fixed bug 5: added stubs for various Gray stream functions called
2699 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2700 installed, at least appropriate type errors are generated
2701 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2702 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2703 consistently in DEFMETHOD forms.
2704 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2705 it doesn't seem to affect SBCL after all
2706 * The C runtime system now builds with better optimization and many
2707 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2709 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2711 * The system is now under CVS at SourceForge (instead of the
2712 CVS repository on my home machine).
2713 * The new signal handling code has been tweaked to treat register
2714 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2715 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2716 downstream, e.g. in debug-int.lisp, has implicit dependencies
2717 on the unsignedness of integer representation of machine words,
2718 and that caused the system to bomb out with infinite regress
2719 when trying to recover from type errors involving signed values,
2720 e.g. (BUTLAST '(1 2 3) -1).)
2721 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2722 (This was one of the bugs Peter Van Eynde reported back in July.)
2723 * The system now uses code inspired by Colin Walters' O(N)
2724 implementation of MAP (from the cmucl-imp@cons.org mailing
2725 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2726 inline the MAP operation, and there is more than one
2727 sequence argument to the MAP call (so that it can't just
2728 do ETYPECASE once and for all based on the type of the
2729 single sequence argument). (The old non-inline implementation
2730 of the general M-argument sequence-of-length-N case required
2731 O(M*N*N) time when any of the sequence arguments were LISTs.)
2732 * The QUIT :UNIX-CODE keyword argument has been renamed to
2733 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2734 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2736 * Raymond Wiker's patches to port dynamic loading from Linux to
2737 FreeBSD have been added.
2738 * The BUGS file is now more nearly up to date, thanks in large part
2739 to Martin Atzmueller's review of it.
2740 * The debugger now flushes standard output streams before it begins
2741 its output ("debugger invoked" and so forth).
2742 * The core version number and fasl file version number have both
2743 been incremented, because of incompatible changes in the layout
2745 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2746 used to not be called for a saved Lisp image.)
2747 * Martin Atzmueller's version of a patch to fix a compiler crash,
2748 as posted on sbcl-devel 13 September 2000, has been installed.
2749 * Instead of installing Martin Atzmueller's patch for the
2750 compiler transform for SUBSEQ, I deleted the compiler transform,
2751 and transforms for some similar consing operations.
2752 * A bug in signal handling which kept TRACE from working on OpenBSD
2754 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2755 away by constant folding
2756 * The system now defines its address space constants in one place
2757 (in the Lisp sources), and propagates them automatically elsewhere
2758 (through GENESIS and the sbcl.h file). Therefore, patching the
2759 address map is less unnecessarily tedious and error-prone. The
2760 Lisp names of address space constants have also been systematized.
2761 * CVS tags like dollar-Header-dollar have been removed from
2762 the sources, because they have never saved me trouble and
2763 they've been source of trouble working with patches and other
2764 diff-related operations.
2765 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2766 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2768 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2770 * The system has been ported to OpenBSD.
2771 * The system now compiles with a simple "sh make.sh" on the systems
2772 that it's supported on. I.e., now you no longer need to tweak
2773 text in the target-features.lisp-expr and symlinks in src/runtime/
2774 by hand, the make.sh takes care of it for you.
2775 * The system is no longer so grossly inefficient when compiling code
2776 involving vectors implemented as general (not simple) vectors (VECTOR T),
2777 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2778 which dares to use the various sequence functions on non-simple
2779 vectors, takes less of a performance hit.
2780 * There is now a primitive type predicate VECTOR-T-P
2781 to test for the (VECTOR T) type, so that e.g.
2782 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2783 can now be compiled with some semblance of efficiency. (The old code
2784 turned the type declaration into a full call to %TYPEP at runtime!)
2785 * AREF on (VECTOR T) is still not fast, since it's still compiled
2786 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2787 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2788 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2789 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2790 without making full calls to SUBTYPEP at runtime.
2791 (Some analogous efficiency issues for non-simple vectors specialized to
2792 element types other than T, or for non-simple multidimensional arrays,
2793 have not been addressed. They could almost certainly be handled the
2794 same way if anyone is motivated to do so.)
2795 * The changes in array handling break binary compatibility, so
2796 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2797 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2798 instead of (VALUES T T).
2799 * By following the instructions that Dan Barlow posted to sbcl-devel
2800 on 2 July 2000, I was able to enable primitive dynamic object
2801 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2802 functionality is not implemented (since it calls ld to resolve
2803 library references automatically, requiring RUN-PROGRAM for its
2804 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2805 try to resolve library references) is now supported.
2806 * The system now flushes the standard output streams when it terminates,
2807 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2808 them at several other probably-convenient times, e.g. in each pass of
2809 the toplevel read-eval-print loop, and after evaluating a form given
2810 as an "--eval" command-line option. (These changes were motivated by a
2811 discussion of stream flushing issues on cmucl-imp in August 2000.)
2812 * The source transform for TYPEP of array types no longer assumes
2813 that an array whose element type is a not-yet-defined type
2814 is implemented as an array of T, but instead punts, so that the
2815 type will be interpreted at runtime.
2816 * There is now some support for cross-compiling in make.sh: each of
2817 the phases of make.sh has its own script. (This should be transparent
2818 to people doing ordinary, non-cross-compile builds.)
2819 * Since my laptop doesn't have hundreds of megabytes of memory like
2820 my desktop machine, I became more motivated to do some items on
2821 my to-do list in order to reduce the size of the system a little:
2822 ** Arrange for various needed-only-at-cold-init things to be
2823 uninterned after cold init. To support this, those things have
2824 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2825 symbols with such names are now uninterned after cold init).
2826 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2827 fasl files for cold load.
2828 ** Remove the old compiler structure pooling code (which used to
2829 be conditional on the target feature :SB-ALLOC) completely.
2830 ** Redo the representation of some data in cold init to be more compact.
2831 (I also looked into supporting byte compiled code at bootstrap time,
2832 which would probably reduce the size of the system a lot, but that
2833 looked too complicated, so I punted for now.)
2834 * The maximum signal nesting depth in the src/runtime/ support code has
2835 been reduced from 4096 to 256. (I don't know any reason for the very
2836 large old value. If the new smaller value turns out to break something,
2837 I'll probably just bump it back up.)
2838 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2840 * Many, many bugs reported by Peter Van Eynde have been added to
2841 the BUGS list; some have even been fixed.
2842 * While enabling dynamic object file loading, I tried to make the
2843 code easier to understand, renaming various functions and variables
2844 with less ambiguous names, and changing some function calling
2845 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2846 * While trying to figure out how to do the OpenBSD port, I tried to
2847 clean up some of the code in src/runtime/. In particular, I dropped
2848 support for non-POSIX signal handling, added various comments,
2849 tweaked the code to reduce the number of compilation warnings, and
2850 renamed some files to increase consistency.
2851 * To support the new automatic configuration functionality in make.sh,
2852 the source file target-features.lisp-expr has been replaced with the
2853 source file base-target-features.lisp-expr and the machine-generated
2854 file local-target-features.lisp-expr.
2855 * fixed a stupid quoting error in make.sh so that using CMU CL
2856 "lisp -batch" as cross-compilation host works again
2858 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2860 * DESCRIBE no longer tries to call itself recursively to describe
2861 bound/fbound values, so that it no longer fails on symbols which are
2862 bound to themselves (like keywords, T, and NIL).
2863 * DESCRIBE now works on generic functions.
2864 * The printer now prints less-screwed-up representations of closures
2865 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2866 * A private symbol is used instead of the :EMPTY keyword previously
2867 used to mark empty slots in hash tables. Thus
2868 (DEFVAR *HT* (MAKE-HASH-TABLE))
2869 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2870 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2871 now does what ANSI says that it should. (You can still get
2872 similar noncompliant behavior if bang on the hash table
2873 implementation with all the symbols you get back from
2874 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2875 This breaks binary compatibility, since tests for equality to
2876 :EMPTY are wired into things like the macroexpansion of
2877 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2879 * There's now a minimal placeholder implementation for CL:STEP,
2880 as required by ANSI.
2881 * An obscure bug in the interaction of the normal compiler, the byte
2882 compiler, inlining, and structure predicates has been patched
2883 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2884 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2886 * Missing ordinary arguments in a macro call are now detected even
2887 when the macro lambda list contains &KEY or &REST.
2888 * The debugger no longer complains about encountering the top of the
2889 stack when you type "FRAME 0" to explicitly instruct it to go to
2890 the top of the stack. And it now prints the frame you request even
2891 if it's the current frame (instead of saying "You are here.").
2892 * As specified by ANSI, the system now always prints keywords
2893 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2895 * The default initial SIZE of HASH-TABLEs is now smaller.
2896 * Type information from CLOS class dispatch is now propagated
2897 into DEFMETHOD bodies, so that e.g.
2898 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2900 is now basically equivalent to
2901 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2902 (DECLARE (TYPE SINGLE-FLOAT X))
2904 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2905 operation, without having to do run-time type dispatch.
2906 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2907 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2908 * Since I don't seem to be making big file reorganizations very often
2909 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2910 is overflowing my ability to conveniently back them up), I've finally
2911 checked the system into CVS. (The CVS repository is on my home system,
2912 not at SourceForge -- putting it on SourceForge might come later.)
2913 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2914 high-level GC-NOTIFY-FOO functions send their output. (There's
2915 still very little control of where low-level verbose GC functions
2916 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2917 less than it used to -- the GC-NOTIFY-FOO functions are now under
2918 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2919 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2920 in only one place in the source code, and propagates it automatically
2921 everywhere that it's needed. Thus e.g. when I bump the version from
2922 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2923 * The C source files now include boilerplate legalese and documentation
2924 at the head of each file (just as the Lisp source files already did).
2925 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2926 to his page will be replaced with a link to his new CLiki service.
2928 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2930 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2931 * The build process now looks for GNU make under the default name "gmake",
2932 instead of "make" as it used to. If GNU make is not available as "gmake"
2933 on your system, you can change this default behavior by setting the
2934 GNUMAKE environment variable.
2935 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2936 can build without error under CMU CL.
2938 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2940 * There is now a partial SBCL user manual (with some new text and some
2941 text cribbed from the CMU CL manual).
2942 * The beginnings of a profiler have been added (starting with the
2943 CMU CL profiler and simplifying and cleaning up). Eventually the
2944 main interface should be through the TRACE macro, but for now,
2945 it's still accessed through vaguely CMU-CL-style functions and macros
2946 exported from the package SB-PROFILE.
2947 * Some problems left over from porting CMU CL to the new
2948 cross-compilation bootstrap process have been cleaned up:
2949 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2950 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2951 part when getting rid of the compiler *BACKEND* stuff.)
2952 ** The value of *NULL-TYPE* was screwed up, because it was
2953 being initialized before the type system knew the final
2954 definition of the 'NULL type. This screwed up several key
2955 optimizations in the compiler, causing inefficiency in all sorts
2956 of places. (I found it because I wanted to understand why
2957 GET-INTERNAL-RUN-TIME was consing.)
2958 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2959 PROCLAIM FTYPE information. Unfortunately this broke binary
2960 compatibility again, since now the forms output by DEFGENERIC
2961 to refer to functions which didn't exist in 0.6.3.
2962 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2963 can use the new (as of 0.6.3) transform for SOME into MAP into
2965 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2966 values to (INTEGER 0 1000000), so that the time code will no longer
2967 occasionally get blown up by Linux returning 1000000 microseconds
2968 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2969 its output conform to the ANSI spec. (Alas, this makes its output
2970 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2971 * A full call to MAP NIL with a single sequence argument no longer conses.
2972 * fixes to problems pointed out by Martin Atzmueller:
2973 * The manual page no longer talks about multiprocessing as though
2974 it were currently supported.
2975 * The ILISP support patches have been removed from the distribution,
2976 because as of version 5.10.1, ILISP now supports SBCL without us
2977 having to maintain patches.
2978 * added a modified version of Raymond Toy's recent CMU CL patch for
2979 EQUALP comparison of HASH-TABLE
2981 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2983 * The system still can't cross-compile itself with
2984 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2985 entails), but at least it can compile more of itself that way
2986 than it used to be able to, and various buglets which were uncovered
2987 by trying to cross-compile itself that way have now been fixed.
2988 * This release breaks binary compatibility again. This time
2989 at least I've incremented the FASL file format version to 2, so that the
2990 problem can be detected reliably instead of just causing weird errors.
2991 * various new style warnings:
2992 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2993 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2994 ** using the lexical binding of a variable named in the *FOO* style
2995 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2996 as specified by ANSI.
2997 * *RANDOM-STATE* is no longer automatically initialized from
2998 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2999 default behavior of the system is to repeat its behavior every time
3000 it's run. If you'd like to change this behavior, you can always
3001 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3002 old convention there was no comparably easy way to get the system to
3003 repeat its behavior every time it was run.
3004 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3005 FTYPE declarations has been removed, in favor of their ANSI
3006 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3007 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3008 the types of their sequence arguments can be determined at compile time.
3009 This is done through a new open code expansion for MAP which eliminates
3010 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3011 argument types can be determined at compile time.
3012 * The optimizer now transforms COERCE into an identity operation when it
3013 can prove that the coerced object is already of the correct type. (This
3014 can be a win for machine generated code, including the output of other
3015 optimization transforms, such as the MAP transform above.)
3016 * Credit information has been moved from source file headers into CREDITS.
3017 * Source file headers have been made more standard.
3018 * The CASE macro now compiles without complaining even when it has
3021 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3023 * (Note that the way that the PCL macroexpansions were rewritten
3024 to accommodate the change in DEFGENERIC below breaks binary
3025 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3026 not run under sbcl-0.6.2. Once we get out of alpha releases,
3027 i.e. hit release 1.0.0, we'll probably try to maintain binary
3028 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3029 and sbcl-1.4.4. Until then, however, it might be fairly common
3030 for maintenance releases to break binary compatibility.)
3031 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3033 * The reporting of unhandled signals has been changed to print some
3034 explanatory text as well as the report form. (Previously only
3035 the report form was printed.)
3036 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3037 it defines, so that the compiler no longer issues undefined function
3038 warnings for compiled-but-not-yet-loaded generic functions.
3039 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3040 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3041 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3042 section of the ANSI spec are supported.
3043 * The "" nickname for the "KEYWORD" package has been removed.
3044 The reader still handles symbol tokens which begin with a package marker
3045 as keywords, but it doesn't expose its mechanism for doing so in the
3046 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3047 * The system now issues STYLE-WARNINGs for contradictory TYPE
3048 proclamations. (Warnings for contradictory FTYPE proclamations would
3049 be nice too, but those can't be done usefully unless the type system
3050 is made smarter about FUNCTION types.)
3051 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3052 systematized, so that "*target-*.lisp is supposed to exist only on the
3053 target and imply that there's a related file which exists on the
3054 host, and *host-*.lisp is supposed to exist only on the host and imply
3055 that there's a related file which exists on the target. This involves a
3056 lot of renaming. Hopefully the acute confusion caused by the renaming
3057 will be justified by the reduction in chronic confusion..
3058 ** runtime-type.lisp -> early-target-type.lisp
3059 ** target-type.lisp -> late-target-type.lisp
3060 ** early-host-format.lisp -> early-format.lisp
3061 ** late-host-format.lisp -> late-format.lisp
3062 ** host-error.lisp -> misc-error.lisp
3063 ** early-error.lisp -> early-target-error.lisp
3064 ** late-error.lisp -> late-target-error.lisp
3065 ** host-defboot.lisp -> early-defboot.lisp
3066 ** code/misc.lisp -> code/target-misc.lisp
3067 ** code/host-misc.lisp -> code/misc.lisp
3068 ** code/numbers.lisp -> code/target-numbers.lisp
3069 ** code/early-numbers.lisp -> numbers.lisp
3070 ** early-host-type.lisp -> early-type.lisp
3071 ** late-host-type.lisp -> late-type.lisp
3072 ** host-typep.lisp -> typep.lisp
3073 ** load.lisp -> target-load.lisp
3074 ** host-load.lisp -> load.lisp
3075 ** host-disassem.lisp -> disassem.lisp
3076 ** host-insts.lisp -> insts.lisp
3077 ** byte-comp.lisp -> target-byte-comp.lisp
3078 ** host-byte-comp.lisp -> byte-comp.lisp
3079 ** host-signal.lisp -> signal.lisp
3080 ** host-defstruct.lisp -> defstruct.lisp
3081 ** late-target-type.lisp -> deftypes-for-target.lisp
3082 Furthermore, several other previously target-only files foo.lisp (e.g.
3083 hash-table.lisp and random.lisp) have been split into a target-and-host
3084 foo.lisp file and a target-only target-foo.lisp file, with their key type
3085 definitions in the target-and-host part, so that the cross-compiler will
3086 know more about target types.
3087 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3088 gone away. In their place are various *BACKEND-FOO* variables
3089 corresponding to the slots of the old structure.
3090 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3091 into the target SBCL has been fixed.
3092 * The chill.lisp system for loading cold code into a running SBCL
3094 * Support for the CMU CL "scavenger hook" extension has been removed.
3095 (It was undocumented and unused in the CMU CL sources that SBCL was
3096 derived from, and stale in sbcl-0.6.1.)
3097 * Various errors in the cross-compiler type system were detected
3098 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3099 (enabling various consistency checks). Many of them were fixed,
3100 but some hard problems remain, so the compiler is back to
3101 running without *TYPE-SYSTEM-INITIALIZED* for now.
3102 * As part of the cross-compiler type system cleanup, I implemented
3103 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3104 * I have started adding UNCROSS calls throughout the type system
3105 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3106 on cross-compiler input files will be able to go away, and various
3108 * CONSTANTP now returns true for quoted forms (as explicitly required
3111 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3113 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3114 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3115 (MAKE-LIST -1). In the longer term, it ought to become true
3116 that declarations are assertions even at SAFETY 1. For now, it's not
3117 quite true even at SAFETY 3, but it's at least more nearly true..
3118 (Note that this change seems to increases the size of the system by
3119 O(5%) and to decrease the speed of the compiler by 20% or more.)
3120 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3121 for the problem of printing dozens of lines of distracting information
3122 about low-level system machinery as part of the top stack frame
3123 on entry to the debugger when an undefined function was called.
3124 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3125 is not reset to COMMON-LISP-USER.
3126 * Compilation of stuff related to dyncount.lisp has been made conditional
3127 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3128 smaller. The various dyncount-related symbols have been moved into
3129 a new "SB-DYNCOUNT" package.
3130 * tty-inspect.lisp has been renamed to inspect.lisp.
3131 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3132 feature has gone away. (When we eventually port to other flavors of
3133 libc and/or Unix, we'll try to make the differences between flavors
3134 invisible at the user level.)
3135 * Various other *FEATURES* tags, and/or their associated conditionals,
3136 have been removed if obsolescent, or given better documentation, or
3137 sometimes given more-mnemonic names.
3139 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3141 * tidied up "make.sh" script
3142 * tidied up system directory structure
3143 * better "clean.sh" behavior
3144 * added doc/FOR-CMUCL-DEVELOPERS
3145 * many many small tweaks to output format, e.g. removing possibly-confusing
3146 trailing #\. character in DESCRIBE-INSTANCE
3147 * (EQUALP #\A 'A) no longer signals an error.
3148 * new hashing code, including EQUALP hashing
3149 * tidied up Lisp initialization and toplevel
3150 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3151 * command line argument processing
3152 * added POSIX-GETENV function to deal with Unix-ish environment variables
3153 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3154 terminating SBCL on EOF
3155 * non-verbose GC by default
3156 * There is no more "sbcl" shell script; the sbcl file is now the C
3157 runtime executable (just like CMU CL).
3158 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3160 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3161 symbol and package information as Lisp native symbol and package objects
3162 instead of strings naming symbols and strings naming packages. This way,
3163 whenever packages are renamed (as in warm init), debug information is
3164 transformed along with everything else.
3165 * tweaked the optimization policy declarations which control the building
3166 of SBCL itself. Now, among other things, the system no longer saves
3167 source location debugging information. (This helps two problems at once
3168 by reducing SBCL size and by keeping SBCL from trying to look for its
3169 sources -- which may not exist -- when reporting errors.)
3170 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3171 debugging and testing purposes
3172 * cleaned up printing, making the printer call PRINT-OBJECT for
3173 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3174 methods, giving nearly-ANSI behavior
3175 * converted almost all special variables to use *FOO* naming convention
3176 * deleted PARSE-TIME functionality, since it can be done portably
3177 * moved some files out of cold init into warm init
3178 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3180 * regularized formatting of source files
3181 * added an install.sh script
3182 * fixed ridiculous memory usage of cross-compiler by making
3183 compiler/alloc.lisp not try to do pooling unless it can hook
3184 itself into the GC of the cross-compilation host. Now the system
3185 builds nicely on my old laptop.
3186 * added :SB-ALLOC in target-features.lisp-expr
3187 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3188 was not implemented)
3189 * re-did condition handling and note reporting in the compiler. Notes
3190 are no longer handled by signalling conditions. Style warnings
3191 and warnings are handled more correctly and reported in such a way
3192 that it's easy to find one or the other in your output (so that you
3193 can e.g. figure out which of many problems caused COMPILE-FILE to
3195 * changed the severity of several compiler warnings from full WARNING
3196 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3197 compiler note reporting so that it doesn't use the condition system
3198 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3199 COMPILE-FILE command)
3200 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3201 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3202 no longer supported, since their implementation was deeply intertwingled
3203 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3204 * removed bogus "support" for compiler macros named (SETF FOO), and
3205 removed the compiler macro for SETF INFO (but only after making a fool
3206 of myself on the cmucl-imp mailing list by posting a bogus patch for
3207 DEFINE-COMPILER-MACRO..)
3208 * Compiled files containing forms which have side effects on the Lisp
3209 reader (such as DEFPACKAGE forms) are now handled more correctly.
3210 (Compiler queuing of top level lambdas has been suppressed by setting
3211 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3212 * deleted various currently-unused source files, e.g. gengc.lisp. They
3213 may be added back at some point e.g. when porting to other architectures,
3214 but until they are it's distracting to distribute them and to try to
3216 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3217 there were so many of them they're just distractions, and UNCROSS is
3218 known to be able to handle the current sources
3219 * moved PROFILE functionality into TRACE, so that it will be clear
3220 how the wrapping and unwrapping of functions when you profile them
3221 interacts with the wrapping and unwrapping of functions when you
3222 trace them. (Actually, the functionality isn't there yet, but at least
3223 the interface specification is there. Hopefully, the functionality will
3224 arrive with some future maintenance release.)
3225 * removed host-oops.lisp
3226 * changed signature of QUIT function to allow UNIX-CODE argument
3227 * fixed READ-SEQUENCE bug
3228 * tweaked verbose GC output so that it looks more like the progress
3229 output that ANSI specifies for functions like LOAD
3230 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3231 * added <http://sbcl.sourceforge.com> to the banner information printed by