1 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2 * fixed cross-compiler leakages that prevented building a 32-bit
3 target with a 64-bit host compiler.
4 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
5 opened with :IF-EXISTS :OVERWRITE.
6 * fixed bug 281: error for an invalid qualifier in a short-form method
7 combination method is not signalled until the faulty method is called.
8 * bug fix: iteration variable type inferrer failed to deal with open
9 intervals. (reported by Alan Shields)
10 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
11 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
12 or more runtime options were provided to the sbcl binary.
13 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
14 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
15 * optimization: there's now a fast-path for fixnum arguments in the
16 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
17 * optimization: the code generated on x86-64 is more compact thanks
18 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
19 * bug fix: floating-point exception handling now partly works on
20 x86-64. (Thanks to James Knight)
21 * improvement to the MIPS backend from Thiemo Seufer: C-style
22 64-bit long long arguments and return values to alien functions
24 * fixed some bugs revealed by Paul Dietz' test suite:
25 ** the type-error signalled from WARN has a filled-in DATUM slot.
26 ** the type-error required when a stream is not associated with
27 a file has the stream as its datum.
28 ** type-errors on single-floats on x86-64 no longer have
29 :INVALID-OBJECT as the datum
30 ** the type-errors signalled for invalid function names now have
31 a correct expected type
32 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
33 for structures defined with a :TYPE. Documentation strings for
34 typed structures are no longer immediately discarded
35 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
36 broadcast streams and synonym streams. FILE-LENGTH now also works
39 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
40 * incompatible change: the --noprogrammer option, deprecated since
41 version 0.7.5, has been removed. Please use the equivalent
42 --disable-debugger option instead.
43 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
45 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
46 inoperational for a while now) has been completely removed.
47 * null lexical environments are now printed as #<NULL-LEXENV>,
48 significantly reducing the amount of clutter in typical backtraces.
49 * documentation on weak pointers, finalization, and after GC hooks
50 has been added to the manual.
51 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
52 as other specialized array types, is much faster. SUBSEQ and
53 COPY-SEQ on such arrays have also been sped up.
54 * optimization: EQL is now more efficient when at least other argument
55 is known to be of type (OR FIXNUM (NOT NUMBER)).
56 * fixed bug: the runtime is now less vulnerable to changes in the
57 size of the SBCL object on OS X, and virtual memory is reserved for
58 all spaces that need to be at a fixed address.
59 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
60 * fixed bug: finalizers and after GC hooks that cause consing are now
62 * fixed bug: compiler error messages and summaries are now printed to
63 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
64 * fixed inference of the upper bound of an iteration variable.
65 (reported by Rajat Datta).
66 * fixed bug 211e: calling local functions with duplicated constant
67 keyword argument no longer causes a bogus style warning about an
69 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
70 ftype declaration to be disregarded. (reported by Dave Roberts)
71 * fixed bug 373: caused by erronous compilation of references to alien
72 variables in the runtime on ppc/darwin.
73 * fixed bug 376: CONJUGATE type deriver.
74 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
75 (reported by Baughn on #lisp)
76 * a cross-compiler bug on non-x86 platforms has been identified and
77 fixed. (thanks to Bruno Haible)
78 * improvements to the MIPS runtime code for increased stability.
79 (thanks to Thiemo Seufer)
80 * increased the maximimum compact environment size to allow
81 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
82 * partial workaround for bug 135: don't name the function we're
83 calling for hairy cases of EVAL, so as not to accumulate one environment
84 entry per call to EVAL. (reported by Kevin Reid)
85 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
86 * optimization: type testing for non-vector arrays should be faster.
87 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
89 * bug fix: setting 31st element of a bit vector to zero did not work
91 * bug fix: redefining a class definition which failed due to a
92 previous accessor / function clash now works (but see BUGS entry
93 #380 for more problems in this area). (thanks to Zach Beane)
94 * the long form of DEFINE-METHOD-COMBINATION disables method group
95 checking when given a single method group with pattern *. (thanks
97 * on x86 compiler supports stack allocation of results of simple
98 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
99 * support for building and running on Mac OS X 10.4 "Tiger" has been added
100 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
101 * fixed some bugs related to Unicode integration:
102 ** the restarts for recovering from input and output encoding
103 errors only appear when there is in fact such an error to
105 * fixed some bugs revealed by Paul Dietz' test suite:
106 ** MISC.549 and similar: late transformation of unsafe type
107 assertions into derived types caused unexpected code
109 ** SCALE-FLOAT type deriver is less wrong.
110 ** type derivers for EXP, LOG and similar functions compute result
111 types for complex arguments better.
112 ** (MISC.563) CONJUGATE type deriver works for very restricted
114 ** out-of-line type testers for character strings are available.
115 ** EQUAL compiler transform understands specialness of objects
117 ** accessing double-floats stored on the stack now works on x86-64.
118 ** debugger internals could sometimes create invalid lispobjs,
119 resulting in GC crashes.
120 ** MISC.548: type check weakening can convert required type into
122 ** initialization forms for bindings are not in scope of free special
125 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
126 * incompatible change: thread support for non-NPTL systems has
127 been removed - locking is buggy and unreliable. A threaded
128 SBCL build will now warn at startup and refuse to create
129 new threads, unless futex support is detected in the kernel
130 * incompatible change: the top level REPL now has only an ABORT
131 restart associated with it, not TOPLEVEL and ABORT as it used to.
132 TOP and TOPLEVEL are now available as debugger commands for
133 returning to the top level.
134 * incompatible change: forms evaluated in the REPL now use the
135 global optimization policy.
136 * incompatible change: user- and system-initialization files are
137 no longer processed with LOAD, but by READ and EVAL; hence the
138 global optimization policy, startup package, readtable, etc,
140 * The .fasl file format number has been incremented because of
141 various incompatible changes.
142 * internal entry point details and argument counts no longer appear
143 in backtraces unless explicitly requested by setting
144 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
145 * built-in and standard functions no longer have names like "top
146 level local call to FOO".
147 * fixed bug 32: functions defined in non-null lexical environments
148 now have more legible printed representation
149 * fixed bug 33: functions defined in non-null lexical environemnts
150 are now more amenable to inspection by INSPECT.
151 * workaround for bug 354: XEPs no longer appear in backtraces unless
152 explicitly requested.
153 * fixed bug: receiving the signal which results from attempting to
154 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
155 SIGBUS on other *BSDs) on architectures where the C stack is also
156 the Lisp stack (x86 and x86-64) from foreign code no longer leads
157 to debugger confusion or wild instability. (reported by Cheuksan
159 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
160 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
161 (reported by Lutz Euler)
162 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
163 compiled code. (reported by Paul Dietz)
164 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
165 specialized parameter to a method to something that is not TYPEP
166 the specializer is now possible.
167 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
168 face of package deletion.
169 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
170 pathname. (reported by David Tolpin on c.l.l)
171 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
172 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
173 STRING-STREAM and FILE-STREAM, respectively.
174 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
175 than previously; in addition, its test suite is now run on build.
176 (thanks to Luke Gorrie)
177 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
178 * optimization: added a immediate representation for single-floats
180 * fixed some bugs revealed by Paul Dietz' test suite:
181 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
182 ** attempting to create a package with a colliding nickname causes
183 correctable errors to be signalled.
184 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
185 ** several x86-64 backend bugs related to sign-extension of immediate
188 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
189 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
190 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
192 * fixed loading of multiply forward-referenced layouts.
193 (thanks to Cheuksan Wang)
194 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
195 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
196 (see COMPILE-FILE documentation for details of :PRINT option)
197 * fixed bugs 19 and 317: fixed-format floating point printing is
198 more accurate. This also fixes a bug reported by Adam Warner
199 related to the ~@F format directive.
200 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
202 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
203 dispatch table if the from-char is a dispatch macro character.
204 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
205 by the wordsize no longer ignore the last word. (reported by Lutz
207 * fixed bug in type checking of dynamic-extent variables. (reported
209 * optimization: sequence traversal functions use their freedom to
210 coerce function designators to functions.
211 * optimization: code with many calls to CLOS methods specialized on
212 CLOS classes has had redundant type checks removed.
213 * optimization: added declarations to speed up operations that access
214 the internal character database (for example STRING-UPCASE)
215 * optimization: comparison operations between floats and sufficiently small
216 fixnums no longer create extra rationals
217 * fixed some bugs related to Unicode integration:
218 ** portions of multibyte characters at the end of buffers for
219 character-based file input are correctly transferred to the
220 start of the buffer at the next read.
221 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
222 passing it through to OPEN.
223 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
224 argument given to internal calls to OPEN.
225 ** the built-in comment readers (introduced by character sequences
226 ";" and "#|") are more forgiving to encoding errors; they will
227 STYLE-WARN and then attempt to resync the stream at a character
228 boundary. (thanks to Teemu Kalvas)
229 * fixed some bugs in the x86-64 port:
230 ** Negative short int return values from c-calls are sign-extended
232 ** The stack is aligned to 16-bytes for c-calls, as required by
233 the ABI. (thanks to Cheuksan Wang)
234 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
235 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
236 for enums. (thanks to Vincent Arkesteijn)
237 ** Multiple small optimizations and bugfixes for floating point
239 * fixed some bugs revealed by Paul Dietz' test suite:
240 ** Space, Tab, Linefeed, Return and Page have the invalid
241 secondary constituent character trait.
242 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
244 ** WITH-INPUT-FROM-STRING should only update the index place on
246 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
248 ** Bit-array operations (BIT-AND and similar) worked incorrectly
249 with one-dimensional arrays with fill pointers.
250 ** TYPE-OF failed on a complex with an integer realpart and a
252 ** compiler failure during type inference for the code of form
253 (IF (EQL X (THE ...)) ...) (MISC.535).
255 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
256 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
257 platform. The port supports 61-bit fixnums, large memory spaces
258 and reloading shared object files.
259 * enhancement: saving cores with foreign code loaded is now
260 supported on ppc/Darwin in addition to the previously supported
262 * enhancement: the statistical profiler now walks deeper into the
263 call stack for more meaningful call-graphs and accrued time
264 reports (x86/x86-64 only). It also now reports time spent in
266 * enhancement: it is now possible to trace most individual methods
267 of a generic function in addition to tracing the generic function
269 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
271 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
272 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
273 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
274 produces an error. (thanks to Vincent Arkesteijn)
275 * bug fix: NAMESTRING on pathnames with :WILD components in their
276 directories works correctly. (thanks to Artem V. Andreev)
277 * fixed bug 125: compiler preserves identity of closures. (reported
279 * bug fixed: functions with &REST arguments sometimes failed with
280 "Undefined function" when compiled with (DEBUG 3). (reported by
282 * bug fix: overflow during compiling of setting element of a bit
283 vector with constant index and value. (reported by Timmy Douglas)
284 * build fix: fixed the dependence on *LOAD-PATHNAME* and
285 *COMPILE-FILE-PATHNAME* being absolute pathnames.
286 * on x86 compiler partially supports stack allocation of dynamic-extent
288 * GO and RETURN-FROM do not check the extent of their exit points
289 when compiled with SAFETY 0.
290 * fixed some bugs related to Unicode integration:
291 ** encoding and decoding errors are now much more robustly
292 handled; it should now be possible to recover even from invalid
293 input or output to the terminal. (thanks to Teemu Kalvas)
294 ** provided a first cut at implementing STRING-TO-OCTETS and
295 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
296 ** altered the SB-MD5 contributed module slightly, changing the
297 interface just enough for it to be supportable for builds where
298 lisp characters are not eight bits.
299 * fixed some bugs revealed by Paul Dietz' test suite:
300 ** the FORMATTER-generated functions for ~V[ conditionals require
301 the correct number of arguments.
302 ** READ-FROM-STRING returns the mandated second value when applied
303 to displaced strings.
304 ** the #\Rubout and #\Backspace characters are treated as invalid
305 constituent characters by the tokenizer.
307 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
308 * new feature: reloading changed shared object files with
309 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
310 * new feature: references to foreign variables and functions
311 can now be compiled and loaded before the corresponding shared
312 object file is loaded, as long as the foreign definitions are
313 available at runtime.
314 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
315 supported. (thanks to Dan Debertin)
316 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
317 just keywords. (thanks to Vincent Arkesteijn)
318 * fixed bug #331: structure-class instances corresponding to
319 DEFSTRUCT forms are now created eagerly.
320 * fixed bug #345: backtraces from calls to undefined functions work
321 on x86 as well. Related bug #61 is now also partially fixed on x86
322 (backtraces from throws to unknown catch tags.)
323 * bug fix: lambda-list parsing is now stricter vrt. order and number
324 of lambda-list keywords.
325 * bug fix: as specified by AMOP, an error is signalled if a
326 class-option appears multiple times in a DEFCLASS form. (reported
328 * bug fix: RANDOM can be compiled when the compiler derives the type
329 of its numeric argument as a disjoint set of small integers.
330 (reported by Paul Dietz)
331 * bug fix: starting a core saved with shared objects loaded when
332 those objects are not available no longer causes threaded SBCL to
333 hang. (reported by Sean Ross)
334 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
336 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
337 argument, which must always be NIL. (reported by Kalle Niemitalo)
338 * bug fix: printing 1.0d+23 no longer results in an error.
339 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
340 * bug fix: structure slot setters preserve evaluation order. (thanks
342 * fixed some bugs related to Unicode integration:
343 ** RUN-PROGRAM can allow its child to take input from a Lisp
344 stream. (reported by Stefan Scholl)
345 ** ASDF-INSTALL successfully downloads .tar.gz files in all
346 locales. (reported by Ken Causey)
347 * fixed some bugs revealed by Paul Dietz' test suite:
348 ** INCF, DECF and REMF evaluate their place form as specified in
350 ** FORMATTER expands ~{ ~} iteration directives with V or #
351 parameters correctly.
352 ** FORMATTER deals with the ~@[ ~] conditional directive where the
353 consequent uses no arguments correctly.
354 ** the system has a partial understanding of the (COMPLEX RATIO)
356 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
359 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
360 * new feature: a build-time option (controlled by the :SB-UNICODE
361 keyword feature, enabled by default) for building the system with
362 support for the entire 21-bit character space defined by the
364 * new feature: the system now has rudimentary external-format
365 support; the primary user-visible change at this time is that
366 characters with the high bit set (such as lower-case-e-acute) will
367 print correctly to a terminal in a UTF-8 environment.
368 * minor incompatible change: BASE-CHAR no longer names a class;
369 however, CHARACTER continues to do so, as required by ANSI.
370 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
371 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
372 has been moved to the SB-EXT package (temporarily re-exported from
374 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
375 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
377 * fixed bug #351: better error-handlind and reporting for malformed
379 * fixed bug #350: bignum-printing is now more memory-efficient,
380 allowing printing of very large bignums, eg. (expt 2 10000000).
381 (reported by Bruno Haible)
382 * fixed bug #302: better primitive-type selection for intersection
384 * fixed bug #308: non-graphic characters now all have names, as
385 required. (reported by Bruno Haible)
386 * bug fix: redefining a class with different superclasses now correctly
387 removes it from the direct-subclasses of its previous superclasses.
388 (reported by David Morse)
389 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
390 new value now works. (reported by Bruno Haible)
391 * bug fix: correct canonicalization of multiple non-standard slot
392 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
393 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
394 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
395 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
396 now exists, an signals an error.
397 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
398 during macroexpansion for non-iteration variables as well. (reported
399 by Bruno Haible for CMUCL)
400 * bug fix: Cyclic structures and unprintable objects in compiler
401 messages no longer cause errors. (reported by Bruno Haible)
402 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
403 and READ-FROM-STRING all now return a primary value of NIL if
404 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
405 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
406 (reported by Bruno Haible for CMUCL)
407 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
408 do not designate a condition. (reported by Bruno Haible for
410 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
411 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
412 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
413 permitting resolution in favour of any of the conflicting symbols.
414 (reported by Bruno Haible for CMUCL)
415 * bug fix: EQUAL compiler optimizations is less aggressive on
416 strings which can potentially compare true despite having distinct
417 specialized array element types.
418 * bug fix: unit enumerations can be defined without dividing by
419 zero. (thanks to Vincent Arkesteijn)
420 * FORMAT compile-time argument count checking has been enhanced.
421 (report from Bruno Haible for CMUCL)
422 * a partial workaround for the bug 262: the compiler does not try to
423 inline-expand a local function doing RETURN-FROM from a deleted
424 BLOCK. (thanks to Peter Denno for the bug report and to David
425 Wragg for the simple test case)
426 * fixed some bugs revealed by Paul Dietz' test suite:
427 ** RENAME-PACKAGE allows all package designators as new package
429 ** constraint propagation and lambda variable substitution are
430 more cautious in dealing with partially deleted code.
431 ** compiler sometimes lost reoptimization passes.
432 ** CERROR, given a condition as condition designator, passes its
433 remaining arguments to the continue format control without
435 ** Case-altering FORMAT directives work correctly on non-ASCII
437 ** The REAL type specifier handles bounds outside the floating
438 point ranges without signalling FLOATING-POINT-OVERFLOW.
439 ** Functions with IR1-transformations can create intercomponent
440 references to global functions.
441 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
443 ** FORMAT ~R treats a nil value for its first parameter correctly.
445 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
446 * enhancement: saving cores with foreign code loaded is now
447 supported on x86/NetBSD and sparc/Linux in addition to the previously
449 * bug fix: on some platforms repeated installations caused multiple
450 copies of HTML documentation to be installed -- should not happen
451 any more. (reported by Stefan Scholl)
452 * bug fix: parsing self-recursive alien record types multiple times
453 no longer causes infinite recursion. (reported by Thomas F. Burdick,
454 original patch by Helmut Eller for CMUCL)
455 * bug fix: stack-exhaustion detection works now on NetBSD as well.
456 (thanks to Richard Kreuter)
457 * bug fix: defining classes whose accessors are methods on existing
458 generic functions in other (locked) packages no longer signals
459 bogus package lock violations. (reported by François-René Rideau)
460 * bug fix: special variables as DEFMETHOD parameters no longer have
461 associated bogus type declarations. (reported by David Wragg and
463 * bug fix: read-write consistency on streams of element-type
464 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
465 * bug fix: redefiniton of the only method of a generic function with
466 no DEFGENERIC no longer emits a full WARNING. In addition,
467 redefinition of generic functions with no DEFGENERIC to an
468 incompatible lambda list now signals an error. (thanks to Zach
470 * bug fix: DEFGENERIC now works even when there's a function of the
471 same name in an enclosing lexical environment. (thanks to Zach
473 * fixed compiler failure, caused by instrumenting code during
474 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
475 * optimization: added loop analysis and improved register allocation
476 * fixed some bugs revealed by Paul Dietz' test suite:
477 ** POSITION on displaced vectors with non-zero displacement
478 returns the right answer.
479 ** (SIMPLE-STRING) is a valid type specifier for sequence
481 ** *PRINT-LEVEL* handling for slotless structures is pedantically
483 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
484 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
486 ** The justification version of the FORMAT ~< directive treats
487 non-zero minpad parameter correctly.
489 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
490 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
491 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
492 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
493 the supported interface.
494 * new feature: Single-stepping of code compiled with DEBUG 2 or
495 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
496 * new feature: saving cores with foreign code loaded is now
497 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
498 Timothy Moore's work for CMUCL)
499 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
500 parameters to * instead of NIL if no initform is supplied.
501 (reported by Johan Bockgaard)
502 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
503 correctly when FUNCALL appears as the car of the form. Note:
504 despite this FUNCALL forms are not currently subject to
505 compiler-macro expansion. (port of Raymond Toy's fix for the
506 same from CMUCL, reported by Johan Bockgaard)
507 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
508 (thanks for Teemu Kalvas)
509 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
510 case instead of terminating the iteration (thanks for Julian
511 Squires, Sean Champ and Raymond Toy)
512 * bug fix: incorrect expansion of defgeneric that caused a style
513 warning. (thanks for Zach Beane)
514 * bug fix: slot accessor effective method computation works properly
515 for classes with multiple non-standard applicable methods on
516 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
517 * on x86 compiler supports stack allocation of results of LIST and
518 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
519 CMUCL implementation by Gerd Moellmann)
520 * fixed some bugs revealed by Paul Dietz' test suite:
521 ** FORMAT strings with both the ~<~:;~> form of the justification
522 directive and pretty-printing directives cause an error.
524 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
525 * incompatible change: the internal functions
526 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
527 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
528 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
529 instead of the old functions.
530 * new feature: on platforms where "dladdr" is available foreign
531 function names now appear in backtraces. (based on Helmut Eller's
533 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
534 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
536 * documentation: ASDF manual is now distributed with SBCL.
537 * bug fix: SBCL can now load its contributed modules with REQUIRE
538 even if the system-provided entries have been removed from
539 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
541 * bug fix: backtraces involving undefined functions or assembly
542 routines are more informative. (thanks to Brian Downing)
543 * bug fix: mutually referent alien structures now work correctly.
544 (reported by Rick Taube)
545 * bug fix: structures defined by WITH-ALIEN can be referred to
546 within other definitions in the same WITH-ALIEN.
547 * bug fix: division operators (MOD, TRUNCATE and the like) with
548 constant zero divisors and integer dividends no longer generate
550 * bug fix: provide default methods for INPUT-STREAM-P and
551 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
552 * bug fix: improve the detection and resolution of MOP metacycles
553 (where effective method computation depends on the generic function
554 having its effective method computed). (reported by Bruno Haible)
555 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
556 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
557 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
558 * bug fix: pathnames with relative directory components can be
559 represented relative to default pathnames.
560 * optimization: in taking the GCD of bignums, reduce the two bignums
561 to approximately the same size (using Euclid's algorithm) before
562 applying the more sophisticated binary GCD. (thanks to Juho
564 * optimization: COUNT on bitvectors now operates word-at-a-time.
565 * optimization: ASH with a positive, but not necessarily constant,
566 (leftwards) shift, when in a modular context, is compiled to a
568 * fixed some bugs revealed by Paul Dietz' test suite:
569 ** FORMAT variable parameters ("~V<char>") are defaulted properly
570 if the corresponding argument is NIL.
571 ** FORMAT directives accepting a minpad parameter treat negative
573 ** literal commas as character parameters to FORMAT directives are
575 ** literal spaces directly after ~<Newline> directives within a
576 format-logical-block (~:< ~@:>) do not induce :FILL-style
577 conditional newlines.
578 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
579 as specified: it no longer includes conditional newlines).
580 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
582 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
583 :PER-LINE-PREFIX argument does not evaluate to a string.
585 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
586 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
587 the manual for details; for now, package locks can be disabled by
588 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
589 if no major problems are found then it is likely that they will be
590 compiled in unconditionally.
591 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
592 unsupported operators on all platforms. To load a shared library
593 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
594 object file, link it into a shared library outside of SBCL and
595 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
597 * bug workaround: ROOM T and the GC were not working together
598 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
599 reporting facility aren't preserved. That reporting has been
600 disabled, so now until and unless someone figures out how to make it
601 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
602 * minor incompatible change: as threatened around sbcl-0.8.0, the
603 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
604 The symbols are also exported from SB-PCL for backwards
605 compatibility, but more so than before SB-PCL should be treated as
606 an implementation-internal package.
607 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
609 * the SB-SPROF contrib now works on (most) non-x86 architectures.
610 It is known as of this release not to work on the Alpha, however.
611 * fixed bug #167: errors signalled due to illegal syntax in method
612 bodies are now more legible.
613 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
614 designators and can hence be used with TYPEP.
615 * fixed bug #333: CHECK-TYPE now ensures that the type error
616 signalled, if any, has the right object to be accessed by
617 TYPE-ERROR-DATUM. (reported by Tony Martinez)
618 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
619 5.1.2.3 for multiple-value place subforms. (reported by Kalle
621 * fixed bug #334: programmatic addition of slots using specialized
622 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
623 and :ALLOCATION :CLASS slots.
624 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
625 float exponents. (rereported by Peter Seibel)
626 * fixed a bug: #\Space (and other whitespace characters) are no
627 longer considered to be macro characters in standard syntax by
629 * fixed bug: initialization of condition class metaobjects no longer
630 causes an instance of the condition to be created. (reported by
632 * fixed bug: it is now possible to have slots such that
633 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
634 description is neither :INSTANCE nor :CLASS.
635 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
636 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
637 SLOT-VALUE-USING-CLASS), and no longer causes errors with
638 non-standard SLOT-DEFINITION-ALLOCATION values.
639 * fixed bugs: various race conditions handling exiting threads.
640 CL-PPCRE's thread torture test now passes.
641 * fixed arguably-a-bug: GC time proportional to number of threads in
642 system even when most of them are idle
643 * optimization: improved performance of BIT and SBIT on bit-vectors.
644 * .fasl file incompatibility: The fasl file version number has
645 been incremented because of changes associated with package locks.
647 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
648 * minor incompatible change: the system no longer provides
649 optimization hints (conditions which are TYPEP
650 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
651 mode; these hints are still emitted when higher SPEED optimization
653 * new contrib module: a sampling profiler (profiling by statistical
654 sampling, rather than by instrumenting functions) is available as
655 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
656 * the behaviour of the standard function ED is now customizeable by
657 third parties through a hook variable: see ED's documentation
658 string for information on the protocol.
659 * the compiler no longer emits efficiency notes for (FUNCALL X)
660 when the type of X is uncertain under default optimization
662 * fixed bug 276: mutating a binding of a specialized parameter to a
663 method to something that is not TYPEP the specializer is now
665 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
666 exhibit the required behaviour on the x86 platform. (thanks to
667 Peter van Eynde, Eric Marsden and Bruno Haible)
668 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
669 even for difficult arguments. (reported by Peter Graves)
670 * fixed bug 141a: the backquote printer now descends quoted
672 * fixed another bug in backquote printing: no more destructive
673 modification of the form's list structure. (reported by Brian
675 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
676 move between its address being taken and the call to
677 interrupt_thread, fixing a crashing race condition.
678 * the SB-POSIX contrib implementation has been adjusted so that it
679 no longer exhibits ridiculously poor performance when constructing
680 instances corresponding to C structs.
682 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
683 * minor incompatible change: the sb-grovel contrib now treats C
684 structures as alien (in the sense of SB-ALIEN) objects rather than
685 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
686 has implications for memory management of client code
687 (sb-grovel-returned objects must now be manually managed) and for
688 type safety (alien objects now have full types).
689 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
690 used to control emission of compiler diagnostics, rather than the
691 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
692 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
693 quality should be considered deprecated.
694 * install.sh now installs the user manual as well
695 * (not quite a new documentable feature, but worth considering in
696 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
697 beginnings of a semantically meaningful condition hierarchy is
698 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
700 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
701 designator as the defaults argument.
702 * fixed bug: Displaced arrays whose displaced-to array has become
703 too small now cause ARRAY-DIMENSION to signal an error, providing
704 for safer bounds-checking. (reported by Bruno Haible)
705 * fixed bug: DEFCLASS slot definitions with identical :READER and
706 :WRITER names now signal a reasonable error. (reported by Thomas
708 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
710 * fixed bug: Invalid slot specification errors now print correctly.
711 (thanks to Zach Beane)
712 * fixed bug 320: Shared to local slot value transfers in class
713 redefinitions now happen corrently from superclasses as
714 well. (reported by Bruno Haible)
715 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
717 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
718 improper lists correctly. (reported by Bruno Haible)
719 * fixed bug 313: source-transform for <fun-name> was erroneously
720 applied to a call of a value of a variable with name <fun-name>.
721 (reported by Antonio Menezes Leitao)
722 * fixed bug 307: The obsolete instance protocol ensures that
723 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
724 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
725 remove dead unknown-values globs from the middle of the stack.
726 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
727 easier to resume long computations after using *BREAK-ON-SIGNALS*
728 to diagnose and fix failures. (thanks to Nikodemus Siivola)
729 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
730 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
731 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
732 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
733 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
735 * fixed a (fairly theoretical) bug in string printing: if
736 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
737 does not have array-element-type equal to the most general string
739 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
740 function name. (thanks to Nikodemus Siivola)
741 * fixed bug: compile-time format string checker failed on
742 non-closed ~{. (reported by Thomas F Burdick)
743 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
744 don't warn on legal loop code involving more than one aggregate
745 boolean. (thanks to Nikodemus Siivola)
746 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
747 in keyword position to MAKE-PACKAGE was not being reported as
749 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
750 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
752 * fixed bugs 280 and 312: the checking for multiple definitions in a
753 file is less likely to become confused by uses of inline
755 * fixed bug: the #S reader macro performs the keyword coercion
756 specified for slot names. (reported by Kalle Niemitalo)
757 * fixed bug: lambda lists may contain symbols whose names start with
758 & but are not lambda-list-keywords; their occurrence triggers a
760 * fixed bug 321: define-method-combination argument lambda lists do
761 not cause invalid code to be generated when &optional or &aux
762 variables have default values. (reported by Bruno Haible)
763 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
764 have CLOS classes; this appears also to have fixed reported
765 instability in stack exhaustion detection.
766 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
767 MUFFLE-WARNING when no associated restart is present is now
769 * optimization: rearranged the expansion of various defining macros
770 so that each expands into only one top-level form in a
771 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
773 * optimization: used a previously little-used slot in symbols to
774 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
776 * fixed some bugs revealed by Paul Dietz' test suite:
777 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
779 ** Fixed an optimization bug related to inheritance of initforms
780 from local to shared slots.
781 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
782 ** CAST optimizer forgot to flush argument derived type.
783 ** print/read consistency on floats is now orders of magnitude
784 more likely. (thanks also to Bruno Haible for a similar report
786 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
787 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
788 for objects of type REAL. Make it so.
789 ** SXHASH is specified (infelicitously) to respect similarity,
790 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
791 it so. (thanks to Markus Ziegler)
792 ** on the Alpha, the compiler succeeds in compiling functions
793 returning a known number of arguments greater than 63.
794 ** fixed handling of invalid NIL arguments in keyword position
796 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
798 ** provided a readably-printable representation for RANDOM-STATE
800 ** ensured that pathnames, where they have a namestring, always
801 print using #P"..." syntax.
803 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
804 * Support for the forthcoming 2.0 version of the NetBSD kernel
805 running on x86 hardware has been added. (thanks to Perry
806 E. Metzger most immediately, and others for their past work)
807 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
808 current release version; SBCL's OpenBSD support had been broken
809 since about the time of OpenBSD's switch to ELF binary format.)
810 * [placeholder for DX summary]
811 ** user code with &REST lists declared dynamic-extent, under high
812 speed or space and low safety and debug optimization policy.
813 * The manual has been converted to Texinfo format and the debugger
814 chapter from the cmucl manual has been added.
815 * A facility has been added to extract documentation strings from
816 sbcl and store them as Texinfo-formatted snippets for inclusion in
817 the manual (via Texinfo's @include directive)
818 * bug fix: compiler emitted division in optimized DEREF. (thanks for
819 the test case to Dave Roberts)
820 * bug fix: multidimensional simple arrays loaded from FASLs had fill
821 pointers. (reported by Sean Ross)
822 * bug fix: PROFILE output is printed nicely even for large numerical
823 values. (thanks to Zach Beane)
824 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
825 greater than 32 handle EOF correctly.
826 * bug fix: on X86 an immediate argument of the IMUL instruction is
827 correctly printed in disassembly. (thanks to Lutz Euler)
828 * bug fix: class slots in redefined classes preserve their old
829 values. (thanks to Bruno Haible and Nikodemus Siivola)
830 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
831 succeeds. (reported by Marco Baringer)
832 * fixed some bugs revealed by Paul Dietz' test suite:
833 ** READ-SEQUENCE now works on ECHO-STREAMs.
834 ** RATIONALIZE works more according to its specification. (thanks
837 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
838 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
839 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
840 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
841 should matter to you only if you rebind the printer control
842 variables and then find you want different bindings in the
843 debugger than in the ordinary execution of your program.)
844 * The runtime build system has been tweaked to support building
845 (on SPARC/SunOS) using a C compiler which invokes Sun's own
846 assembler and linker. (thanks to Nikodemus Siivola)
847 * Unbound, undefined, undeclared variables now trigger full
848 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
849 more often programmer error than deliberate exploitation of undefined
851 * optimization: the hash algorithm for strings has changed to one
852 that is less vulnerable to spurious collisions. (thanks to Juho
854 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
855 needless bounds checking. (thanks to Juho Snellman)
856 * optimization: implemented multiplication as a modular
857 (UNSIGNED-BYTE 32) operation on the PPC backend.
858 * fixed some bugs revealed by Paul Dietz' test suite:
859 ** ADJUST-ARRAY now copies the datum in a zero rank array if
861 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
862 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
863 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
865 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
867 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
869 ** ECHO-STREAMs no longer attempt to echo the end of file value to
870 their output stream on EOF from read.
871 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
872 have been read to end-of-file.
873 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
875 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
876 description of determination of which consecutive characters
878 ** Printing the "Space" character with escaping on now yields "#\\ ",
879 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
880 ** Reading floating-point numbers with *READ-BASE* set to a number
881 less than 10 works correctly.
882 ** Reading floating-point numbers with *READ-BASE* set to a number
883 more than 10 works correctly.
884 ** Printing with *PRINT-READABLY* targets the standard readtable, not
885 the readtable currently in effect.
887 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
888 * minor incompatible change: parsing of namestrings on a physical
889 (Unix) host has changed; numbers after the final #\. in a
890 namestring are no longer interpreted as a version field. This is
891 intented to be largely invisible to the user, except that the
892 meaning of the namestring "*.*.*" has changed: it now refers to a
893 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
894 should usually be replaced by
895 (make-pathname :name :wild :type :wild :version :wild)
896 with the added benefit that this is more likely to be portable.
897 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
898 option to OPEN now signals an error if the file being opened
899 exists; this may have an impact on existing code.
900 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
901 signals received, on the PowerPC platforms (both Linux and
902 Darwin). (thanks to Pierre Mai for pointing out the location of
904 * several fixes on OS X: The system now builds and runs cleanly on
905 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
906 (thanks to Brian Mastenbrook)
907 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
908 second-resolution: integer multiples of 1/3600 between -24 and 24.
909 (thanks to Vincent Arkesteijn)
910 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
911 type when called with 1 argument; PEEK-CHAR checked type of
912 PEEK-TYPE only after having read first character from a
913 stream. (reported by Peter Graves)
914 * bug fix: the garbage collector now has much better locality
915 behaviour, and in particular no longer treats all memory as being
916 exhausted when many small objects point to each other in a deeply
918 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
919 recognized as being TYPEP their class.
920 * bug fix: the PUSHNEW documentation string has been corrected.
921 (thanks to Vincent Arkesteijn)
922 * bug fix: defaulting of the value for the last of an atypically
923 large number of multiple values being bound was not being
924 performed correctly on the Alpha or PPC platforms
925 * optimization: implemented multiplication as a modular
926 (UNSIGNED-BYTE 32) operation on the x86 backend.
927 * optimization: SEARCH on simple-base-strings can now be open-coded.
928 (see also contrib/compiler-extras.lisp for inspiration for
929 teaching the compiler about the Boyer-Moore algorithm).
930 * value, returned by MAX (and MIN) called with several EQUALP, but
931 not EQL, arguments now does not depend on compiler settings.
932 * fixed some bugs revealed by Paul Dietz' test suite:
933 ** in stack analysis liveness information is propagated from
934 non-local entry points.
935 ** pathwise CAST removing failed when the CAST node did not start
937 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
938 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
940 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
941 host is already defined.
942 ** RENAME-FILE works on streams instead of signalling an internal
944 ** PEEK-CHAR uses the current readtable when determining whether
945 or not a character is whitespace.
946 ** MERGE-PATHNAMES handles the case when the pathname does not
947 specify a name while the default-pathname specifies a version.
948 ** Pathnames now stand a better chance of respecting print/read
950 ** Attempting to use standardized file system operators with a
951 pathname with invalid :DIRECTORY components signals a
953 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
955 ** DIRECTORY on logical pathnames is more correct.
956 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
957 signal a TYPE-ERROR if their argument is not a stream.
958 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
959 designator argument does not designate a stream.
960 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
961 examining the synonym.
962 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
964 ** OPEN and WITH-OPEN-STREAM allow opening streams with
965 element-type larger than ([UN]SIGNED-BYTE 32).
967 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
968 * When built with the :SB-FUTEX feature, threaded builds now take
969 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
970 for faster/more reliable mutex and condition variable support.
971 * Incompatible change (but one you probably shouldn't have been using
972 anyway): the interface and code for arbitrating between multiple
973 threads in the same user session has been redesigned.
974 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
975 error when called without an explicit environment argument.
976 (thanks to Nikodemus Siivola)
977 * bug fix: buffered :DIRECTION :IO streams are less likely to become
978 confused about their position. (thanks to Adam Warner and Gerd
980 * bug fix: Pretty printing backquoted forms with unquotations in the
981 argument list position of various code constructs such as LAMBDA
982 now works correctly. (reported by Paul Dietz)
983 * bug fix: Pretty printing unquotations no longer loses all
984 stream position information.
985 * optimization: performance of string output streams is now less
986 poor for multiple small sequence writes.
987 * optimization: performance of CSUBTYPEP in the presence of complex
988 expressions involving CONS and NOT many times has been improved.
989 (reported by Paul Dietz)
990 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
992 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
993 streams for the REPL, for the debugger, and so on. (thanks to
995 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
996 * fixed some bugs revealed by Paul Dietz' test suite:
997 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
998 is ignored if it is too small, rather than propagating through
999 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1000 ** extremely complex negations of CONS types were not being
1001 sufficiently canonicalized, leading to inconsistencies in
1003 ** VALUES tranformer lost derived type.
1005 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1006 * fixed a bootstrapping bug: the build process no longer assumes
1007 that the various BOOLE-related constants have the same value in
1008 host and target lisps. (noted by Paul Dietz' test suite on an
1009 SBCL binary built from CLISP)
1010 * The system can now be dynamically linked on the MIPS platform,
1011 which enables dynamic loading of foreign code from Lisp. (thanks
1012 to Ralf Baechle for discussions on the MIPS ABI)
1013 * The system now records debugging information for its own source
1014 files in a filesystem-position-independent manner, relative to
1015 the "SYS" logical host.
1016 * fixed a compiler bug: MV-LET convertion did not check references
1017 to the "max args" entry point. (reported by Brian Downing)
1018 * tweaked disassembly notes to be less confident about proclaiming
1019 some instruction as an LRA. (thanks to Brian Downing)
1020 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1021 now each have their own history, command character, and other
1022 characteristics. (thanks to David Lichteblau)
1023 * fixed some bugs revealed by Paul Dietz' test suite:
1024 ** compiler failure in compiling LOGAND expressions including a
1026 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1027 shift greater than 32.
1028 ** FLUSH-DEST did not mark blocks for type check regeneration.
1029 ** HANDLER-CASE failed to accept declarations in handler clauses
1030 in some circumstances.
1032 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1033 * New code in contrib/sb-introspect (still probably not entirely
1034 stable yet) provides some support for smart Lisp development
1035 environments like SLIME.
1036 * The conditions signalled for errors occurring when loading .fasl
1037 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1038 in a way which should help ASDF recover gracefully.
1039 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1040 stuff has been cleaned up. If you code contrib/ stuff, this might
1041 affect you, and you can look at contrib/README, contrib/STANDARDS,
1042 and/or the 0.8.4.27 diff to check.
1043 * In full calls the compiler now does not generate checks for declared
1044 argument types for all arguments.
1045 * various threading fixes
1046 ** and some experimental patches which didn't make it into
1047 the main tree for this release, but which are shipped in
1048 contrib/experimental-thread.patch as a possible fix for some
1049 failures (deadlock, spinning...) in GC-intensive multithreaded
1051 * fixed PPC build problem (source code incompatibility of different
1052 library versions): added offsetof() hackery which attempts to divine
1053 where glibc maintainers put uc_mcontext today
1054 * fixed bug 282: compiler does not trust type assertions while passing
1055 arguments to a full call.
1056 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1057 &OPTIONAL VALUES type parameter.
1058 * fix bug 214: algorithm for noting rejected templates is now more
1059 similar to that of template seletion. (also reported by rydis on
1061 * fixed bug 141b: printing backquoted information readably and prettily
1062 inserts a space where necessary.
1063 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1064 (CHARACTER 10) are now reported as errors, rather than propagated
1065 as unknown types. (reported by piso on #lisp)
1066 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1067 with values NIL and :ERROR. (thanks to Milan Zamazal)
1068 * fixed bug 191c: CLOS now does proper keyword argument checking as
1069 described in CLHS 7.6.5 and 7.6.5.1.
1070 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1071 longer raise an error; further, using a list as a for-as-arithmetic
1072 counter now raises a meaningful error.
1073 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1074 now understood by sequence creation functions such as MAKE-SEQUENCE
1076 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1077 a STRING-INPUT-STREAM.
1078 * compiler enhancement: SIGNUM is now better able to derive the type
1080 * type declarations inside WITH-SLOTS are checked. (reported by
1082 * fixed some bugs revealed by Paul Dietz' test suite:
1083 ** incorrect optimization of TRUNCATE for a positive first
1084 argument and negative second.
1085 ** compiler failure in let-convertion during flushing dead code.
1086 ** compiler failure while deriving type of TRUNCATE on an
1087 interval, containing 0.
1088 ** ASH of a negative bignum by a negative bignum count now returns
1090 ** intersection of CONS types now canonicalizes properly, fixing
1091 inconsistencies in SUBTYPEP.
1093 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1094 * incompatible change: The --disable-debugger command line
1095 option now clobbers the debugger at a more fundamental
1096 level, by redefining #'INVOKE-DEBUGGER instead of by
1097 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1098 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1099 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1100 enter the debugger REPL and then suffer recursive errors
1101 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1102 while under the new system, BREAK in --disable-debugger mode
1103 terminates the system just as an unhandled error would.
1104 * fixed compiler performance when processing loops with a step >1;
1105 * bug fix: DOCUMENTATION now retrieves generic function
1106 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1107 support has been systematized, and now supports the methods
1108 specified by ANSI, along with a default method and a method for
1109 slot documentation. (reported by Nathan Froyd)
1110 * bug fix: effective methods associated with a generic function are
1111 no longer cached over a change of that generic function's method
1112 combination. (reported by Andreas Fuchs)
1113 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1114 names a non-existent directory. (thanks to Andreas Fuchs)
1115 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1116 the PPC platform, lead to stack corruption; this has been fixed.
1117 (reported by Rainer Joswig)
1118 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1119 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1120 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1121 arithmetic in general, is now much more reliable on the PPC
1123 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1124 platform now returns the right answer.
1125 * optimization: restored some effective method precomputation in
1126 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1127 precomputation is now tunable.
1128 * optimization: compiler-internal data structure use has been
1129 reviewed, and changes have been made that should improve the
1130 performance of the compiler by about 20%.
1131 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1132 simple-base-strings and simple-bit-vectors is improved.
1133 * optimization: the optimization of 32-bit logical and arithmetic
1134 functions introduced in version 0.8.3 on the x86 has been
1135 implemented on the mips, ppc and sparc platforms; an
1136 implementation of the same facility, but for 64-bit arithmetic,
1137 has been added for the alpha.
1138 * microoptimization: the compiler is better able to make use of the
1139 x86 LEA instruction for multiplication by constants.
1140 * bug fix: in some situations compiler did not report usage of
1141 generic arithmetic in (SPEED 3) policy.
1142 * bug 145b fix: compiler used wrong type specifier while converting
1143 MEMBER-types to numeric.
1144 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1146 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1147 index argument now works properly on the Alpha platform.
1148 * bug fix: floating point exception treatment on the Alpha platform
1150 * bug fix: FILE-POSITION works much better on string input and
1151 output streams. (thanks to Nikodemus Siivola)
1152 * bug fix: many threading/garbage collection symptoms sorted.
1153 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1154 might be pseudo-atomic.
1155 * internal change: Stopping for GC is now done with signals not
1156 ptrace. GC is now done in whichever thread wanted it, instead of
1158 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1160 * bug fix: result form in DO is not contained in the implicit
1162 * incompatible change: ICR structure is changed; the value part of
1163 CONTINUATION is now called LVAR; corresponding functions are
1164 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1165 * added type deriver for ISQRT (thanks to Robert E. Brown).
1166 * bug fix: better support for loading from the command line when an
1167 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1169 * fixed some bugs revealed by Paul Dietz' test suite:
1170 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1171 ** ROUND and FROUND now give the right answer when given very
1172 small float arguments.
1173 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1175 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1176 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1177 ** LCM should return a non-negative integer.
1178 ** PARSE-INTEGER returned the index of a terminator instead of the
1179 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1180 ** PARSE-INTEGER returned an incorrect index being applied to a
1182 ** LCM with two arguments of 0 returns 0 rather than signalling
1184 ** unsigned addition of a 32-bit constant with the high bit set no
1185 longer causes an internal compiler error on the x86.
1186 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1187 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1188 with negative last argument.
1189 ** byte specifiers with zero size and position no longer cause
1190 an error during type derivation.
1191 ** bignum multiplication on the Alpha platform now returns the
1193 * porting: The system now builds on SuSE AMD64, although it still
1194 generates a 32-bit binary.
1195 * .fasl file incompatibility: The fasl file version number has
1196 been incremented (because of the changes to internal compiler
1197 data structures referred to above).
1199 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1200 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1201 more accurately, on the Darwin kernel running on PowerPC hardware.
1202 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1203 * Compiler code deletion notes now signal a condition of type
1204 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1205 an associated MUFFLE-WARNING restart.
1206 * The compiler now performs limited argument count validation of
1207 constant format strings in FORMAT, and where appropriate in ERROR,
1208 CERROR and WARN. (thanks to Gerd Moellmann)
1209 * New ASDF-INSTALL contrib can be used for automatic download and
1210 installation of third-party Lisp code from CCLAN or other sites
1212 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1213 which forces another thread to execute a function supplied by the
1215 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1216 now accept and act upon their :ELEMENT-TYPE keyword argument.
1217 (reported by Martin Atzmueller, Edi Weitz)
1218 * bug fix: FILE-POSITION now accepts position designators up to
1219 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1220 is the greater. (thanks to Patrik Nordebo)
1221 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1222 Kalle Olavi Niemitalo)
1223 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1224 printable. (reported by Eric Marsden)
1225 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1226 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1227 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1228 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1229 (thanks to Stig Sandoe)
1230 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1231 UNREAD-CHAR now correctly includes the unread character in the
1232 target sequence. (thanks to Gerd Moellmann)
1233 * bug fix in threaded builds: the system can now be suspended and
1234 resumed by shell job control with minimal disruption.
1235 * bug fixes in times and timezones >2038AD
1236 * better handling of "where is GNU make?" problem in build scripts
1237 (thanks to Nikodemus Siivola)
1238 * new optimization: inside a named function any reference to a
1239 function with the same name is considered to be a self-reference;
1240 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1241 optimization quality.
1242 * new optimization on x86: logical functions and + now have
1243 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1244 used when the result is truncated to 32 bits.
1245 * VALUES declaration is partially enabled.
1246 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1247 * fixed some bugs revealed by Paul Dietz' test suite:
1248 ** The system now obeys the constraint imposed by
1249 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1250 types form a lattice under type intersection.
1251 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1252 ** ASSOC now ignores NIL elements in an alist.
1253 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1254 and (1+ MOST-POSITIVE-FIXNUM) answers.
1255 ** The addition of a method with invalid qualifiers to a generic
1256 function does not cause an error to be signalled immediately;
1257 a warning is signalled, and the error is generated only on
1258 calling the generic function.
1259 * changed .fasl file version number, in part to add type codes for
1260 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1261 obscure ANSI requirements
1263 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1264 * fixed bug 148: failure to inline-expand a local function left
1265 garbage, confusing the compiler.
1266 * fixed bugs 3cd: structure slot readers perform type check if the
1267 slot can have an invalid value (i.e. it is either not initialized
1268 or can be written with a less specific slot writer).
1269 * bug fix: the compiler now traps array references to elements off
1270 the end of an array; previously, the bounds checking in some
1271 circumstances could go off-by-one.
1272 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1274 * type declarations for array element types now obey the description
1275 on the CLHS page "Declaration TYPE", as per discussions on
1276 sbcl-help around 2003-05-08. This means that a declaration
1277 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1278 declaration, all references to BAR will be asserted or assumed
1279 (with THE, so dependent on compiler policy) to involve objects of
1280 type FOO. Note that no such declaration is implied in
1281 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1282 * declared types of functions from the "Conditions"
1283 chapter. (reported by Paul Dietz)
1284 * bug fix: CERROR accepts a function as its first argument.
1285 * bug fix: NTH an NTHCDR accept a bignum as index
1286 arguments. (reported by Adam Warner)
1287 * optimization: character compare routines now optimize comparing
1288 against a constant character. (reported by Gilbert Baumann)
1289 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1290 argument now works properly on the MIPS platform.
1291 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1292 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1293 declared to have a functional type, does not cause an error
1295 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1296 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1297 * fixed bug: if last continuation of a deleted block has a
1298 destination, this destination should be deleted too. (reported by
1300 * fixed a bug in the bootstrap process: the host compiler's values
1301 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1302 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1303 test case from Patrik Nordebo)
1304 * improved the ability of the disassembler on the PPC platform to
1305 provide helpful disassembly notes.
1306 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1307 the class in more cases than previously.
1308 * bug fix: FILE-POSITION now understands :START and :END for
1309 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1310 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1311 8) data. (thanks to David Lichteblau)
1312 * bug fix: it is possible to add a method to a generic function
1313 without lambda list.
1314 * bug fix: reader failed to signal END-OF-FILE inside an
1315 object representation. (reported by Nikodemus Siivola)
1316 * fixed some bugs revealed by Paul Dietz' test suite:
1317 ** LAST and [N]BUTLAST should accept a bignum.
1318 ** condition slot accessors are methods.
1319 ** (VECTOR NIL) is a subtype of STRING.
1321 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1322 * minor incompatible change: some nonsensical specialized lambda
1323 lists (used in DEFMETHOD) which were previously ignored now signal
1325 * minor incompatible change: the system is now aware of the types of
1326 variables in the COMMON-LISP package, and will signal errors for
1327 most violations of these type constraints (where previously they
1328 were silently accepted).
1329 * minor incompatible change: COMPILE-FILE now uses the freedom
1330 afforded (ANSI 3.2.2.3) to use derived function types for
1331 functions defined in the same file. This also permits the system
1332 to warn on static type mismatches and function
1333 redefinition. (Currently it does not work with high DEBUG level.)
1334 * minor incompatible change: VALUES declaration is disabled.
1335 * When issuing notes, the compiler now signals a condition of type
1336 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1337 restart for use in user handlers. It is expected that the
1338 COMPILER-NOTE condition will eventually become a condition
1339 supertype to a hierarchy of note types, which will then be
1340 handleable in a similar fashion. However, at the moment, no such
1341 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1342 but it's an implementation detail, not a classification for the
1344 * Changes in type checking closed the following bugs:
1345 ** type checking of unused values (192b, 194d, 203);
1346 ** template selection based on unsafe type assertions (192c, 236);
1347 ** type checking in branches (194bc).
1348 * A short form of VALUES type specifier has ANSI meaning (it has
1349 increased the number of situations when SBCL cannot perform type
1351 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1352 keywords or constants is permissible.
1353 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1354 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1355 classes. (thanks to Antonio Martinez)
1356 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1357 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1358 argument to a pretty-print stream. Instead, it leaves any such
1359 operations to DESCRIBE-OBJECT methods.
1360 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1362 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1363 lambda lists. (thanks to David Lichteblau)
1364 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1365 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1366 (thanks to Gerd Moellmann)
1367 * fixed bug 246: increased compilation speed of long
1368 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1370 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1372 * DEFCONSTANT now throws a condition of type
1373 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1374 constant to a non-EQL value; CONTINUE and ABORT restarts
1375 respectively change and preserve the value.
1376 * fixed bug 63: The code walker, part of the implementation of CLOS,
1377 is now better at handling symbol macros.
1378 * bug fix: There is no longer an internal implementation type named
1379 CL:LENGTH. (reported by Raymond Toy)
1380 * bug fix: In macro-like defining macros/special operators the
1381 implicit block does not enclose the lambda list.
1382 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1383 enclosed in parentheses) are not suitable as type specifiers, and
1384 their use properly signals an error now.
1385 * bug fix: An argument count mismatch for a type specifier in code
1386 being compiled no longer causes an unhandled error at compile
1387 time, but signals a compile-time warning.
1388 * fixed simple vector readable printing
1389 * bug fix: DESCRIBE takes more care over whether the class
1390 precedence list slot of a class is bound before accessing it.
1391 (reported by Markus Krummenacker)
1392 * bug fix: FORMATTER can successfully compile pretty-printer format
1393 strings which use variants of the ~* directive inside.
1394 * bug fix: SEARCH now applies its TEST predicate to the elements of
1395 the arguments in the correct order. (thanks to Wolfhard Buss)
1396 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1397 in type inference. (thanks to Robert E. Brown)
1398 * bug fix: Reading in symbols with an explicit package name of ""
1399 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1400 * fixed some bugs revealed by Paul Dietz' test suite:
1401 ** NIL is now allowed as a structure slot name.
1402 ** Arbitrary numbers, not just REALs, are allowed in certain
1403 circumstances in LOOP for-as-arithmetic clauses.
1404 ** Multiple class redefinitions before slot access no longer
1405 causes a type error.
1406 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1407 association between the name and a class.
1408 ** Generic functions with non-standard method-combination and over
1409 five methods all of which return constants no longer return NIL
1410 after the first few invocations. (thanks to Gerd Moellmann)
1411 ** CALL-NEXT-METHOD with no arguments now passes the original
1412 values of the arguments, even in the presence of assignment.
1413 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1415 ** DISASSEMBLE works with closures and funcallable instances.
1416 ** ADD-METHOD now returns the generic function, not the new method.
1417 ** FIND-METHOD signals an error if the lengths of the specializers
1418 is incompatible with the generic function, even if the ERRORP
1420 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1421 which its argument is a member.
1422 ** DEFCLASS only redefines the class named by its class-name
1423 argument if that name is the proper name of the class;
1424 otherwise, it creates a new class.
1425 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1426 of the UNBOUND-SLOT condition to the name of the slot.
1427 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1428 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1429 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1430 SLOT-MAKUNBOUND in the specified fashion.
1432 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1433 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1434 cross-compilation host. As a consequence, we can now bootstrap our
1435 way up to SBCL starting with a bare gcc toolchain and human-readable
1436 source code (first the source to CLISP, then the source to SBCL).
1437 * A contributed module containing a partial implementation of the
1438 simple-streams interface has been included. (thanks to Rudi
1440 * A contributed module implementing the RFC1321 Message Digest
1441 Algorithm, known as MD5, has been included.
1442 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1443 no longer has any effect, as the code controlled by this feature
1444 has been deleted. (As far as we know, no-one has ever built using
1445 this feature, and its semantics were confused in any case).
1446 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1447 work on conditions (as required by the ANSI specification),
1448 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1449 expected behaviour on conditions. Users should note, however,
1450 that such behaviour is not required by the ANSI specification,
1451 and so use of this behaviour may render their code unportable.
1452 * fixed some bugs revealed by Paul Dietz' test suite:
1453 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1455 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1456 ** SXHASH on condition objects no longer returns NIL.
1457 ** :ALLOCATION :CLASS slots are better treated; their values are
1458 updated on class redefinition, and initforms inherited from
1459 superclasses are applied.
1460 ** REMOVE-METHOD returns its generic function argument even when
1461 no method was removed.
1462 ** SHARED-INITIALIZE now initializes the values of the requested
1463 slots, including those with :ALLOCATION :CLASS.
1464 ** ALLOCATE-INSTANCE now works on structure classes defined via
1465 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1467 ** SLOT-EXISTS-P now works on conditions, as well as structures
1469 ** MAKE-LOAD-FORM now has the required methods on
1470 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1471 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1472 keyword, and now implements the SLOT-NAMES argument.
1473 ** methods with &OPTIONAL arguments no longer allow too many
1474 arguments to be passed in the call without error.
1475 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1476 option is consistent with the required arguments of the generic
1477 function lambda list.
1478 * bug fix: REQUIRE accepts a string designator. (Thanks to
1480 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1481 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1482 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1483 * bug fix: The compiler now checks for duplicated variables in macro
1485 * bug fix: SETQ on globals returns the correct value.
1486 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1487 not a COMPILER-ERROR (followed by some other strange error on
1488 choosing the CONTINUE restart).
1489 * bug fix: make.sh and friends are now more consistent in the way that
1490 they look for GNU "make".
1492 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1493 * experimental native threads support (on x86 Linux >=2.4 only).
1494 This is not compiled in by default: you need to add :SB-THREAD to
1495 the target features. See the "Beyond ANSI" chapter of the manual
1497 * fix for longstanding nonANSIism: The old distinction between
1498 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1499 The return value from CL:FIND-CLASS is now a CLOS class, and
1500 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1501 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1502 * An interface to the MetaObject Protocol, as described in Kiczales,
1503 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1504 MIT Press, 1991, is available from the SB-MOP package.
1505 * incompatible change: the SB-PCL package should now be considered
1506 a private implementation detail, and no longer a semi-private MOP
1508 * minor incompatible change: due to rearrangement for threads, the
1509 control stack and binding stack are now allocated at arbitrary
1510 addresses instead of being hardcoded per-port. Users affected by
1511 this probably have to be doing advanced things with shared
1512 libraries, and will know who they are.
1513 * minor incompatible change: Previously, all --eval forms used were
1514 processed with READ before any of them were processed with EVAL.
1515 Now each --eval form is processed with both READ and EVAL before
1516 the next --eval form is processed. (Thus package operations like
1517 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1518 work as the user might reasonably expect.)
1519 * minor incompatible change: *STANDARD-INPUT* is now only an
1520 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1522 * minor incompatible change: Y-OR-N-P is now character-oriented,
1523 not line oriented. Also, YES-OR-NO-P now works without errors.
1524 (thanks to Antonio Martinez)
1525 * sb-aclrepl module improvements: an integrated inspector, added
1526 repl features, and a bug fix to :trace command.
1527 * Known functions, which cannot be open coded by the backend, are
1528 considered to be able to check types of their arguments. (fixing
1529 a bug report by Nathan J. Froyd)
1530 * fixed a bug in computing method discriminating functions: It is
1531 now possible to define methods specialized on classes which have
1532 forward-referenced superclasses. (thanks to Gerd Moellmann)
1533 * fixed evaluation order in optional entries (reported by Gilbert
1535 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1536 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1537 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1538 required-by-AMOP NAME argument, as well as CLASS and
1539 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1540 * fixed bug 20: DEFMETHOD can define methods using names that are
1541 not the proper names of classes to designate class specializers.
1542 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1543 Unix file descriptors, instead of blowing up. (thanks to Antonio
1545 * Garbage collection refactoring: user-visible change is that a
1546 call to the GC function during WITHOUT-GCING will not do garbage
1547 collection until the end of the WITHOUT-GCING. If you were doing
1548 this you were probably losing anyway.
1549 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1550 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1551 TYPEP the latter but not the former.
1552 * The compiler issues a full WARNING for calls to undefined functions
1553 with names from the CL package.
1554 * MAP-INTO for a vector destination is open coded. (reported by
1555 Brian Downing on c.l.l)
1556 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1557 documentation string.
1558 * fixed some bugs revealed by Paul Dietz' test suite:
1559 ** COPY-ALIST now signals an error if its argument is a dotted
1561 ** Condition slots are now accessed more correctly in the presence
1562 of multiple initargs for a given slot.
1563 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1564 exclude restarts of the same name associated with a different
1566 ** DEFCLASS of forward-referenced classes with another
1567 forward-referenced class in the superclasses list no longer
1569 ** Condition slots are now initialized once each, not multiple
1570 times. (thanks to Gerd Moellmann)
1571 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1573 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1574 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1575 arguments contain duplicated elements.
1576 ** RESTART-CASE understands local macros.
1577 ** RESTART-CASE associates exactly its own restarts with a condition.
1578 ** ENDP in safe mode checks its argument to be of type LIST.
1579 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1580 in question is unbound.
1581 ** Optimized MAKE-INSTANCE functions no longer cause internal
1582 assertion failures in the presence of duplicate initargs.
1583 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1584 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1585 propagated even in the ctor optimized implementation of
1587 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1589 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1590 * a better implementation of SXHASH on (simple) bit vectors,
1591 measured both in execution speed and in distribution of results
1592 over the positive fixnums, has been installed. Likewise, a better
1593 implementation of EQUAL for simple bit vectors is now available.
1594 * fixed CEILING optimization for a divisor of form 2^k.
1595 * fixed bug 240 (emitting extra style warnings "using the lexical
1596 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1597 by Antonio Martinez)
1598 * fixed SXHASH, giving different results for NIL depending on type
1599 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1600 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1601 variable. (found by Rolf Wester)
1602 * SBCL does not ignore type declarations for special
1603 variables. (reported by rif on c.l.l 2003-03-05)
1604 * some bug fixes in contrib/sb-aclrepl/
1605 * fixed some bugs revealed by Paul Dietz' test suite:
1606 ** a bug in the CONS type specifier, whereby the CAR and CDR
1607 types got intertwined, has been fixed;
1608 ** the type system is now able to reason about the interaction
1609 between INTEGER and RATIO types more completely;
1610 ** APPEND, [N]REVERSE and NRECONC check that those their
1611 arguments, which must be proper lists, are really so;
1612 ** An array specialized to be unable to hold elements has been
1613 implemented, as required -- yes, really -- by ANSI;
1614 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1615 on malformed property lists;
1617 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1618 * incompatible packaging change: in line with Unix convention,
1619 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1620 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1621 * REQUIRE and PROVIDE are now optionally capable of doing something
1622 useful. See the documentation string for REQUIRE.
1623 * infrastructure for a managed SBCL contrib system: contributed
1624 modules in this release include:
1625 ** the ASDF system definition facility;
1626 ** an interface to the BSD Sockets API;
1627 ** an ACL-like convenience interface to the repl;
1628 (thanks to Kevin Rosenberg)
1629 ** an implementation of ROTATE-BYTE, with efficient implementation
1631 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1632 gives a closer approximation to the right answer than previously.
1633 (thanks to Raymond Toy)
1634 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1635 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1636 optional environment arguments, as required by ANSI.
1637 * fixed bugs in other functions taking environment objects, allowing
1638 calls with an explicit NIL environment argument to be compiled
1640 * fixed bug 228: primary return values from
1641 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1642 COMPILE or FUNCTION.
1643 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1644 :TYPE will no longer signal errors on innocuous objects.
1645 * fixed bug 231b: SETQ is better at respecting type declarations in
1646 the lexical environment.
1647 * fixed a bug in DEFCLASS: classes named by symbols with no or
1648 unprintable packages can now be defined.
1649 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1650 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1651 * fixed compiler failure related to checking types of functions.
1652 (reported by Robert E. Brown)
1653 * the compiler is now much more consistent in its error-checking
1654 treatment of bounding index arguments to sequence functions: in
1655 (SAFETY 3) code, errors will be signalled in almost all cases if
1656 invalid sequence bounding indices are passed to functions defined
1657 by ANSI to operate on sequences.
1658 * fixed a bug in the build procedure: documentation of SBCL-specific
1659 packages is now preserved and available in the final Lisp image.
1660 * lifted FDEFINITION lookup out of loops in the implementation of
1661 many list operations. (thanks to Robert E. Brown)
1662 * fixed a bug in the reader: the #n# reader macro now works for
1663 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1664 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1665 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1666 E. Brown and Tony Martinez respectively)
1667 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1668 SLOT-BOUNDP: in particular, it is now able to optimize them much
1669 better, and is now not vulnerable to having packages renamed.
1670 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1671 always called when a slot is not present in an instance. (thanks
1673 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1674 longer signal an error on unreading a character following EOF on
1675 the previous constituent stream. (thanks to Tony Martinez)
1676 * fixed some bugs revealed by Paul Dietz' test suite:
1677 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1678 not just nonnegative fixnums;
1679 ** the logical bit-array operators such as BIT-AND now accept an
1680 explicit NIL for their "opt-arg" argument (to indicate a
1681 freshly-consed result bit-array);
1682 ** ELT now signals an error on an invalid sequence index in safe
1684 ** the type system is now cleverer about negations of numeric
1685 types, and consequently understands the BIGNUM and RATIO types
1687 ** the type system is now cleverer about the interaction between
1688 INTEGER and RATIO types: while bugs still remain, many more
1689 cases are accurately computed;
1690 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1691 if it is in the last clause;
1692 ** CONSTANTLY now correctly returns a side-effect-free function in
1694 ** DECLARE is no longer treated as a special-operator; in
1695 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1696 * incremented fasl file version number due to the change in the
1697 DEFSTRUCT-SLOT-DESCRIPTION structure.
1699 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1700 * minor incompatible change: code processed by the "interpreter" or
1701 EVAL now has a compilation optimization policy of (DEBUG 2)
1702 (changed from (DEBUG 1)) to improve debuggability of interactive
1703 development, and to allow the use of the debug RETURN command in
1705 * an experimental implementation of the RETURN command for the
1706 debugger has been included. (thanks to Frederik Kuivinen)
1707 * fixed bug 62: constraints were not propagated into a loop.
1708 * fixed bug in embedded calls of SORT (reported and investigated by
1710 * fixed some bugs revealed by Paul Dietz' test suite:
1711 ** printing and reading of arrays with some dimensions having
1712 length 0 (thanks to Gerd Moellmann);
1713 ** BOA constructor with &AUX argument without a default value does
1714 not cause a type error;
1715 ** CONSTANTP now returns true for all self-evaluating objects.
1717 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1718 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1719 accessors that are related by inheritance, as specified in the
1720 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1721 Valtteri Vuorikoski)
1722 * The compiler is now able to inline functions that were defined in
1723 a complex lexical environment (e.g. inside a MACROLET).
1724 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1726 * Support for the upcoming FreeBSD-5.0 release has been included.
1727 (thanks to Dag-Erling Smorgrav)
1728 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1729 effect when it is not in a toplevel context.
1730 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1731 stand a better chance of being correct. (thanks to Gerd
1733 * fixed bug in COERCE, which now signals an error on coercing a
1734 rational to a bounded real type which excludes the expected
1736 * The compiler is now able to derive types more accurately from the
1737 COERCE and COMPILE functions.
1738 * fixed bug 223: functional binding is considered to be constant
1739 only for symbols in the CL package.
1740 * fixed bug 231: SETQ did not check the type of a variable being set
1741 (reported by Robert E. Brown)
1742 * A new optimization for MAKE-INSTANCE has been included, fixing
1743 various bugs (including relating to :ALLOCATION :CLASS slots and
1744 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1746 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1747 ** As required by ANSI, LOOP now disallows anonymous collection
1748 clauses such as COLLECT I in conjunction with aggregate boolean
1749 clauses such as THEREIS (= I 1);
1750 ** LOOP now signals an error when any variable is reused in the
1751 same loop (including the potentially useful construct analogous
1752 to WITH A = 1 WITH A = (1+ A);
1753 ** IT is only a special loop symbol within the first clause of a
1754 conditional loop clause;
1755 ** LOOP with a typed iteration variable over a hashtable now
1756 signals a type error iff it should.
1757 * fixed some other bugs revealed by Paul Dietz' test suite:
1758 ** FILE-STREAM now names the class previously known as FD-STREAM;
1759 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1760 argument) no longer signals an error;
1761 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1762 of the slot symbol, rather than using the current package
1763 ((:CONC-NAME "") continues to intern the slot's name in the
1765 * incremented fasl file version number, because of the incompatible
1766 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1767 of the new implementation of DEFINE-COMPILER-MACRO.
1769 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1770 * Support for building SBCL for MIPS platforms running in
1771 little-endian mode has now been checked in, and basic
1772 functionality on said platforms verified.
1773 * minor incompatible change: PCL now records the pathname of a file
1774 in which methods and the like are defined, rather than its
1776 * minor incompatible change: TRUENAME now considers the truename of
1777 a file naming a directory to be the pathname with :DIRECTORY
1778 component indicating that directory.
1779 * minor incompatible change: a NAMED clause in the extended form of
1780 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1781 reason for the previous behaviour is unclear.
1782 * more systematization and improvement of CLOS and MOP conformance
1783 in PCL (thanks to Gerd Moellman and Pierre Mai):
1784 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1786 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1787 primary methods with no specializers;
1788 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1790 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1791 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1792 FORWARD-REFERENCED-CLASSes; error reporting on
1793 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1795 ** SXHASH on CLOS instances now uses a slot internal to the
1796 instance to return different numbers on distinct instances,
1797 while preserving the same return value through invocations of
1799 ** DEFMETHOD signals errors when methods with longer incongruent
1800 lambda lists are added to generic functions;
1801 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1802 CLASS, as specified in AMOP;
1803 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1804 on the order returned by the primary method for classes of
1805 class STANDARD-CLASS;
1806 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1807 * fixed some bugs shown by Paul Dietz' test suite:
1808 ** DOLIST puts its body in TAGBODY;
1809 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1811 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1812 value producing form;
1813 ** if more variables are given to PROGV than values, extra
1814 variables are bound and made to have no value;
1815 ** NSUBSTITUTE on list arguments gets the right answer with
1817 ** ELT signals an error of type TYPE-ERROR when the index argument
1818 is not a valid sequence index;
1819 ** LOOP signals (at macroexpansion time) an error of type
1820 PROGRAM-ERROR when duplicate variable names are found;
1821 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1822 ** FUNCALL of special-operators now cause an error of type
1824 ** PSETQ now works as required in the presence of side-effecting
1825 symbol-macro places;
1826 ** NCONC accepts any object as its last argument;
1827 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1829 ** loop-for-as-package does not require a package to be explicitely
1831 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1833 * fixed bug 166: compiler preserves "there is a way to go"
1834 invariant when deleting code.
1835 * fixed bug 172: macro lambda lists with required arguments after
1836 &REST arguments now cause an error to be signalled. (thanks to
1838 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1839 bug. (thanks to Matthew Danish)
1840 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1842 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1843 arguments are lexically rebound. (thanks to Gerd Moellmann and
1845 * fixed bug 194: error messages are now more informative when there
1846 is no primary method applicable in a call to a generic
1847 function. (thanks to Gerd Moellmann)
1848 * fixed bug in command line argument checking (thanks to Julian
1850 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1851 (thanks to Matthew Danish)
1852 * incremented fasl file version number, because of the
1853 SXHASH-related changes in the layout of CLOS data structures
1855 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1856 * minor incompatible change: The runtime (the Unix executable named
1857 "sbcl") is now much pickier about the .core files it will load.
1858 Essentially it now requires .core files to descend from the same
1859 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1860 as the runtime does. (The intent is to prevent the crashes which
1861 can occur, and which can even be reported as mysterious failures,
1862 when people patch the sources or change the build parameters
1863 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1864 sbcl and .core files.)
1865 * fixed bug: VALUES-LIST is no longer optimized away.
1866 * fixed bug 142: The FFI conversion of C string values to Lisp
1867 string values no longer conses excessively. (thanks to Nathan
1868 Froyd porting Raymond Toy's fix to CMU CL)
1869 * began to systematize and improve MOP conformance in PCL (thanks to
1870 Nathan Froyd, Gerd Moellman and Pierre Mai):
1871 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1873 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1874 ** FINALIZE-INHERITANCE is now called on class finalization;
1875 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1876 argument precedence order.
1877 * fixed bug 202: The compiler no longer fails on functions whose
1878 derived types contradict their declared type.
1879 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1880 so it can be non-toplevel.
1881 * The fasl file version number has changed (because of the new
1882 implementation of DEFMACRO).
1883 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1884 safe code, that any length requirement by their type-specifier
1885 argument is valid. The exceptions to this are described in bug
1887 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1888 functions now check the types of their inputs as required by ANSI.
1889 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1890 introduced symbol is DECLAREd to be SPECIAL.
1891 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1892 * fixed bug: :COUNT argument to sequence functions may be negative
1893 * fixed bug: body of DO-SYMBOLS may contain declarations
1894 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1895 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1896 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1897 symbol macro only once
1898 * fixed printing of call frame when argument list is unavailable
1899 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1900 * compiler no longer signals WARNING on unknown keyword
1903 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1904 * A beta-quality port to the mips architecture running Linux,
1905 based on the old CMUCL backend, has been made. It has been tested
1906 on a big-endian kernel, and works sufficiently well to be able to
1907 rebuild itself; it has not been tested in little-endian mode.
1908 * fixed an inconsistency between gencgc.c and purify.c which made
1909 dumping/loading .core files unreliable
1910 * fixed bug 120a: The compiler now deals correctly with IFs where
1911 the consequent is the same as the alternative, instead of
1912 misderiving the return type. (thanks to Alexey Dejneka)
1913 * fixed bug 113: Logical pathnames are now dumpable (the logical
1914 host is resolved at load-time, throwing an error if it is not
1916 * fixed bug 174: FORMAT's error message is slightly clearer when a
1917 non-printing character is used in a format directive.
1918 * fixed several bugs in compiler checking of type declarations, i.e.
1919 violations of the Python "declarations are assertions" principle
1920 (thanks to Alexey Dejneka)
1921 * fixed several bugs in PCL's error checking (thanks to Gerd
1923 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1925 * fixed bug in compilation of functions as first class values
1926 (thanks to Antonio Martinez-Shotton)
1927 * The compiler's handling TYPE-ERRORs which it can prove will
1928 inevitably happen at runtime has been cleaned up and corrected
1929 in several ways. (thanks to Alexey Dejneka)
1930 * improved argument type checking for various basic arithmetic
1931 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1932 lavished on them in the past that they can be compiled in many
1933 ways in different special cases
1934 * fixed bug 181: compiler checks validity of user supplied type
1936 * cleaned up code flushing in optimization: Function calls which
1937 should signal errors for safety purposes (e.g. which ANSI says
1938 should signal errors when their arguments are of incorrect type)
1939 are no longer optimized away.
1940 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1941 * incremented fasl file version number, because changes in the
1942 implementation of sequence functions like COERCE caused
1943 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1944 in old inline expansions) to become undefined. (Actually these
1945 changes were later undone, so we might very well be binary
1946 compatible with 0.7.7 after all, but leaving the version number
1947 incremented seemed like the simplest and most conservative
1950 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1951 * An alpha-quality port to the parisc architecture running Linux,
1952 based on the old CMUCL backend, has been made. This, even more so
1953 than the other backends, should be considered still a work in
1954 progress; known problems include that the Linux kernel in 64-bit
1955 mode does not propagate the correct sigcontext structure to
1956 userspace, and consequently SBCL on a parisc64 kernel will not
1958 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1959 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1960 suppressed.) Also now that the compiler is looking at declarations
1961 in the environment, it checks optimization declarations as well,
1962 and suppresses inlining when (> DEBUG SPEED).
1963 * More fixes have been made to treatment of floating point exception
1964 treatment and other Unix signals. In particular, floating point
1965 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1966 * The detection and handling of control stack exhaustion (infinite
1967 or very deeply nested recursion) has changed. Stack exhaustion
1968 detection is now done by write-protecting pages at the OS level
1969 and applies at all optimization settings; when found, a
1970 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1971 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1972 be caught using IGNORE-ERRORS.
1973 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1974 correctly when transforming recursive LABELS functions to LETs.
1975 (thanks to Alexey Dejneka)
1976 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1977 that are names of constants or global variables.
1978 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1979 alien routines with docstrings.
1980 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1981 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1983 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1984 (thanks to Eric Marsden)
1985 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1986 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1987 * Bug fix: The SPARC backend can now compile functions involving
1988 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1989 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1990 to a functions expecting a PATHNAME-DESIGNATOR.
1991 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1992 lambda lists. (thanks to Alexey Dejneka)
1993 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1994 functions so that EXPORTing the name of the class doesn't cause
1995 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1996 to Antonio Martinez for reporting this)
1997 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1998 constant-folded without failing an assertion. (thanks to Einar
1999 Floystad Dorum for reporting this)
2000 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2001 types are now dealt with more correctly by the compiler.
2002 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2003 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2004 behaviour analogous to RENAME-FILE. This puts its behaviour more
2005 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2007 * The fasl file version number has changed again. (because of the
2008 bug fix involving the names of PCL MAKE-INSTANCE functions)
2010 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2011 * bug fix: Floating point exceptions are treated much more
2012 consistently on the x86/Linux and PPC/Linux platforms.
2013 * Array initialization with :INITIAL-ELEMENT is now much faster for
2014 cases when the compiler cannot open code the array creation, but
2015 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2016 array accesses have also seen a speed increase.
2017 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2018 specified. (This was at the root of some bad interactions between
2019 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2021 * bug fix: Internal error arguments for undefined functions are now
2022 computed correctly on the PPC/Linux platform.
2023 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2024 Raymond Toy's patch for CMU CL)
2025 * Support for the Solaris 9 operating environment has been included
2026 (thanks to Daniel Merritt)
2027 * A very ugly but hopefully complete draft of the missing FFI chapter
2028 of the manual has been created by reformatting the corresponding
2029 CMU CL manual chapter into (currently very ugly and incoherent)
2030 DocBook and bringing it up to date for SBCL behavior. Thus, the
2031 manual is now essentially complete, at least by my extreme
2032 once-and-only-once standards, whereby it's acceptable to refer to
2033 the doc strings of SB-EXT functions as the primary documentation.
2034 * The fasl file version number has changed again, due to cleanup of
2035 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2036 is no longer a static symbol.)
2038 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2039 * SBCL now builds with OpenMCL (version 0.12) as the
2040 cross-compilation host; also, more progress has been made toward
2041 bootstrapping under CLISP.
2042 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2044 * bug 158 fixed: The compiler can now deal with integer loop
2045 increments different from 1; fixing this turned out also to fix
2047 * bug 169 fixed: no more bogus warnings about using lexical bindings
2048 despite the presence of perfectly good SPECIAL declarations (thanks
2049 to David Lichteblau)
2050 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2051 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2052 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2054 * bug 180 fixed: Method combination specifications no longer ignore
2055 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2056 * bug fix: Structure type predicate functions now check their argument
2057 count as they should.
2058 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2059 correctly. (thanks to Pierre Mai)
2060 * minor incompatible change: The --noprogrammer option is deprecated
2061 in favor of the new --disable-debugger option, which is very similar.
2062 (The major difference is that it takes effect at a slightly different
2063 time at startup, causing handling of errors in --sysinit and
2064 --userinit files will be affected differently.) The
2065 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2066 been added to allow this functionality to be controlled from ordinary
2067 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2068 maintainers, who might want to run non-interactive scripts to
2069 build SBCL cores which will later be used interactively.)
2070 * minor incompatible change: The LOAD function no longer, when given
2071 a wild pathname to load, loads all files matching that pathname.
2072 Instead, an error of type FILE-ERROR is signalled.
2074 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2075 * bug 147 fixed: The compiler preserves its block link/count
2076 invariants more correctly now so that it doesn't crash. (thanks
2078 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2080 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2081 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2082 no-value-for-this-character as NIL (as specified by ANSI).
2083 * HOST-NAMESTRING on physical pathnames now returns a string that is
2084 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2085 (thanks to Christophe Rhodes)
2086 * The Alpha port handles icache flushing more correctly. (thanks to
2088 * More progress has been made toward bootstrapping under CLISP. (thanks
2089 to Christophe Rhodes)
2090 * The fasl file format has changed again, because dynamic loading
2091 on OpenBSD (which has non-ELF object files) motivated some cleanups
2092 in the way that foreign symbols are transformed and passed around.
2093 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2094 is no longer treated as whitespace by the reader, but instead as
2095 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
2096 |A
\x7fB|, instead of A as it used to.
2098 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2099 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2100 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2101 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2102 this as of 0.7.1.45, but was left out of the previous news section
2103 (thanks to Dan Barlow)
2104 * SBCL now runs on the Solaris operating system on SPARC architectures
2105 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2106 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2107 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2108 * SPARC backend cleanups, allowing builds of cores optimized for V8
2109 and V9 SPARCS, and also emission of code targeted to a particular
2110 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2112 * SBCL is closer to bootstrapping under CLISP, thanks to various
2113 fixes by Christophe Rhodes.
2114 * The fasl file format has changed again, to allow the compiler's
2115 INFO database to support symbol macros.
2116 * The user manual (in doc/) is formatted into HTML more nicely.
2117 (thanks to coreythomas)
2118 * The system is smarter about SUBTYPEP relationships, especially
2119 those involving NOT types (including types such as ATOM which are
2120 represented internally using NOT types). Thus SUBTYPEP is less
2121 likely to return (VALUES NIL NIL) in general, and in particular
2122 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2124 * The fasl file format has changed again, because the internal
2125 representation of types now includes a new slot to support the new
2126 SUBTYPEP-of-NOT-types logic.
2127 * (not a change in the main branch of SBCL, but a related prototype
2128 which can hopefully be merged into the main branch of SBCL in the
2129 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2130 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2131 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2132 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2133 dispatch macro character combinations. (thanks to Alexey Dejneka)
2134 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2135 that superficially look like logical namestrings correctly.
2136 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2138 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2139 default supplied-p) for &optional and &key arguments. (thanks to
2142 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2143 * incompatible change: The compiler is now less aggressive about
2144 tail call optimization, doing it only when (> SPACE DEBUG) or
2145 (> SPEED DEBUG). (This is an incompatible change because there are
2146 programs which relied on the old CMU-CL-style behavior to optimize
2147 away their unbounded recursion which will now die of stack overflow.)
2148 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2149 for non-GENCGC systems has been increased to 20M (since that
2150 seems much closer to the likely performance optimum for modern
2151 systems than the old 4M value was)
2152 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2153 and *DEBUG-PRINT-LEVEL*
2154 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2155 of CMU CL's support for SPARC, and various endianness and other
2156 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2157 * new syntactic sugar for the Unix command line: --load foo.bar is now
2158 an alternate notation for --eval '(load "foo.bar")'.
2160 ** The system now detects stack overflow and handles it gracefully,
2161 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2162 optimization settings. (This is a good thing in general, and
2163 its introduction in this version should be particularly timely
2164 for anyone whose code fails because of suppression of tail
2166 ** The system now hunts for the C variable "environ" in a more
2167 devious way, to avoid segfaults when the C library version differs
2168 between compile time and run time. (thanks to Christophe Rhodes)
2169 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2170 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2171 ** The compiler no longer issues bogus style warnings for undefined
2172 classes in the same source file as the DEFCLASSes which defined
2173 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2175 ** fixes in CONDITION class precedence list for undefined function
2176 errors (thanks to Alexei Dejneka)
2177 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2178 correctly. (thanks to Dan Barlow)
2179 ** portability fixes aiming at bootstrapping under CLISP (thanks
2180 to Dave McDonald and Christophe Rhodes)
2181 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2182 ** fixes in type translation and and type inference (thanks to
2184 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2185 ** various fixes in the new ports (thanks to Dan Barlow)
2186 * several changes related to debugging:
2187 ** suppression of tail recursion, as noted above
2188 ** stack overflow detection, as noted above
2189 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2190 is now the default. (For some time encapsulation has been more
2191 reliable than the breakpoint-based :ENCAPSULATE NIL
2192 implementation, at least on X86 systems; and I just noticed that
2193 encapsulation also seems closer to the spirit of the ANSI
2196 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2198 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2199 up properly again. (There was a packaging bug in 0.7.0 which
2200 left their definitions in SB-SYS::LOAD-FOREIGN and
2201 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2202 vital for most things which interface to C-level interfaces,
2203 like extensions working with sockets or databases or
2204 Perl-compatible regexes or whatever, and the need to fix
2205 this bug is the main reason that 0.7.1 was released so
2207 ** DEFGENERIC is now choosier about the methods it redefines, so that
2208 reLOADing a previously-LOADed file containing DEFGENERICs does
2209 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2210 cycle now works as it should. (thanks to Alexey Dejneka)
2211 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2212 fixed by Christophe Rhodes. (He actually submitted this patch
2213 months ago, and I delayed until after 0.7.0.)
2214 ** Bug 111 (internal compiler confusion about runtime checks on
2215 FUNCTION types) was fixed by Alexey Dejneka.
2216 * Some internal cleanups (getting rid of variables which aren't
2217 needed now that the byte interpreter is gone) caused the fasl
2218 file format number to change again.
2220 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2221 * major incompatible change: The default fasl file extension, i.e. the
2222 default extension for files produced by COMPILE-FILE, has changed
2223 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2225 ** There are many changes in the implementation of the compiler.
2226 SBCL is now essentially a compiler-only implementation of ANSI
2227 Common Lisp. EVAL still "interprets" a few special cases, but
2228 almost all the interesting cases are handled by creating
2229 a LAMBDA expression, calling COMPILE on it, then calling
2230 FUNCALL on the result.
2231 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2232 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2233 Since the code is newer, there might still be some new bugs
2234 (though not as many as before Martin Atzmueller's fixes:-). But
2235 the new code is substantially simpler and clearer, and hopefully
2236 any remaining bugs will be simpler, less fundamental, and more
2237 fixable then the bugs in the old code.
2238 ** The revised compiler is still a little unsteady on its feet.
2240 *** The debugging information it produces (particularly the names
2241 of FUNCTION objects) is sometimes much less useful than what
2242 the old compiler produced.
2243 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2244 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2245 within a MACROLET) has been temporarily weakened.
2246 ** There are new compiler optimizations for various functions:
2247 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2248 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2249 *** the math functions TRUNCATE, FLOOR, and CEILING
2250 *** the function-of-all-trades COERCE
2251 Mostly these should be transparent, but there's one
2252 potentially-annoying problem (bug 117): when the compiler
2253 inline-expands a function and does type analysis on the result,
2254 it can create control paths which have type mismatches, and
2255 when it can't prove that those control paths aren't taken,
2256 it will issue WARNINGs about the type mismatches. This is
2257 a particular problem in practice for the new sequence functions.
2258 It's not clear how this should be fixed, and for now, a
2259 workaround is given in the entry for 117 in the BUGS file.
2260 ** (Because of the interaction between the two previous items --
2261 occasional inlining problems and new inline expansions -- some
2262 of the new sequence function optimizations won't really kick in
2263 completely until debugging information, and then inlining, are
2264 straightened out in some future version.)
2265 * minor incompatible changes:
2266 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2267 DIRECTORY no longer implicitly promotes NIL slots of its
2268 pathname argument to :WILD. In particular, when you ask for the
2269 contents of a directory (which you used to be able to do without
2270 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2271 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2272 ** changes in behavior that ANSI explicitly defines to be
2273 implementation dependent:
2274 *** The new compiler-only implementation still conforms with ANSI,
2275 but acts a little different than before. Besides the obvious
2276 changes in performance tradeoffs (that the cost per form passed
2277 to EVAL has gone up, and the cost per form executed by EVAL
2278 has gone down), the behavior of the system changes a little
2279 because there are no longer any interpreted function objects.
2280 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2281 e.g. doing COMPILE on the output of interactive DEFUN is
2283 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2285 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2286 has changed from (:CL) to NIL.
2287 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2288 e.g. #\Newline and #\Tab, don't have names specified in the
2289 ANSI Common Lisp standard, is now based on their ASCII symbolic
2290 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2291 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2292 are no longer used for output.
2293 ** changes in internal implementation constants:
2294 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2295 4 million. (If your application spends a lot of time GCing and
2296 you have a lot of RAM, you might want to experiment with
2297 increasing it even more.)
2298 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2299 However, almost all old code should still continue to work without
2300 immediate update, as SB-C-CALL is now a (deprecated) nickname
2302 ** Old operator names in the style DEF-FOO are now deprecated in
2303 favor of new corresponding names DEFINE-FOO, for consistency with
2304 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2305 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2306 internal symbols, but a few supported extensions like
2307 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2308 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2309 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2310 etc. as you get deeper into recursive calls to the debugger
2311 command loop, instead of the old "5]", "5]]", "5]]]"
2312 sequence. (I was motivated to do this when squabbles between
2313 ILISP and SBCL left me very deeply nested in the debugger. In the
2314 short term, this change will probably provoke more ILISP/SBCL
2315 squabbles, but hopefully it will be an improvement in the long run.)
2316 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2317 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2318 function, since declaiming return types involving aliens is
2319 (1) annoyingly messy to do by hand and (2) vital to efficient
2320 compilation of code which calls such functions.
2321 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2322 longer reexported by the SB-EXT package. They're solely useful
2323 for alien code, so it seems more logical that you should get
2324 them from the SB-ALIEN package, not in SB-EXT.
2325 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2326 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2327 features. Instead, the code that they used to control is always
2328 built into the system.
2329 * many other bug fixes
2330 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2331 advantage of the new EVAL-WHEN stuff and to clean them up in
2332 general, and they are now more ANSI-compliant in a number of
2333 ways. Martin Atzmueller is responsible for a lot of this.
2334 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2336 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2337 *** correct ERROR type for various file operations
2338 *** some fixes for Lisp streams
2339 *** DEFMETHOD syntax checking
2340 *** changing old weird representation of debug information as
2341 strings (which, among their other deficiencies, don't transform
2342 correctly when you rename packages, and don't change their
2343 print representation when you change things like *PACKAGE*
2344 and *PRINT-LENGTH*) to symbols and lists of symbols
2345 He also made several improvements and fixed several bugs in DESCRIBE.
2346 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2348 *** misbehavior of WRITE-STRING/WRITE-LINE
2349 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2350 and several other LOOP problems as well
2351 *** DIRECTORY when similar filenames are present
2352 *** DEFGENERIC with :METHOD options
2353 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2354 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2355 *** argument ordering in FIND with :TEST option
2356 *** mishandled package designator argument in APROPOS-LIST
2357 *** various problems in the backquote readmacro
2358 *** a bug in APROPOS
2359 *** probably some others that I'm not describing very well here,
2360 since the CVS log documents them by reference to sbcl-devel
2361 messages, and the SourceForge archives aren't working well.:-(
2362 ** Dan Barlow improved the Alpha port (and is making progress on the
2363 PPC port, for those of you who think different).
2364 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2365 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2366 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2367 continues to work on the SPARC port (for those of you in a position
2368 to look down upon our little PC-compatible boxes from a great height).
2369 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2370 and uses that copy, rather than the current dynamic value, when
2371 it's trying to decide whether to truncate output. Thus e.g.
2372 (let ((*print-lines* 50))
2373 (pprint-logical-block (stream nil)
2375 (let ((*print-lines* 8))
2376 (print (aref possiblybigthings i) stream)))))
2377 should now truncate the logical block only at 50 lines, instead of
2378 often truncating it at 8 lines, as it did before.
2379 * The doc/cmucl/ directory, containing old CMU CL documentation
2380 from the time of the fork, is no longer part of the base system.
2381 SourceForge has shut down its anonymous FTP service, and with it
2382 my original plan for distributing the old CMU CL documentation
2383 there. For now, if you need these files you can download an old
2384 SBCL source release and extract them from it.
2385 * The fasl file version number changed again, for dozens of reasons,
2386 some of which are apparent above.
2388 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2389 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2390 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2391 patch, so that DISASSEMBLE output is much nicer.
2392 * The code in the SB-PROFILE package now seems reasonably stable.
2393 I still haven't decided what the final interface should look like
2394 (I'd like PROFILE to interact cleanly with TRACE, since both
2395 facilities use function encapsulation) but if you have a need
2396 for profiling now, you can probably use it successfully with
2397 the current CMU-CL-style interface.
2398 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2399 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2400 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2401 link now returns the dangling link itself, and for similar
2402 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2403 link itself. (In these cases the old code signalled an error and
2404 looped endlessly, respectively.) Thus, DIRECTORY now works even
2405 in the presence of dangling and cyclic symbolic links.
2406 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2407 is now a supported extension again, since the consensus on
2408 sbcl-devel was that it can be useful for ordinary development
2409 work, not just for debugging SBCL itself.
2410 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2411 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2412 that currently-defined functions might be redefined later with
2413 different return types.
2414 * Hash tables can be printed readably, as inspired by CMU CL code
2415 of Eric Marsden and SBCL code of Martin Atzmueller.
2416 * better error handling in CLOS method combination, thanks to
2417 Martin Atzmueller porting Pierre Mai's CMU CL patches
2418 * more overflow fixes for >16Mbyte I/O buffers
2419 * A bug in READ has been fixed, so that now a single Ctrl-D
2420 character suffices to cause end-of-file on character streams.
2421 In particular, now you only need one Ctrl-D at the command
2422 line (not two) to exit SBCL.
2423 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2425 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2426 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2427 does the right thing.
2428 * The implementation of some type tests, especially for CONDITION
2429 types, is now tidier and maybe faster, due to CMU CL code
2430 originally by Douglas Crosher, ported by Martin Atzmueller.
2431 * Some math functions have been fixed, and there are new
2432 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2433 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2434 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2435 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2436 likely to be built into sbcl-0.7.0.)
2437 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2438 CL, ported to SBCL by Martin Atzmueller)
2439 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2440 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2441 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2442 * There's a new slam.sh hack to shorten the edit/compile/debug
2443 cycle for low-level changes to SBCL itself, and a new
2444 :SB-AFTER-XC-CORE target feature to control the generation of
2445 the after-xc.core file needed by slam.sh.
2446 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2447 COMPILE-FILE is no longer supported, so that now every function
2448 gets an entry point, so that block compilation looks a little
2449 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2450 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2451 returns the number of bytes consed since the system started,
2452 rather than the number consed since the first time the function
2453 was called. (The new definition parallels ANSI functions like
2454 CL:GET-INTERNAL-RUN-TIME.)
2455 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2456 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2457 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2458 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2459 * Fasl file version numbers are now independent of the target CPU,
2460 since historically most system changes which required version
2461 number changes have affected all CPUs equally. Similarly,
2462 the byte fasl file version is now equal to the ordinary
2465 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2466 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2467 is no longer recognized. I apologize for this, because it was
2468 listed in SB-EXT as a supported extension, but I found that
2469 its existing behavior was poorly specified, as well as incorrectly
2470 specified, and it looked like too much of a mess to straighten it
2471 out. I have enough on my hands trying to get ANSI stuff to work..
2472 * many patches ported from CMU CL by Martin Atzmueller, with
2473 half a dozen bug fixes in pretty-printing and the debugger, and
2474 half a dozen others elsewhere
2475 * fixed bug 13: Floating point infinities are now supported again.
2476 They might still be a little bit flaky, but thanks to bug reports
2477 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2478 as flaky as they were.
2479 * The --noprogrammer command line option is now supported. (Its
2480 behavior is slightly different in detail from what the old man
2481 page claimed it would do, but it's still appropriate under the
2482 same circumstances that the man page talks about.)
2483 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2484 are now supported, and enabled by default. Thus, the compiler can
2485 handle many floating point and complex operations much less
2486 inefficiently. (Thus e.g. you can implement a complex FFT
2488 * The compiler now detects type mismatches between DECLAIM FTYPE
2489 and DEFUN better, and implements CHECK-TYPE more correctly, and
2490 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2491 thanks to patches from Martin Atzmueller.
2492 * various fixes to make the cross-compiler more portable to
2493 ANSI-conforming-but-different cross-compilation hosts (notably
2494 Lispworks for Windows, following bug reports from Arthur Lemmens)
2495 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2496 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2497 fixed thanks to Pierre Mai's CMU CL patches.
2498 * fixes to keep the system from overflowing internal counters when
2499 it tries to use i/o buffers larger than 16M bytes
2500 * fixed bug 45a: Various internal functions required to support
2501 complex special functions have been merged from CMU CL sources.
2502 (When I was first setting up SBCL, I misunderstood a compile-time
2503 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2504 * improved support for type intersection and union, fixing bug 12
2505 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2506 more obscure bugs as well
2507 * some steps toward byte-compiling non-performance-critical
2508 parts of the system, courtesy of patches from Martin Atzmueller
2509 * Christophe Rhodes has made some debian packages of sbcl at
2510 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2511 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2512 stable, but are nonetheless usable. When he's ready, I'd be happy
2513 to add them to the SourceForge "File Releases" section. (And if
2514 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2515 * new fasl file format version number (because of changes in
2516 internal representation of (OR ..) types to accommodate the new
2517 support for (AND ..) types, among other things)
2519 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2520 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2522 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2524 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2525 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2526 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2527 * DESCRIBE now works on structure objects again.
2528 * Most function call argument type mismatches are now handled as
2529 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2530 know whether the function will be redefined before the call is
2531 executed. (The compiler could flag local calls with full WARNINGs,
2532 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2533 it doesn't keep track of enough information to know whether calls
2534 are local in this sense.)
2535 * Compiler output is now more verbose, with messages truncated
2536 later than before. (There should be some supported way for users
2537 to override the default verbosity, but I haven't decided how to
2538 provide it yet, so this behavior is still controlled by the internal
2539 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2540 src/compiler/ir1util.lisp.)
2541 * Fasl file format version numbers have increased again, because
2542 support for the Gray streams extension changes the layout of the
2543 system's STREAM objects.
2544 * The Gray subclassable streams extension now works, thanks to a
2545 patch from Martin Atzmueller.
2546 * The full LOAD-FOREIGN extension (not just the primitive
2547 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2548 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2549 but like most other programs, it defaults to copying the Unix
2550 environment from the original process instead of starting the
2551 new process in an empty environment.
2552 * Extensions which manipulate the Unix environment now support
2553 an :ENVIRONMENT keyword option which doesn't smash case or
2554 do other bad things. The CMU-CL-style :ENV option is retained
2555 for porting convenience.
2556 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2557 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2559 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2561 * A patch from Martin Atzmueller seems to have solved the SIGINT
2562 problem, and as far as we know, signal-handling now works cleanly.
2563 (If you find any new bugs, please report them!)
2564 * The system no longer defaults Lisp source file names to types
2565 ".l", ".cl", or ".lsp", but only to ".lisp".
2566 * The compiler no longer uses special default file extensions for
2567 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2568 seems to expect a single default extension for all compiled code,
2569 and there's no compelling reason to try to stretch the standard
2570 to allow two different extensions.) Instead, byte-compiled files
2571 default to the same extension as native-compiled files.
2572 * Fasl file format version numbers have increased again, because
2573 a rearrangement of internal implementation packages made some
2574 dumped symbols in old fasl files unreadable in new cores.
2575 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2576 many fewer weird special cases.
2577 * Bug #17 (differing COMPILE-FILE behavior between logical and
2578 physical pathnames) has been fixed, and some related misbehavior too,
2579 thanks to a patch from Martin Atzmueller.
2580 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2581 by Tim Moore, ported to SBCL by Martin Atzmueller.
2582 * Martin Atzmueller fixed several filesystem-related problems,
2583 including bug #36, in part by porting CMU CL patches, which were
2584 written in part by Paul Werkowski.
2585 * More compiler warnings in src/runtime/ are gone, thanks to
2586 more patches from Martin Atzmueller.
2587 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2590 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2592 * DESCRIBE now works on CONDITION objects.
2593 * The debugger now handles errors which arise when trying to print
2594 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2596 * The build system now uses an additional file, customize-target-features.lisp,
2597 to allow local modifications to the target *FEATURES* list. (The point of
2598 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2599 debugging features enabled, without having to worry about propagating it
2600 into everyone's system when I do a "cvs update".) When no
2601 customize-target-features.lisp file exists, the target *FEATURES* list
2602 should be constructed the same way as before.
2603 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2604 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2605 warn about behavior which is undefined under the ANSI spec. Note: This
2606 is specified by ANSI, but it's not very popular with programmers.
2607 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2608 macro in the SBCL sources for an example of a workaround which you
2609 might use to make portable ANSI-standard code which does what you want.
2610 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2611 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2612 when it's used as a non-toplevel form. (This is required in order
2613 to implement the DEFCONSTANT-EQX macro.)
2614 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2615 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2616 funny things, due to bugs in the implementation of EVAL-WHEN
2617 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2618 non-toplevel DEFCONSTANTs are uncommon.)
2619 * The core file version number and fasl file version number have been
2620 incremented, because the old noncompliant DEFCONSTANT behavior involved
2621 calling functions which no longer exist, and because I also took the
2622 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2623 * fixed bug 1 (error handling before read-eval-print loop starts), and
2624 redid debugger restarts and related debugger commands somewhat while
2626 ** The QUIT debugger command is gone, since it did something
2627 rather different than the SB-EXT:QUIT command, and since it never
2628 worked properly outside the main toplevel read/eval/print loop.
2629 Invoking the new TOPLEVEL restart provides the same functionality.
2630 ** The GO debugger command is also gone, since you can just invoke
2631 the CONTINUE restart directly instead.
2632 ** The TOP debugger command is also gone, since it's redundant with the
2633 FRAME 0 command, and since it interfered with abbreviations for the
2635 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2637 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2638 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2639 some porting work by Martin Atzmueller)
2640 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2641 with a boatload of patches which he ported to SBCL. Now that those
2643 ** The system tries to make sure that its low-priority messages
2644 are prefixed by semicolons, to help people who like to use
2645 syntax highlighting in their ILISP buffer. (This patch
2646 was originally due to Raymond Toy.)
2647 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2648 patches originally written by Raymond Toy.
2649 ** The compiler understands coercion between single-value and
2650 multiple-VALUES type expressions better, getting rid of some very
2651 weird behavior, thanks to patches originally by Robert MacLachlan
2652 and Douglas Crosher.
2653 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2654 lambda lists, thanks to a patch originally by Pierre Mai.
2655 ** The system no longer bogusly warns about "abbreviated type
2657 ** The compiler gets less confused by inlining and RETURN-FROM,
2658 thanks to some patches originally by Tim Moore.
2659 ** The system no longer hangs when dumping circular lists to fasl
2660 files, thanks to a patch originally from Douglas Crosher.
2661 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2662 undefined function error.
2663 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2664 in structure slot initforms) for now, documented workaround instead:-|
2665 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2666 * fixed bug 5: added stubs for various Gray stream functions called
2667 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2668 installed, at least appropriate type errors are generated
2669 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2670 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2671 consistently in DEFMETHOD forms.
2672 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2673 it doesn't seem to affect SBCL after all
2674 * The C runtime system now builds with better optimization and many
2675 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2677 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2679 * The system is now under CVS at SourceForge (instead of the
2680 CVS repository on my home machine).
2681 * The new signal handling code has been tweaked to treat register
2682 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2683 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2684 downstream, e.g. in debug-int.lisp, has implicit dependencies
2685 on the unsignedness of integer representation of machine words,
2686 and that caused the system to bomb out with infinite regress
2687 when trying to recover from type errors involving signed values,
2688 e.g. (BUTLAST '(1 2 3) -1).)
2689 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2690 (This was one of the bugs Peter Van Eynde reported back in July.)
2691 * The system now uses code inspired by Colin Walters' O(N)
2692 implementation of MAP (from the cmucl-imp@cons.org mailing
2693 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2694 inline the MAP operation, and there is more than one
2695 sequence argument to the MAP call (so that it can't just
2696 do ETYPECASE once and for all based on the type of the
2697 single sequence argument). (The old non-inline implementation
2698 of the general M-argument sequence-of-length-N case required
2699 O(M*N*N) time when any of the sequence arguments were LISTs.)
2700 * The QUIT :UNIX-CODE keyword argument has been renamed to
2701 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2702 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2704 * Raymond Wiker's patches to port dynamic loading from Linux to
2705 FreeBSD have been added.
2706 * The BUGS file is now more nearly up to date, thanks in large part
2707 to Martin Atzmueller's review of it.
2708 * The debugger now flushes standard output streams before it begins
2709 its output ("debugger invoked" and so forth).
2710 * The core version number and fasl file version number have both
2711 been incremented, because of incompatible changes in the layout
2713 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2714 used to not be called for a saved Lisp image.)
2715 * Martin Atzmueller's version of a patch to fix a compiler crash,
2716 as posted on sbcl-devel 13 September 2000, has been installed.
2717 * Instead of installing Martin Atzmueller's patch for the
2718 compiler transform for SUBSEQ, I deleted the compiler transform,
2719 and transforms for some similar consing operations.
2720 * A bug in signal handling which kept TRACE from working on OpenBSD
2722 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2723 away by constant folding
2724 * The system now defines its address space constants in one place
2725 (in the Lisp sources), and propagates them automatically elsewhere
2726 (through GENESIS and the sbcl.h file). Therefore, patching the
2727 address map is less unnecessarily tedious and error-prone. The
2728 Lisp names of address space constants have also been systematized.
2729 * CVS tags like dollar-Header-dollar have been removed from
2730 the sources, because they have never saved me trouble and
2731 they've been source of trouble working with patches and other
2732 diff-related operations.
2733 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2734 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2736 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2738 * The system has been ported to OpenBSD.
2739 * The system now compiles with a simple "sh make.sh" on the systems
2740 that it's supported on. I.e., now you no longer need to tweak
2741 text in the target-features.lisp-expr and symlinks in src/runtime/
2742 by hand, the make.sh takes care of it for you.
2743 * The system is no longer so grossly inefficient when compiling code
2744 involving vectors implemented as general (not simple) vectors (VECTOR T),
2745 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2746 which dares to use the various sequence functions on non-simple
2747 vectors, takes less of a performance hit.
2748 * There is now a primitive type predicate VECTOR-T-P
2749 to test for the (VECTOR T) type, so that e.g.
2750 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2751 can now be compiled with some semblance of efficiency. (The old code
2752 turned the type declaration into a full call to %TYPEP at runtime!)
2753 * AREF on (VECTOR T) is still not fast, since it's still compiled
2754 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2755 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2756 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2757 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2758 without making full calls to SUBTYPEP at runtime.
2759 (Some analogous efficiency issues for non-simple vectors specialized to
2760 element types other than T, or for non-simple multidimensional arrays,
2761 have not been addressed. They could almost certainly be handled the
2762 same way if anyone is motivated to do so.)
2763 * The changes in array handling break binary compatibility, so
2764 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2765 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2766 instead of (VALUES T T).
2767 * By following the instructions that Dan Barlow posted to sbcl-devel
2768 on 2 July 2000, I was able to enable primitive dynamic object
2769 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2770 functionality is not implemented (since it calls ld to resolve
2771 library references automatically, requiring RUN-PROGRAM for its
2772 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2773 try to resolve library references) is now supported.
2774 * The system now flushes the standard output streams when it terminates,
2775 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2776 them at several other probably-convenient times, e.g. in each pass of
2777 the toplevel read-eval-print loop, and after evaluating a form given
2778 as an "--eval" command-line option. (These changes were motivated by a
2779 discussion of stream flushing issues on cmucl-imp in August 2000.)
2780 * The source transform for TYPEP of array types no longer assumes
2781 that an array whose element type is a not-yet-defined type
2782 is implemented as an array of T, but instead punts, so that the
2783 type will be interpreted at runtime.
2784 * There is now some support for cross-compiling in make.sh: each of
2785 the phases of make.sh has its own script. (This should be transparent
2786 to people doing ordinary, non-cross-compile builds.)
2787 * Since my laptop doesn't have hundreds of megabytes of memory like
2788 my desktop machine, I became more motivated to do some items on
2789 my to-do list in order to reduce the size of the system a little:
2790 ** Arrange for various needed-only-at-cold-init things to be
2791 uninterned after cold init. To support this, those things have
2792 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2793 symbols with such names are now uninterned after cold init).
2794 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2795 fasl files for cold load.
2796 ** Remove the old compiler structure pooling code (which used to
2797 be conditional on the target feature :SB-ALLOC) completely.
2798 ** Redo the representation of some data in cold init to be more compact.
2799 (I also looked into supporting byte compiled code at bootstrap time,
2800 which would probably reduce the size of the system a lot, but that
2801 looked too complicated, so I punted for now.)
2802 * The maximum signal nesting depth in the src/runtime/ support code has
2803 been reduced from 4096 to 256. (I don't know any reason for the very
2804 large old value. If the new smaller value turns out to break something,
2805 I'll probably just bump it back up.)
2806 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2808 * Many, many bugs reported by Peter Van Eynde have been added to
2809 the BUGS list; some have even been fixed.
2810 * While enabling dynamic object file loading, I tried to make the
2811 code easier to understand, renaming various functions and variables
2812 with less ambiguous names, and changing some function calling
2813 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2814 * While trying to figure out how to do the OpenBSD port, I tried to
2815 clean up some of the code in src/runtime/. In particular, I dropped
2816 support for non-POSIX signal handling, added various comments,
2817 tweaked the code to reduce the number of compilation warnings, and
2818 renamed some files to increase consistency.
2819 * To support the new automatic configuration functionality in make.sh,
2820 the source file target-features.lisp-expr has been replaced with the
2821 source file base-target-features.lisp-expr and the machine-generated
2822 file local-target-features.lisp-expr.
2823 * fixed a stupid quoting error in make.sh so that using CMU CL
2824 "lisp -batch" as cross-compilation host works again
2826 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2828 * DESCRIBE no longer tries to call itself recursively to describe
2829 bound/fbound values, so that it no longer fails on symbols which are
2830 bound to themselves (like keywords, T, and NIL).
2831 * DESCRIBE now works on generic functions.
2832 * The printer now prints less-screwed-up representations of closures
2833 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2834 * A private symbol is used instead of the :EMPTY keyword previously
2835 used to mark empty slots in hash tables. Thus
2836 (DEFVAR *HT* (MAKE-HASH-TABLE))
2837 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2838 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2839 now does what ANSI says that it should. (You can still get
2840 similar noncompliant behavior if bang on the hash table
2841 implementation with all the symbols you get back from
2842 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2843 This breaks binary compatibility, since tests for equality to
2844 :EMPTY are wired into things like the macroexpansion of
2845 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2847 * There's now a minimal placeholder implementation for CL:STEP,
2848 as required by ANSI.
2849 * An obscure bug in the interaction of the normal compiler, the byte
2850 compiler, inlining, and structure predicates has been patched
2851 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2852 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2854 * Missing ordinary arguments in a macro call are now detected even
2855 when the macro lambda list contains &KEY or &REST.
2856 * The debugger no longer complains about encountering the top of the
2857 stack when you type "FRAME 0" to explicitly instruct it to go to
2858 the top of the stack. And it now prints the frame you request even
2859 if it's the current frame (instead of saying "You are here.").
2860 * As specified by ANSI, the system now always prints keywords
2861 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2863 * The default initial SIZE of HASH-TABLEs is now smaller.
2864 * Type information from CLOS class dispatch is now propagated
2865 into DEFMETHOD bodies, so that e.g.
2866 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2868 is now basically equivalent to
2869 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2870 (DECLARE (TYPE SINGLE-FLOAT X))
2872 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2873 operation, without having to do run-time type dispatch.
2874 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2875 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2876 * Since I don't seem to be making big file reorganizations very often
2877 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2878 is overflowing my ability to conveniently back them up), I've finally
2879 checked the system into CVS. (The CVS repository is on my home system,
2880 not at SourceForge -- putting it on SourceForge might come later.)
2881 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2882 high-level GC-NOTIFY-FOO functions send their output. (There's
2883 still very little control of where low-level verbose GC functions
2884 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2885 less than it used to -- the GC-NOTIFY-FOO functions are now under
2886 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2887 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2888 in only one place in the source code, and propagates it automatically
2889 everywhere that it's needed. Thus e.g. when I bump the version from
2890 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2891 * The C source files now include boilerplate legalese and documentation
2892 at the head of each file (just as the Lisp source files already did).
2893 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2894 to his page will be replaced with a link to his new CLiki service.
2896 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2898 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2899 * The build process now looks for GNU make under the default name "gmake",
2900 instead of "make" as it used to. If GNU make is not available as "gmake"
2901 on your system, you can change this default behavior by setting the
2902 GNUMAKE environment variable.
2903 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2904 can build without error under CMU CL.
2906 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2908 * There is now a partial SBCL user manual (with some new text and some
2909 text cribbed from the CMU CL manual).
2910 * The beginnings of a profiler have been added (starting with the
2911 CMU CL profiler and simplifying and cleaning up). Eventually the
2912 main interface should be through the TRACE macro, but for now,
2913 it's still accessed through vaguely CMU-CL-style functions and macros
2914 exported from the package SB-PROFILE.
2915 * Some problems left over from porting CMU CL to the new
2916 cross-compilation bootstrap process have been cleaned up:
2917 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2918 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2919 part when getting rid of the compiler *BACKEND* stuff.)
2920 ** The value of *NULL-TYPE* was screwed up, because it was
2921 being initialized before the type system knew the final
2922 definition of the 'NULL type. This screwed up several key
2923 optimizations in the compiler, causing inefficiency in all sorts
2924 of places. (I found it because I wanted to understand why
2925 GET-INTERNAL-RUN-TIME was consing.)
2926 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2927 PROCLAIM FTYPE information. Unfortunately this broke binary
2928 compatibility again, since now the forms output by DEFGENERIC
2929 to refer to functions which didn't exist in 0.6.3.
2930 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2931 can use the new (as of 0.6.3) transform for SOME into MAP into
2933 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2934 values to (INTEGER 0 1000000), so that the time code will no longer
2935 occasionally get blown up by Linux returning 1000000 microseconds
2936 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2937 its output conform to the ANSI spec. (Alas, this makes its output
2938 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2939 * A full call to MAP NIL with a single sequence argument no longer conses.
2940 * fixes to problems pointed out by Martin Atzmueller:
2941 * The manual page no longer talks about multiprocessing as though
2942 it were currently supported.
2943 * The ILISP support patches have been removed from the distribution,
2944 because as of version 5.10.1, ILISP now supports SBCL without us
2945 having to maintain patches.
2946 * added a modified version of Raymond Toy's recent CMU CL patch for
2947 EQUALP comparison of HASH-TABLE
2949 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2951 * The system still can't cross-compile itself with
2952 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2953 entails), but at least it can compile more of itself that way
2954 than it used to be able to, and various buglets which were uncovered
2955 by trying to cross-compile itself that way have now been fixed.
2956 * This release breaks binary compatibility again. This time
2957 at least I've incremented the FASL file format version to 2, so that the
2958 problem can be detected reliably instead of just causing weird errors.
2959 * various new style warnings:
2960 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2961 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2962 ** using the lexical binding of a variable named in the *FOO* style
2963 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2964 as specified by ANSI.
2965 * *RANDOM-STATE* is no longer automatically initialized from
2966 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2967 default behavior of the system is to repeat its behavior every time
2968 it's run. If you'd like to change this behavior, you can always
2969 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2970 old convention there was no comparably easy way to get the system to
2971 repeat its behavior every time it was run.
2972 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2973 FTYPE declarations has been removed, in favor of their ANSI
2974 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2975 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2976 the types of their sequence arguments can be determined at compile time.
2977 This is done through a new open code expansion for MAP which eliminates
2978 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2979 argument types can be determined at compile time.
2980 * The optimizer now transforms COERCE into an identity operation when it
2981 can prove that the coerced object is already of the correct type. (This
2982 can be a win for machine generated code, including the output of other
2983 optimization transforms, such as the MAP transform above.)
2984 * Credit information has been moved from source file headers into CREDITS.
2985 * Source file headers have been made more standard.
2986 * The CASE macro now compiles without complaining even when it has
2989 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2991 * (Note that the way that the PCL macroexpansions were rewritten
2992 to accommodate the change in DEFGENERIC below breaks binary
2993 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2994 not run under sbcl-0.6.2. Once we get out of alpha releases,
2995 i.e. hit release 1.0.0, we'll probably try to maintain binary
2996 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2997 and sbcl-1.4.4. Until then, however, it might be fairly common
2998 for maintenance releases to break binary compatibility.)
2999 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3001 * The reporting of unhandled signals has been changed to print some
3002 explanatory text as well as the report form. (Previously only
3003 the report form was printed.)
3004 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3005 it defines, so that the compiler no longer issues undefined function
3006 warnings for compiled-but-not-yet-loaded generic functions.
3007 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3008 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3009 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3010 section of the ANSI spec are supported.
3011 * The "" nickname for the "KEYWORD" package has been removed.
3012 The reader still handles symbol tokens which begin with a package marker
3013 as keywords, but it doesn't expose its mechanism for doing so in the
3014 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3015 * The system now issues STYLE-WARNINGs for contradictory TYPE
3016 proclamations. (Warnings for contradictory FTYPE proclamations would
3017 be nice too, but those can't be done usefully unless the type system
3018 is made smarter about FUNCTION types.)
3019 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3020 systematized, so that "*target-*.lisp is supposed to exist only on the
3021 target and imply that there's a related file which exists on the
3022 host, and *host-*.lisp is supposed to exist only on the host and imply
3023 that there's a related file which exists on the target. This involves a
3024 lot of renaming. Hopefully the acute confusion caused by the renaming
3025 will be justified by the reduction in chronic confusion..
3026 ** runtime-type.lisp -> early-target-type.lisp
3027 ** target-type.lisp -> late-target-type.lisp
3028 ** early-host-format.lisp -> early-format.lisp
3029 ** late-host-format.lisp -> late-format.lisp
3030 ** host-error.lisp -> misc-error.lisp
3031 ** early-error.lisp -> early-target-error.lisp
3032 ** late-error.lisp -> late-target-error.lisp
3033 ** host-defboot.lisp -> early-defboot.lisp
3034 ** code/misc.lisp -> code/target-misc.lisp
3035 ** code/host-misc.lisp -> code/misc.lisp
3036 ** code/numbers.lisp -> code/target-numbers.lisp
3037 ** code/early-numbers.lisp -> numbers.lisp
3038 ** early-host-type.lisp -> early-type.lisp
3039 ** late-host-type.lisp -> late-type.lisp
3040 ** host-typep.lisp -> typep.lisp
3041 ** load.lisp -> target-load.lisp
3042 ** host-load.lisp -> load.lisp
3043 ** host-disassem.lisp -> disassem.lisp
3044 ** host-insts.lisp -> insts.lisp
3045 ** byte-comp.lisp -> target-byte-comp.lisp
3046 ** host-byte-comp.lisp -> byte-comp.lisp
3047 ** host-signal.lisp -> signal.lisp
3048 ** host-defstruct.lisp -> defstruct.lisp
3049 ** late-target-type.lisp -> deftypes-for-target.lisp
3050 Furthermore, several other previously target-only files foo.lisp (e.g.
3051 hash-table.lisp and random.lisp) have been split into a target-and-host
3052 foo.lisp file and a target-only target-foo.lisp file, with their key type
3053 definitions in the target-and-host part, so that the cross-compiler will
3054 know more about target types.
3055 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3056 gone away. In their place are various *BACKEND-FOO* variables
3057 corresponding to the slots of the old structure.
3058 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3059 into the target SBCL has been fixed.
3060 * The chill.lisp system for loading cold code into a running SBCL
3062 * Support for the CMU CL "scavenger hook" extension has been removed.
3063 (It was undocumented and unused in the CMU CL sources that SBCL was
3064 derived from, and stale in sbcl-0.6.1.)
3065 * Various errors in the cross-compiler type system were detected
3066 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3067 (enabling various consistency checks). Many of them were fixed,
3068 but some hard problems remain, so the compiler is back to
3069 running without *TYPE-SYSTEM-INITIALIZED* for now.
3070 * As part of the cross-compiler type system cleanup, I implemented
3071 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3072 * I have started adding UNCROSS calls throughout the type system
3073 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3074 on cross-compiler input files will be able to go away, and various
3076 * CONSTANTP now returns true for quoted forms (as explicitly required
3079 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3081 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3082 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3083 (MAKE-LIST -1). In the longer term, it ought to become true
3084 that declarations are assertions even at SAFETY 1. For now, it's not
3085 quite true even at SAFETY 3, but it's at least more nearly true..
3086 (Note that this change seems to increases the size of the system by
3087 O(5%) and to decrease the speed of the compiler by 20% or more.)
3088 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3089 for the problem of printing dozens of lines of distracting information
3090 about low-level system machinery as part of the top stack frame
3091 on entry to the debugger when an undefined function was called.
3092 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3093 is not reset to COMMON-LISP-USER.
3094 * Compilation of stuff related to dyncount.lisp has been made conditional
3095 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3096 smaller. The various dyncount-related symbols have been moved into
3097 a new "SB-DYNCOUNT" package.
3098 * tty-inspect.lisp has been renamed to inspect.lisp.
3099 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3100 feature has gone away. (When we eventually port to other flavors of
3101 libc and/or Unix, we'll try to make the differences between flavors
3102 invisible at the user level.)
3103 * Various other *FEATURES* tags, and/or their associated conditionals,
3104 have been removed if obsolescent, or given better documentation, or
3105 sometimes given more-mnemonic names.
3107 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3109 * tidied up "make.sh" script
3110 * tidied up system directory structure
3111 * better "clean.sh" behavior
3112 * added doc/FOR-CMUCL-DEVELOPERS
3113 * many many small tweaks to output format, e.g. removing possibly-confusing
3114 trailing #\. character in DESCRIBE-INSTANCE
3115 * (EQUALP #\A 'A) no longer signals an error.
3116 * new hashing code, including EQUALP hashing
3117 * tidied up Lisp initialization and toplevel
3118 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3119 * command line argument processing
3120 * added POSIX-GETENV function to deal with Unix-ish environment variables
3121 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3122 terminating SBCL on EOF
3123 * non-verbose GC by default
3124 * There is no more "sbcl" shell script; the sbcl file is now the C
3125 runtime executable (just like CMU CL).
3126 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3128 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3129 symbol and package information as Lisp native symbol and package objects
3130 instead of strings naming symbols and strings naming packages. This way,
3131 whenever packages are renamed (as in warm init), debug information is
3132 transformed along with everything else.
3133 * tweaked the optimization policy declarations which control the building
3134 of SBCL itself. Now, among other things, the system no longer saves
3135 source location debugging information. (This helps two problems at once
3136 by reducing SBCL size and by keeping SBCL from trying to look for its
3137 sources -- which may not exist -- when reporting errors.)
3138 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3139 debugging and testing purposes
3140 * cleaned up printing, making the printer call PRINT-OBJECT for
3141 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3142 methods, giving nearly-ANSI behavior
3143 * converted almost all special variables to use *FOO* naming convention
3144 * deleted PARSE-TIME functionality, since it can be done portably
3145 * moved some files out of cold init into warm init
3146 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3148 * regularized formatting of source files
3149 * added an install.sh script
3150 * fixed ridiculous memory usage of cross-compiler by making
3151 compiler/alloc.lisp not try to do pooling unless it can hook
3152 itself into the GC of the cross-compilation host. Now the system
3153 builds nicely on my old laptop.
3154 * added :SB-ALLOC in target-features.lisp-expr
3155 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3156 was not implemented)
3157 * re-did condition handling and note reporting in the compiler. Notes
3158 are no longer handled by signalling conditions. Style warnings
3159 and warnings are handled more correctly and reported in such a way
3160 that it's easy to find one or the other in your output (so that you
3161 can e.g. figure out which of many problems caused COMPILE-FILE to
3163 * changed the severity of several compiler warnings from full WARNING
3164 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3165 compiler note reporting so that it doesn't use the condition system
3166 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3167 COMPILE-FILE command)
3168 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3169 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3170 no longer supported, since their implementation was deeply intertwingled
3171 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3172 * removed bogus "support" for compiler macros named (SETF FOO), and
3173 removed the compiler macro for SETF INFO (but only after making a fool
3174 of myself on the cmucl-imp mailing list by posting a bogus patch for
3175 DEFINE-COMPILER-MACRO..)
3176 * Compiled files containing forms which have side effects on the Lisp
3177 reader (such as DEFPACKAGE forms) are now handled more correctly.
3178 (Compiler queuing of top level lambdas has been suppressed by setting
3179 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3180 * deleted various currently-unused source files, e.g. gengc.lisp. They
3181 may be added back at some point e.g. when porting to other architectures,
3182 but until they are it's distracting to distribute them and to try to
3184 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3185 there were so many of them they're just distractions, and UNCROSS is
3186 known to be able to handle the current sources
3187 * moved PROFILE functionality into TRACE, so that it will be clear
3188 how the wrapping and unwrapping of functions when you profile them
3189 interacts with the wrapping and unwrapping of functions when you
3190 trace them. (Actually, the functionality isn't there yet, but at least
3191 the interface specification is there. Hopefully, the functionality will
3192 arrive with some future maintenance release.)
3193 * removed host-oops.lisp
3194 * changed signature of QUIT function to allow UNIX-CODE argument
3195 * fixed READ-SEQUENCE bug
3196 * tweaked verbose GC output so that it looks more like the progress
3197 output that ANSI specifies for functions like LOAD
3198 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3199 * added <http://sbcl.sourceforge.com> to the banner information printed by