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 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
10 pathname is a directory pathname.
11 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
12 * fixed some lockups due to gc/thread interaction
13 * dynamic space size on PPC has been increased to 768Mb. (thanks to
15 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
16 :METACLASS argument in addition to a class name. (reported by
17 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellman)
18 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
20 * bug fix: more cleanups to the floating point exception handling on
21 x86-64 (thanks to James Knight)
22 * optimization: improved the MIPS versions of generic (in the
23 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
24 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
25 the wrong file descriptor; implementation of SOCKET-OPEN-P.
26 (thanks to Tony Martinez)
27 * fixed some bugs revealed by Paul Dietz' test suite:
28 ** invalid dotted lists no longer raise a read error when
30 ** signal an error if a symbol that names a declaration is used
31 as the name of a type, or vice versa
32 ** allow using the (declare (typespec var*)) abbreviation for
33 (declare (type typespec var*)) with all type specifiers
34 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
35 ** always return NIL from PPRINT-POP when OBJECT is NIL
36 ** don't signal errors when pretty-printing malformed LABELS,
37 FLET or MACROLET forms
39 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
40 * fixed cross-compiler leakages that prevented building a 32-bit
41 target with a 64-bit host compiler.
42 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
43 opened with :IF-EXISTS :OVERWRITE.
44 * fixed bug 281: error for an invalid qualifier in a short-form method
45 combination method is not signalled until the faulty method is called.
46 * bug fix: iteration variable type inferrer failed to deal with open
47 intervals. (reported by Alan Shields)
48 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
49 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
50 or more runtime options were provided to the sbcl binary.
51 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
52 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
53 * optimization: there's now a fast-path for fixnum arguments in the
54 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
55 * optimization: the code generated on x86-64 is more compact thanks
56 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
57 * bug fix: floating-point exception handling now partly works on
58 x86-64. (Thanks to James Knight)
59 * improvement to the MIPS backend from Thiemo Seufer: C-style
60 64-bit long long arguments and return values to alien functions
62 * fixed some bugs revealed by Paul Dietz' test suite:
63 ** the type-error signalled from WARN has a filled-in DATUM slot.
64 ** the type-error required when a stream is not associated with
65 a file has the stream as its datum.
66 ** type-errors on single-floats on x86-64 no longer have
67 :INVALID-OBJECT as the datum
68 ** the type-errors signalled for invalid function names now have
69 a correct expected type
70 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
71 for structures defined with a :TYPE. Documentation strings for
72 typed structures are no longer immediately discarded
73 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
74 broadcast streams and synonym streams. FILE-LENGTH now also works
77 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
78 * incompatible change: the --noprogrammer option, deprecated since
79 version 0.7.5, has been removed. Please use the equivalent
80 --disable-debugger option instead.
81 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
83 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
84 inoperational for a while now) has been completely removed.
85 * null lexical environments are now printed as #<NULL-LEXENV>,
86 significantly reducing the amount of clutter in typical backtraces.
87 * documentation on weak pointers, finalization, and after GC hooks
88 has been added to the manual.
89 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
90 as other specialized array types, is much faster. SUBSEQ and
91 COPY-SEQ on such arrays have also been sped up.
92 * optimization: EQL is now more efficient when at least other argument
93 is known to be of type (OR FIXNUM (NOT NUMBER)).
94 * fixed bug: the runtime is now less vulnerable to changes in the
95 size of the SBCL object on OS X, and virtual memory is reserved for
96 all spaces that need to be at a fixed address.
97 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
98 * fixed bug: finalizers and after GC hooks that cause consing are now
100 * fixed bug: compiler error messages and summaries are now printed to
101 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
102 * fixed inference of the upper bound of an iteration variable.
103 (reported by Rajat Datta).
104 * fixed bug 211e: calling local functions with duplicated constant
105 keyword argument no longer causes a bogus style warning about an
107 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
108 ftype declaration to be disregarded. (reported by Dave Roberts)
109 * fixed bug 373: caused by erronous compilation of references to alien
110 variables in the runtime on ppc/darwin.
111 * fixed bug 376: CONJUGATE type deriver.
112 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
113 (reported by Baughn on #lisp)
114 * a cross-compiler bug on non-x86 platforms has been identified and
115 fixed. (thanks to Bruno Haible)
116 * improvements to the MIPS runtime code for increased stability.
117 (thanks to Thiemo Seufer)
118 * increased the maximimum compact environment size to allow
119 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
120 * partial workaround for bug 135: don't name the function we're
121 calling for hairy cases of EVAL, so as not to accumulate one environment
122 entry per call to EVAL. (reported by Kevin Reid)
123 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
124 * optimization: type testing for non-vector arrays should be faster.
125 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
127 * bug fix: setting 31st element of a bit vector to zero did not work
129 * bug fix: redefining a class definition which failed due to a
130 previous accessor / function clash now works (but see BUGS entry
131 #380 for more problems in this area). (thanks to Zach Beane)
132 * the long form of DEFINE-METHOD-COMBINATION disables method group
133 checking when given a single method group with pattern *. (thanks
135 * on x86 compiler supports stack allocation of results of simple
136 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
137 * support for building and running on Mac OS X 10.4 "Tiger" has been added
138 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
139 * fixed some bugs related to Unicode integration:
140 ** the restarts for recovering from input and output encoding
141 errors only appear when there is in fact such an error to
143 * fixed some bugs revealed by Paul Dietz' test suite:
144 ** MISC.549 and similar: late transformation of unsafe type
145 assertions into derived types caused unexpected code
147 ** SCALE-FLOAT type deriver is less wrong.
148 ** type derivers for EXP, LOG and similar functions compute result
149 types for complex arguments better.
150 ** (MISC.563) CONJUGATE type deriver works for very restricted
152 ** out-of-line type testers for character strings are available.
153 ** EQUAL compiler transform understands specialness of objects
155 ** accessing double-floats stored on the stack now works on x86-64.
156 ** debugger internals could sometimes create invalid lispobjs,
157 resulting in GC crashes.
158 ** MISC.548: type check weakening can convert required type into
160 ** initialization forms for bindings are not in scope of free special
163 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
164 * incompatible change: thread support for non-NPTL systems has
165 been removed - locking is buggy and unreliable. A threaded
166 SBCL build will now warn at startup and refuse to create
167 new threads, unless futex support is detected in the kernel
168 * incompatible change: the top level REPL now has only an ABORT
169 restart associated with it, not TOPLEVEL and ABORT as it used to.
170 TOP and TOPLEVEL are now available as debugger commands for
171 returning to the top level.
172 * incompatible change: forms evaluated in the REPL now use the
173 global optimization policy.
174 * incompatible change: user- and system-initialization files are
175 no longer processed with LOAD, but by READ and EVAL; hence the
176 global optimization policy, startup package, readtable, etc,
178 * The .fasl file format number has been incremented because of
179 various incompatible changes.
180 * internal entry point details and argument counts no longer appear
181 in backtraces unless explicitly requested by setting
182 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
183 * built-in and standard functions no longer have names like "top
184 level local call to FOO".
185 * fixed bug 32: functions defined in non-null lexical environments
186 now have more legible printed representation
187 * fixed bug 33: functions defined in non-null lexical environemnts
188 are now more amenable to inspection by INSPECT.
189 * workaround for bug 354: XEPs no longer appear in backtraces unless
190 explicitly requested.
191 * fixed bug: receiving the signal which results from attempting to
192 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
193 SIGBUS on other *BSDs) on architectures where the C stack is also
194 the Lisp stack (x86 and x86-64) from foreign code no longer leads
195 to debugger confusion or wild instability. (reported by Cheuksan
197 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
198 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
199 (reported by Lutz Euler)
200 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
201 compiled code. (reported by Paul Dietz)
202 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
203 specialized parameter to a method to something that is not TYPEP
204 the specializer is now possible.
205 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
206 face of package deletion.
207 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
208 pathname. (reported by David Tolpin on c.l.l)
209 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
210 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
211 STRING-STREAM and FILE-STREAM, respectively.
212 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
213 than previously; in addition, its test suite is now run on build.
214 (thanks to Luke Gorrie)
215 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
216 * optimization: added a immediate representation for single-floats
218 * fixed some bugs revealed by Paul Dietz' test suite:
219 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
220 ** attempting to create a package with a colliding nickname causes
221 correctable errors to be signalled.
222 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
223 ** several x86-64 backend bugs related to sign-extension of immediate
226 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
227 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
228 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
230 * fixed loading of multiply forward-referenced layouts.
231 (thanks to Cheuksan Wang)
232 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
233 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
234 (see COMPILE-FILE documentation for details of :PRINT option)
235 * fixed bugs 19 and 317: fixed-format floating point printing is
236 more accurate. This also fixes a bug reported by Adam Warner
237 related to the ~@F format directive.
238 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
240 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
241 dispatch table if the from-char is a dispatch macro character.
242 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
243 by the wordsize no longer ignore the last word. (reported by Lutz
245 * fixed bug in type checking of dynamic-extent variables. (reported
247 * optimization: sequence traversal functions use their freedom to
248 coerce function designators to functions.
249 * optimization: code with many calls to CLOS methods specialized on
250 CLOS classes has had redundant type checks removed.
251 * optimization: added declarations to speed up operations that access
252 the internal character database (for example STRING-UPCASE)
253 * optimization: comparison operations between floats and sufficiently small
254 fixnums no longer create extra rationals
255 * fixed some bugs related to Unicode integration:
256 ** portions of multibyte characters at the end of buffers for
257 character-based file input are correctly transferred to the
258 start of the buffer at the next read.
259 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
260 passing it through to OPEN.
261 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
262 argument given to internal calls to OPEN.
263 ** the built-in comment readers (introduced by character sequences
264 ";" and "#|") are more forgiving to encoding errors; they will
265 STYLE-WARN and then attempt to resync the stream at a character
266 boundary. (thanks to Teemu Kalvas)
267 * fixed some bugs in the x86-64 port:
268 ** Negative short int return values from c-calls are sign-extended
270 ** The stack is aligned to 16-bytes for c-calls, as required by
271 the ABI. (thanks to Cheuksan Wang)
272 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
273 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
274 for enums. (thanks to Vincent Arkesteijn)
275 ** Multiple small optimizations and bugfixes for floating point
277 * fixed some bugs revealed by Paul Dietz' test suite:
278 ** Space, Tab, Linefeed, Return and Page have the invalid
279 secondary constituent character trait.
280 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
282 ** WITH-INPUT-FROM-STRING should only update the index place on
284 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
286 ** Bit-array operations (BIT-AND and similar) worked incorrectly
287 with one-dimensional arrays with fill pointers.
288 ** TYPE-OF failed on a complex with an integer realpart and a
290 ** compiler failure during type inference for the code of form
291 (IF (EQL X (THE ...)) ...) (MISC.535).
293 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
294 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
295 platform. The port supports 61-bit fixnums, large memory spaces
296 and reloading shared object files.
297 * enhancement: saving cores with foreign code loaded is now
298 supported on ppc/Darwin in addition to the previously supported
300 * enhancement: the statistical profiler now walks deeper into the
301 call stack for more meaningful call-graphs and accrued time
302 reports (x86/x86-64 only). It also now reports time spent in
304 * enhancement: it is now possible to trace most individual methods
305 of a generic function in addition to tracing the generic function
307 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
309 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
310 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
311 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
312 produces an error. (thanks to Vincent Arkesteijn)
313 * bug fix: NAMESTRING on pathnames with :WILD components in their
314 directories works correctly. (thanks to Artem V. Andreev)
315 * fixed bug 125: compiler preserves identity of closures. (reported
317 * bug fixed: functions with &REST arguments sometimes failed with
318 "Undefined function" when compiled with (DEBUG 3). (reported by
320 * bug fix: overflow during compiling of setting element of a bit
321 vector with constant index and value. (reported by Timmy Douglas)
322 * build fix: fixed the dependence on *LOAD-PATHNAME* and
323 *COMPILE-FILE-PATHNAME* being absolute pathnames.
324 * on x86 compiler partially supports stack allocation of dynamic-extent
326 * GO and RETURN-FROM do not check the extent of their exit points
327 when compiled with SAFETY 0.
328 * fixed some bugs related to Unicode integration:
329 ** encoding and decoding errors are now much more robustly
330 handled; it should now be possible to recover even from invalid
331 input or output to the terminal. (thanks to Teemu Kalvas)
332 ** provided a first cut at implementing STRING-TO-OCTETS and
333 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
334 ** altered the SB-MD5 contributed module slightly, changing the
335 interface just enough for it to be supportable for builds where
336 lisp characters are not eight bits.
337 * fixed some bugs revealed by Paul Dietz' test suite:
338 ** the FORMATTER-generated functions for ~V[ conditionals require
339 the correct number of arguments.
340 ** READ-FROM-STRING returns the mandated second value when applied
341 to displaced strings.
342 ** the #\Rubout and #\Backspace characters are treated as invalid
343 constituent characters by the tokenizer.
345 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
346 * new feature: reloading changed shared object files with
347 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
348 * new feature: references to foreign variables and functions
349 can now be compiled and loaded before the corresponding shared
350 object file is loaded, as long as the foreign definitions are
351 available at runtime.
352 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
353 supported. (thanks to Dan Debertin)
354 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
355 just keywords. (thanks to Vincent Arkesteijn)
356 * fixed bug #331: structure-class instances corresponding to
357 DEFSTRUCT forms are now created eagerly.
358 * fixed bug #345: backtraces from calls to undefined functions work
359 on x86 as well. Related bug #61 is now also partially fixed on x86
360 (backtraces from throws to unknown catch tags.)
361 * bug fix: lambda-list parsing is now stricter vrt. order and number
362 of lambda-list keywords.
363 * bug fix: as specified by AMOP, an error is signalled if a
364 class-option appears multiple times in a DEFCLASS form. (reported
366 * bug fix: RANDOM can be compiled when the compiler derives the type
367 of its numeric argument as a disjoint set of small integers.
368 (reported by Paul Dietz)
369 * bug fix: starting a core saved with shared objects loaded when
370 those objects are not available no longer causes threaded SBCL to
371 hang. (reported by Sean Ross)
372 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
374 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
375 argument, which must always be NIL. (reported by Kalle Niemitalo)
376 * bug fix: printing 1.0d+23 no longer results in an error.
377 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
378 * bug fix: structure slot setters preserve evaluation order. (thanks
380 * fixed some bugs related to Unicode integration:
381 ** RUN-PROGRAM can allow its child to take input from a Lisp
382 stream. (reported by Stefan Scholl)
383 ** ASDF-INSTALL successfully downloads .tar.gz files in all
384 locales. (reported by Ken Causey)
385 * fixed some bugs revealed by Paul Dietz' test suite:
386 ** INCF, DECF and REMF evaluate their place form as specified in
388 ** FORMATTER expands ~{ ~} iteration directives with V or #
389 parameters correctly.
390 ** FORMATTER deals with the ~@[ ~] conditional directive where the
391 consequent uses no arguments correctly.
392 ** the system has a partial understanding of the (COMPLEX RATIO)
394 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
397 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
398 * new feature: a build-time option (controlled by the :SB-UNICODE
399 keyword feature, enabled by default) for building the system with
400 support for the entire 21-bit character space defined by the
402 * new feature: the system now has rudimentary external-format
403 support; the primary user-visible change at this time is that
404 characters with the high bit set (such as lower-case-e-acute) will
405 print correctly to a terminal in a UTF-8 environment.
406 * minor incompatible change: BASE-CHAR no longer names a class;
407 however, CHARACTER continues to do so, as required by ANSI.
408 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
409 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
410 has been moved to the SB-EXT package (temporarily re-exported from
412 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
413 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
415 * fixed bug #351: better error-handlind and reporting for malformed
417 * fixed bug #350: bignum-printing is now more memory-efficient,
418 allowing printing of very large bignums, eg. (expt 2 10000000).
419 (reported by Bruno Haible)
420 * fixed bug #302: better primitive-type selection for intersection
422 * fixed bug #308: non-graphic characters now all have names, as
423 required. (reported by Bruno Haible)
424 * bug fix: redefining a class with different superclasses now correctly
425 removes it from the direct-subclasses of its previous superclasses.
426 (reported by David Morse)
427 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
428 new value now works. (reported by Bruno Haible)
429 * bug fix: correct canonicalization of multiple non-standard slot
430 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
431 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
432 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
433 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
434 now exists, an signals an error.
435 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
436 during macroexpansion for non-iteration variables as well. (reported
437 by Bruno Haible for CMUCL)
438 * bug fix: Cyclic structures and unprintable objects in compiler
439 messages no longer cause errors. (reported by Bruno Haible)
440 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
441 and READ-FROM-STRING all now return a primary value of NIL if
442 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
443 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
444 (reported by Bruno Haible for CMUCL)
445 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
446 do not designate a condition. (reported by Bruno Haible for
448 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
449 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
450 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
451 permitting resolution in favour of any of the conflicting symbols.
452 (reported by Bruno Haible for CMUCL)
453 * bug fix: EQUAL compiler optimizations is less aggressive on
454 strings which can potentially compare true despite having distinct
455 specialized array element types.
456 * bug fix: unit enumerations can be defined without dividing by
457 zero. (thanks to Vincent Arkesteijn)
458 * FORMAT compile-time argument count checking has been enhanced.
459 (report from Bruno Haible for CMUCL)
460 * a partial workaround for the bug 262: the compiler does not try to
461 inline-expand a local function doing RETURN-FROM from a deleted
462 BLOCK. (thanks to Peter Denno for the bug report and to David
463 Wragg for the simple test case)
464 * fixed some bugs revealed by Paul Dietz' test suite:
465 ** RENAME-PACKAGE allows all package designators as new package
467 ** constraint propagation and lambda variable substitution are
468 more cautious in dealing with partially deleted code.
469 ** compiler sometimes lost reoptimization passes.
470 ** CERROR, given a condition as condition designator, passes its
471 remaining arguments to the continue format control without
473 ** Case-altering FORMAT directives work correctly on non-ASCII
475 ** The REAL type specifier handles bounds outside the floating
476 point ranges without signalling FLOATING-POINT-OVERFLOW.
477 ** Functions with IR1-transformations can create intercomponent
478 references to global functions.
479 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
481 ** FORMAT ~R treats a nil value for its first parameter correctly.
483 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
484 * enhancement: saving cores with foreign code loaded is now
485 supported on x86/NetBSD and sparc/Linux in addition to the previously
487 * bug fix: on some platforms repeated installations caused multiple
488 copies of HTML documentation to be installed -- should not happen
489 any more. (reported by Stefan Scholl)
490 * bug fix: parsing self-recursive alien record types multiple times
491 no longer causes infinite recursion. (reported by Thomas F. Burdick,
492 original patch by Helmut Eller for CMUCL)
493 * bug fix: stack-exhaustion detection works now on NetBSD as well.
494 (thanks to Richard Kreuter)
495 * bug fix: defining classes whose accessors are methods on existing
496 generic functions in other (locked) packages no longer signals
497 bogus package lock violations. (reported by François-René Rideau)
498 * bug fix: special variables as DEFMETHOD parameters no longer have
499 associated bogus type declarations. (reported by David Wragg and
501 * bug fix: read-write consistency on streams of element-type
502 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
503 * bug fix: redefiniton of the only method of a generic function with
504 no DEFGENERIC no longer emits a full WARNING. In addition,
505 redefinition of generic functions with no DEFGENERIC to an
506 incompatible lambda list now signals an error. (thanks to Zach
508 * bug fix: DEFGENERIC now works even when there's a function of the
509 same name in an enclosing lexical environment. (thanks to Zach
511 * fixed compiler failure, caused by instrumenting code during
512 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
513 * optimization: added loop analysis and improved register allocation
514 * fixed some bugs revealed by Paul Dietz' test suite:
515 ** POSITION on displaced vectors with non-zero displacement
516 returns the right answer.
517 ** (SIMPLE-STRING) is a valid type specifier for sequence
519 ** *PRINT-LEVEL* handling for slotless structures is pedantically
521 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
522 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
524 ** The justification version of the FORMAT ~< directive treats
525 non-zero minpad parameter correctly.
527 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
528 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
529 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
530 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
531 the supported interface.
532 * new feature: Single-stepping of code compiled with DEBUG 2 or
533 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
534 * new feature: saving cores with foreign code loaded is now
535 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
536 Timothy Moore's work for CMUCL)
537 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
538 parameters to * instead of NIL if no initform is supplied.
539 (reported by Johan Bockgaard)
540 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
541 correctly when FUNCALL appears as the car of the form. Note:
542 despite this FUNCALL forms are not currently subject to
543 compiler-macro expansion. (port of Raymond Toy's fix for the
544 same from CMUCL, reported by Johan Bockgaard)
545 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
546 (thanks for Teemu Kalvas)
547 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
548 case instead of terminating the iteration (thanks for Julian
549 Squires, Sean Champ and Raymond Toy)
550 * bug fix: incorrect expansion of defgeneric that caused a style
551 warning. (thanks for Zach Beane)
552 * bug fix: slot accessor effective method computation works properly
553 for classes with multiple non-standard applicable methods on
554 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
555 * on x86 compiler supports stack allocation of results of LIST and
556 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
557 CMUCL implementation by Gerd Moellmann)
558 * fixed some bugs revealed by Paul Dietz' test suite:
559 ** FORMAT strings with both the ~<~:;~> form of the justification
560 directive and pretty-printing directives cause an error.
562 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
563 * incompatible change: the internal functions
564 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
565 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
566 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
567 instead of the old functions.
568 * new feature: on platforms where "dladdr" is available foreign
569 function names now appear in backtraces. (based on Helmut Eller's
571 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
572 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
574 * documentation: ASDF manual is now distributed with SBCL.
575 * bug fix: SBCL can now load its contributed modules with REQUIRE
576 even if the system-provided entries have been removed from
577 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
579 * bug fix: backtraces involving undefined functions or assembly
580 routines are more informative. (thanks to Brian Downing)
581 * bug fix: mutually referent alien structures now work correctly.
582 (reported by Rick Taube)
583 * bug fix: structures defined by WITH-ALIEN can be referred to
584 within other definitions in the same WITH-ALIEN.
585 * bug fix: division operators (MOD, TRUNCATE and the like) with
586 constant zero divisors and integer dividends no longer generate
588 * bug fix: provide default methods for INPUT-STREAM-P and
589 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
590 * bug fix: improve the detection and resolution of MOP metacycles
591 (where effective method computation depends on the generic function
592 having its effective method computed). (reported by Bruno Haible)
593 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
594 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
595 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
596 * bug fix: pathnames with relative directory components can be
597 represented relative to default pathnames.
598 * optimization: in taking the GCD of bignums, reduce the two bignums
599 to approximately the same size (using Euclid's algorithm) before
600 applying the more sophisticated binary GCD. (thanks to Juho
602 * optimization: COUNT on bitvectors now operates word-at-a-time.
603 * optimization: ASH with a positive, but not necessarily constant,
604 (leftwards) shift, when in a modular context, is compiled to a
606 * fixed some bugs revealed by Paul Dietz' test suite:
607 ** FORMAT variable parameters ("~V<char>") are defaulted properly
608 if the corresponding argument is NIL.
609 ** FORMAT directives accepting a minpad parameter treat negative
611 ** literal commas as character parameters to FORMAT directives are
613 ** literal spaces directly after ~<Newline> directives within a
614 format-logical-block (~:< ~@:>) do not induce :FILL-style
615 conditional newlines.
616 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
617 as specified: it no longer includes conditional newlines).
618 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
620 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
621 :PER-LINE-PREFIX argument does not evaluate to a string.
623 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
624 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
625 the manual for details; for now, package locks can be disabled by
626 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
627 if no major problems are found then it is likely that they will be
628 compiled in unconditionally.
629 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
630 unsupported operators on all platforms. To load a shared library
631 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
632 object file, link it into a shared library outside of SBCL and
633 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
635 * bug workaround: ROOM T and the GC were not working together
636 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
637 reporting facility aren't preserved. That reporting has been
638 disabled, so now until and unless someone figures out how to make it
639 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
640 * minor incompatible change: as threatened around sbcl-0.8.0, the
641 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
642 The symbols are also exported from SB-PCL for backwards
643 compatibility, but more so than before SB-PCL should be treated as
644 an implementation-internal package.
645 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
647 * the SB-SPROF contrib now works on (most) non-x86 architectures.
648 It is known as of this release not to work on the Alpha, however.
649 * fixed bug #167: errors signalled due to illegal syntax in method
650 bodies are now more legible.
651 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
652 designators and can hence be used with TYPEP.
653 * fixed bug #333: CHECK-TYPE now ensures that the type error
654 signalled, if any, has the right object to be accessed by
655 TYPE-ERROR-DATUM. (reported by Tony Martinez)
656 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
657 5.1.2.3 for multiple-value place subforms. (reported by Kalle
659 * fixed bug #334: programmatic addition of slots using specialized
660 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
661 and :ALLOCATION :CLASS slots.
662 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
663 float exponents. (rereported by Peter Seibel)
664 * fixed a bug: #\Space (and other whitespace characters) are no
665 longer considered to be macro characters in standard syntax by
667 * fixed bug: initialization of condition class metaobjects no longer
668 causes an instance of the condition to be created. (reported by
670 * fixed bug: it is now possible to have slots such that
671 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
672 description is neither :INSTANCE nor :CLASS.
673 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
674 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
675 SLOT-VALUE-USING-CLASS), and no longer causes errors with
676 non-standard SLOT-DEFINITION-ALLOCATION values.
677 * fixed bugs: various race conditions handling exiting threads.
678 CL-PPCRE's thread torture test now passes.
679 * fixed arguably-a-bug: GC time proportional to number of threads in
680 system even when most of them are idle
681 * optimization: improved performance of BIT and SBIT on bit-vectors.
682 * .fasl file incompatibility: The fasl file version number has
683 been incremented because of changes associated with package locks.
685 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
686 * minor incompatible change: the system no longer provides
687 optimization hints (conditions which are TYPEP
688 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
689 mode; these hints are still emitted when higher SPEED optimization
691 * new contrib module: a sampling profiler (profiling by statistical
692 sampling, rather than by instrumenting functions) is available as
693 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
694 * the behaviour of the standard function ED is now customizeable by
695 third parties through a hook variable: see ED's documentation
696 string for information on the protocol.
697 * the compiler no longer emits efficiency notes for (FUNCALL X)
698 when the type of X is uncertain under default optimization
700 * fixed bug 276: mutating a binding of a specialized parameter to a
701 method to something that is not TYPEP the specializer is now
703 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
704 exhibit the required behaviour on the x86 platform. (thanks to
705 Peter van Eynde, Eric Marsden and Bruno Haible)
706 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
707 even for difficult arguments. (reported by Peter Graves)
708 * fixed bug 141a: the backquote printer now descends quoted
710 * fixed another bug in backquote printing: no more destructive
711 modification of the form's list structure. (reported by Brian
713 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
714 move between its address being taken and the call to
715 interrupt_thread, fixing a crashing race condition.
716 * the SB-POSIX contrib implementation has been adjusted so that it
717 no longer exhibits ridiculously poor performance when constructing
718 instances corresponding to C structs.
720 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
721 * minor incompatible change: the sb-grovel contrib now treats C
722 structures as alien (in the sense of SB-ALIEN) objects rather than
723 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
724 has implications for memory management of client code
725 (sb-grovel-returned objects must now be manually managed) and for
726 type safety (alien objects now have full types).
727 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
728 used to control emission of compiler diagnostics, rather than the
729 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
730 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
731 quality should be considered deprecated.
732 * install.sh now installs the user manual as well
733 * (not quite a new documentable feature, but worth considering in
734 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
735 beginnings of a semantically meaningful condition hierarchy is
736 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
738 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
739 designator as the defaults argument.
740 * fixed bug: Displaced arrays whose displaced-to array has become
741 too small now cause ARRAY-DIMENSION to signal an error, providing
742 for safer bounds-checking. (reported by Bruno Haible)
743 * fixed bug: DEFCLASS slot definitions with identical :READER and
744 :WRITER names now signal a reasonable error. (reported by Thomas
746 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
748 * fixed bug: Invalid slot specification errors now print correctly.
749 (thanks to Zach Beane)
750 * fixed bug 320: Shared to local slot value transfers in class
751 redefinitions now happen corrently from superclasses as
752 well. (reported by Bruno Haible)
753 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
755 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
756 improper lists correctly. (reported by Bruno Haible)
757 * fixed bug 313: source-transform for <fun-name> was erroneously
758 applied to a call of a value of a variable with name <fun-name>.
759 (reported by Antonio Menezes Leitao)
760 * fixed bug 307: The obsolete instance protocol ensures that
761 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
762 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
763 remove dead unknown-values globs from the middle of the stack.
764 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
765 easier to resume long computations after using *BREAK-ON-SIGNALS*
766 to diagnose and fix failures. (thanks to Nikodemus Siivola)
767 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
768 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
769 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
770 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
771 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
773 * fixed a (fairly theoretical) bug in string printing: if
774 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
775 does not have array-element-type equal to the most general string
777 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
778 function name. (thanks to Nikodemus Siivola)
779 * fixed bug: compile-time format string checker failed on
780 non-closed ~{. (reported by Thomas F Burdick)
781 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
782 don't warn on legal loop code involving more than one aggregate
783 boolean. (thanks to Nikodemus Siivola)
784 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
785 in keyword position to MAKE-PACKAGE was not being reported as
787 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
788 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
790 * fixed bugs 280 and 312: the checking for multiple definitions in a
791 file is less likely to become confused by uses of inline
793 * fixed bug: the #S reader macro performs the keyword coercion
794 specified for slot names. (reported by Kalle Niemitalo)
795 * fixed bug: lambda lists may contain symbols whose names start with
796 & but are not lambda-list-keywords; their occurrence triggers a
798 * fixed bug 321: define-method-combination argument lambda lists do
799 not cause invalid code to be generated when &optional or &aux
800 variables have default values. (reported by Bruno Haible)
801 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
802 have CLOS classes; this appears also to have fixed reported
803 instability in stack exhaustion detection.
804 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
805 MUFFLE-WARNING when no associated restart is present is now
807 * optimization: rearranged the expansion of various defining macros
808 so that each expands into only one top-level form in a
809 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
811 * optimization: used a previously little-used slot in symbols to
812 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
814 * fixed some bugs revealed by Paul Dietz' test suite:
815 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
817 ** Fixed an optimization bug related to inheritance of initforms
818 from local to shared slots.
819 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
820 ** CAST optimizer forgot to flush argument derived type.
821 ** print/read consistency on floats is now orders of magnitude
822 more likely. (thanks also to Bruno Haible for a similar report
824 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
825 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
826 for objects of type REAL. Make it so.
827 ** SXHASH is specified (infelicitously) to respect similarity,
828 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
829 it so. (thanks to Markus Ziegler)
830 ** on the Alpha, the compiler succeeds in compiling functions
831 returning a known number of arguments greater than 63.
832 ** fixed handling of invalid NIL arguments in keyword position
834 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
836 ** provided a readably-printable representation for RANDOM-STATE
838 ** ensured that pathnames, where they have a namestring, always
839 print using #P"..." syntax.
841 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
842 * Support for the forthcoming 2.0 version of the NetBSD kernel
843 running on x86 hardware has been added. (thanks to Perry
844 E. Metzger most immediately, and others for their past work)
845 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
846 current release version; SBCL's OpenBSD support had been broken
847 since about the time of OpenBSD's switch to ELF binary format.)
848 * [placeholder for DX summary]
849 ** user code with &REST lists declared dynamic-extent, under high
850 speed or space and low safety and debug optimization policy.
851 * The manual has been converted to Texinfo format and the debugger
852 chapter from the cmucl manual has been added.
853 * A facility has been added to extract documentation strings from
854 sbcl and store them as Texinfo-formatted snippets for inclusion in
855 the manual (via Texinfo's @include directive)
856 * bug fix: compiler emitted division in optimized DEREF. (thanks for
857 the test case to Dave Roberts)
858 * bug fix: multidimensional simple arrays loaded from FASLs had fill
859 pointers. (reported by Sean Ross)
860 * bug fix: PROFILE output is printed nicely even for large numerical
861 values. (thanks to Zach Beane)
862 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
863 greater than 32 handle EOF correctly.
864 * bug fix: on X86 an immediate argument of the IMUL instruction is
865 correctly printed in disassembly. (thanks to Lutz Euler)
866 * bug fix: class slots in redefined classes preserve their old
867 values. (thanks to Bruno Haible and Nikodemus Siivola)
868 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
869 succeeds. (reported by Marco Baringer)
870 * fixed some bugs revealed by Paul Dietz' test suite:
871 ** READ-SEQUENCE now works on ECHO-STREAMs.
872 ** RATIONALIZE works more according to its specification. (thanks
875 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
876 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
877 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
878 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
879 should matter to you only if you rebind the printer control
880 variables and then find you want different bindings in the
881 debugger than in the ordinary execution of your program.)
882 * The runtime build system has been tweaked to support building
883 (on SPARC/SunOS) using a C compiler which invokes Sun's own
884 assembler and linker. (thanks to Nikodemus Siivola)
885 * Unbound, undefined, undeclared variables now trigger full
886 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
887 more often programmer error than deliberate exploitation of undefined
889 * optimization: the hash algorithm for strings has changed to one
890 that is less vulnerable to spurious collisions. (thanks to Juho
892 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
893 needless bounds checking. (thanks to Juho Snellman)
894 * optimization: implemented multiplication as a modular
895 (UNSIGNED-BYTE 32) operation on the PPC backend.
896 * fixed some bugs revealed by Paul Dietz' test suite:
897 ** ADJUST-ARRAY now copies the datum in a zero rank array if
899 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
900 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
901 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
903 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
905 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
907 ** ECHO-STREAMs no longer attempt to echo the end of file value to
908 their output stream on EOF from read.
909 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
910 have been read to end-of-file.
911 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
913 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
914 description of determination of which consecutive characters
916 ** Printing the "Space" character with escaping on now yields "#\\ ",
917 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
918 ** Reading floating-point numbers with *READ-BASE* set to a number
919 less than 10 works correctly.
920 ** Reading floating-point numbers with *READ-BASE* set to a number
921 more than 10 works correctly.
922 ** Printing with *PRINT-READABLY* targets the standard readtable, not
923 the readtable currently in effect.
925 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
926 * minor incompatible change: parsing of namestrings on a physical
927 (Unix) host has changed; numbers after the final #\. in a
928 namestring are no longer interpreted as a version field. This is
929 intented to be largely invisible to the user, except that the
930 meaning of the namestring "*.*.*" has changed: it now refers to a
931 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
932 should usually be replaced by
933 (make-pathname :name :wild :type :wild :version :wild)
934 with the added benefit that this is more likely to be portable.
935 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
936 option to OPEN now signals an error if the file being opened
937 exists; this may have an impact on existing code.
938 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
939 signals received, on the PowerPC platforms (both Linux and
940 Darwin). (thanks to Pierre Mai for pointing out the location of
942 * several fixes on OS X: The system now builds and runs cleanly on
943 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
944 (thanks to Brian Mastenbrook)
945 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
946 second-resolution: integer multiples of 1/3600 between -24 and 24.
947 (thanks to Vincent Arkesteijn)
948 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
949 type when called with 1 argument; PEEK-CHAR checked type of
950 PEEK-TYPE only after having read first character from a
951 stream. (reported by Peter Graves)
952 * bug fix: the garbage collector now has much better locality
953 behaviour, and in particular no longer treats all memory as being
954 exhausted when many small objects point to each other in a deeply
956 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
957 recognized as being TYPEP their class.
958 * bug fix: the PUSHNEW documentation string has been corrected.
959 (thanks to Vincent Arkesteijn)
960 * bug fix: defaulting of the value for the last of an atypically
961 large number of multiple values being bound was not being
962 performed correctly on the Alpha or PPC platforms
963 * optimization: implemented multiplication as a modular
964 (UNSIGNED-BYTE 32) operation on the x86 backend.
965 * optimization: SEARCH on simple-base-strings can now be open-coded.
966 (see also contrib/compiler-extras.lisp for inspiration for
967 teaching the compiler about the Boyer-Moore algorithm).
968 * value, returned by MAX (and MIN) called with several EQUALP, but
969 not EQL, arguments now does not depend on compiler settings.
970 * fixed some bugs revealed by Paul Dietz' test suite:
971 ** in stack analysis liveness information is propagated from
972 non-local entry points.
973 ** pathwise CAST removing failed when the CAST node did not start
975 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
976 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
978 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
979 host is already defined.
980 ** RENAME-FILE works on streams instead of signalling an internal
982 ** PEEK-CHAR uses the current readtable when determining whether
983 or not a character is whitespace.
984 ** MERGE-PATHNAMES handles the case when the pathname does not
985 specify a name while the default-pathname specifies a version.
986 ** Pathnames now stand a better chance of respecting print/read
988 ** Attempting to use standardized file system operators with a
989 pathname with invalid :DIRECTORY components signals a
991 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
993 ** DIRECTORY on logical pathnames is more correct.
994 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
995 signal a TYPE-ERROR if their argument is not a stream.
996 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
997 designator argument does not designate a stream.
998 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
999 examining the synonym.
1000 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1002 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1003 element-type larger than ([UN]SIGNED-BYTE 32).
1005 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1006 * When built with the :SB-FUTEX feature, threaded builds now take
1007 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1008 for faster/more reliable mutex and condition variable support.
1009 * Incompatible change (but one you probably shouldn't have been using
1010 anyway): the interface and code for arbitrating between multiple
1011 threads in the same user session has been redesigned.
1012 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1013 error when called without an explicit environment argument.
1014 (thanks to Nikodemus Siivola)
1015 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1016 confused about their position. (thanks to Adam Warner and Gerd
1018 * bug fix: Pretty printing backquoted forms with unquotations in the
1019 argument list position of various code constructs such as LAMBDA
1020 now works correctly. (reported by Paul Dietz)
1021 * bug fix: Pretty printing unquotations no longer loses all
1022 stream position information.
1023 * optimization: performance of string output streams is now less
1024 poor for multiple small sequence writes.
1025 * optimization: performance of CSUBTYPEP in the presence of complex
1026 expressions involving CONS and NOT many times has been improved.
1027 (reported by Paul Dietz)
1028 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1030 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1031 streams for the REPL, for the debugger, and so on. (thanks to
1033 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1034 * fixed some bugs revealed by Paul Dietz' test suite:
1035 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1036 is ignored if it is too small, rather than propagating through
1037 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1038 ** extremely complex negations of CONS types were not being
1039 sufficiently canonicalized, leading to inconsistencies in
1041 ** VALUES tranformer lost derived type.
1043 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1044 * fixed a bootstrapping bug: the build process no longer assumes
1045 that the various BOOLE-related constants have the same value in
1046 host and target lisps. (noted by Paul Dietz' test suite on an
1047 SBCL binary built from CLISP)
1048 * The system can now be dynamically linked on the MIPS platform,
1049 which enables dynamic loading of foreign code from Lisp. (thanks
1050 to Ralf Baechle for discussions on the MIPS ABI)
1051 * The system now records debugging information for its own source
1052 files in a filesystem-position-independent manner, relative to
1053 the "SYS" logical host.
1054 * fixed a compiler bug: MV-LET convertion did not check references
1055 to the "max args" entry point. (reported by Brian Downing)
1056 * tweaked disassembly notes to be less confident about proclaiming
1057 some instruction as an LRA. (thanks to Brian Downing)
1058 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1059 now each have their own history, command character, and other
1060 characteristics. (thanks to David Lichteblau)
1061 * fixed some bugs revealed by Paul Dietz' test suite:
1062 ** compiler failure in compiling LOGAND expressions including a
1064 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1065 shift greater than 32.
1066 ** FLUSH-DEST did not mark blocks for type check regeneration.
1067 ** HANDLER-CASE failed to accept declarations in handler clauses
1068 in some circumstances.
1070 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1071 * New code in contrib/sb-introspect (still probably not entirely
1072 stable yet) provides some support for smart Lisp development
1073 environments like SLIME.
1074 * The conditions signalled for errors occurring when loading .fasl
1075 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1076 in a way which should help ASDF recover gracefully.
1077 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1078 stuff has been cleaned up. If you code contrib/ stuff, this might
1079 affect you, and you can look at contrib/README, contrib/STANDARDS,
1080 and/or the 0.8.4.27 diff to check.
1081 * In full calls the compiler now does not generate checks for declared
1082 argument types for all arguments.
1083 * various threading fixes
1084 ** and some experimental patches which didn't make it into
1085 the main tree for this release, but which are shipped in
1086 contrib/experimental-thread.patch as a possible fix for some
1087 failures (deadlock, spinning...) in GC-intensive multithreaded
1089 * fixed PPC build problem (source code incompatibility of different
1090 library versions): added offsetof() hackery which attempts to divine
1091 where glibc maintainers put uc_mcontext today
1092 * fixed bug 282: compiler does not trust type assertions while passing
1093 arguments to a full call.
1094 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1095 &OPTIONAL VALUES type parameter.
1096 * fix bug 214: algorithm for noting rejected templates is now more
1097 similar to that of template seletion. (also reported by rydis on
1099 * fixed bug 141b: printing backquoted information readably and prettily
1100 inserts a space where necessary.
1101 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1102 (CHARACTER 10) are now reported as errors, rather than propagated
1103 as unknown types. (reported by piso on #lisp)
1104 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1105 with values NIL and :ERROR. (thanks to Milan Zamazal)
1106 * fixed bug 191c: CLOS now does proper keyword argument checking as
1107 described in CLHS 7.6.5 and 7.6.5.1.
1108 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1109 longer raise an error; further, using a list as a for-as-arithmetic
1110 counter now raises a meaningful error.
1111 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1112 now understood by sequence creation functions such as MAKE-SEQUENCE
1114 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1115 a STRING-INPUT-STREAM.
1116 * compiler enhancement: SIGNUM is now better able to derive the type
1118 * type declarations inside WITH-SLOTS are checked. (reported by
1120 * fixed some bugs revealed by Paul Dietz' test suite:
1121 ** incorrect optimization of TRUNCATE for a positive first
1122 argument and negative second.
1123 ** compiler failure in let-convertion during flushing dead code.
1124 ** compiler failure while deriving type of TRUNCATE on an
1125 interval, containing 0.
1126 ** ASH of a negative bignum by a negative bignum count now returns
1128 ** intersection of CONS types now canonicalizes properly, fixing
1129 inconsistencies in SUBTYPEP.
1131 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1132 * incompatible change: The --disable-debugger command line
1133 option now clobbers the debugger at a more fundamental
1134 level, by redefining #'INVOKE-DEBUGGER instead of by
1135 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1136 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1137 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1138 enter the debugger REPL and then suffer recursive errors
1139 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1140 while under the new system, BREAK in --disable-debugger mode
1141 terminates the system just as an unhandled error would.
1142 * fixed compiler performance when processing loops with a step >1;
1143 * bug fix: DOCUMENTATION now retrieves generic function
1144 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1145 support has been systematized, and now supports the methods
1146 specified by ANSI, along with a default method and a method for
1147 slot documentation. (reported by Nathan Froyd)
1148 * bug fix: effective methods associated with a generic function are
1149 no longer cached over a change of that generic function's method
1150 combination. (reported by Andreas Fuchs)
1151 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1152 names a non-existent directory. (thanks to Andreas Fuchs)
1153 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1154 the PPC platform, lead to stack corruption; this has been fixed.
1155 (reported by Rainer Joswig)
1156 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1157 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1158 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1159 arithmetic in general, is now much more reliable on the PPC
1161 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1162 platform now returns the right answer.
1163 * optimization: restored some effective method precomputation in
1164 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1165 precomputation is now tunable.
1166 * optimization: compiler-internal data structure use has been
1167 reviewed, and changes have been made that should improve the
1168 performance of the compiler by about 20%.
1169 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1170 simple-base-strings and simple-bit-vectors is improved.
1171 * optimization: the optimization of 32-bit logical and arithmetic
1172 functions introduced in version 0.8.3 on the x86 has been
1173 implemented on the mips, ppc and sparc platforms; an
1174 implementation of the same facility, but for 64-bit arithmetic,
1175 has been added for the alpha.
1176 * microoptimization: the compiler is better able to make use of the
1177 x86 LEA instruction for multiplication by constants.
1178 * bug fix: in some situations compiler did not report usage of
1179 generic arithmetic in (SPEED 3) policy.
1180 * bug 145b fix: compiler used wrong type specifier while converting
1181 MEMBER-types to numeric.
1182 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1184 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1185 index argument now works properly on the Alpha platform.
1186 * bug fix: floating point exception treatment on the Alpha platform
1188 * bug fix: FILE-POSITION works much better on string input and
1189 output streams. (thanks to Nikodemus Siivola)
1190 * bug fix: many threading/garbage collection symptoms sorted.
1191 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1192 might be pseudo-atomic.
1193 * internal change: Stopping for GC is now done with signals not
1194 ptrace. GC is now done in whichever thread wanted it, instead of
1196 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1198 * bug fix: result form in DO is not contained in the implicit
1200 * incompatible change: ICR structure is changed; the value part of
1201 CONTINUATION is now called LVAR; corresponding functions are
1202 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1203 * added type deriver for ISQRT (thanks to Robert E. Brown).
1204 * bug fix: better support for loading from the command line when an
1205 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1207 * fixed some bugs revealed by Paul Dietz' test suite:
1208 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1209 ** ROUND and FROUND now give the right answer when given very
1210 small float arguments.
1211 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1213 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1214 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1215 ** LCM should return a non-negative integer.
1216 ** PARSE-INTEGER returned the index of a terminator instead of the
1217 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1218 ** PARSE-INTEGER returned an incorrect index being applied to a
1220 ** LCM with two arguments of 0 returns 0 rather than signalling
1222 ** unsigned addition of a 32-bit constant with the high bit set no
1223 longer causes an internal compiler error on the x86.
1224 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1225 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1226 with negative last argument.
1227 ** byte specifiers with zero size and position no longer cause
1228 an error during type derivation.
1229 ** bignum multiplication on the Alpha platform now returns the
1231 * porting: The system now builds on SuSE AMD64, although it still
1232 generates a 32-bit binary.
1233 * .fasl file incompatibility: The fasl file version number has
1234 been incremented (because of the changes to internal compiler
1235 data structures referred to above).
1237 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1238 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1239 more accurately, on the Darwin kernel running on PowerPC hardware.
1240 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1241 * Compiler code deletion notes now signal a condition of type
1242 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1243 an associated MUFFLE-WARNING restart.
1244 * The compiler now performs limited argument count validation of
1245 constant format strings in FORMAT, and where appropriate in ERROR,
1246 CERROR and WARN. (thanks to Gerd Moellmann)
1247 * New ASDF-INSTALL contrib can be used for automatic download and
1248 installation of third-party Lisp code from CCLAN or other sites
1250 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1251 which forces another thread to execute a function supplied by the
1253 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1254 now accept and act upon their :ELEMENT-TYPE keyword argument.
1255 (reported by Martin Atzmueller, Edi Weitz)
1256 * bug fix: FILE-POSITION now accepts position designators up to
1257 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1258 is the greater. (thanks to Patrik Nordebo)
1259 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1260 Kalle Olavi Niemitalo)
1261 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1262 printable. (reported by Eric Marsden)
1263 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1264 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1265 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1266 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1267 (thanks to Stig Sandoe)
1268 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1269 UNREAD-CHAR now correctly includes the unread character in the
1270 target sequence. (thanks to Gerd Moellmann)
1271 * bug fix in threaded builds: the system can now be suspended and
1272 resumed by shell job control with minimal disruption.
1273 * bug fixes in times and timezones >2038AD
1274 * better handling of "where is GNU make?" problem in build scripts
1275 (thanks to Nikodemus Siivola)
1276 * new optimization: inside a named function any reference to a
1277 function with the same name is considered to be a self-reference;
1278 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1279 optimization quality.
1280 * new optimization on x86: logical functions and + now have
1281 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1282 used when the result is truncated to 32 bits.
1283 * VALUES declaration is partially enabled.
1284 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1285 * fixed some bugs revealed by Paul Dietz' test suite:
1286 ** The system now obeys the constraint imposed by
1287 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1288 types form a lattice under type intersection.
1289 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1290 ** ASSOC now ignores NIL elements in an alist.
1291 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1292 and (1+ MOST-POSITIVE-FIXNUM) answers.
1293 ** The addition of a method with invalid qualifiers to a generic
1294 function does not cause an error to be signalled immediately;
1295 a warning is signalled, and the error is generated only on
1296 calling the generic function.
1297 * changed .fasl file version number, in part to add type codes for
1298 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1299 obscure ANSI requirements
1301 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1302 * fixed bug 148: failure to inline-expand a local function left
1303 garbage, confusing the compiler.
1304 * fixed bugs 3cd: structure slot readers perform type check if the
1305 slot can have an invalid value (i.e. it is either not initialized
1306 or can be written with a less specific slot writer).
1307 * bug fix: the compiler now traps array references to elements off
1308 the end of an array; previously, the bounds checking in some
1309 circumstances could go off-by-one.
1310 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1312 * type declarations for array element types now obey the description
1313 on the CLHS page "Declaration TYPE", as per discussions on
1314 sbcl-help around 2003-05-08. This means that a declaration
1315 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1316 declaration, all references to BAR will be asserted or assumed
1317 (with THE, so dependent on compiler policy) to involve objects of
1318 type FOO. Note that no such declaration is implied in
1319 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1320 * declared types of functions from the "Conditions"
1321 chapter. (reported by Paul Dietz)
1322 * bug fix: CERROR accepts a function as its first argument.
1323 * bug fix: NTH an NTHCDR accept a bignum as index
1324 arguments. (reported by Adam Warner)
1325 * optimization: character compare routines now optimize comparing
1326 against a constant character. (reported by Gilbert Baumann)
1327 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1328 argument now works properly on the MIPS platform.
1329 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1330 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1331 declared to have a functional type, does not cause an error
1333 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1334 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1335 * fixed bug: if last continuation of a deleted block has a
1336 destination, this destination should be deleted too. (reported by
1338 * fixed a bug in the bootstrap process: the host compiler's values
1339 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1340 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1341 test case from Patrik Nordebo)
1342 * improved the ability of the disassembler on the PPC platform to
1343 provide helpful disassembly notes.
1344 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1345 the class in more cases than previously.
1346 * bug fix: FILE-POSITION now understands :START and :END for
1347 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1348 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1349 8) data. (thanks to David Lichteblau)
1350 * bug fix: it is possible to add a method to a generic function
1351 without lambda list.
1352 * bug fix: reader failed to signal END-OF-FILE inside an
1353 object representation. (reported by Nikodemus Siivola)
1354 * fixed some bugs revealed by Paul Dietz' test suite:
1355 ** LAST and [N]BUTLAST should accept a bignum.
1356 ** condition slot accessors are methods.
1357 ** (VECTOR NIL) is a subtype of STRING.
1359 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1360 * minor incompatible change: some nonsensical specialized lambda
1361 lists (used in DEFMETHOD) which were previously ignored now signal
1363 * minor incompatible change: the system is now aware of the types of
1364 variables in the COMMON-LISP package, and will signal errors for
1365 most violations of these type constraints (where previously they
1366 were silently accepted).
1367 * minor incompatible change: COMPILE-FILE now uses the freedom
1368 afforded (ANSI 3.2.2.3) to use derived function types for
1369 functions defined in the same file. This also permits the system
1370 to warn on static type mismatches and function
1371 redefinition. (Currently it does not work with high DEBUG level.)
1372 * minor incompatible change: VALUES declaration is disabled.
1373 * When issuing notes, the compiler now signals a condition of type
1374 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1375 restart for use in user handlers. It is expected that the
1376 COMPILER-NOTE condition will eventually become a condition
1377 supertype to a hierarchy of note types, which will then be
1378 handleable in a similar fashion. However, at the moment, no such
1379 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1380 but it's an implementation detail, not a classification for the
1382 * Changes in type checking closed the following bugs:
1383 ** type checking of unused values (192b, 194d, 203);
1384 ** template selection based on unsafe type assertions (192c, 236);
1385 ** type checking in branches (194bc).
1386 * A short form of VALUES type specifier has ANSI meaning (it has
1387 increased the number of situations when SBCL cannot perform type
1389 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1390 keywords or constants is permissible.
1391 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1392 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1393 classes. (thanks to Antonio Martinez)
1394 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1395 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1396 argument to a pretty-print stream. Instead, it leaves any such
1397 operations to DESCRIBE-OBJECT methods.
1398 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1400 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1401 lambda lists. (thanks to David Lichteblau)
1402 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1403 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1404 (thanks to Gerd Moellmann)
1405 * fixed bug 246: increased compilation speed of long
1406 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1408 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1410 * DEFCONSTANT now throws a condition of type
1411 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1412 constant to a non-EQL value; CONTINUE and ABORT restarts
1413 respectively change and preserve the value.
1414 * fixed bug 63: The code walker, part of the implementation of CLOS,
1415 is now better at handling symbol macros.
1416 * bug fix: There is no longer an internal implementation type named
1417 CL:LENGTH. (reported by Raymond Toy)
1418 * bug fix: In macro-like defining macros/special operators the
1419 implicit block does not enclose the lambda list.
1420 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1421 enclosed in parentheses) are not suitable as type specifiers, and
1422 their use properly signals an error now.
1423 * bug fix: An argument count mismatch for a type specifier in code
1424 being compiled no longer causes an unhandled error at compile
1425 time, but signals a compile-time warning.
1426 * fixed simple vector readable printing
1427 * bug fix: DESCRIBE takes more care over whether the class
1428 precedence list slot of a class is bound before accessing it.
1429 (reported by Markus Krummenacker)
1430 * bug fix: FORMATTER can successfully compile pretty-printer format
1431 strings which use variants of the ~* directive inside.
1432 * bug fix: SEARCH now applies its TEST predicate to the elements of
1433 the arguments in the correct order. (thanks to Wolfhard Buss)
1434 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1435 in type inference. (thanks to Robert E. Brown)
1436 * bug fix: Reading in symbols with an explicit package name of ""
1437 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1438 * fixed some bugs revealed by Paul Dietz' test suite:
1439 ** NIL is now allowed as a structure slot name.
1440 ** Arbitrary numbers, not just REALs, are allowed in certain
1441 circumstances in LOOP for-as-arithmetic clauses.
1442 ** Multiple class redefinitions before slot access no longer
1443 causes a type error.
1444 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1445 association between the name and a class.
1446 ** Generic functions with non-standard method-combination and over
1447 five methods all of which return constants no longer return NIL
1448 after the first few invocations. (thanks to Gerd Moellmann)
1449 ** CALL-NEXT-METHOD with no arguments now passes the original
1450 values of the arguments, even in the presence of assignment.
1451 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1453 ** DISASSEMBLE works with closures and funcallable instances.
1454 ** ADD-METHOD now returns the generic function, not the new method.
1455 ** FIND-METHOD signals an error if the lengths of the specializers
1456 is incompatible with the generic function, even if the ERRORP
1458 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1459 which its argument is a member.
1460 ** DEFCLASS only redefines the class named by its class-name
1461 argument if that name is the proper name of the class;
1462 otherwise, it creates a new class.
1463 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1464 of the UNBOUND-SLOT condition to the name of the slot.
1465 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1466 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1467 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1468 SLOT-MAKUNBOUND in the specified fashion.
1470 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1471 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1472 cross-compilation host. As a consequence, we can now bootstrap our
1473 way up to SBCL starting with a bare gcc toolchain and human-readable
1474 source code (first the source to CLISP, then the source to SBCL).
1475 * A contributed module containing a partial implementation of the
1476 simple-streams interface has been included. (thanks to Rudi
1478 * A contributed module implementing the RFC1321 Message Digest
1479 Algorithm, known as MD5, has been included.
1480 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1481 no longer has any effect, as the code controlled by this feature
1482 has been deleted. (As far as we know, no-one has ever built using
1483 this feature, and its semantics were confused in any case).
1484 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1485 work on conditions (as required by the ANSI specification),
1486 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1487 expected behaviour on conditions. Users should note, however,
1488 that such behaviour is not required by the ANSI specification,
1489 and so use of this behaviour may render their code unportable.
1490 * fixed some bugs revealed by Paul Dietz' test suite:
1491 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1493 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1494 ** SXHASH on condition objects no longer returns NIL.
1495 ** :ALLOCATION :CLASS slots are better treated; their values are
1496 updated on class redefinition, and initforms inherited from
1497 superclasses are applied.
1498 ** REMOVE-METHOD returns its generic function argument even when
1499 no method was removed.
1500 ** SHARED-INITIALIZE now initializes the values of the requested
1501 slots, including those with :ALLOCATION :CLASS.
1502 ** ALLOCATE-INSTANCE now works on structure classes defined via
1503 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1505 ** SLOT-EXISTS-P now works on conditions, as well as structures
1507 ** MAKE-LOAD-FORM now has the required methods on
1508 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1509 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1510 keyword, and now implements the SLOT-NAMES argument.
1511 ** methods with &OPTIONAL arguments no longer allow too many
1512 arguments to be passed in the call without error.
1513 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1514 option is consistent with the required arguments of the generic
1515 function lambda list.
1516 * bug fix: REQUIRE accepts a string designator. (Thanks to
1518 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1519 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1520 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1521 * bug fix: The compiler now checks for duplicated variables in macro
1523 * bug fix: SETQ on globals returns the correct value.
1524 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1525 not a COMPILER-ERROR (followed by some other strange error on
1526 choosing the CONTINUE restart).
1527 * bug fix: make.sh and friends are now more consistent in the way that
1528 they look for GNU "make".
1530 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1531 * experimental native threads support (on x86 Linux >=2.4 only).
1532 This is not compiled in by default: you need to add :SB-THREAD to
1533 the target features. See the "Beyond ANSI" chapter of the manual
1535 * fix for longstanding nonANSIism: The old distinction between
1536 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1537 The return value from CL:FIND-CLASS is now a CLOS class, and
1538 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1539 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1540 * An interface to the MetaObject Protocol, as described in Kiczales,
1541 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1542 MIT Press, 1991, is available from the SB-MOP package.
1543 * incompatible change: the SB-PCL package should now be considered
1544 a private implementation detail, and no longer a semi-private MOP
1546 * minor incompatible change: due to rearrangement for threads, the
1547 control stack and binding stack are now allocated at arbitrary
1548 addresses instead of being hardcoded per-port. Users affected by
1549 this probably have to be doing advanced things with shared
1550 libraries, and will know who they are.
1551 * minor incompatible change: Previously, all --eval forms used were
1552 processed with READ before any of them were processed with EVAL.
1553 Now each --eval form is processed with both READ and EVAL before
1554 the next --eval form is processed. (Thus package operations like
1555 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1556 work as the user might reasonably expect.)
1557 * minor incompatible change: *STANDARD-INPUT* is now only an
1558 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1560 * minor incompatible change: Y-OR-N-P is now character-oriented,
1561 not line oriented. Also, YES-OR-NO-P now works without errors.
1562 (thanks to Antonio Martinez)
1563 * sb-aclrepl module improvements: an integrated inspector, added
1564 repl features, and a bug fix to :trace command.
1565 * Known functions, which cannot be open coded by the backend, are
1566 considered to be able to check types of their arguments. (fixing
1567 a bug report by Nathan J. Froyd)
1568 * fixed a bug in computing method discriminating functions: It is
1569 now possible to define methods specialized on classes which have
1570 forward-referenced superclasses. (thanks to Gerd Moellmann)
1571 * fixed evaluation order in optional entries (reported by Gilbert
1573 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1574 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1575 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1576 required-by-AMOP NAME argument, as well as CLASS and
1577 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1578 * fixed bug 20: DEFMETHOD can define methods using names that are
1579 not the proper names of classes to designate class specializers.
1580 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1581 Unix file descriptors, instead of blowing up. (thanks to Antonio
1583 * Garbage collection refactoring: user-visible change is that a
1584 call to the GC function during WITHOUT-GCING will not do garbage
1585 collection until the end of the WITHOUT-GCING. If you were doing
1586 this you were probably losing anyway.
1587 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1588 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1589 TYPEP the latter but not the former.
1590 * The compiler issues a full WARNING for calls to undefined functions
1591 with names from the CL package.
1592 * MAP-INTO for a vector destination is open coded. (reported by
1593 Brian Downing on c.l.l)
1594 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1595 documentation string.
1596 * fixed some bugs revealed by Paul Dietz' test suite:
1597 ** COPY-ALIST now signals an error if its argument is a dotted
1599 ** Condition slots are now accessed more correctly in the presence
1600 of multiple initargs for a given slot.
1601 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1602 exclude restarts of the same name associated with a different
1604 ** DEFCLASS of forward-referenced classes with another
1605 forward-referenced class in the superclasses list no longer
1607 ** Condition slots are now initialized once each, not multiple
1608 times. (thanks to Gerd Moellmann)
1609 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1611 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1612 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1613 arguments contain duplicated elements.
1614 ** RESTART-CASE understands local macros.
1615 ** RESTART-CASE associates exactly its own restarts with a condition.
1616 ** ENDP in safe mode checks its argument to be of type LIST.
1617 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1618 in question is unbound.
1619 ** Optimized MAKE-INSTANCE functions no longer cause internal
1620 assertion failures in the presence of duplicate initargs.
1621 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1622 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1623 propagated even in the ctor optimized implementation of
1625 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1627 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1628 * a better implementation of SXHASH on (simple) bit vectors,
1629 measured both in execution speed and in distribution of results
1630 over the positive fixnums, has been installed. Likewise, a better
1631 implementation of EQUAL for simple bit vectors is now available.
1632 * fixed CEILING optimization for a divisor of form 2^k.
1633 * fixed bug 240 (emitting extra style warnings "using the lexical
1634 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1635 by Antonio Martinez)
1636 * fixed SXHASH, giving different results for NIL depending on type
1637 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1638 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1639 variable. (found by Rolf Wester)
1640 * SBCL does not ignore type declarations for special
1641 variables. (reported by rif on c.l.l 2003-03-05)
1642 * some bug fixes in contrib/sb-aclrepl/
1643 * fixed some bugs revealed by Paul Dietz' test suite:
1644 ** a bug in the CONS type specifier, whereby the CAR and CDR
1645 types got intertwined, has been fixed;
1646 ** the type system is now able to reason about the interaction
1647 between INTEGER and RATIO types more completely;
1648 ** APPEND, [N]REVERSE and NRECONC check that those their
1649 arguments, which must be proper lists, are really so;
1650 ** An array specialized to be unable to hold elements has been
1651 implemented, as required -- yes, really -- by ANSI;
1652 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1653 on malformed property lists;
1655 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1656 * incompatible packaging change: in line with Unix convention,
1657 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1658 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1659 * REQUIRE and PROVIDE are now optionally capable of doing something
1660 useful. See the documentation string for REQUIRE.
1661 * infrastructure for a managed SBCL contrib system: contributed
1662 modules in this release include:
1663 ** the ASDF system definition facility;
1664 ** an interface to the BSD Sockets API;
1665 ** an ACL-like convenience interface to the repl;
1666 (thanks to Kevin Rosenberg)
1667 ** an implementation of ROTATE-BYTE, with efficient implementation
1669 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1670 gives a closer approximation to the right answer than previously.
1671 (thanks to Raymond Toy)
1672 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1673 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1674 optional environment arguments, as required by ANSI.
1675 * fixed bugs in other functions taking environment objects, allowing
1676 calls with an explicit NIL environment argument to be compiled
1678 * fixed bug 228: primary return values from
1679 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1680 COMPILE or FUNCTION.
1681 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1682 :TYPE will no longer signal errors on innocuous objects.
1683 * fixed bug 231b: SETQ is better at respecting type declarations in
1684 the lexical environment.
1685 * fixed a bug in DEFCLASS: classes named by symbols with no or
1686 unprintable packages can now be defined.
1687 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1688 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1689 * fixed compiler failure related to checking types of functions.
1690 (reported by Robert E. Brown)
1691 * the compiler is now much more consistent in its error-checking
1692 treatment of bounding index arguments to sequence functions: in
1693 (SAFETY 3) code, errors will be signalled in almost all cases if
1694 invalid sequence bounding indices are passed to functions defined
1695 by ANSI to operate on sequences.
1696 * fixed a bug in the build procedure: documentation of SBCL-specific
1697 packages is now preserved and available in the final Lisp image.
1698 * lifted FDEFINITION lookup out of loops in the implementation of
1699 many list operations. (thanks to Robert E. Brown)
1700 * fixed a bug in the reader: the #n# reader macro now works for
1701 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1702 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1703 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1704 E. Brown and Tony Martinez respectively)
1705 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1706 SLOT-BOUNDP: in particular, it is now able to optimize them much
1707 better, and is now not vulnerable to having packages renamed.
1708 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1709 always called when a slot is not present in an instance. (thanks
1711 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1712 longer signal an error on unreading a character following EOF on
1713 the previous constituent stream. (thanks to Tony Martinez)
1714 * fixed some bugs revealed by Paul Dietz' test suite:
1715 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1716 not just nonnegative fixnums;
1717 ** the logical bit-array operators such as BIT-AND now accept an
1718 explicit NIL for their "opt-arg" argument (to indicate a
1719 freshly-consed result bit-array);
1720 ** ELT now signals an error on an invalid sequence index in safe
1722 ** the type system is now cleverer about negations of numeric
1723 types, and consequently understands the BIGNUM and RATIO types
1725 ** the type system is now cleverer about the interaction between
1726 INTEGER and RATIO types: while bugs still remain, many more
1727 cases are accurately computed;
1728 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1729 if it is in the last clause;
1730 ** CONSTANTLY now correctly returns a side-effect-free function in
1732 ** DECLARE is no longer treated as a special-operator; in
1733 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1734 * incremented fasl file version number due to the change in the
1735 DEFSTRUCT-SLOT-DESCRIPTION structure.
1737 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1738 * minor incompatible change: code processed by the "interpreter" or
1739 EVAL now has a compilation optimization policy of (DEBUG 2)
1740 (changed from (DEBUG 1)) to improve debuggability of interactive
1741 development, and to allow the use of the debug RETURN command in
1743 * an experimental implementation of the RETURN command for the
1744 debugger has been included. (thanks to Frederik Kuivinen)
1745 * fixed bug 62: constraints were not propagated into a loop.
1746 * fixed bug in embedded calls of SORT (reported and investigated by
1748 * fixed some bugs revealed by Paul Dietz' test suite:
1749 ** printing and reading of arrays with some dimensions having
1750 length 0 (thanks to Gerd Moellmann);
1751 ** BOA constructor with &AUX argument without a default value does
1752 not cause a type error;
1753 ** CONSTANTP now returns true for all self-evaluating objects.
1755 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1756 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1757 accessors that are related by inheritance, as specified in the
1758 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1759 Valtteri Vuorikoski)
1760 * The compiler is now able to inline functions that were defined in
1761 a complex lexical environment (e.g. inside a MACROLET).
1762 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1764 * Support for the upcoming FreeBSD-5.0 release has been included.
1765 (thanks to Dag-Erling Smorgrav)
1766 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1767 effect when it is not in a toplevel context.
1768 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1769 stand a better chance of being correct. (thanks to Gerd
1771 * fixed bug in COERCE, which now signals an error on coercing a
1772 rational to a bounded real type which excludes the expected
1774 * The compiler is now able to derive types more accurately from the
1775 COERCE and COMPILE functions.
1776 * fixed bug 223: functional binding is considered to be constant
1777 only for symbols in the CL package.
1778 * fixed bug 231: SETQ did not check the type of a variable being set
1779 (reported by Robert E. Brown)
1780 * A new optimization for MAKE-INSTANCE has been included, fixing
1781 various bugs (including relating to :ALLOCATION :CLASS slots and
1782 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1784 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1785 ** As required by ANSI, LOOP now disallows anonymous collection
1786 clauses such as COLLECT I in conjunction with aggregate boolean
1787 clauses such as THEREIS (= I 1);
1788 ** LOOP now signals an error when any variable is reused in the
1789 same loop (including the potentially useful construct analogous
1790 to WITH A = 1 WITH A = (1+ A);
1791 ** IT is only a special loop symbol within the first clause of a
1792 conditional loop clause;
1793 ** LOOP with a typed iteration variable over a hashtable now
1794 signals a type error iff it should.
1795 * fixed some other bugs revealed by Paul Dietz' test suite:
1796 ** FILE-STREAM now names the class previously known as FD-STREAM;
1797 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1798 argument) no longer signals an error;
1799 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1800 of the slot symbol, rather than using the current package
1801 ((:CONC-NAME "") continues to intern the slot's name in the
1803 * incremented fasl file version number, because of the incompatible
1804 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1805 of the new implementation of DEFINE-COMPILER-MACRO.
1807 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1808 * Support for building SBCL for MIPS platforms running in
1809 little-endian mode has now been checked in, and basic
1810 functionality on said platforms verified.
1811 * minor incompatible change: PCL now records the pathname of a file
1812 in which methods and the like are defined, rather than its
1814 * minor incompatible change: TRUENAME now considers the truename of
1815 a file naming a directory to be the pathname with :DIRECTORY
1816 component indicating that directory.
1817 * minor incompatible change: a NAMED clause in the extended form of
1818 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1819 reason for the previous behaviour is unclear.
1820 * more systematization and improvement of CLOS and MOP conformance
1821 in PCL (thanks to Gerd Moellman and Pierre Mai):
1822 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1824 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1825 primary methods with no specializers;
1826 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1828 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1829 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1830 FORWARD-REFERENCED-CLASSes; error reporting on
1831 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1833 ** SXHASH on CLOS instances now uses a slot internal to the
1834 instance to return different numbers on distinct instances,
1835 while preserving the same return value through invocations of
1837 ** DEFMETHOD signals errors when methods with longer incongruent
1838 lambda lists are added to generic functions;
1839 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1840 CLASS, as specified in AMOP;
1841 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1842 on the order returned by the primary method for classes of
1843 class STANDARD-CLASS;
1844 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1845 * fixed some bugs shown by Paul Dietz' test suite:
1846 ** DOLIST puts its body in TAGBODY;
1847 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1849 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1850 value producing form;
1851 ** if more variables are given to PROGV than values, extra
1852 variables are bound and made to have no value;
1853 ** NSUBSTITUTE on list arguments gets the right answer with
1855 ** ELT signals an error of type TYPE-ERROR when the index argument
1856 is not a valid sequence index;
1857 ** LOOP signals (at macroexpansion time) an error of type
1858 PROGRAM-ERROR when duplicate variable names are found;
1859 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1860 ** FUNCALL of special-operators now cause an error of type
1862 ** PSETQ now works as required in the presence of side-effecting
1863 symbol-macro places;
1864 ** NCONC accepts any object as its last argument;
1865 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1867 ** loop-for-as-package does not require a package to be explicitely
1869 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1871 * fixed bug 166: compiler preserves "there is a way to go"
1872 invariant when deleting code.
1873 * fixed bug 172: macro lambda lists with required arguments after
1874 &REST arguments now cause an error to be signalled. (thanks to
1876 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1877 bug. (thanks to Matthew Danish)
1878 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1880 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1881 arguments are lexically rebound. (thanks to Gerd Moellmann and
1883 * fixed bug 194: error messages are now more informative when there
1884 is no primary method applicable in a call to a generic
1885 function. (thanks to Gerd Moellmann)
1886 * fixed bug in command line argument checking (thanks to Julian
1888 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1889 (thanks to Matthew Danish)
1890 * incremented fasl file version number, because of the
1891 SXHASH-related changes in the layout of CLOS data structures
1893 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1894 * minor incompatible change: The runtime (the Unix executable named
1895 "sbcl") is now much pickier about the .core files it will load.
1896 Essentially it now requires .core files to descend from the same
1897 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1898 as the runtime does. (The intent is to prevent the crashes which
1899 can occur, and which can even be reported as mysterious failures,
1900 when people patch the sources or change the build parameters
1901 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1902 sbcl and .core files.)
1903 * fixed bug: VALUES-LIST is no longer optimized away.
1904 * fixed bug 142: The FFI conversion of C string values to Lisp
1905 string values no longer conses excessively. (thanks to Nathan
1906 Froyd porting Raymond Toy's fix to CMU CL)
1907 * began to systematize and improve MOP conformance in PCL (thanks to
1908 Nathan Froyd, Gerd Moellman and Pierre Mai):
1909 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1911 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1912 ** FINALIZE-INHERITANCE is now called on class finalization;
1913 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1914 argument precedence order.
1915 * fixed bug 202: The compiler no longer fails on functions whose
1916 derived types contradict their declared type.
1917 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1918 so it can be non-toplevel.
1919 * The fasl file version number has changed (because of the new
1920 implementation of DEFMACRO).
1921 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1922 safe code, that any length requirement by their type-specifier
1923 argument is valid. The exceptions to this are described in bug
1925 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1926 functions now check the types of their inputs as required by ANSI.
1927 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1928 introduced symbol is DECLAREd to be SPECIAL.
1929 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1930 * fixed bug: :COUNT argument to sequence functions may be negative
1931 * fixed bug: body of DO-SYMBOLS may contain declarations
1932 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1933 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1934 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1935 symbol macro only once
1936 * fixed printing of call frame when argument list is unavailable
1937 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1938 * compiler no longer signals WARNING on unknown keyword
1941 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1942 * A beta-quality port to the mips architecture running Linux,
1943 based on the old CMUCL backend, has been made. It has been tested
1944 on a big-endian kernel, and works sufficiently well to be able to
1945 rebuild itself; it has not been tested in little-endian mode.
1946 * fixed an inconsistency between gencgc.c and purify.c which made
1947 dumping/loading .core files unreliable
1948 * fixed bug 120a: The compiler now deals correctly with IFs where
1949 the consequent is the same as the alternative, instead of
1950 misderiving the return type. (thanks to Alexey Dejneka)
1951 * fixed bug 113: Logical pathnames are now dumpable (the logical
1952 host is resolved at load-time, throwing an error if it is not
1954 * fixed bug 174: FORMAT's error message is slightly clearer when a
1955 non-printing character is used in a format directive.
1956 * fixed several bugs in compiler checking of type declarations, i.e.
1957 violations of the Python "declarations are assertions" principle
1958 (thanks to Alexey Dejneka)
1959 * fixed several bugs in PCL's error checking (thanks to Gerd
1961 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1963 * fixed bug in compilation of functions as first class values
1964 (thanks to Antonio Martinez-Shotton)
1965 * The compiler's handling TYPE-ERRORs which it can prove will
1966 inevitably happen at runtime has been cleaned up and corrected
1967 in several ways. (thanks to Alexey Dejneka)
1968 * improved argument type checking for various basic arithmetic
1969 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1970 lavished on them in the past that they can be compiled in many
1971 ways in different special cases
1972 * fixed bug 181: compiler checks validity of user supplied type
1974 * cleaned up code flushing in optimization: Function calls which
1975 should signal errors for safety purposes (e.g. which ANSI says
1976 should signal errors when their arguments are of incorrect type)
1977 are no longer optimized away.
1978 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1979 * incremented fasl file version number, because changes in the
1980 implementation of sequence functions like COERCE caused
1981 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1982 in old inline expansions) to become undefined. (Actually these
1983 changes were later undone, so we might very well be binary
1984 compatible with 0.7.7 after all, but leaving the version number
1985 incremented seemed like the simplest and most conservative
1988 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1989 * An alpha-quality port to the parisc architecture running Linux,
1990 based on the old CMUCL backend, has been made. This, even more so
1991 than the other backends, should be considered still a work in
1992 progress; known problems include that the Linux kernel in 64-bit
1993 mode does not propagate the correct sigcontext structure to
1994 userspace, and consequently SBCL on a parisc64 kernel will not
1996 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1997 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1998 suppressed.) Also now that the compiler is looking at declarations
1999 in the environment, it checks optimization declarations as well,
2000 and suppresses inlining when (> DEBUG SPEED).
2001 * More fixes have been made to treatment of floating point exception
2002 treatment and other Unix signals. In particular, floating point
2003 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2004 * The detection and handling of control stack exhaustion (infinite
2005 or very deeply nested recursion) has changed. Stack exhaustion
2006 detection is now done by write-protecting pages at the OS level
2007 and applies at all optimization settings; when found, a
2008 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2009 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2010 be caught using IGNORE-ERRORS.
2011 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2012 correctly when transforming recursive LABELS functions to LETs.
2013 (thanks to Alexey Dejneka)
2014 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2015 that are names of constants or global variables.
2016 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2017 alien routines with docstrings.
2018 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2019 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2021 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2022 (thanks to Eric Marsden)
2023 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2024 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2025 * Bug fix: The SPARC backend can now compile functions involving
2026 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2027 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2028 to a functions expecting a PATHNAME-DESIGNATOR.
2029 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2030 lambda lists. (thanks to Alexey Dejneka)
2031 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2032 functions so that EXPORTing the name of the class doesn't cause
2033 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2034 to Antonio Martinez for reporting this)
2035 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2036 constant-folded without failing an assertion. (thanks to Einar
2037 Floystad Dorum for reporting this)
2038 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2039 types are now dealt with more correctly by the compiler.
2040 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2041 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2042 behaviour analogous to RENAME-FILE. This puts its behaviour more
2043 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2045 * The fasl file version number has changed again. (because of the
2046 bug fix involving the names of PCL MAKE-INSTANCE functions)
2048 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2049 * bug fix: Floating point exceptions are treated much more
2050 consistently on the x86/Linux and PPC/Linux platforms.
2051 * Array initialization with :INITIAL-ELEMENT is now much faster for
2052 cases when the compiler cannot open code the array creation, but
2053 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2054 array accesses have also seen a speed increase.
2055 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2056 specified. (This was at the root of some bad interactions between
2057 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2059 * bug fix: Internal error arguments for undefined functions are now
2060 computed correctly on the PPC/Linux platform.
2061 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2062 Raymond Toy's patch for CMU CL)
2063 * Support for the Solaris 9 operating environment has been included
2064 (thanks to Daniel Merritt)
2065 * A very ugly but hopefully complete draft of the missing FFI chapter
2066 of the manual has been created by reformatting the corresponding
2067 CMU CL manual chapter into (currently very ugly and incoherent)
2068 DocBook and bringing it up to date for SBCL behavior. Thus, the
2069 manual is now essentially complete, at least by my extreme
2070 once-and-only-once standards, whereby it's acceptable to refer to
2071 the doc strings of SB-EXT functions as the primary documentation.
2072 * The fasl file version number has changed again, due to cleanup of
2073 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2074 is no longer a static symbol.)
2076 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2077 * SBCL now builds with OpenMCL (version 0.12) as the
2078 cross-compilation host; also, more progress has been made toward
2079 bootstrapping under CLISP.
2080 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2082 * bug 158 fixed: The compiler can now deal with integer loop
2083 increments different from 1; fixing this turned out also to fix
2085 * bug 169 fixed: no more bogus warnings about using lexical bindings
2086 despite the presence of perfectly good SPECIAL declarations (thanks
2087 to David Lichteblau)
2088 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2089 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2090 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2092 * bug 180 fixed: Method combination specifications no longer ignore
2093 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2094 * bug fix: Structure type predicate functions now check their argument
2095 count as they should.
2096 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2097 correctly. (thanks to Pierre Mai)
2098 * minor incompatible change: The --noprogrammer option is deprecated
2099 in favor of the new --disable-debugger option, which is very similar.
2100 (The major difference is that it takes effect at a slightly different
2101 time at startup, causing handling of errors in --sysinit and
2102 --userinit files will be affected differently.) The
2103 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2104 been added to allow this functionality to be controlled from ordinary
2105 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2106 maintainers, who might want to run non-interactive scripts to
2107 build SBCL cores which will later be used interactively.)
2108 * minor incompatible change: The LOAD function no longer, when given
2109 a wild pathname to load, loads all files matching that pathname.
2110 Instead, an error of type FILE-ERROR is signalled.
2112 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2113 * bug 147 fixed: The compiler preserves its block link/count
2114 invariants more correctly now so that it doesn't crash. (thanks
2116 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2118 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2119 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2120 no-value-for-this-character as NIL (as specified by ANSI).
2121 * HOST-NAMESTRING on physical pathnames now returns a string that is
2122 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2123 (thanks to Christophe Rhodes)
2124 * The Alpha port handles icache flushing more correctly. (thanks to
2126 * More progress has been made toward bootstrapping under CLISP. (thanks
2127 to Christophe Rhodes)
2128 * The fasl file format has changed again, because dynamic loading
2129 on OpenBSD (which has non-ELF object files) motivated some cleanups
2130 in the way that foreign symbols are transformed and passed around.
2131 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2132 is no longer treated as whitespace by the reader, but instead as
2133 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2134 |A
\7fB|, instead of A as it used to.
2136 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2137 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2138 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2139 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2140 this as of 0.7.1.45, but was left out of the previous news section
2141 (thanks to Dan Barlow)
2142 * SBCL now runs on the Solaris operating system on SPARC architectures
2143 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2144 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2145 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2146 * SPARC backend cleanups, allowing builds of cores optimized for V8
2147 and V9 SPARCS, and also emission of code targeted to a particular
2148 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2150 * SBCL is closer to bootstrapping under CLISP, thanks to various
2151 fixes by Christophe Rhodes.
2152 * The fasl file format has changed again, to allow the compiler's
2153 INFO database to support symbol macros.
2154 * The user manual (in doc/) is formatted into HTML more nicely.
2155 (thanks to coreythomas)
2156 * The system is smarter about SUBTYPEP relationships, especially
2157 those involving NOT types (including types such as ATOM which are
2158 represented internally using NOT types). Thus SUBTYPEP is less
2159 likely to return (VALUES NIL NIL) in general, and in particular
2160 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2162 * The fasl file format has changed again, because the internal
2163 representation of types now includes a new slot to support the new
2164 SUBTYPEP-of-NOT-types logic.
2165 * (not a change in the main branch of SBCL, but a related prototype
2166 which can hopefully be merged into the main branch of SBCL in the
2167 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2168 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2169 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2170 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2171 dispatch macro character combinations. (thanks to Alexey Dejneka)
2172 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2173 that superficially look like logical namestrings correctly.
2174 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2176 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2177 default supplied-p) for &optional and &key arguments. (thanks to
2180 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2181 * incompatible change: The compiler is now less aggressive about
2182 tail call optimization, doing it only when (> SPACE DEBUG) or
2183 (> SPEED DEBUG). (This is an incompatible change because there are
2184 programs which relied on the old CMU-CL-style behavior to optimize
2185 away their unbounded recursion which will now die of stack overflow.)
2186 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2187 for non-GENCGC systems has been increased to 20M (since that
2188 seems much closer to the likely performance optimum for modern
2189 systems than the old 4M value was)
2190 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2191 and *DEBUG-PRINT-LEVEL*
2192 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2193 of CMU CL's support for SPARC, and various endianness and other
2194 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2195 * new syntactic sugar for the Unix command line: --load foo.bar is now
2196 an alternate notation for --eval '(load "foo.bar")'.
2198 ** The system now detects stack overflow and handles it gracefully,
2199 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2200 optimization settings. (This is a good thing in general, and
2201 its introduction in this version should be particularly timely
2202 for anyone whose code fails because of suppression of tail
2204 ** The system now hunts for the C variable "environ" in a more
2205 devious way, to avoid segfaults when the C library version differs
2206 between compile time and run time. (thanks to Christophe Rhodes)
2207 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2208 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2209 ** The compiler no longer issues bogus style warnings for undefined
2210 classes in the same source file as the DEFCLASSes which defined
2211 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2213 ** fixes in CONDITION class precedence list for undefined function
2214 errors (thanks to Alexei Dejneka)
2215 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2216 correctly. (thanks to Dan Barlow)
2217 ** portability fixes aiming at bootstrapping under CLISP (thanks
2218 to Dave McDonald and Christophe Rhodes)
2219 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2220 ** fixes in type translation and and type inference (thanks to
2222 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2223 ** various fixes in the new ports (thanks to Dan Barlow)
2224 * several changes related to debugging:
2225 ** suppression of tail recursion, as noted above
2226 ** stack overflow detection, as noted above
2227 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2228 is now the default. (For some time encapsulation has been more
2229 reliable than the breakpoint-based :ENCAPSULATE NIL
2230 implementation, at least on X86 systems; and I just noticed that
2231 encapsulation also seems closer to the spirit of the ANSI
2234 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2236 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2237 up properly again. (There was a packaging bug in 0.7.0 which
2238 left their definitions in SB-SYS::LOAD-FOREIGN and
2239 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2240 vital for most things which interface to C-level interfaces,
2241 like extensions working with sockets or databases or
2242 Perl-compatible regexes or whatever, and the need to fix
2243 this bug is the main reason that 0.7.1 was released so
2245 ** DEFGENERIC is now choosier about the methods it redefines, so that
2246 reLOADing a previously-LOADed file containing DEFGENERICs does
2247 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2248 cycle now works as it should. (thanks to Alexey Dejneka)
2249 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2250 fixed by Christophe Rhodes. (He actually submitted this patch
2251 months ago, and I delayed until after 0.7.0.)
2252 ** Bug 111 (internal compiler confusion about runtime checks on
2253 FUNCTION types) was fixed by Alexey Dejneka.
2254 * Some internal cleanups (getting rid of variables which aren't
2255 needed now that the byte interpreter is gone) caused the fasl
2256 file format number to change again.
2258 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2259 * major incompatible change: The default fasl file extension, i.e. the
2260 default extension for files produced by COMPILE-FILE, has changed
2261 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2263 ** There are many changes in the implementation of the compiler.
2264 SBCL is now essentially a compiler-only implementation of ANSI
2265 Common Lisp. EVAL still "interprets" a few special cases, but
2266 almost all the interesting cases are handled by creating
2267 a LAMBDA expression, calling COMPILE on it, then calling
2268 FUNCALL on the result.
2269 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2270 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2271 Since the code is newer, there might still be some new bugs
2272 (though not as many as before Martin Atzmueller's fixes:-). But
2273 the new code is substantially simpler and clearer, and hopefully
2274 any remaining bugs will be simpler, less fundamental, and more
2275 fixable then the bugs in the old code.
2276 ** The revised compiler is still a little unsteady on its feet.
2278 *** The debugging information it produces (particularly the names
2279 of FUNCTION objects) is sometimes much less useful than what
2280 the old compiler produced.
2281 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2282 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2283 within a MACROLET) has been temporarily weakened.
2284 ** There are new compiler optimizations for various functions:
2285 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2286 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2287 *** the math functions TRUNCATE, FLOOR, and CEILING
2288 *** the function-of-all-trades COERCE
2289 Mostly these should be transparent, but there's one
2290 potentially-annoying problem (bug 117): when the compiler
2291 inline-expands a function and does type analysis on the result,
2292 it can create control paths which have type mismatches, and
2293 when it can't prove that those control paths aren't taken,
2294 it will issue WARNINGs about the type mismatches. This is
2295 a particular problem in practice for the new sequence functions.
2296 It's not clear how this should be fixed, and for now, a
2297 workaround is given in the entry for 117 in the BUGS file.
2298 ** (Because of the interaction between the two previous items --
2299 occasional inlining problems and new inline expansions -- some
2300 of the new sequence function optimizations won't really kick in
2301 completely until debugging information, and then inlining, are
2302 straightened out in some future version.)
2303 * minor incompatible changes:
2304 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2305 DIRECTORY no longer implicitly promotes NIL slots of its
2306 pathname argument to :WILD. In particular, when you ask for the
2307 contents of a directory (which you used to be able to do without
2308 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2309 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2310 ** changes in behavior that ANSI explicitly defines to be
2311 implementation dependent:
2312 *** The new compiler-only implementation still conforms with ANSI,
2313 but acts a little different than before. Besides the obvious
2314 changes in performance tradeoffs (that the cost per form passed
2315 to EVAL has gone up, and the cost per form executed by EVAL
2316 has gone down), the behavior of the system changes a little
2317 because there are no longer any interpreted function objects.
2318 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2319 e.g. doing COMPILE on the output of interactive DEFUN is
2321 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2323 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2324 has changed from (:CL) to NIL.
2325 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2326 e.g. #\Newline and #\Tab, don't have names specified in the
2327 ANSI Common Lisp standard, is now based on their ASCII symbolic
2328 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2329 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2330 are no longer used for output.
2331 ** changes in internal implementation constants:
2332 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2333 4 million. (If your application spends a lot of time GCing and
2334 you have a lot of RAM, you might want to experiment with
2335 increasing it even more.)
2336 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2337 However, almost all old code should still continue to work without
2338 immediate update, as SB-C-CALL is now a (deprecated) nickname
2340 ** Old operator names in the style DEF-FOO are now deprecated in
2341 favor of new corresponding names DEFINE-FOO, for consistency with
2342 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2343 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2344 internal symbols, but a few supported extensions like
2345 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2346 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2347 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2348 etc. as you get deeper into recursive calls to the debugger
2349 command loop, instead of the old "5]", "5]]", "5]]]"
2350 sequence. (I was motivated to do this when squabbles between
2351 ILISP and SBCL left me very deeply nested in the debugger. In the
2352 short term, this change will probably provoke more ILISP/SBCL
2353 squabbles, but hopefully it will be an improvement in the long run.)
2354 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2355 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2356 function, since declaiming return types involving aliens is
2357 (1) annoyingly messy to do by hand and (2) vital to efficient
2358 compilation of code which calls such functions.
2359 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2360 longer reexported by the SB-EXT package. They're solely useful
2361 for alien code, so it seems more logical that you should get
2362 them from the SB-ALIEN package, not in SB-EXT.
2363 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2364 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2365 features. Instead, the code that they used to control is always
2366 built into the system.
2367 * many other bug fixes
2368 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2369 advantage of the new EVAL-WHEN stuff and to clean them up in
2370 general, and they are now more ANSI-compliant in a number of
2371 ways. Martin Atzmueller is responsible for a lot of this.
2372 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2374 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2375 *** correct ERROR type for various file operations
2376 *** some fixes for Lisp streams
2377 *** DEFMETHOD syntax checking
2378 *** changing old weird representation of debug information as
2379 strings (which, among their other deficiencies, don't transform
2380 correctly when you rename packages, and don't change their
2381 print representation when you change things like *PACKAGE*
2382 and *PRINT-LENGTH*) to symbols and lists of symbols
2383 He also made several improvements and fixed several bugs in DESCRIBE.
2384 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2386 *** misbehavior of WRITE-STRING/WRITE-LINE
2387 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2388 and several other LOOP problems as well
2389 *** DIRECTORY when similar filenames are present
2390 *** DEFGENERIC with :METHOD options
2391 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2392 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2393 *** argument ordering in FIND with :TEST option
2394 *** mishandled package designator argument in APROPOS-LIST
2395 *** various problems in the backquote readmacro
2396 *** a bug in APROPOS
2397 *** probably some others that I'm not describing very well here,
2398 since the CVS log documents them by reference to sbcl-devel
2399 messages, and the SourceForge archives aren't working well.:-(
2400 ** Dan Barlow improved the Alpha port (and is making progress on the
2401 PPC port, for those of you who think different).
2402 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2403 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2404 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2405 continues to work on the SPARC port (for those of you in a position
2406 to look down upon our little PC-compatible boxes from a great height).
2407 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2408 and uses that copy, rather than the current dynamic value, when
2409 it's trying to decide whether to truncate output. Thus e.g.
2410 (let ((*print-lines* 50))
2411 (pprint-logical-block (stream nil)
2413 (let ((*print-lines* 8))
2414 (print (aref possiblybigthings i) stream)))))
2415 should now truncate the logical block only at 50 lines, instead of
2416 often truncating it at 8 lines, as it did before.
2417 * The doc/cmucl/ directory, containing old CMU CL documentation
2418 from the time of the fork, is no longer part of the base system.
2419 SourceForge has shut down its anonymous FTP service, and with it
2420 my original plan for distributing the old CMU CL documentation
2421 there. For now, if you need these files you can download an old
2422 SBCL source release and extract them from it.
2423 * The fasl file version number changed again, for dozens of reasons,
2424 some of which are apparent above.
2426 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2427 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2428 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2429 patch, so that DISASSEMBLE output is much nicer.
2430 * The code in the SB-PROFILE package now seems reasonably stable.
2431 I still haven't decided what the final interface should look like
2432 (I'd like PROFILE to interact cleanly with TRACE, since both
2433 facilities use function encapsulation) but if you have a need
2434 for profiling now, you can probably use it successfully with
2435 the current CMU-CL-style interface.
2436 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2437 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2438 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2439 link now returns the dangling link itself, and for similar
2440 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2441 link itself. (In these cases the old code signalled an error and
2442 looped endlessly, respectively.) Thus, DIRECTORY now works even
2443 in the presence of dangling and cyclic symbolic links.
2444 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2445 is now a supported extension again, since the consensus on
2446 sbcl-devel was that it can be useful for ordinary development
2447 work, not just for debugging SBCL itself.
2448 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2449 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2450 that currently-defined functions might be redefined later with
2451 different return types.
2452 * Hash tables can be printed readably, as inspired by CMU CL code
2453 of Eric Marsden and SBCL code of Martin Atzmueller.
2454 * better error handling in CLOS method combination, thanks to
2455 Martin Atzmueller porting Pierre Mai's CMU CL patches
2456 * more overflow fixes for >16Mbyte I/O buffers
2457 * A bug in READ has been fixed, so that now a single Ctrl-D
2458 character suffices to cause end-of-file on character streams.
2459 In particular, now you only need one Ctrl-D at the command
2460 line (not two) to exit SBCL.
2461 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2463 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2464 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2465 does the right thing.
2466 * The implementation of some type tests, especially for CONDITION
2467 types, is now tidier and maybe faster, due to CMU CL code
2468 originally by Douglas Crosher, ported by Martin Atzmueller.
2469 * Some math functions have been fixed, and there are new
2470 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2471 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2472 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2473 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2474 likely to be built into sbcl-0.7.0.)
2475 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2476 CL, ported to SBCL by Martin Atzmueller)
2477 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2478 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2479 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2480 * There's a new slam.sh hack to shorten the edit/compile/debug
2481 cycle for low-level changes to SBCL itself, and a new
2482 :SB-AFTER-XC-CORE target feature to control the generation of
2483 the after-xc.core file needed by slam.sh.
2484 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2485 COMPILE-FILE is no longer supported, so that now every function
2486 gets an entry point, so that block compilation looks a little
2487 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2488 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2489 returns the number of bytes consed since the system started,
2490 rather than the number consed since the first time the function
2491 was called. (The new definition parallels ANSI functions like
2492 CL:GET-INTERNAL-RUN-TIME.)
2493 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2494 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2495 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2496 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2497 * Fasl file version numbers are now independent of the target CPU,
2498 since historically most system changes which required version
2499 number changes have affected all CPUs equally. Similarly,
2500 the byte fasl file version is now equal to the ordinary
2503 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2504 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2505 is no longer recognized. I apologize for this, because it was
2506 listed in SB-EXT as a supported extension, but I found that
2507 its existing behavior was poorly specified, as well as incorrectly
2508 specified, and it looked like too much of a mess to straighten it
2509 out. I have enough on my hands trying to get ANSI stuff to work..
2510 * many patches ported from CMU CL by Martin Atzmueller, with
2511 half a dozen bug fixes in pretty-printing and the debugger, and
2512 half a dozen others elsewhere
2513 * fixed bug 13: Floating point infinities are now supported again.
2514 They might still be a little bit flaky, but thanks to bug reports
2515 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2516 as flaky as they were.
2517 * The --noprogrammer command line option is now supported. (Its
2518 behavior is slightly different in detail from what the old man
2519 page claimed it would do, but it's still appropriate under the
2520 same circumstances that the man page talks about.)
2521 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2522 are now supported, and enabled by default. Thus, the compiler can
2523 handle many floating point and complex operations much less
2524 inefficiently. (Thus e.g. you can implement a complex FFT
2526 * The compiler now detects type mismatches between DECLAIM FTYPE
2527 and DEFUN better, and implements CHECK-TYPE more correctly, and
2528 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2529 thanks to patches from Martin Atzmueller.
2530 * various fixes to make the cross-compiler more portable to
2531 ANSI-conforming-but-different cross-compilation hosts (notably
2532 Lispworks for Windows, following bug reports from Arthur Lemmens)
2533 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2534 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2535 fixed thanks to Pierre Mai's CMU CL patches.
2536 * fixes to keep the system from overflowing internal counters when
2537 it tries to use i/o buffers larger than 16M bytes
2538 * fixed bug 45a: Various internal functions required to support
2539 complex special functions have been merged from CMU CL sources.
2540 (When I was first setting up SBCL, I misunderstood a compile-time
2541 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2542 * improved support for type intersection and union, fixing bug 12
2543 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2544 more obscure bugs as well
2545 * some steps toward byte-compiling non-performance-critical
2546 parts of the system, courtesy of patches from Martin Atzmueller
2547 * Christophe Rhodes has made some debian packages of sbcl at
2548 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2549 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2550 stable, but are nonetheless usable. When he's ready, I'd be happy
2551 to add them to the SourceForge "File Releases" section. (And if
2552 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2553 * new fasl file format version number (because of changes in
2554 internal representation of (OR ..) types to accommodate the new
2555 support for (AND ..) types, among other things)
2557 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2558 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2560 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2562 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2563 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2564 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2565 * DESCRIBE now works on structure objects again.
2566 * Most function call argument type mismatches are now handled as
2567 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2568 know whether the function will be redefined before the call is
2569 executed. (The compiler could flag local calls with full WARNINGs,
2570 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2571 it doesn't keep track of enough information to know whether calls
2572 are local in this sense.)
2573 * Compiler output is now more verbose, with messages truncated
2574 later than before. (There should be some supported way for users
2575 to override the default verbosity, but I haven't decided how to
2576 provide it yet, so this behavior is still controlled by the internal
2577 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2578 src/compiler/ir1util.lisp.)
2579 * Fasl file format version numbers have increased again, because
2580 support for the Gray streams extension changes the layout of the
2581 system's STREAM objects.
2582 * The Gray subclassable streams extension now works, thanks to a
2583 patch from Martin Atzmueller.
2584 * The full LOAD-FOREIGN extension (not just the primitive
2585 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2586 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2587 but like most other programs, it defaults to copying the Unix
2588 environment from the original process instead of starting the
2589 new process in an empty environment.
2590 * Extensions which manipulate the Unix environment now support
2591 an :ENVIRONMENT keyword option which doesn't smash case or
2592 do other bad things. The CMU-CL-style :ENV option is retained
2593 for porting convenience.
2594 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2595 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2597 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2599 * A patch from Martin Atzmueller seems to have solved the SIGINT
2600 problem, and as far as we know, signal-handling now works cleanly.
2601 (If you find any new bugs, please report them!)
2602 * The system no longer defaults Lisp source file names to types
2603 ".l", ".cl", or ".lsp", but only to ".lisp".
2604 * The compiler no longer uses special default file extensions for
2605 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2606 seems to expect a single default extension for all compiled code,
2607 and there's no compelling reason to try to stretch the standard
2608 to allow two different extensions.) Instead, byte-compiled files
2609 default to the same extension as native-compiled files.
2610 * Fasl file format version numbers have increased again, because
2611 a rearrangement of internal implementation packages made some
2612 dumped symbols in old fasl files unreadable in new cores.
2613 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2614 many fewer weird special cases.
2615 * Bug #17 (differing COMPILE-FILE behavior between logical and
2616 physical pathnames) has been fixed, and some related misbehavior too,
2617 thanks to a patch from Martin Atzmueller.
2618 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2619 by Tim Moore, ported to SBCL by Martin Atzmueller.
2620 * Martin Atzmueller fixed several filesystem-related problems,
2621 including bug #36, in part by porting CMU CL patches, which were
2622 written in part by Paul Werkowski.
2623 * More compiler warnings in src/runtime/ are gone, thanks to
2624 more patches from Martin Atzmueller.
2625 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2628 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2630 * DESCRIBE now works on CONDITION objects.
2631 * The debugger now handles errors which arise when trying to print
2632 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2634 * The build system now uses an additional file, customize-target-features.lisp,
2635 to allow local modifications to the target *FEATURES* list. (The point of
2636 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2637 debugging features enabled, without having to worry about propagating it
2638 into everyone's system when I do a "cvs update".) When no
2639 customize-target-features.lisp file exists, the target *FEATURES* list
2640 should be constructed the same way as before.
2641 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2642 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2643 warn about behavior which is undefined under the ANSI spec. Note: This
2644 is specified by ANSI, but it's not very popular with programmers.
2645 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2646 macro in the SBCL sources for an example of a workaround which you
2647 might use to make portable ANSI-standard code which does what you want.
2648 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2649 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2650 when it's used as a non-toplevel form. (This is required in order
2651 to implement the DEFCONSTANT-EQX macro.)
2652 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2653 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2654 funny things, due to bugs in the implementation of EVAL-WHEN
2655 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2656 non-toplevel DEFCONSTANTs are uncommon.)
2657 * The core file version number and fasl file version number have been
2658 incremented, because the old noncompliant DEFCONSTANT behavior involved
2659 calling functions which no longer exist, and because I also took the
2660 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2661 * fixed bug 1 (error handling before read-eval-print loop starts), and
2662 redid debugger restarts and related debugger commands somewhat while
2664 ** The QUIT debugger command is gone, since it did something
2665 rather different than the SB-EXT:QUIT command, and since it never
2666 worked properly outside the main toplevel read/eval/print loop.
2667 Invoking the new TOPLEVEL restart provides the same functionality.
2668 ** The GO debugger command is also gone, since you can just invoke
2669 the CONTINUE restart directly instead.
2670 ** The TOP debugger command is also gone, since it's redundant with the
2671 FRAME 0 command, and since it interfered with abbreviations for the
2673 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2675 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2676 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2677 some porting work by Martin Atzmueller)
2678 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2679 with a boatload of patches which he ported to SBCL. Now that those
2681 ** The system tries to make sure that its low-priority messages
2682 are prefixed by semicolons, to help people who like to use
2683 syntax highlighting in their ILISP buffer. (This patch
2684 was originally due to Raymond Toy.)
2685 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2686 patches originally written by Raymond Toy.
2687 ** The compiler understands coercion between single-value and
2688 multiple-VALUES type expressions better, getting rid of some very
2689 weird behavior, thanks to patches originally by Robert MacLachlan
2690 and Douglas Crosher.
2691 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2692 lambda lists, thanks to a patch originally by Pierre Mai.
2693 ** The system no longer bogusly warns about "abbreviated type
2695 ** The compiler gets less confused by inlining and RETURN-FROM,
2696 thanks to some patches originally by Tim Moore.
2697 ** The system no longer hangs when dumping circular lists to fasl
2698 files, thanks to a patch originally from Douglas Crosher.
2699 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2700 undefined function error.
2701 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2702 in structure slot initforms) for now, documented workaround instead:-|
2703 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2704 * fixed bug 5: added stubs for various Gray stream functions called
2705 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2706 installed, at least appropriate type errors are generated
2707 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2708 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2709 consistently in DEFMETHOD forms.
2710 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2711 it doesn't seem to affect SBCL after all
2712 * The C runtime system now builds with better optimization and many
2713 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2715 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2717 * The system is now under CVS at SourceForge (instead of the
2718 CVS repository on my home machine).
2719 * The new signal handling code has been tweaked to treat register
2720 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2721 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2722 downstream, e.g. in debug-int.lisp, has implicit dependencies
2723 on the unsignedness of integer representation of machine words,
2724 and that caused the system to bomb out with infinite regress
2725 when trying to recover from type errors involving signed values,
2726 e.g. (BUTLAST '(1 2 3) -1).)
2727 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2728 (This was one of the bugs Peter Van Eynde reported back in July.)
2729 * The system now uses code inspired by Colin Walters' O(N)
2730 implementation of MAP (from the cmucl-imp@cons.org mailing
2731 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2732 inline the MAP operation, and there is more than one
2733 sequence argument to the MAP call (so that it can't just
2734 do ETYPECASE once and for all based on the type of the
2735 single sequence argument). (The old non-inline implementation
2736 of the general M-argument sequence-of-length-N case required
2737 O(M*N*N) time when any of the sequence arguments were LISTs.)
2738 * The QUIT :UNIX-CODE keyword argument has been renamed to
2739 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2740 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2742 * Raymond Wiker's patches to port dynamic loading from Linux to
2743 FreeBSD have been added.
2744 * The BUGS file is now more nearly up to date, thanks in large part
2745 to Martin Atzmueller's review of it.
2746 * The debugger now flushes standard output streams before it begins
2747 its output ("debugger invoked" and so forth).
2748 * The core version number and fasl file version number have both
2749 been incremented, because of incompatible changes in the layout
2751 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2752 used to not be called for a saved Lisp image.)
2753 * Martin Atzmueller's version of a patch to fix a compiler crash,
2754 as posted on sbcl-devel 13 September 2000, has been installed.
2755 * Instead of installing Martin Atzmueller's patch for the
2756 compiler transform for SUBSEQ, I deleted the compiler transform,
2757 and transforms for some similar consing operations.
2758 * A bug in signal handling which kept TRACE from working on OpenBSD
2760 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2761 away by constant folding
2762 * The system now defines its address space constants in one place
2763 (in the Lisp sources), and propagates them automatically elsewhere
2764 (through GENESIS and the sbcl.h file). Therefore, patching the
2765 address map is less unnecessarily tedious and error-prone. The
2766 Lisp names of address space constants have also been systematized.
2767 * CVS tags like dollar-Header-dollar have been removed from
2768 the sources, because they have never saved me trouble and
2769 they've been source of trouble working with patches and other
2770 diff-related operations.
2771 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2772 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2774 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2776 * The system has been ported to OpenBSD.
2777 * The system now compiles with a simple "sh make.sh" on the systems
2778 that it's supported on. I.e., now you no longer need to tweak
2779 text in the target-features.lisp-expr and symlinks in src/runtime/
2780 by hand, the make.sh takes care of it for you.
2781 * The system is no longer so grossly inefficient when compiling code
2782 involving vectors implemented as general (not simple) vectors (VECTOR T),
2783 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2784 which dares to use the various sequence functions on non-simple
2785 vectors, takes less of a performance hit.
2786 * There is now a primitive type predicate VECTOR-T-P
2787 to test for the (VECTOR T) type, so that e.g.
2788 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2789 can now be compiled with some semblance of efficiency. (The old code
2790 turned the type declaration into a full call to %TYPEP at runtime!)
2791 * AREF on (VECTOR T) is still not fast, since it's still compiled
2792 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2793 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2794 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2795 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2796 without making full calls to SUBTYPEP at runtime.
2797 (Some analogous efficiency issues for non-simple vectors specialized to
2798 element types other than T, or for non-simple multidimensional arrays,
2799 have not been addressed. They could almost certainly be handled the
2800 same way if anyone is motivated to do so.)
2801 * The changes in array handling break binary compatibility, so
2802 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2803 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2804 instead of (VALUES T T).
2805 * By following the instructions that Dan Barlow posted to sbcl-devel
2806 on 2 July 2000, I was able to enable primitive dynamic object
2807 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2808 functionality is not implemented (since it calls ld to resolve
2809 library references automatically, requiring RUN-PROGRAM for its
2810 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2811 try to resolve library references) is now supported.
2812 * The system now flushes the standard output streams when it terminates,
2813 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2814 them at several other probably-convenient times, e.g. in each pass of
2815 the toplevel read-eval-print loop, and after evaluating a form given
2816 as an "--eval" command-line option. (These changes were motivated by a
2817 discussion of stream flushing issues on cmucl-imp in August 2000.)
2818 * The source transform for TYPEP of array types no longer assumes
2819 that an array whose element type is a not-yet-defined type
2820 is implemented as an array of T, but instead punts, so that the
2821 type will be interpreted at runtime.
2822 * There is now some support for cross-compiling in make.sh: each of
2823 the phases of make.sh has its own script. (This should be transparent
2824 to people doing ordinary, non-cross-compile builds.)
2825 * Since my laptop doesn't have hundreds of megabytes of memory like
2826 my desktop machine, I became more motivated to do some items on
2827 my to-do list in order to reduce the size of the system a little:
2828 ** Arrange for various needed-only-at-cold-init things to be
2829 uninterned after cold init. To support this, those things have
2830 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2831 symbols with such names are now uninterned after cold init).
2832 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2833 fasl files for cold load.
2834 ** Remove the old compiler structure pooling code (which used to
2835 be conditional on the target feature :SB-ALLOC) completely.
2836 ** Redo the representation of some data in cold init to be more compact.
2837 (I also looked into supporting byte compiled code at bootstrap time,
2838 which would probably reduce the size of the system a lot, but that
2839 looked too complicated, so I punted for now.)
2840 * The maximum signal nesting depth in the src/runtime/ support code has
2841 been reduced from 4096 to 256. (I don't know any reason for the very
2842 large old value. If the new smaller value turns out to break something,
2843 I'll probably just bump it back up.)
2844 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2846 * Many, many bugs reported by Peter Van Eynde have been added to
2847 the BUGS list; some have even been fixed.
2848 * While enabling dynamic object file loading, I tried to make the
2849 code easier to understand, renaming various functions and variables
2850 with less ambiguous names, and changing some function calling
2851 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2852 * While trying to figure out how to do the OpenBSD port, I tried to
2853 clean up some of the code in src/runtime/. In particular, I dropped
2854 support for non-POSIX signal handling, added various comments,
2855 tweaked the code to reduce the number of compilation warnings, and
2856 renamed some files to increase consistency.
2857 * To support the new automatic configuration functionality in make.sh,
2858 the source file target-features.lisp-expr has been replaced with the
2859 source file base-target-features.lisp-expr and the machine-generated
2860 file local-target-features.lisp-expr.
2861 * fixed a stupid quoting error in make.sh so that using CMU CL
2862 "lisp -batch" as cross-compilation host works again
2864 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2866 * DESCRIBE no longer tries to call itself recursively to describe
2867 bound/fbound values, so that it no longer fails on symbols which are
2868 bound to themselves (like keywords, T, and NIL).
2869 * DESCRIBE now works on generic functions.
2870 * The printer now prints less-screwed-up representations of closures
2871 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2872 * A private symbol is used instead of the :EMPTY keyword previously
2873 used to mark empty slots in hash tables. Thus
2874 (DEFVAR *HT* (MAKE-HASH-TABLE))
2875 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2876 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2877 now does what ANSI says that it should. (You can still get
2878 similar noncompliant behavior if bang on the hash table
2879 implementation with all the symbols you get back from
2880 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2881 This breaks binary compatibility, since tests for equality to
2882 :EMPTY are wired into things like the macroexpansion of
2883 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2885 * There's now a minimal placeholder implementation for CL:STEP,
2886 as required by ANSI.
2887 * An obscure bug in the interaction of the normal compiler, the byte
2888 compiler, inlining, and structure predicates has been patched
2889 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2890 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2892 * Missing ordinary arguments in a macro call are now detected even
2893 when the macro lambda list contains &KEY or &REST.
2894 * The debugger no longer complains about encountering the top of the
2895 stack when you type "FRAME 0" to explicitly instruct it to go to
2896 the top of the stack. And it now prints the frame you request even
2897 if it's the current frame (instead of saying "You are here.").
2898 * As specified by ANSI, the system now always prints keywords
2899 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2901 * The default initial SIZE of HASH-TABLEs is now smaller.
2902 * Type information from CLOS class dispatch is now propagated
2903 into DEFMETHOD bodies, so that e.g.
2904 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2906 is now basically equivalent to
2907 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2908 (DECLARE (TYPE SINGLE-FLOAT X))
2910 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2911 operation, without having to do run-time type dispatch.
2912 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2913 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2914 * Since I don't seem to be making big file reorganizations very often
2915 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2916 is overflowing my ability to conveniently back them up), I've finally
2917 checked the system into CVS. (The CVS repository is on my home system,
2918 not at SourceForge -- putting it on SourceForge might come later.)
2919 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2920 high-level GC-NOTIFY-FOO functions send their output. (There's
2921 still very little control of where low-level verbose GC functions
2922 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2923 less than it used to -- the GC-NOTIFY-FOO functions are now under
2924 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2925 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2926 in only one place in the source code, and propagates it automatically
2927 everywhere that it's needed. Thus e.g. when I bump the version from
2928 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2929 * The C source files now include boilerplate legalese and documentation
2930 at the head of each file (just as the Lisp source files already did).
2931 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2932 to his page will be replaced with a link to his new CLiki service.
2934 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2936 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2937 * The build process now looks for GNU make under the default name "gmake",
2938 instead of "make" as it used to. If GNU make is not available as "gmake"
2939 on your system, you can change this default behavior by setting the
2940 GNUMAKE environment variable.
2941 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2942 can build without error under CMU CL.
2944 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2946 * There is now a partial SBCL user manual (with some new text and some
2947 text cribbed from the CMU CL manual).
2948 * The beginnings of a profiler have been added (starting with the
2949 CMU CL profiler and simplifying and cleaning up). Eventually the
2950 main interface should be through the TRACE macro, but for now,
2951 it's still accessed through vaguely CMU-CL-style functions and macros
2952 exported from the package SB-PROFILE.
2953 * Some problems left over from porting CMU CL to the new
2954 cross-compilation bootstrap process have been cleaned up:
2955 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2956 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2957 part when getting rid of the compiler *BACKEND* stuff.)
2958 ** The value of *NULL-TYPE* was screwed up, because it was
2959 being initialized before the type system knew the final
2960 definition of the 'NULL type. This screwed up several key
2961 optimizations in the compiler, causing inefficiency in all sorts
2962 of places. (I found it because I wanted to understand why
2963 GET-INTERNAL-RUN-TIME was consing.)
2964 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2965 PROCLAIM FTYPE information. Unfortunately this broke binary
2966 compatibility again, since now the forms output by DEFGENERIC
2967 to refer to functions which didn't exist in 0.6.3.
2968 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2969 can use the new (as of 0.6.3) transform for SOME into MAP into
2971 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2972 values to (INTEGER 0 1000000), so that the time code will no longer
2973 occasionally get blown up by Linux returning 1000000 microseconds
2974 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2975 its output conform to the ANSI spec. (Alas, this makes its output
2976 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2977 * A full call to MAP NIL with a single sequence argument no longer conses.
2978 * fixes to problems pointed out by Martin Atzmueller:
2979 * The manual page no longer talks about multiprocessing as though
2980 it were currently supported.
2981 * The ILISP support patches have been removed from the distribution,
2982 because as of version 5.10.1, ILISP now supports SBCL without us
2983 having to maintain patches.
2984 * added a modified version of Raymond Toy's recent CMU CL patch for
2985 EQUALP comparison of HASH-TABLE
2987 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2989 * The system still can't cross-compile itself with
2990 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2991 entails), but at least it can compile more of itself that way
2992 than it used to be able to, and various buglets which were uncovered
2993 by trying to cross-compile itself that way have now been fixed.
2994 * This release breaks binary compatibility again. This time
2995 at least I've incremented the FASL file format version to 2, so that the
2996 problem can be detected reliably instead of just causing weird errors.
2997 * various new style warnings:
2998 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2999 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3000 ** using the lexical binding of a variable named in the *FOO* style
3001 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3002 as specified by ANSI.
3003 * *RANDOM-STATE* is no longer automatically initialized from
3004 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3005 default behavior of the system is to repeat its behavior every time
3006 it's run. If you'd like to change this behavior, you can always
3007 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3008 old convention there was no comparably easy way to get the system to
3009 repeat its behavior every time it was run.
3010 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3011 FTYPE declarations has been removed, in favor of their ANSI
3012 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3013 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3014 the types of their sequence arguments can be determined at compile time.
3015 This is done through a new open code expansion for MAP which eliminates
3016 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3017 argument types can be determined at compile time.
3018 * The optimizer now transforms COERCE into an identity operation when it
3019 can prove that the coerced object is already of the correct type. (This
3020 can be a win for machine generated code, including the output of other
3021 optimization transforms, such as the MAP transform above.)
3022 * Credit information has been moved from source file headers into CREDITS.
3023 * Source file headers have been made more standard.
3024 * The CASE macro now compiles without complaining even when it has
3027 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3029 * (Note that the way that the PCL macroexpansions were rewritten
3030 to accommodate the change in DEFGENERIC below breaks binary
3031 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3032 not run under sbcl-0.6.2. Once we get out of alpha releases,
3033 i.e. hit release 1.0.0, we'll probably try to maintain binary
3034 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3035 and sbcl-1.4.4. Until then, however, it might be fairly common
3036 for maintenance releases to break binary compatibility.)
3037 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3039 * The reporting of unhandled signals has been changed to print some
3040 explanatory text as well as the report form. (Previously only
3041 the report form was printed.)
3042 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3043 it defines, so that the compiler no longer issues undefined function
3044 warnings for compiled-but-not-yet-loaded generic functions.
3045 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3046 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3047 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3048 section of the ANSI spec are supported.
3049 * The "" nickname for the "KEYWORD" package has been removed.
3050 The reader still handles symbol tokens which begin with a package marker
3051 as keywords, but it doesn't expose its mechanism for doing so in the
3052 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3053 * The system now issues STYLE-WARNINGs for contradictory TYPE
3054 proclamations. (Warnings for contradictory FTYPE proclamations would
3055 be nice too, but those can't be done usefully unless the type system
3056 is made smarter about FUNCTION types.)
3057 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3058 systematized, so that "*target-*.lisp is supposed to exist only on the
3059 target and imply that there's a related file which exists on the
3060 host, and *host-*.lisp is supposed to exist only on the host and imply
3061 that there's a related file which exists on the target. This involves a
3062 lot of renaming. Hopefully the acute confusion caused by the renaming
3063 will be justified by the reduction in chronic confusion..
3064 ** runtime-type.lisp -> early-target-type.lisp
3065 ** target-type.lisp -> late-target-type.lisp
3066 ** early-host-format.lisp -> early-format.lisp
3067 ** late-host-format.lisp -> late-format.lisp
3068 ** host-error.lisp -> misc-error.lisp
3069 ** early-error.lisp -> early-target-error.lisp
3070 ** late-error.lisp -> late-target-error.lisp
3071 ** host-defboot.lisp -> early-defboot.lisp
3072 ** code/misc.lisp -> code/target-misc.lisp
3073 ** code/host-misc.lisp -> code/misc.lisp
3074 ** code/numbers.lisp -> code/target-numbers.lisp
3075 ** code/early-numbers.lisp -> numbers.lisp
3076 ** early-host-type.lisp -> early-type.lisp
3077 ** late-host-type.lisp -> late-type.lisp
3078 ** host-typep.lisp -> typep.lisp
3079 ** load.lisp -> target-load.lisp
3080 ** host-load.lisp -> load.lisp
3081 ** host-disassem.lisp -> disassem.lisp
3082 ** host-insts.lisp -> insts.lisp
3083 ** byte-comp.lisp -> target-byte-comp.lisp
3084 ** host-byte-comp.lisp -> byte-comp.lisp
3085 ** host-signal.lisp -> signal.lisp
3086 ** host-defstruct.lisp -> defstruct.lisp
3087 ** late-target-type.lisp -> deftypes-for-target.lisp
3088 Furthermore, several other previously target-only files foo.lisp (e.g.
3089 hash-table.lisp and random.lisp) have been split into a target-and-host
3090 foo.lisp file and a target-only target-foo.lisp file, with their key type
3091 definitions in the target-and-host part, so that the cross-compiler will
3092 know more about target types.
3093 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3094 gone away. In their place are various *BACKEND-FOO* variables
3095 corresponding to the slots of the old structure.
3096 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3097 into the target SBCL has been fixed.
3098 * The chill.lisp system for loading cold code into a running SBCL
3100 * Support for the CMU CL "scavenger hook" extension has been removed.
3101 (It was undocumented and unused in the CMU CL sources that SBCL was
3102 derived from, and stale in sbcl-0.6.1.)
3103 * Various errors in the cross-compiler type system were detected
3104 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3105 (enabling various consistency checks). Many of them were fixed,
3106 but some hard problems remain, so the compiler is back to
3107 running without *TYPE-SYSTEM-INITIALIZED* for now.
3108 * As part of the cross-compiler type system cleanup, I implemented
3109 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3110 * I have started adding UNCROSS calls throughout the type system
3111 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3112 on cross-compiler input files will be able to go away, and various
3114 * CONSTANTP now returns true for quoted forms (as explicitly required
3117 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3119 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3120 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3121 (MAKE-LIST -1). In the longer term, it ought to become true
3122 that declarations are assertions even at SAFETY 1. For now, it's not
3123 quite true even at SAFETY 3, but it's at least more nearly true..
3124 (Note that this change seems to increases the size of the system by
3125 O(5%) and to decrease the speed of the compiler by 20% or more.)
3126 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3127 for the problem of printing dozens of lines of distracting information
3128 about low-level system machinery as part of the top stack frame
3129 on entry to the debugger when an undefined function was called.
3130 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3131 is not reset to COMMON-LISP-USER.
3132 * Compilation of stuff related to dyncount.lisp has been made conditional
3133 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3134 smaller. The various dyncount-related symbols have been moved into
3135 a new "SB-DYNCOUNT" package.
3136 * tty-inspect.lisp has been renamed to inspect.lisp.
3137 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3138 feature has gone away. (When we eventually port to other flavors of
3139 libc and/or Unix, we'll try to make the differences between flavors
3140 invisible at the user level.)
3141 * Various other *FEATURES* tags, and/or their associated conditionals,
3142 have been removed if obsolescent, or given better documentation, or
3143 sometimes given more-mnemonic names.
3145 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3147 * tidied up "make.sh" script
3148 * tidied up system directory structure
3149 * better "clean.sh" behavior
3150 * added doc/FOR-CMUCL-DEVELOPERS
3151 * many many small tweaks to output format, e.g. removing possibly-confusing
3152 trailing #\. character in DESCRIBE-INSTANCE
3153 * (EQUALP #\A 'A) no longer signals an error.
3154 * new hashing code, including EQUALP hashing
3155 * tidied up Lisp initialization and toplevel
3156 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3157 * command line argument processing
3158 * added POSIX-GETENV function to deal with Unix-ish environment variables
3159 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3160 terminating SBCL on EOF
3161 * non-verbose GC by default
3162 * There is no more "sbcl" shell script; the sbcl file is now the C
3163 runtime executable (just like CMU CL).
3164 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3166 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3167 symbol and package information as Lisp native symbol and package objects
3168 instead of strings naming symbols and strings naming packages. This way,
3169 whenever packages are renamed (as in warm init), debug information is
3170 transformed along with everything else.
3171 * tweaked the optimization policy declarations which control the building
3172 of SBCL itself. Now, among other things, the system no longer saves
3173 source location debugging information. (This helps two problems at once
3174 by reducing SBCL size and by keeping SBCL from trying to look for its
3175 sources -- which may not exist -- when reporting errors.)
3176 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3177 debugging and testing purposes
3178 * cleaned up printing, making the printer call PRINT-OBJECT for
3179 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3180 methods, giving nearly-ANSI behavior
3181 * converted almost all special variables to use *FOO* naming convention
3182 * deleted PARSE-TIME functionality, since it can be done portably
3183 * moved some files out of cold init into warm init
3184 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3186 * regularized formatting of source files
3187 * added an install.sh script
3188 * fixed ridiculous memory usage of cross-compiler by making
3189 compiler/alloc.lisp not try to do pooling unless it can hook
3190 itself into the GC of the cross-compilation host. Now the system
3191 builds nicely on my old laptop.
3192 * added :SB-ALLOC in target-features.lisp-expr
3193 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3194 was not implemented)
3195 * re-did condition handling and note reporting in the compiler. Notes
3196 are no longer handled by signalling conditions. Style warnings
3197 and warnings are handled more correctly and reported in such a way
3198 that it's easy to find one or the other in your output (so that you
3199 can e.g. figure out which of many problems caused COMPILE-FILE to
3201 * changed the severity of several compiler warnings from full WARNING
3202 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3203 compiler note reporting so that it doesn't use the condition system
3204 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3205 COMPILE-FILE command)
3206 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3207 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3208 no longer supported, since their implementation was deeply intertwingled
3209 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3210 * removed bogus "support" for compiler macros named (SETF FOO), and
3211 removed the compiler macro for SETF INFO (but only after making a fool
3212 of myself on the cmucl-imp mailing list by posting a bogus patch for
3213 DEFINE-COMPILER-MACRO..)
3214 * Compiled files containing forms which have side effects on the Lisp
3215 reader (such as DEFPACKAGE forms) are now handled more correctly.
3216 (Compiler queuing of top level lambdas has been suppressed by setting
3217 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3218 * deleted various currently-unused source files, e.g. gengc.lisp. They
3219 may be added back at some point e.g. when porting to other architectures,
3220 but until they are it's distracting to distribute them and to try to
3222 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3223 there were so many of them they're just distractions, and UNCROSS is
3224 known to be able to handle the current sources
3225 * moved PROFILE functionality into TRACE, so that it will be clear
3226 how the wrapping and unwrapping of functions when you profile them
3227 interacts with the wrapping and unwrapping of functions when you
3228 trace them. (Actually, the functionality isn't there yet, but at least
3229 the interface specification is there. Hopefully, the functionality will
3230 arrive with some future maintenance release.)
3231 * removed host-oops.lisp
3232 * changed signature of QUIT function to allow UNIX-CODE argument
3233 * fixed READ-SEQUENCE bug
3234 * tweaked verbose GC output so that it looks more like the progress
3235 output that ANSI specifies for functions like LOAD
3236 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3237 * added <http://sbcl.sourceforge.com> to the banner information printed by