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 works on x86-64.
20 (Thanks to James Knight)
21 * fixed some bugs revealed by Paul Dietz' test suite:
22 ** the type-error signalled from WARN has a filled-in DATUM slot.
23 ** the type-error required when a stream is not associated with
24 a file has the stream as its datum.
25 ** type-errors on single-floats on x86-64 no longer have
26 :INVALID-OBJECT as the datum
27 ** the type-errors signalled for invalid function names now have
28 a correct expected type
29 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
30 for structures defined with a :TYPE. Documentation strings for
31 typed structures are no longer immediately discarded
32 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
33 broadcast streams and synonym streams. FILE-LENGTH now also works
36 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
37 * incompatible change: the --noprogrammer option, deprecated since
38 version 0.7.5, has been removed. Please use the equivalent
39 --disable-debugger option instead.
40 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
42 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
43 inoperational for a while now) has been completely removed.
44 * null lexical environments are now printed as #<NULL-LEXENV>,
45 significantly reducing the amount of clutter in typical backtraces.
46 * documentation on weak pointers, finalization, and after GC hooks
47 has been added to the manual.
48 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
49 as other specialized array types, is much faster. SUBSEQ and
50 COPY-SEQ on such arrays have also been sped up.
51 * optimization: EQL is now more efficient when at least other argument
52 is known to be of type (OR FIXNUM (NOT NUMBER)).
53 * fixed bug: the runtime is now less vulnerable to changes in the
54 size of the SBCL object on OS X, and virtual memory is reserved for
55 all spaces that need to be at a fixed address.
56 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
57 * fixed bug: finalizers and after GC hooks that cause consing are now
59 * fixed bug: compiler error messages and summaries are now printed to
60 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
61 * fixed inference of the upper bound of an iteration variable.
62 (reported by Rajat Datta).
63 * fixed bug 211e: calling local functions with duplicated constant
64 keyword argument no longer causes a bogus style warning about an
66 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
67 ftype declaration to be disregarded. (reported by Dave Roberts)
68 * fixed bug 373: caused by erronous compilation of references to alien
69 variables in the runtime on ppc/darwin.
70 * fixed bug 376: CONJUGATE type deriver.
71 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
72 (reported by Baughn on #lisp)
73 * a cross-compiler bug on non-x86 platforms has been identified and
74 fixed. (thanks to Bruno Haible)
75 * improvements to the MIPS runtime code for increased stability.
76 (thanks to Thiemo Seufer)
77 * increased the maximimum compact environment size to allow
78 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
79 * partial workaround for bug 135: don't name the function we're
80 calling for hairy cases of EVAL, so as not to accumulate one environment
81 entry per call to EVAL. (reported by Kevin Reid)
82 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
83 * optimization: type testing for non-vector arrays should be faster.
84 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
86 * bug fix: setting 31st element of a bit vector to zero did not work
88 * bug fix: redefining a class definition which failed due to a
89 previous accessor / function clash now works (but see BUGS entry
90 #380 for more problems in this area). (thanks to Zach Beane)
91 * the long form of DEFINE-METHOD-COMBINATION disables method group
92 checking when given a single method group with pattern *. (thanks
94 * on x86 compiler supports stack allocation of results of simple
95 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
96 * support for building and running on Mac OS X 10.4 "Tiger" has been added
97 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
98 * fixed some bugs related to Unicode integration:
99 ** the restarts for recovering from input and output encoding
100 errors only appear when there is in fact such an error to
102 * fixed some bugs revealed by Paul Dietz' test suite:
103 ** MISC.549 and similar: late transformation of unsafe type
104 assertions into derived types caused unexpected code
106 ** SCALE-FLOAT type deriver is less wrong.
107 ** type derivers for EXP, LOG and similar functions compute result
108 types for complex arguments better.
109 ** (MISC.563) CONJUGATE type deriver works for very restricted
111 ** out-of-line type testers for character strings are available.
112 ** EQUAL compiler transform understands specialness of objects
114 ** accessing double-floats stored on the stack now works on x86-64.
115 ** debugger internals could sometimes create invalid lispobjs,
116 resulting in GC crashes.
117 ** MISC.548: type check weakening can convert required type into
119 ** initialization forms for bindings are not in scope of free special
122 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
123 * incompatible change: thread support for non-NPTL systems has
124 been removed - locking is buggy and unreliable. A threaded
125 SBCL build will now warn at startup and refuse to create
126 new threads, unless futex support is detected in the kernel
127 * incompatible change: the top level REPL now has only an ABORT
128 restart associated with it, not TOPLEVEL and ABORT as it used to.
129 TOP and TOPLEVEL are now available as debugger commands for
130 returning to the top level.
131 * incompatible change: forms evaluated in the REPL now use the
132 global optimization policy.
133 * incompatible change: user- and system-initialization files are
134 no longer processed with LOAD, but by READ and EVAL; hence the
135 global optimization policy, startup package, readtable, etc,
137 * The .fasl file format number has been incremented because of
138 various incompatible changes.
139 * internal entry point details and argument counts no longer appear
140 in backtraces unless explicitly requested by setting
141 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
142 * built-in and standard functions no longer have names like "top
143 level local call to FOO".
144 * fixed bug 32: functions defined in non-null lexical environments
145 now have more legible printed representation
146 * fixed bug 33: functions defined in non-null lexical environemnts
147 are now more amenable to inspection by INSPECT.
148 * workaround for bug 354: XEPs no longer appear in backtraces unless
149 explicitly requested.
150 * fixed bug: receiving the signal which results from attempting to
151 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
152 SIGBUS on other *BSDs) on architectures where the C stack is also
153 the Lisp stack (x86 and x86-64) from foreign code no longer leads
154 to debugger confusion or wild instability. (reported by Cheuksan
156 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
157 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
158 (reported by Lutz Euler)
159 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
160 compiled code. (reported by Paul Dietz)
161 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
162 specialized parameter to a method to something that is not TYPEP
163 the specializer is now possible.
164 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
165 face of package deletion.
166 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
167 pathname. (reported by David Tolpin on c.l.l)
168 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
169 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
170 STRING-STREAM and FILE-STREAM, respectively.
171 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
172 than previously; in addition, its test suite is now run on build.
173 (thanks to Luke Gorrie)
174 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
175 * optimization: added a immediate representation for single-floats
177 * fixed some bugs revealed by Paul Dietz' test suite:
178 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
179 ** attempting to create a package with a colliding nickname causes
180 correctable errors to be signalled.
181 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
182 ** several x86-64 backend bugs related to sign-extension of immediate
185 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
186 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
187 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
189 * fixed loading of multiply forward-referenced layouts.
190 (thanks to Cheuksan Wang)
191 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
192 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
193 (see COMPILE-FILE documentation for details of :PRINT option)
194 * fixed bugs 19 and 317: fixed-format floating point printing is
195 more accurate. This also fixes a bug reported by Adam Warner
196 related to the ~@F format directive.
197 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
199 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
200 dispatch table if the from-char is a dispatch macro character.
201 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
202 by the wordsize no longer ignore the last word. (reported by Lutz
204 * fixed bug in type checking of dynamic-extent variables. (reported
206 * optimization: sequence traversal functions use their freedom to
207 coerce function designators to functions.
208 * optimization: code with many calls to CLOS methods specialized on
209 CLOS classes has had redundant type checks removed.
210 * optimization: added declarations to speed up operations that access
211 the internal character database (for example STRING-UPCASE)
212 * optimization: comparison operations between floats and sufficiently small
213 fixnums no longer create extra rationals
214 * fixed some bugs related to Unicode integration:
215 ** portions of multibyte characters at the end of buffers for
216 character-based file input are correctly transferred to the
217 start of the buffer at the next read.
218 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
219 passing it through to OPEN.
220 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
221 argument given to internal calls to OPEN.
222 ** the built-in comment readers (introduced by character sequences
223 ";" and "#|") are more forgiving to encoding errors; they will
224 STYLE-WARN and then attempt to resync the stream at a character
225 boundary. (thanks to Teemu Kalvas)
226 * fixed some bugs in the x86-64 port:
227 ** Negative short int return values from c-calls are sign-extended
229 ** The stack is aligned to 16-bytes for c-calls, as required by
230 the ABI. (thanks to Cheuksan Wang)
231 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
232 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
233 for enums. (thanks to Vincent Arkesteijn)
234 ** Multiple small optimizations and bugfixes for floating point
236 * fixed some bugs revealed by Paul Dietz' test suite:
237 ** Space, Tab, Linefeed, Return and Page have the invalid
238 secondary constituent character trait.
239 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
241 ** WITH-INPUT-FROM-STRING should only update the index place on
243 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
245 ** Bit-array operations (BIT-AND and similar) worked incorrectly
246 with one-dimensional arrays with fill pointers.
247 ** TYPE-OF failed on a complex with an integer realpart and a
249 ** compiler failure during type inference for the code of form
250 (IF (EQL X (THE ...)) ...) (MISC.535).
252 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
253 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
254 platform. The port supports 61-bit fixnums, large memory spaces
255 and reloading shared object files.
256 * enhancement: saving cores with foreign code loaded is now
257 supported on ppc/Darwin in addition to the previously supported
259 * enhancement: the statistical profiler now walks deeper into the
260 call stack for more meaningful call-graphs and accrued time
261 reports (x86/x86-64 only). It also now reports time spent in
263 * enhancement: it is now possible to trace most individual methods
264 of a generic function in addition to tracing the generic function
266 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
268 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
269 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
270 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
271 produces an error. (thanks to Vincent Arkesteijn)
272 * bug fix: NAMESTRING on pathnames with :WILD components in their
273 directories works correctly. (thanks to Artem V. Andreev)
274 * fixed bug 125: compiler preserves identity of closures. (reported
276 * bug fixed: functions with &REST arguments sometimes failed with
277 "Undefined function" when compiled with (DEBUG 3). (reported by
279 * bug fix: overflow during compiling of setting element of a bit
280 vector with constant index and value. (reported by Timmy Douglas)
281 * build fix: fixed the dependence on *LOAD-PATHNAME* and
282 *COMPILE-FILE-PATHNAME* being absolute pathnames.
283 * on x86 compiler partially supports stack allocation of dynamic-extent
285 * GO and RETURN-FROM do not check the extent of their exit points
286 when compiled with SAFETY 0.
287 * fixed some bugs related to Unicode integration:
288 ** encoding and decoding errors are now much more robustly
289 handled; it should now be possible to recover even from invalid
290 input or output to the terminal. (thanks to Teemu Kalvas)
291 ** provided a first cut at implementing STRING-TO-OCTETS and
292 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
293 ** altered the SB-MD5 contributed module slightly, changing the
294 interface just enough for it to be supportable for builds where
295 lisp characters are not eight bits.
296 * fixed some bugs revealed by Paul Dietz' test suite:
297 ** the FORMATTER-generated functions for ~V[ conditionals require
298 the correct number of arguments.
299 ** READ-FROM-STRING returns the mandated second value when applied
300 to displaced strings.
301 ** the #\Rubout and #\Backspace characters are treated as invalid
302 constituent characters by the tokenizer.
304 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
305 * new feature: reloading changed shared object files with
306 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
307 * new feature: references to foreign variables and functions
308 can now be compiled and loaded before the corresponding shared
309 object file is loaded, as long as the foreign definitions are
310 available at runtime.
311 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
312 supported. (thanks to Dan Debertin)
313 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
314 just keywords. (thanks to Vincent Arkesteijn)
315 * fixed bug #331: structure-class instances corresponding to
316 DEFSTRUCT forms are now created eagerly.
317 * fixed bug #345: backtraces from calls to undefined functions work
318 on x86 as well. Related bug #61 is now also partially fixed on x86
319 (backtraces from throws to unknown catch tags.)
320 * bug fix: lambda-list parsing is now stricter vrt. order and number
321 of lambda-list keywords.
322 * bug fix: as specified by AMOP, an error is signalled if a
323 class-option appears multiple times in a DEFCLASS form. (reported
325 * bug fix: RANDOM can be compiled when the compiler derives the type
326 of its numeric argument as a disjoint set of small integers.
327 (reported by Paul Dietz)
328 * bug fix: starting a core saved with shared objects loaded when
329 those objects are not available no longer causes threaded SBCL to
330 hang. (reported by Sean Ross)
331 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
333 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
334 argument, which must always be NIL. (reported by Kalle Niemitalo)
335 * bug fix: printing 1.0d+23 no longer results in an error.
336 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
337 * bug fix: structure slot setters preserve evaluation order. (thanks
339 * fixed some bugs related to Unicode integration:
340 ** RUN-PROGRAM can allow its child to take input from a Lisp
341 stream. (reported by Stefan Scholl)
342 ** ASDF-INSTALL successfully downloads .tar.gz files in all
343 locales. (reported by Ken Causey)
344 * fixed some bugs revealed by Paul Dietz' test suite:
345 ** INCF, DECF and REMF evaluate their place form as specified in
347 ** FORMATTER expands ~{ ~} iteration directives with V or #
348 parameters correctly.
349 ** FORMATTER deals with the ~@[ ~] conditional directive where the
350 consequent uses no arguments correctly.
351 ** the system has a partial understanding of the (COMPLEX RATIO)
353 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
356 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
357 * new feature: a build-time option (controlled by the :SB-UNICODE
358 keyword feature, enabled by default) for building the system with
359 support for the entire 21-bit character space defined by the
361 * new feature: the system now has rudimentary external-format
362 support; the primary user-visible change at this time is that
363 characters with the high bit set (such as lower-case-e-acute) will
364 print correctly to a terminal in a UTF-8 environment.
365 * minor incompatible change: BASE-CHAR no longer names a class;
366 however, CHARACTER continues to do so, as required by ANSI.
367 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
368 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
369 has been moved to the SB-EXT package (temporarily re-exported from
371 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
372 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
374 * fixed bug #351: better error-handlind and reporting for malformed
376 * fixed bug #350: bignum-printing is now more memory-efficient,
377 allowing printing of very large bignums, eg. (expt 2 10000000).
378 (reported by Bruno Haible)
379 * fixed bug #302: better primitive-type selection for intersection
381 * fixed bug #308: non-graphic characters now all have names, as
382 required. (reported by Bruno Haible)
383 * bug fix: redefining a class with different superclasses now correctly
384 removes it from the direct-subclasses of its previous superclasses.
385 (reported by David Morse)
386 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
387 new value now works. (reported by Bruno Haible)
388 * bug fix: correct canonicalization of multiple non-standard slot
389 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
390 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
391 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
392 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
393 now exists, an signals an error.
394 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
395 during macroexpansion for non-iteration variables as well. (reported
396 by Bruno Haible for CMUCL)
397 * bug fix: Cyclic structures and unprintable objects in compiler
398 messages no longer cause errors. (reported by Bruno Haible)
399 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
400 and READ-FROM-STRING all now return a primary value of NIL if
401 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
402 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
403 (reported by Bruno Haible for CMUCL)
404 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
405 do not designate a condition. (reported by Bruno Haible for
407 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
408 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
409 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
410 permitting resolution in favour of any of the conflicting symbols.
411 (reported by Bruno Haible for CMUCL)
412 * bug fix: EQUAL compiler optimizations is less aggressive on
413 strings which can potentially compare true despite having distinct
414 specialized array element types.
415 * bug fix: unit enumerations can be defined without dividing by
416 zero. (thanks to Vincent Arkesteijn)
417 * FORMAT compile-time argument count checking has been enhanced.
418 (report from Bruno Haible for CMUCL)
419 * a partial workaround for the bug 262: the compiler does not try to
420 inline-expand a local function doing RETURN-FROM from a deleted
421 BLOCK. (thanks to Peter Denno for the bug report and to David
422 Wragg for the simple test case)
423 * fixed some bugs revealed by Paul Dietz' test suite:
424 ** RENAME-PACKAGE allows all package designators as new package
426 ** constraint propagation and lambda variable substitution are
427 more cautious in dealing with partially deleted code.
428 ** compiler sometimes lost reoptimization passes.
429 ** CERROR, given a condition as condition designator, passes its
430 remaining arguments to the continue format control without
432 ** Case-altering FORMAT directives work correctly on non-ASCII
434 ** The REAL type specifier handles bounds outside the floating
435 point ranges without signalling FLOATING-POINT-OVERFLOW.
436 ** Functions with IR1-transformations can create intercomponent
437 references to global functions.
438 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
440 ** FORMAT ~R treats a nil value for its first parameter correctly.
442 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
443 * enhancement: saving cores with foreign code loaded is now
444 supported on x86/NetBSD and sparc/Linux in addition to the previously
446 * bug fix: on some platforms repeated installations caused multiple
447 copies of HTML documentation to be installed -- should not happen
448 any more. (reported by Stefan Scholl)
449 * bug fix: parsing self-recursive alien record types multiple times
450 no longer causes infinite recursion. (reported by Thomas F. Burdick,
451 original patch by Helmut Eller for CMUCL)
452 * bug fix: stack-exhaustion detection works now on NetBSD as well.
453 (thanks to Richard Kreuter)
454 * bug fix: defining classes whose accessors are methods on existing
455 generic functions in other (locked) packages no longer signals
456 bogus package lock violations. (reported by François-René Rideau)
457 * bug fix: special variables as DEFMETHOD parameters no longer have
458 associated bogus type declarations. (reported by David Wragg and
460 * bug fix: read-write consistency on streams of element-type
461 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
462 * bug fix: redefiniton of the only method of a generic function with
463 no DEFGENERIC no longer emits a full WARNING. In addition,
464 redefinition of generic functions with no DEFGENERIC to an
465 incompatible lambda list now signals an error. (thanks to Zach
467 * bug fix: DEFGENERIC now works even when there's a function of the
468 same name in an enclosing lexical environment. (thanks to Zach
470 * fixed compiler failure, caused by instrumenting code during
471 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
472 * optimization: added loop analysis and improved register allocation
473 * fixed some bugs revealed by Paul Dietz' test suite:
474 ** POSITION on displaced vectors with non-zero displacement
475 returns the right answer.
476 ** (SIMPLE-STRING) is a valid type specifier for sequence
478 ** *PRINT-LEVEL* handling for slotless structures is pedantically
480 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
481 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
483 ** The justification version of the FORMAT ~< directive treats
484 non-zero minpad parameter correctly.
486 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
487 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
488 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
489 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
490 the supported interface.
491 * new feature: Single-stepping of code compiled with DEBUG 2 or
492 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
493 * new feature: saving cores with foreign code loaded is now
494 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
495 Timothy Moore's work for CMUCL)
496 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
497 parameters to * instead of NIL if no initform is supplied.
498 (reported by Johan Bockgaard)
499 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
500 correctly when FUNCALL appears as the car of the form. Note:
501 despite this FUNCALL forms are not currently subject to
502 compiler-macro expansion. (port of Raymond Toy's fix for the
503 same from CMUCL, reported by Johan Bockgaard)
504 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
505 (thanks for Teemu Kalvas)
506 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
507 case instead of terminating the iteration (thanks for Julian
508 Squires, Sean Champ and Raymond Toy)
509 * bug fix: incorrect expansion of defgeneric that caused a style
510 warning. (thanks for Zach Beane)
511 * bug fix: slot accessor effective method computation works properly
512 for classes with multiple non-standard applicable methods on
513 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
514 * on x86 compiler supports stack allocation of results of LIST and
515 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
516 CMUCL implementation by Gerd Moellmann)
517 * fixed some bugs revealed by Paul Dietz' test suite:
518 ** FORMAT strings with both the ~<~:;~> form of the justification
519 directive and pretty-printing directives cause an error.
521 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
522 * incompatible change: the internal functions
523 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
524 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
525 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
526 instead of the old functions.
527 * new feature: on platforms where "dladdr" is available foreign
528 function names now appear in backtraces. (based on Helmut Eller's
530 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
531 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
533 * documentation: ASDF manual is now distributed with SBCL.
534 * bug fix: SBCL can now load its contributed modules with REQUIRE
535 even if the system-provided entries have been removed from
536 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
538 * bug fix: backtraces involving undefined functions or assembly
539 routines are more informative. (thanks to Brian Downing)
540 * bug fix: mutually referent alien structures now work correctly.
541 (reported by Rick Taube)
542 * bug fix: structures defined by WITH-ALIEN can be referred to
543 within other definitions in the same WITH-ALIEN.
544 * bug fix: division operators (MOD, TRUNCATE and the like) with
545 constant zero divisors and integer dividends no longer generate
547 * bug fix: provide default methods for INPUT-STREAM-P and
548 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
549 * bug fix: improve the detection and resolution of MOP metacycles
550 (where effective method computation depends on the generic function
551 having its effective method computed). (reported by Bruno Haible)
552 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
553 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
554 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
555 * bug fix: pathnames with relative directory components can be
556 represented relative to default pathnames.
557 * optimization: in taking the GCD of bignums, reduce the two bignums
558 to approximately the same size (using Euclid's algorithm) before
559 applying the more sophisticated binary GCD. (thanks to Juho
561 * optimization: COUNT on bitvectors now operates word-at-a-time.
562 * optimization: ASH with a positive, but not necessarily constant,
563 (leftwards) shift, when in a modular context, is compiled to a
565 * fixed some bugs revealed by Paul Dietz' test suite:
566 ** FORMAT variable parameters ("~V<char>") are defaulted properly
567 if the corresponding argument is NIL.
568 ** FORMAT directives accepting a minpad parameter treat negative
570 ** literal commas as character parameters to FORMAT directives are
572 ** literal spaces directly after ~<Newline> directives within a
573 format-logical-block (~:< ~@:>) do not induce :FILL-style
574 conditional newlines.
575 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
576 as specified: it no longer includes conditional newlines).
577 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
579 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
580 :PER-LINE-PREFIX argument does not evaluate to a string.
582 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
583 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
584 the manual for details; for now, package locks can be disabled by
585 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
586 if no major problems are found then it is likely that they will be
587 compiled in unconditionally.
588 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
589 unsupported operators on all platforms. To load a shared library
590 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
591 object file, link it into a shared library outside of SBCL and
592 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
594 * bug workaround: ROOM T and the GC were not working together
595 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
596 reporting facility aren't preserved. That reporting has been
597 disabled, so now until and unless someone figures out how to make it
598 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
599 * minor incompatible change: as threatened around sbcl-0.8.0, the
600 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
601 The symbols are also exported from SB-PCL for backwards
602 compatibility, but more so than before SB-PCL should be treated as
603 an implementation-internal package.
604 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
606 * the SB-SPROF contrib now works on (most) non-x86 architectures.
607 It is known as of this release not to work on the Alpha, however.
608 * fixed bug #167: errors signalled due to illegal syntax in method
609 bodies are now more legible.
610 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
611 designators and can hence be used with TYPEP.
612 * fixed bug #333: CHECK-TYPE now ensures that the type error
613 signalled, if any, has the right object to be accessed by
614 TYPE-ERROR-DATUM. (reported by Tony Martinez)
615 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
616 5.1.2.3 for multiple-value place subforms. (reported by Kalle
618 * fixed bug #334: programmatic addition of slots using specialized
619 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
620 and :ALLOCATION :CLASS slots.
621 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
622 float exponents. (rereported by Peter Seibel)
623 * fixed a bug: #\Space (and other whitespace characters) are no
624 longer considered to be macro characters in standard syntax by
626 * fixed bug: initialization of condition class metaobjects no longer
627 causes an instance of the condition to be created. (reported by
629 * fixed bug: it is now possible to have slots such that
630 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
631 description is neither :INSTANCE nor :CLASS.
632 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
633 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
634 SLOT-VALUE-USING-CLASS), and no longer causes errors with
635 non-standard SLOT-DEFINITION-ALLOCATION values.
636 * fixed bugs: various race conditions handling exiting threads.
637 CL-PPCRE's thread torture test now passes.
638 * fixed arguably-a-bug: GC time proportional to number of threads in
639 system even when most of them are idle
640 * optimization: improved performance of BIT and SBIT on bit-vectors.
641 * .fasl file incompatibility: The fasl file version number has
642 been incremented because of changes associated with package locks.
644 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
645 * minor incompatible change: the system no longer provides
646 optimization hints (conditions which are TYPEP
647 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
648 mode; these hints are still emitted when higher SPEED optimization
650 * new contrib module: a sampling profiler (profiling by statistical
651 sampling, rather than by instrumenting functions) is available as
652 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
653 * the behaviour of the standard function ED is now customizeable by
654 third parties through a hook variable: see ED's documentation
655 string for information on the protocol.
656 * the compiler no longer emits efficiency notes for (FUNCALL X)
657 when the type of X is uncertain under default optimization
659 * fixed bug 276: mutating a binding of a specialized parameter to a
660 method to something that is not TYPEP the specializer is now
662 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
663 exhibit the required behaviour on the x86 platform. (thanks to
664 Peter van Eynde, Eric Marsden and Bruno Haible)
665 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
666 even for difficult arguments. (reported by Peter Graves)
667 * fixed bug 141a: the backquote printer now descends quoted
669 * fixed another bug in backquote printing: no more destructive
670 modification of the form's list structure. (reported by Brian
672 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
673 move between its address being taken and the call to
674 interrupt_thread, fixing a crashing race condition.
675 * the SB-POSIX contrib implementation has been adjusted so that it
676 no longer exhibits ridiculously poor performance when constructing
677 instances corresponding to C structs.
679 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
680 * minor incompatible change: the sb-grovel contrib now treats C
681 structures as alien (in the sense of SB-ALIEN) objects rather than
682 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
683 has implications for memory management of client code
684 (sb-grovel-returned objects must now be manually managed) and for
685 type safety (alien objects now have full types).
686 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
687 used to control emission of compiler diagnostics, rather than the
688 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
689 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
690 quality should be considered deprecated.
691 * install.sh now installs the user manual as well
692 * (not quite a new documentable feature, but worth considering in
693 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
694 beginnings of a semantically meaningful condition hierarchy is
695 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
697 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
698 designator as the defaults argument.
699 * fixed bug: Displaced arrays whose displaced-to array has become
700 too small now cause ARRAY-DIMENSION to signal an error, providing
701 for safer bounds-checking. (reported by Bruno Haible)
702 * fixed bug: DEFCLASS slot definitions with identical :READER and
703 :WRITER names now signal a reasonable error. (reported by Thomas
705 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
707 * fixed bug: Invalid slot specification errors now print correctly.
708 (thanks to Zach Beane)
709 * fixed bug 320: Shared to local slot value transfers in class
710 redefinitions now happen corrently from superclasses as
711 well. (reported by Bruno Haible)
712 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
714 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
715 improper lists correctly. (reported by Bruno Haible)
716 * fixed bug 313: source-transform for <fun-name> was erroneously
717 applied to a call of a value of a variable with name <fun-name>.
718 (reported by Antonio Menezes Leitao)
719 * fixed bug 307: The obsolete instance protocol ensures that
720 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
721 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
722 remove dead unknown-values globs from the middle of the stack.
723 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
724 easier to resume long computations after using *BREAK-ON-SIGNALS*
725 to diagnose and fix failures. (thanks to Nikodemus Siivola)
726 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
727 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
728 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
729 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
730 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
732 * fixed a (fairly theoretical) bug in string printing: if
733 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
734 does not have array-element-type equal to the most general string
736 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
737 function name. (thanks to Nikodemus Siivola)
738 * fixed bug: compile-time format string checker failed on
739 non-closed ~{. (reported by Thomas F Burdick)
740 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
741 don't warn on legal loop code involving more than one aggregate
742 boolean. (thanks to Nikodemus Siivola)
743 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
744 in keyword position to MAKE-PACKAGE was not being reported as
746 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
747 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
749 * fixed bugs 280 and 312: the checking for multiple definitions in a
750 file is less likely to become confused by uses of inline
752 * fixed bug: the #S reader macro performs the keyword coercion
753 specified for slot names. (reported by Kalle Niemitalo)
754 * fixed bug: lambda lists may contain symbols whose names start with
755 & but are not lambda-list-keywords; their occurrence triggers a
757 * fixed bug 321: define-method-combination argument lambda lists do
758 not cause invalid code to be generated when &optional or &aux
759 variables have default values. (reported by Bruno Haible)
760 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
761 have CLOS classes; this appears also to have fixed reported
762 instability in stack exhaustion detection.
763 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
764 MUFFLE-WARNING when no associated restart is present is now
766 * optimization: rearranged the expansion of various defining macros
767 so that each expands into only one top-level form in a
768 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
770 * optimization: used a previously little-used slot in symbols to
771 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
773 * fixed some bugs revealed by Paul Dietz' test suite:
774 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
776 ** Fixed an optimization bug related to inheritance of initforms
777 from local to shared slots.
778 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
779 ** CAST optimizer forgot to flush argument derived type.
780 ** print/read consistency on floats is now orders of magnitude
781 more likely. (thanks also to Bruno Haible for a similar report
783 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
784 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
785 for objects of type REAL. Make it so.
786 ** SXHASH is specified (infelicitously) to respect similarity,
787 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
788 it so. (thanks to Markus Ziegler)
789 ** on the Alpha, the compiler succeeds in compiling functions
790 returning a known number of arguments greater than 63.
791 ** fixed handling of invalid NIL arguments in keyword position
793 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
795 ** provided a readably-printable representation for RANDOM-STATE
797 ** ensured that pathnames, where they have a namestring, always
798 print using #P"..." syntax.
800 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
801 * Support for the forthcoming 2.0 version of the NetBSD kernel
802 running on x86 hardware has been added. (thanks to Perry
803 E. Metzger most immediately, and others for their past work)
804 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
805 current release version; SBCL's OpenBSD support had been broken
806 since about the time of OpenBSD's switch to ELF binary format.)
807 * [placeholder for DX summary]
808 ** user code with &REST lists declared dynamic-extent, under high
809 speed or space and low safety and debug optimization policy.
810 * The manual has been converted to Texinfo format and the debugger
811 chapter from the cmucl manual has been added.
812 * A facility has been added to extract documentation strings from
813 sbcl and store them as Texinfo-formatted snippets for inclusion in
814 the manual (via Texinfo's @include directive)
815 * bug fix: compiler emitted division in optimized DEREF. (thanks for
816 the test case to Dave Roberts)
817 * bug fix: multidimensional simple arrays loaded from FASLs had fill
818 pointers. (reported by Sean Ross)
819 * bug fix: PROFILE output is printed nicely even for large numerical
820 values. (thanks to Zach Beane)
821 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
822 greater than 32 handle EOF correctly.
823 * bug fix: on X86 an immediate argument of the IMUL instruction is
824 correctly printed in disassembly. (thanks to Lutz Euler)
825 * bug fix: class slots in redefined classes preserve their old
826 values. (thanks to Bruno Haible and Nikodemus Siivola)
827 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
828 succeeds. (reported by Marco Baringer)
829 * fixed some bugs revealed by Paul Dietz' test suite:
830 ** READ-SEQUENCE now works on ECHO-STREAMs.
831 ** RATIONALIZE works more according to its specification. (thanks
834 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
835 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
836 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
837 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
838 should matter to you only if you rebind the printer control
839 variables and then find you want different bindings in the
840 debugger than in the ordinary execution of your program.)
841 * The runtime build system has been tweaked to support building
842 (on SPARC/SunOS) using a C compiler which invokes Sun's own
843 assembler and linker. (thanks to Nikodemus Siivola)
844 * Unbound, undefined, undeclared variables now trigger full
845 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
846 more often programmer error than deliberate exploitation of undefined
848 * optimization: the hash algorithm for strings has changed to one
849 that is less vulnerable to spurious collisions. (thanks to Juho
851 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
852 needless bounds checking. (thanks to Juho Snellman)
853 * optimization: implemented multiplication as a modular
854 (UNSIGNED-BYTE 32) operation on the PPC backend.
855 * fixed some bugs revealed by Paul Dietz' test suite:
856 ** ADJUST-ARRAY now copies the datum in a zero rank array if
858 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
859 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
860 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
862 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
864 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
866 ** ECHO-STREAMs no longer attempt to echo the end of file value to
867 their output stream on EOF from read.
868 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
869 have been read to end-of-file.
870 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
872 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
873 description of determination of which consecutive characters
875 ** Printing the "Space" character with escaping on now yields "#\\ ",
876 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
877 ** Reading floating-point numbers with *READ-BASE* set to a number
878 less than 10 works correctly.
879 ** Reading floating-point numbers with *READ-BASE* set to a number
880 more than 10 works correctly.
881 ** Printing with *PRINT-READABLY* targets the standard readtable, not
882 the readtable currently in effect.
884 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
885 * minor incompatible change: parsing of namestrings on a physical
886 (Unix) host has changed; numbers after the final #\. in a
887 namestring are no longer interpreted as a version field. This is
888 intented to be largely invisible to the user, except that the
889 meaning of the namestring "*.*.*" has changed: it now refers to a
890 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
891 should usually be replaced by
892 (make-pathname :name :wild :type :wild :version :wild)
893 with the added benefit that this is more likely to be portable.
894 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
895 option to OPEN now signals an error if the file being opened
896 exists; this may have an impact on existing code.
897 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
898 signals received, on the PowerPC platforms (both Linux and
899 Darwin). (thanks to Pierre Mai for pointing out the location of
901 * several fixes on OS X: The system now builds and runs cleanly on
902 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
903 (thanks to Brian Mastenbrook)
904 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
905 second-resolution: integer multiples of 1/3600 between -24 and 24.
906 (thanks to Vincent Arkesteijn)
907 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
908 type when called with 1 argument; PEEK-CHAR checked type of
909 PEEK-TYPE only after having read first character from a
910 stream. (reported by Peter Graves)
911 * bug fix: the garbage collector now has much better locality
912 behaviour, and in particular no longer treats all memory as being
913 exhausted when many small objects point to each other in a deeply
915 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
916 recognized as being TYPEP their class.
917 * bug fix: the PUSHNEW documentation string has been corrected.
918 (thanks to Vincent Arkesteijn)
919 * bug fix: defaulting of the value for the last of an atypically
920 large number of multiple values being bound was not being
921 performed correctly on the Alpha or PPC platforms
922 * optimization: implemented multiplication as a modular
923 (UNSIGNED-BYTE 32) operation on the x86 backend.
924 * optimization: SEARCH on simple-base-strings can now be open-coded.
925 (see also contrib/compiler-extras.lisp for inspiration for
926 teaching the compiler about the Boyer-Moore algorithm).
927 * value, returned by MAX (and MIN) called with several EQUALP, but
928 not EQL, arguments now does not depend on compiler settings.
929 * fixed some bugs revealed by Paul Dietz' test suite:
930 ** in stack analysis liveness information is propagated from
931 non-local entry points.
932 ** pathwise CAST removing failed when the CAST node did not start
934 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
935 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
937 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
938 host is already defined.
939 ** RENAME-FILE works on streams instead of signalling an internal
941 ** PEEK-CHAR uses the current readtable when determining whether
942 or not a character is whitespace.
943 ** MERGE-PATHNAMES handles the case when the pathname does not
944 specify a name while the default-pathname specifies a version.
945 ** Pathnames now stand a better chance of respecting print/read
947 ** Attempting to use standardized file system operators with a
948 pathname with invalid :DIRECTORY components signals a
950 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
952 ** DIRECTORY on logical pathnames is more correct.
953 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
954 signal a TYPE-ERROR if their argument is not a stream.
955 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
956 designator argument does not designate a stream.
957 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
958 examining the synonym.
959 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
961 ** OPEN and WITH-OPEN-STREAM allow opening streams with
962 element-type larger than ([UN]SIGNED-BYTE 32).
964 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
965 * When built with the :SB-FUTEX feature, threaded builds now take
966 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
967 for faster/more reliable mutex and condition variable support.
968 * Incompatible change (but one you probably shouldn't have been using
969 anyway): the interface and code for arbitrating between multiple
970 threads in the same user session has been redesigned.
971 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
972 error when called without an explicit environment argument.
973 (thanks to Nikodemus Siivola)
974 * bug fix: buffered :DIRECTION :IO streams are less likely to become
975 confused about their position. (thanks to Adam Warner and Gerd
977 * bug fix: Pretty printing backquoted forms with unquotations in the
978 argument list position of various code constructs such as LAMBDA
979 now works correctly. (reported by Paul Dietz)
980 * bug fix: Pretty printing unquotations no longer loses all
981 stream position information.
982 * optimization: performance of string output streams is now less
983 poor for multiple small sequence writes.
984 * optimization: performance of CSUBTYPEP in the presence of complex
985 expressions involving CONS and NOT many times has been improved.
986 (reported by Paul Dietz)
987 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
989 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
990 streams for the REPL, for the debugger, and so on. (thanks to
992 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
993 * fixed some bugs revealed by Paul Dietz' test suite:
994 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
995 is ignored if it is too small, rather than propagating through
996 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
997 ** extremely complex negations of CONS types were not being
998 sufficiently canonicalized, leading to inconsistencies in
1000 ** VALUES tranformer lost derived type.
1002 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1003 * fixed a bootstrapping bug: the build process no longer assumes
1004 that the various BOOLE-related constants have the same value in
1005 host and target lisps. (noted by Paul Dietz' test suite on an
1006 SBCL binary built from CLISP)
1007 * The system can now be dynamically linked on the MIPS platform,
1008 which enables dynamic loading of foreign code from Lisp. (thanks
1009 to Ralf Baechle for discussions on the MIPS ABI)
1010 * The system now records debugging information for its own source
1011 files in a filesystem-position-independent manner, relative to
1012 the "SYS" logical host.
1013 * fixed a compiler bug: MV-LET convertion did not check references
1014 to the "max args" entry point. (reported by Brian Downing)
1015 * tweaked disassembly notes to be less confident about proclaiming
1016 some instruction as an LRA. (thanks to Brian Downing)
1017 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1018 now each have their own history, command character, and other
1019 characteristics. (thanks to David Lichteblau)
1020 * fixed some bugs revealed by Paul Dietz' test suite:
1021 ** compiler failure in compiling LOGAND expressions including a
1023 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1024 shift greater than 32.
1025 ** FLUSH-DEST did not mark blocks for type check regeneration.
1026 ** HANDLER-CASE failed to accept declarations in handler clauses
1027 in some circumstances.
1029 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1030 * New code in contrib/sb-introspect (still probably not entirely
1031 stable yet) provides some support for smart Lisp development
1032 environments like SLIME.
1033 * The conditions signalled for errors occurring when loading .fasl
1034 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1035 in a way which should help ASDF recover gracefully.
1036 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1037 stuff has been cleaned up. If you code contrib/ stuff, this might
1038 affect you, and you can look at contrib/README, contrib/STANDARDS,
1039 and/or the 0.8.4.27 diff to check.
1040 * In full calls the compiler now does not generate checks for declared
1041 argument types for all arguments.
1042 * various threading fixes
1043 ** and some experimental patches which didn't make it into
1044 the main tree for this release, but which are shipped in
1045 contrib/experimental-thread.patch as a possible fix for some
1046 failures (deadlock, spinning...) in GC-intensive multithreaded
1048 * fixed PPC build problem (source code incompatibility of different
1049 library versions): added offsetof() hackery which attempts to divine
1050 where glibc maintainers put uc_mcontext today
1051 * fixed bug 282: compiler does not trust type assertions while passing
1052 arguments to a full call.
1053 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1054 &OPTIONAL VALUES type parameter.
1055 * fix bug 214: algorithm for noting rejected templates is now more
1056 similar to that of template seletion. (also reported by rydis on
1058 * fixed bug 141b: printing backquoted information readably and prettily
1059 inserts a space where necessary.
1060 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1061 (CHARACTER 10) are now reported as errors, rather than propagated
1062 as unknown types. (reported by piso on #lisp)
1063 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1064 with values NIL and :ERROR. (thanks to Milan Zamazal)
1065 * fixed bug 191c: CLOS now does proper keyword argument checking as
1066 described in CLHS 7.6.5 and 7.6.5.1.
1067 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1068 longer raise an error; further, using a list as a for-as-arithmetic
1069 counter now raises a meaningful error.
1070 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1071 now understood by sequence creation functions such as MAKE-SEQUENCE
1073 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1074 a STRING-INPUT-STREAM.
1075 * compiler enhancement: SIGNUM is now better able to derive the type
1077 * type declarations inside WITH-SLOTS are checked. (reported by
1079 * fixed some bugs revealed by Paul Dietz' test suite:
1080 ** incorrect optimization of TRUNCATE for a positive first
1081 argument and negative second.
1082 ** compiler failure in let-convertion during flushing dead code.
1083 ** compiler failure while deriving type of TRUNCATE on an
1084 interval, containing 0.
1085 ** ASH of a negative bignum by a negative bignum count now returns
1087 ** intersection of CONS types now canonicalizes properly, fixing
1088 inconsistencies in SUBTYPEP.
1090 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1091 * incompatible change: The --disable-debugger command line
1092 option now clobbers the debugger at a more fundamental
1093 level, by redefining #'INVOKE-DEBUGGER instead of by
1094 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1095 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1096 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1097 enter the debugger REPL and then suffer recursive errors
1098 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1099 while under the new system, BREAK in --disable-debugger mode
1100 terminates the system just as an unhandled error would.
1101 * fixed compiler performance when processing loops with a step >1;
1102 * bug fix: DOCUMENTATION now retrieves generic function
1103 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1104 support has been systematized, and now supports the methods
1105 specified by ANSI, along with a default method and a method for
1106 slot documentation. (reported by Nathan Froyd)
1107 * bug fix: effective methods associated with a generic function are
1108 no longer cached over a change of that generic function's method
1109 combination. (reported by Andreas Fuchs)
1110 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1111 names a non-existent directory. (thanks to Andreas Fuchs)
1112 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1113 the PPC platform, lead to stack corruption; this has been fixed.
1114 (reported by Rainer Joswig)
1115 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1116 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1117 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1118 arithmetic in general, is now much more reliable on the PPC
1120 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1121 platform now returns the right answer.
1122 * optimization: restored some effective method precomputation in
1123 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1124 precomputation is now tunable.
1125 * optimization: compiler-internal data structure use has been
1126 reviewed, and changes have been made that should improve the
1127 performance of the compiler by about 20%.
1128 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1129 simple-base-strings and simple-bit-vectors is improved.
1130 * optimization: the optimization of 32-bit logical and arithmetic
1131 functions introduced in version 0.8.3 on the x86 has been
1132 implemented on the mips, ppc and sparc platforms; an
1133 implementation of the same facility, but for 64-bit arithmetic,
1134 has been added for the alpha.
1135 * microoptimization: the compiler is better able to make use of the
1136 x86 LEA instruction for multiplication by constants.
1137 * bug fix: in some situations compiler did not report usage of
1138 generic arithmetic in (SPEED 3) policy.
1139 * bug 145b fix: compiler used wrong type specifier while converting
1140 MEMBER-types to numeric.
1141 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1143 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1144 index argument now works properly on the Alpha platform.
1145 * bug fix: floating point exception treatment on the Alpha platform
1147 * bug fix: FILE-POSITION works much better on string input and
1148 output streams. (thanks to Nikodemus Siivola)
1149 * bug fix: many threading/garbage collection symptoms sorted.
1150 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1151 might be pseudo-atomic.
1152 * internal change: Stopping for GC is now done with signals not
1153 ptrace. GC is now done in whichever thread wanted it, instead of
1155 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1157 * bug fix: result form in DO is not contained in the implicit
1159 * incompatible change: ICR structure is changed; the value part of
1160 CONTINUATION is now called LVAR; corresponding functions are
1161 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1162 * added type deriver for ISQRT (thanks to Robert E. Brown).
1163 * bug fix: better support for loading from the command line when an
1164 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1166 * fixed some bugs revealed by Paul Dietz' test suite:
1167 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1168 ** ROUND and FROUND now give the right answer when given very
1169 small float arguments.
1170 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1172 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1173 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1174 ** LCM should return a non-negative integer.
1175 ** PARSE-INTEGER returned the index of a terminator instead of the
1176 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1177 ** PARSE-INTEGER returned an incorrect index being applied to a
1179 ** LCM with two arguments of 0 returns 0 rather than signalling
1181 ** unsigned addition of a 32-bit constant with the high bit set no
1182 longer causes an internal compiler error on the x86.
1183 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1184 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1185 with negative last argument.
1186 ** byte specifiers with zero size and position no longer cause
1187 an error during type derivation.
1188 ** bignum multiplication on the Alpha platform now returns the
1190 * porting: The system now builds on SuSE AMD64, although it still
1191 generates a 32-bit binary.
1192 * .fasl file incompatibility: The fasl file version number has
1193 been incremented (because of the changes to internal compiler
1194 data structures referred to above).
1196 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1197 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1198 more accurately, on the Darwin kernel running on PowerPC hardware.
1199 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1200 * Compiler code deletion notes now signal a condition of type
1201 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1202 an associated MUFFLE-WARNING restart.
1203 * The compiler now performs limited argument count validation of
1204 constant format strings in FORMAT, and where appropriate in ERROR,
1205 CERROR and WARN. (thanks to Gerd Moellmann)
1206 * New ASDF-INSTALL contrib can be used for automatic download and
1207 installation of third-party Lisp code from CCLAN or other sites
1209 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1210 which forces another thread to execute a function supplied by the
1212 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1213 now accept and act upon their :ELEMENT-TYPE keyword argument.
1214 (reported by Martin Atzmueller, Edi Weitz)
1215 * bug fix: FILE-POSITION now accepts position designators up to
1216 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1217 is the greater. (thanks to Patrik Nordebo)
1218 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1219 Kalle Olavi Niemitalo)
1220 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1221 printable. (reported by Eric Marsden)
1222 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1223 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1224 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1225 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1226 (thanks to Stig Sandoe)
1227 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1228 UNREAD-CHAR now correctly includes the unread character in the
1229 target sequence. (thanks to Gerd Moellmann)
1230 * bug fix in threaded builds: the system can now be suspended and
1231 resumed by shell job control with minimal disruption.
1232 * bug fixes in times and timezones >2038AD
1233 * better handling of "where is GNU make?" problem in build scripts
1234 (thanks to Nikodemus Siivola)
1235 * new optimization: inside a named function any reference to a
1236 function with the same name is considered to be a self-reference;
1237 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1238 optimization quality.
1239 * new optimization on x86: logical functions and + now have
1240 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1241 used when the result is truncated to 32 bits.
1242 * VALUES declaration is partially enabled.
1243 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1244 * fixed some bugs revealed by Paul Dietz' test suite:
1245 ** The system now obeys the constraint imposed by
1246 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1247 types form a lattice under type intersection.
1248 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1249 ** ASSOC now ignores NIL elements in an alist.
1250 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1251 and (1+ MOST-POSITIVE-FIXNUM) answers.
1252 ** The addition of a method with invalid qualifiers to a generic
1253 function does not cause an error to be signalled immediately;
1254 a warning is signalled, and the error is generated only on
1255 calling the generic function.
1256 * changed .fasl file version number, in part to add type codes for
1257 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1258 obscure ANSI requirements
1260 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1261 * fixed bug 148: failure to inline-expand a local function left
1262 garbage, confusing the compiler.
1263 * fixed bugs 3cd: structure slot readers perform type check if the
1264 slot can have an invalid value (i.e. it is either not initialized
1265 or can be written with a less specific slot writer).
1266 * bug fix: the compiler now traps array references to elements off
1267 the end of an array; previously, the bounds checking in some
1268 circumstances could go off-by-one.
1269 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1271 * type declarations for array element types now obey the description
1272 on the CLHS page "Declaration TYPE", as per discussions on
1273 sbcl-help around 2003-05-08. This means that a declaration
1274 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1275 declaration, all references to BAR will be asserted or assumed
1276 (with THE, so dependent on compiler policy) to involve objects of
1277 type FOO. Note that no such declaration is implied in
1278 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1279 * declared types of functions from the "Conditions"
1280 chapter. (reported by Paul Dietz)
1281 * bug fix: CERROR accepts a function as its first argument.
1282 * bug fix: NTH an NTHCDR accept a bignum as index
1283 arguments. (reported by Adam Warner)
1284 * optimization: character compare routines now optimize comparing
1285 against a constant character. (reported by Gilbert Baumann)
1286 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1287 argument now works properly on the MIPS platform.
1288 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1289 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1290 declared to have a functional type, does not cause an error
1292 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1293 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1294 * fixed bug: if last continuation of a deleted block has a
1295 destination, this destination should be deleted too. (reported by
1297 * fixed a bug in the bootstrap process: the host compiler's values
1298 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1299 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1300 test case from Patrik Nordebo)
1301 * improved the ability of the disassembler on the PPC platform to
1302 provide helpful disassembly notes.
1303 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1304 the class in more cases than previously.
1305 * bug fix: FILE-POSITION now understands :START and :END for
1306 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1307 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1308 8) data. (thanks to David Lichteblau)
1309 * bug fix: it is possible to add a method to a generic function
1310 without lambda list.
1311 * bug fix: reader failed to signal END-OF-FILE inside an
1312 object representation. (reported by Nikodemus Siivola)
1313 * fixed some bugs revealed by Paul Dietz' test suite:
1314 ** LAST and [N]BUTLAST should accept a bignum.
1315 ** condition slot accessors are methods.
1316 ** (VECTOR NIL) is a subtype of STRING.
1318 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1319 * minor incompatible change: some nonsensical specialized lambda
1320 lists (used in DEFMETHOD) which were previously ignored now signal
1322 * minor incompatible change: the system is now aware of the types of
1323 variables in the COMMON-LISP package, and will signal errors for
1324 most violations of these type constraints (where previously they
1325 were silently accepted).
1326 * minor incompatible change: COMPILE-FILE now uses the freedom
1327 afforded (ANSI 3.2.2.3) to use derived function types for
1328 functions defined in the same file. This also permits the system
1329 to warn on static type mismatches and function
1330 redefinition. (Currently it does not work with high DEBUG level.)
1331 * minor incompatible change: VALUES declaration is disabled.
1332 * When issuing notes, the compiler now signals a condition of type
1333 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1334 restart for use in user handlers. It is expected that the
1335 COMPILER-NOTE condition will eventually become a condition
1336 supertype to a hierarchy of note types, which will then be
1337 handleable in a similar fashion. However, at the moment, no such
1338 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1339 but it's an implementation detail, not a classification for the
1341 * Changes in type checking closed the following bugs:
1342 ** type checking of unused values (192b, 194d, 203);
1343 ** template selection based on unsafe type assertions (192c, 236);
1344 ** type checking in branches (194bc).
1345 * A short form of VALUES type specifier has ANSI meaning (it has
1346 increased the number of situations when SBCL cannot perform type
1348 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1349 keywords or constants is permissible.
1350 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1351 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1352 classes. (thanks to Antonio Martinez)
1353 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1354 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1355 argument to a pretty-print stream. Instead, it leaves any such
1356 operations to DESCRIBE-OBJECT methods.
1357 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1359 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1360 lambda lists. (thanks to David Lichteblau)
1361 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1362 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1363 (thanks to Gerd Moellmann)
1364 * fixed bug 246: increased compilation speed of long
1365 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1367 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1369 * DEFCONSTANT now throws a condition of type
1370 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1371 constant to a non-EQL value; CONTINUE and ABORT restarts
1372 respectively change and preserve the value.
1373 * fixed bug 63: The code walker, part of the implementation of CLOS,
1374 is now better at handling symbol macros.
1375 * bug fix: There is no longer an internal implementation type named
1376 CL:LENGTH. (reported by Raymond Toy)
1377 * bug fix: In macro-like defining macros/special operators the
1378 implicit block does not enclose the lambda list.
1379 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1380 enclosed in parentheses) are not suitable as type specifiers, and
1381 their use properly signals an error now.
1382 * bug fix: An argument count mismatch for a type specifier in code
1383 being compiled no longer causes an unhandled error at compile
1384 time, but signals a compile-time warning.
1385 * fixed simple vector readable printing
1386 * bug fix: DESCRIBE takes more care over whether the class
1387 precedence list slot of a class is bound before accessing it.
1388 (reported by Markus Krummenacker)
1389 * bug fix: FORMATTER can successfully compile pretty-printer format
1390 strings which use variants of the ~* directive inside.
1391 * bug fix: SEARCH now applies its TEST predicate to the elements of
1392 the arguments in the correct order. (thanks to Wolfhard Buss)
1393 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1394 in type inference. (thanks to Robert E. Brown)
1395 * bug fix: Reading in symbols with an explicit package name of ""
1396 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1397 * fixed some bugs revealed by Paul Dietz' test suite:
1398 ** NIL is now allowed as a structure slot name.
1399 ** Arbitrary numbers, not just REALs, are allowed in certain
1400 circumstances in LOOP for-as-arithmetic clauses.
1401 ** Multiple class redefinitions before slot access no longer
1402 causes a type error.
1403 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1404 association between the name and a class.
1405 ** Generic functions with non-standard method-combination and over
1406 five methods all of which return constants no longer return NIL
1407 after the first few invocations. (thanks to Gerd Moellmann)
1408 ** CALL-NEXT-METHOD with no arguments now passes the original
1409 values of the arguments, even in the presence of assignment.
1410 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1412 ** DISASSEMBLE works with closures and funcallable instances.
1413 ** ADD-METHOD now returns the generic function, not the new method.
1414 ** FIND-METHOD signals an error if the lengths of the specializers
1415 is incompatible with the generic function, even if the ERRORP
1417 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1418 which its argument is a member.
1419 ** DEFCLASS only redefines the class named by its class-name
1420 argument if that name is the proper name of the class;
1421 otherwise, it creates a new class.
1422 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1423 of the UNBOUND-SLOT condition to the name of the slot.
1424 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1425 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1426 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1427 SLOT-MAKUNBOUND in the specified fashion.
1429 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1430 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1431 cross-compilation host. As a consequence, we can now bootstrap our
1432 way up to SBCL starting with a bare gcc toolchain and human-readable
1433 source code (first the source to CLISP, then the source to SBCL).
1434 * A contributed module containing a partial implementation of the
1435 simple-streams interface has been included. (thanks to Rudi
1437 * A contributed module implementing the RFC1321 Message Digest
1438 Algorithm, known as MD5, has been included.
1439 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1440 no longer has any effect, as the code controlled by this feature
1441 has been deleted. (As far as we know, no-one has ever built using
1442 this feature, and its semantics were confused in any case).
1443 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1444 work on conditions (as required by the ANSI specification),
1445 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1446 expected behaviour on conditions. Users should note, however,
1447 that such behaviour is not required by the ANSI specification,
1448 and so use of this behaviour may render their code unportable.
1449 * fixed some bugs revealed by Paul Dietz' test suite:
1450 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1452 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1453 ** SXHASH on condition objects no longer returns NIL.
1454 ** :ALLOCATION :CLASS slots are better treated; their values are
1455 updated on class redefinition, and initforms inherited from
1456 superclasses are applied.
1457 ** REMOVE-METHOD returns its generic function argument even when
1458 no method was removed.
1459 ** SHARED-INITIALIZE now initializes the values of the requested
1460 slots, including those with :ALLOCATION :CLASS.
1461 ** ALLOCATE-INSTANCE now works on structure classes defined via
1462 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1464 ** SLOT-EXISTS-P now works on conditions, as well as structures
1466 ** MAKE-LOAD-FORM now has the required methods on
1467 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1468 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1469 keyword, and now implements the SLOT-NAMES argument.
1470 ** methods with &OPTIONAL arguments no longer allow too many
1471 arguments to be passed in the call without error.
1472 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1473 option is consistent with the required arguments of the generic
1474 function lambda list.
1475 * bug fix: REQUIRE accepts a string designator. (Thanks to
1477 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1478 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1479 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1480 * bug fix: The compiler now checks for duplicated variables in macro
1482 * bug fix: SETQ on globals returns the correct value.
1483 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1484 not a COMPILER-ERROR (followed by some other strange error on
1485 choosing the CONTINUE restart).
1486 * bug fix: make.sh and friends are now more consistent in the way that
1487 they look for GNU "make".
1489 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1490 * experimental native threads support (on x86 Linux >=2.4 only).
1491 This is not compiled in by default: you need to add :SB-THREAD to
1492 the target features. See the "Beyond ANSI" chapter of the manual
1494 * fix for longstanding nonANSIism: The old distinction between
1495 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1496 The return value from CL:FIND-CLASS is now a CLOS class, and
1497 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1498 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1499 * An interface to the MetaObject Protocol, as described in Kiczales,
1500 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1501 MIT Press, 1991, is available from the SB-MOP package.
1502 * incompatible change: the SB-PCL package should now be considered
1503 a private implementation detail, and no longer a semi-private MOP
1505 * minor incompatible change: due to rearrangement for threads, the
1506 control stack and binding stack are now allocated at arbitrary
1507 addresses instead of being hardcoded per-port. Users affected by
1508 this probably have to be doing advanced things with shared
1509 libraries, and will know who they are.
1510 * minor incompatible change: Previously, all --eval forms used were
1511 processed with READ before any of them were processed with EVAL.
1512 Now each --eval form is processed with both READ and EVAL before
1513 the next --eval form is processed. (Thus package operations like
1514 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1515 work as the user might reasonably expect.)
1516 * minor incompatible change: *STANDARD-INPUT* is now only an
1517 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1519 * minor incompatible change: Y-OR-N-P is now character-oriented,
1520 not line oriented. Also, YES-OR-NO-P now works without errors.
1521 (thanks to Antonio Martinez)
1522 * sb-aclrepl module improvements: an integrated inspector, added
1523 repl features, and a bug fix to :trace command.
1524 * Known functions, which cannot be open coded by the backend, are
1525 considered to be able to check types of their arguments. (fixing
1526 a bug report by Nathan J. Froyd)
1527 * fixed a bug in computing method discriminating functions: It is
1528 now possible to define methods specialized on classes which have
1529 forward-referenced superclasses. (thanks to Gerd Moellmann)
1530 * fixed evaluation order in optional entries (reported by Gilbert
1532 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1533 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1534 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1535 required-by-AMOP NAME argument, as well as CLASS and
1536 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1537 * fixed bug 20: DEFMETHOD can define methods using names that are
1538 not the proper names of classes to designate class specializers.
1539 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1540 Unix file descriptors, instead of blowing up. (thanks to Antonio
1542 * Garbage collection refactoring: user-visible change is that a
1543 call to the GC function during WITHOUT-GCING will not do garbage
1544 collection until the end of the WITHOUT-GCING. If you were doing
1545 this you were probably losing anyway.
1546 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1547 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1548 TYPEP the latter but not the former.
1549 * The compiler issues a full WARNING for calls to undefined functions
1550 with names from the CL package.
1551 * MAP-INTO for a vector destination is open coded. (reported by
1552 Brian Downing on c.l.l)
1553 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1554 documentation string.
1555 * fixed some bugs revealed by Paul Dietz' test suite:
1556 ** COPY-ALIST now signals an error if its argument is a dotted
1558 ** Condition slots are now accessed more correctly in the presence
1559 of multiple initargs for a given slot.
1560 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1561 exclude restarts of the same name associated with a different
1563 ** DEFCLASS of forward-referenced classes with another
1564 forward-referenced class in the superclasses list no longer
1566 ** Condition slots are now initialized once each, not multiple
1567 times. (thanks to Gerd Moellmann)
1568 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1570 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1571 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1572 arguments contain duplicated elements.
1573 ** RESTART-CASE understands local macros.
1574 ** RESTART-CASE associates exactly its own restarts with a condition.
1575 ** ENDP in safe mode checks its argument to be of type LIST.
1576 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1577 in question is unbound.
1578 ** Optimized MAKE-INSTANCE functions no longer cause internal
1579 assertion failures in the presence of duplicate initargs.
1580 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1581 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1582 propagated even in the ctor optimized implementation of
1584 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1586 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1587 * a better implementation of SXHASH on (simple) bit vectors,
1588 measured both in execution speed and in distribution of results
1589 over the positive fixnums, has been installed. Likewise, a better
1590 implementation of EQUAL for simple bit vectors is now available.
1591 * fixed CEILING optimization for a divisor of form 2^k.
1592 * fixed bug 240 (emitting extra style warnings "using the lexical
1593 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1594 by Antonio Martinez)
1595 * fixed SXHASH, giving different results for NIL depending on type
1596 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1597 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1598 variable. (found by Rolf Wester)
1599 * SBCL does not ignore type declarations for special
1600 variables. (reported by rif on c.l.l 2003-03-05)
1601 * some bug fixes in contrib/sb-aclrepl/
1602 * fixed some bugs revealed by Paul Dietz' test suite:
1603 ** a bug in the CONS type specifier, whereby the CAR and CDR
1604 types got intertwined, has been fixed;
1605 ** the type system is now able to reason about the interaction
1606 between INTEGER and RATIO types more completely;
1607 ** APPEND, [N]REVERSE and NRECONC check that those their
1608 arguments, which must be proper lists, are really so;
1609 ** An array specialized to be unable to hold elements has been
1610 implemented, as required -- yes, really -- by ANSI;
1611 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1612 on malformed property lists;
1614 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1615 * incompatible packaging change: in line with Unix convention,
1616 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1617 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1618 * REQUIRE and PROVIDE are now optionally capable of doing something
1619 useful. See the documentation string for REQUIRE.
1620 * infrastructure for a managed SBCL contrib system: contributed
1621 modules in this release include:
1622 ** the ASDF system definition facility;
1623 ** an interface to the BSD Sockets API;
1624 ** an ACL-like convenience interface to the repl;
1625 (thanks to Kevin Rosenberg)
1626 ** an implementation of ROTATE-BYTE, with efficient implementation
1628 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1629 gives a closer approximation to the right answer than previously.
1630 (thanks to Raymond Toy)
1631 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1632 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1633 optional environment arguments, as required by ANSI.
1634 * fixed bugs in other functions taking environment objects, allowing
1635 calls with an explicit NIL environment argument to be compiled
1637 * fixed bug 228: primary return values from
1638 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1639 COMPILE or FUNCTION.
1640 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1641 :TYPE will no longer signal errors on innocuous objects.
1642 * fixed bug 231b: SETQ is better at respecting type declarations in
1643 the lexical environment.
1644 * fixed a bug in DEFCLASS: classes named by symbols with no or
1645 unprintable packages can now be defined.
1646 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1647 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1648 * fixed compiler failure related to checking types of functions.
1649 (reported by Robert E. Brown)
1650 * the compiler is now much more consistent in its error-checking
1651 treatment of bounding index arguments to sequence functions: in
1652 (SAFETY 3) code, errors will be signalled in almost all cases if
1653 invalid sequence bounding indices are passed to functions defined
1654 by ANSI to operate on sequences.
1655 * fixed a bug in the build procedure: documentation of SBCL-specific
1656 packages is now preserved and available in the final Lisp image.
1657 * lifted FDEFINITION lookup out of loops in the implementation of
1658 many list operations. (thanks to Robert E. Brown)
1659 * fixed a bug in the reader: the #n# reader macro now works for
1660 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1661 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1662 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1663 E. Brown and Tony Martinez respectively)
1664 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1665 SLOT-BOUNDP: in particular, it is now able to optimize them much
1666 better, and is now not vulnerable to having packages renamed.
1667 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1668 always called when a slot is not present in an instance. (thanks
1670 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1671 longer signal an error on unreading a character following EOF on
1672 the previous constituent stream. (thanks to Tony Martinez)
1673 * fixed some bugs revealed by Paul Dietz' test suite:
1674 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1675 not just nonnegative fixnums;
1676 ** the logical bit-array operators such as BIT-AND now accept an
1677 explicit NIL for their "opt-arg" argument (to indicate a
1678 freshly-consed result bit-array);
1679 ** ELT now signals an error on an invalid sequence index in safe
1681 ** the type system is now cleverer about negations of numeric
1682 types, and consequently understands the BIGNUM and RATIO types
1684 ** the type system is now cleverer about the interaction between
1685 INTEGER and RATIO types: while bugs still remain, many more
1686 cases are accurately computed;
1687 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1688 if it is in the last clause;
1689 ** CONSTANTLY now correctly returns a side-effect-free function in
1691 ** DECLARE is no longer treated as a special-operator; in
1692 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1693 * incremented fasl file version number due to the change in the
1694 DEFSTRUCT-SLOT-DESCRIPTION structure.
1696 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1697 * minor incompatible change: code processed by the "interpreter" or
1698 EVAL now has a compilation optimization policy of (DEBUG 2)
1699 (changed from (DEBUG 1)) to improve debuggability of interactive
1700 development, and to allow the use of the debug RETURN command in
1702 * an experimental implementation of the RETURN command for the
1703 debugger has been included. (thanks to Frederik Kuivinen)
1704 * fixed bug 62: constraints were not propagated into a loop.
1705 * fixed bug in embedded calls of SORT (reported and investigated by
1707 * fixed some bugs revealed by Paul Dietz' test suite:
1708 ** printing and reading of arrays with some dimensions having
1709 length 0 (thanks to Gerd Moellmann);
1710 ** BOA constructor with &AUX argument without a default value does
1711 not cause a type error;
1712 ** CONSTANTP now returns true for all self-evaluating objects.
1714 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1715 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1716 accessors that are related by inheritance, as specified in the
1717 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1718 Valtteri Vuorikoski)
1719 * The compiler is now able to inline functions that were defined in
1720 a complex lexical environment (e.g. inside a MACROLET).
1721 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1723 * Support for the upcoming FreeBSD-5.0 release has been included.
1724 (thanks to Dag-Erling Smorgrav)
1725 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1726 effect when it is not in a toplevel context.
1727 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1728 stand a better chance of being correct. (thanks to Gerd
1730 * fixed bug in COERCE, which now signals an error on coercing a
1731 rational to a bounded real type which excludes the expected
1733 * The compiler is now able to derive types more accurately from the
1734 COERCE and COMPILE functions.
1735 * fixed bug 223: functional binding is considered to be constant
1736 only for symbols in the CL package.
1737 * fixed bug 231: SETQ did not check the type of a variable being set
1738 (reported by Robert E. Brown)
1739 * A new optimization for MAKE-INSTANCE has been included, fixing
1740 various bugs (including relating to :ALLOCATION :CLASS slots and
1741 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1743 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1744 ** As required by ANSI, LOOP now disallows anonymous collection
1745 clauses such as COLLECT I in conjunction with aggregate boolean
1746 clauses such as THEREIS (= I 1);
1747 ** LOOP now signals an error when any variable is reused in the
1748 same loop (including the potentially useful construct analogous
1749 to WITH A = 1 WITH A = (1+ A);
1750 ** IT is only a special loop symbol within the first clause of a
1751 conditional loop clause;
1752 ** LOOP with a typed iteration variable over a hashtable now
1753 signals a type error iff it should.
1754 * fixed some other bugs revealed by Paul Dietz' test suite:
1755 ** FILE-STREAM now names the class previously known as FD-STREAM;
1756 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1757 argument) no longer signals an error;
1758 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1759 of the slot symbol, rather than using the current package
1760 ((:CONC-NAME "") continues to intern the slot's name in the
1762 * incremented fasl file version number, because of the incompatible
1763 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1764 of the new implementation of DEFINE-COMPILER-MACRO.
1766 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1767 * Support for building SBCL for MIPS platforms running in
1768 little-endian mode has now been checked in, and basic
1769 functionality on said platforms verified.
1770 * minor incompatible change: PCL now records the pathname of a file
1771 in which methods and the like are defined, rather than its
1773 * minor incompatible change: TRUENAME now considers the truename of
1774 a file naming a directory to be the pathname with :DIRECTORY
1775 component indicating that directory.
1776 * minor incompatible change: a NAMED clause in the extended form of
1777 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1778 reason for the previous behaviour is unclear.
1779 * more systematization and improvement of CLOS and MOP conformance
1780 in PCL (thanks to Gerd Moellman and Pierre Mai):
1781 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1783 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1784 primary methods with no specializers;
1785 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1787 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1788 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1789 FORWARD-REFERENCED-CLASSes; error reporting on
1790 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1792 ** SXHASH on CLOS instances now uses a slot internal to the
1793 instance to return different numbers on distinct instances,
1794 while preserving the same return value through invocations of
1796 ** DEFMETHOD signals errors when methods with longer incongruent
1797 lambda lists are added to generic functions;
1798 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1799 CLASS, as specified in AMOP;
1800 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1801 on the order returned by the primary method for classes of
1802 class STANDARD-CLASS;
1803 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1804 * fixed some bugs shown by Paul Dietz' test suite:
1805 ** DOLIST puts its body in TAGBODY;
1806 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1808 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1809 value producing form;
1810 ** if more variables are given to PROGV than values, extra
1811 variables are bound and made to have no value;
1812 ** NSUBSTITUTE on list arguments gets the right answer with
1814 ** ELT signals an error of type TYPE-ERROR when the index argument
1815 is not a valid sequence index;
1816 ** LOOP signals (at macroexpansion time) an error of type
1817 PROGRAM-ERROR when duplicate variable names are found;
1818 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1819 ** FUNCALL of special-operators now cause an error of type
1821 ** PSETQ now works as required in the presence of side-effecting
1822 symbol-macro places;
1823 ** NCONC accepts any object as its last argument;
1824 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1826 ** loop-for-as-package does not require a package to be explicitely
1828 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1830 * fixed bug 166: compiler preserves "there is a way to go"
1831 invariant when deleting code.
1832 * fixed bug 172: macro lambda lists with required arguments after
1833 &REST arguments now cause an error to be signalled. (thanks to
1835 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1836 bug. (thanks to Matthew Danish)
1837 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1839 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1840 arguments are lexically rebound. (thanks to Gerd Moellmann and
1842 * fixed bug 194: error messages are now more informative when there
1843 is no primary method applicable in a call to a generic
1844 function. (thanks to Gerd Moellmann)
1845 * fixed bug in command line argument checking (thanks to Julian
1847 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1848 (thanks to Matthew Danish)
1849 * incremented fasl file version number, because of the
1850 SXHASH-related changes in the layout of CLOS data structures
1852 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1853 * minor incompatible change: The runtime (the Unix executable named
1854 "sbcl") is now much pickier about the .core files it will load.
1855 Essentially it now requires .core files to descend from the same
1856 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1857 as the runtime does. (The intent is to prevent the crashes which
1858 can occur, and which can even be reported as mysterious failures,
1859 when people patch the sources or change the build parameters
1860 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1861 sbcl and .core files.)
1862 * fixed bug: VALUES-LIST is no longer optimized away.
1863 * fixed bug 142: The FFI conversion of C string values to Lisp
1864 string values no longer conses excessively. (thanks to Nathan
1865 Froyd porting Raymond Toy's fix to CMU CL)
1866 * began to systematize and improve MOP conformance in PCL (thanks to
1867 Nathan Froyd, Gerd Moellman and Pierre Mai):
1868 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1870 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1871 ** FINALIZE-INHERITANCE is now called on class finalization;
1872 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1873 argument precedence order.
1874 * fixed bug 202: The compiler no longer fails on functions whose
1875 derived types contradict their declared type.
1876 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1877 so it can be non-toplevel.
1878 * The fasl file version number has changed (because of the new
1879 implementation of DEFMACRO).
1880 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1881 safe code, that any length requirement by their type-specifier
1882 argument is valid. The exceptions to this are described in bug
1884 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1885 functions now check the types of their inputs as required by ANSI.
1886 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1887 introduced symbol is DECLAREd to be SPECIAL.
1888 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1889 * fixed bug: :COUNT argument to sequence functions may be negative
1890 * fixed bug: body of DO-SYMBOLS may contain declarations
1891 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1892 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1893 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1894 symbol macro only once
1895 * fixed printing of call frame when argument list is unavailable
1896 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1897 * compiler no longer signals WARNING on unknown keyword
1900 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1901 * A beta-quality port to the mips architecture running Linux,
1902 based on the old CMUCL backend, has been made. It has been tested
1903 on a big-endian kernel, and works sufficiently well to be able to
1904 rebuild itself; it has not been tested in little-endian mode.
1905 * fixed an inconsistency between gencgc.c and purify.c which made
1906 dumping/loading .core files unreliable
1907 * fixed bug 120a: The compiler now deals correctly with IFs where
1908 the consequent is the same as the alternative, instead of
1909 misderiving the return type. (thanks to Alexey Dejneka)
1910 * fixed bug 113: Logical pathnames are now dumpable (the logical
1911 host is resolved at load-time, throwing an error if it is not
1913 * fixed bug 174: FORMAT's error message is slightly clearer when a
1914 non-printing character is used in a format directive.
1915 * fixed several bugs in compiler checking of type declarations, i.e.
1916 violations of the Python "declarations are assertions" principle
1917 (thanks to Alexey Dejneka)
1918 * fixed several bugs in PCL's error checking (thanks to Gerd
1920 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1922 * fixed bug in compilation of functions as first class values
1923 (thanks to Antonio Martinez-Shotton)
1924 * The compiler's handling TYPE-ERRORs which it can prove will
1925 inevitably happen at runtime has been cleaned up and corrected
1926 in several ways. (thanks to Alexey Dejneka)
1927 * improved argument type checking for various basic arithmetic
1928 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1929 lavished on them in the past that they can be compiled in many
1930 ways in different special cases
1931 * fixed bug 181: compiler checks validity of user supplied type
1933 * cleaned up code flushing in optimization: Function calls which
1934 should signal errors for safety purposes (e.g. which ANSI says
1935 should signal errors when their arguments are of incorrect type)
1936 are no longer optimized away.
1937 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1938 * incremented fasl file version number, because changes in the
1939 implementation of sequence functions like COERCE caused
1940 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1941 in old inline expansions) to become undefined. (Actually these
1942 changes were later undone, so we might very well be binary
1943 compatible with 0.7.7 after all, but leaving the version number
1944 incremented seemed like the simplest and most conservative
1947 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1948 * An alpha-quality port to the parisc architecture running Linux,
1949 based on the old CMUCL backend, has been made. This, even more so
1950 than the other backends, should be considered still a work in
1951 progress; known problems include that the Linux kernel in 64-bit
1952 mode does not propagate the correct sigcontext structure to
1953 userspace, and consequently SBCL on a parisc64 kernel will not
1955 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1956 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1957 suppressed.) Also now that the compiler is looking at declarations
1958 in the environment, it checks optimization declarations as well,
1959 and suppresses inlining when (> DEBUG SPEED).
1960 * More fixes have been made to treatment of floating point exception
1961 treatment and other Unix signals. In particular, floating point
1962 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1963 * The detection and handling of control stack exhaustion (infinite
1964 or very deeply nested recursion) has changed. Stack exhaustion
1965 detection is now done by write-protecting pages at the OS level
1966 and applies at all optimization settings; when found, a
1967 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1968 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1969 be caught using IGNORE-ERRORS.
1970 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1971 correctly when transforming recursive LABELS functions to LETs.
1972 (thanks to Alexey Dejneka)
1973 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1974 that are names of constants or global variables.
1975 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1976 alien routines with docstrings.
1977 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1978 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1980 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1981 (thanks to Eric Marsden)
1982 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1983 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1984 * Bug fix: The SPARC backend can now compile functions involving
1985 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1986 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1987 to a functions expecting a PATHNAME-DESIGNATOR.
1988 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1989 lambda lists. (thanks to Alexey Dejneka)
1990 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1991 functions so that EXPORTing the name of the class doesn't cause
1992 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1993 to Antonio Martinez for reporting this)
1994 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1995 constant-folded without failing an assertion. (thanks to Einar
1996 Floystad Dorum for reporting this)
1997 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1998 types are now dealt with more correctly by the compiler.
1999 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2000 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2001 behaviour analogous to RENAME-FILE. This puts its behaviour more
2002 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2004 * The fasl file version number has changed again. (because of the
2005 bug fix involving the names of PCL MAKE-INSTANCE functions)
2007 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2008 * bug fix: Floating point exceptions are treated much more
2009 consistently on the x86/Linux and PPC/Linux platforms.
2010 * Array initialization with :INITIAL-ELEMENT is now much faster for
2011 cases when the compiler cannot open code the array creation, but
2012 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2013 array accesses have also seen a speed increase.
2014 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2015 specified. (This was at the root of some bad interactions between
2016 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2018 * bug fix: Internal error arguments for undefined functions are now
2019 computed correctly on the PPC/Linux platform.
2020 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2021 Raymond Toy's patch for CMU CL)
2022 * Support for the Solaris 9 operating environment has been included
2023 (thanks to Daniel Merritt)
2024 * A very ugly but hopefully complete draft of the missing FFI chapter
2025 of the manual has been created by reformatting the corresponding
2026 CMU CL manual chapter into (currently very ugly and incoherent)
2027 DocBook and bringing it up to date for SBCL behavior. Thus, the
2028 manual is now essentially complete, at least by my extreme
2029 once-and-only-once standards, whereby it's acceptable to refer to
2030 the doc strings of SB-EXT functions as the primary documentation.
2031 * The fasl file version number has changed again, due to cleanup of
2032 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2033 is no longer a static symbol.)
2035 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2036 * SBCL now builds with OpenMCL (version 0.12) as the
2037 cross-compilation host; also, more progress has been made toward
2038 bootstrapping under CLISP.
2039 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2041 * bug 158 fixed: The compiler can now deal with integer loop
2042 increments different from 1; fixing this turned out also to fix
2044 * bug 169 fixed: no more bogus warnings about using lexical bindings
2045 despite the presence of perfectly good SPECIAL declarations (thanks
2046 to David Lichteblau)
2047 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2048 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2049 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2051 * bug 180 fixed: Method combination specifications no longer ignore
2052 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2053 * bug fix: Structure type predicate functions now check their argument
2054 count as they should.
2055 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2056 correctly. (thanks to Pierre Mai)
2057 * minor incompatible change: The --noprogrammer option is deprecated
2058 in favor of the new --disable-debugger option, which is very similar.
2059 (The major difference is that it takes effect at a slightly different
2060 time at startup, causing handling of errors in --sysinit and
2061 --userinit files will be affected differently.) The
2062 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2063 been added to allow this functionality to be controlled from ordinary
2064 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2065 maintainers, who might want to run non-interactive scripts to
2066 build SBCL cores which will later be used interactively.)
2067 * minor incompatible change: The LOAD function no longer, when given
2068 a wild pathname to load, loads all files matching that pathname.
2069 Instead, an error of type FILE-ERROR is signalled.
2071 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2072 * bug 147 fixed: The compiler preserves its block link/count
2073 invariants more correctly now so that it doesn't crash. (thanks
2075 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2077 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2078 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2079 no-value-for-this-character as NIL (as specified by ANSI).
2080 * HOST-NAMESTRING on physical pathnames now returns a string that is
2081 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2082 (thanks to Christophe Rhodes)
2083 * The Alpha port handles icache flushing more correctly. (thanks to
2085 * More progress has been made toward bootstrapping under CLISP. (thanks
2086 to Christophe Rhodes)
2087 * The fasl file format has changed again, because dynamic loading
2088 on OpenBSD (which has non-ELF object files) motivated some cleanups
2089 in the way that foreign symbols are transformed and passed around.
2090 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2091 is no longer treated as whitespace by the reader, but instead as
2092 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2093 |A
\7fB|, instead of A as it used to.
2095 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2096 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2097 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2098 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2099 this as of 0.7.1.45, but was left out of the previous news section
2100 (thanks to Dan Barlow)
2101 * SBCL now runs on the Solaris operating system on SPARC architectures
2102 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2103 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2104 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2105 * SPARC backend cleanups, allowing builds of cores optimized for V8
2106 and V9 SPARCS, and also emission of code targeted to a particular
2107 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2109 * SBCL is closer to bootstrapping under CLISP, thanks to various
2110 fixes by Christophe Rhodes.
2111 * The fasl file format has changed again, to allow the compiler's
2112 INFO database to support symbol macros.
2113 * The user manual (in doc/) is formatted into HTML more nicely.
2114 (thanks to coreythomas)
2115 * The system is smarter about SUBTYPEP relationships, especially
2116 those involving NOT types (including types such as ATOM which are
2117 represented internally using NOT types). Thus SUBTYPEP is less
2118 likely to return (VALUES NIL NIL) in general, and in particular
2119 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2121 * The fasl file format has changed again, because the internal
2122 representation of types now includes a new slot to support the new
2123 SUBTYPEP-of-NOT-types logic.
2124 * (not a change in the main branch of SBCL, but a related prototype
2125 which can hopefully be merged into the main branch of SBCL in the
2126 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2127 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2128 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2129 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2130 dispatch macro character combinations. (thanks to Alexey Dejneka)
2131 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2132 that superficially look like logical namestrings correctly.
2133 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2135 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2136 default supplied-p) for &optional and &key arguments. (thanks to
2139 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2140 * incompatible change: The compiler is now less aggressive about
2141 tail call optimization, doing it only when (> SPACE DEBUG) or
2142 (> SPEED DEBUG). (This is an incompatible change because there are
2143 programs which relied on the old CMU-CL-style behavior to optimize
2144 away their unbounded recursion which will now die of stack overflow.)
2145 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2146 for non-GENCGC systems has been increased to 20M (since that
2147 seems much closer to the likely performance optimum for modern
2148 systems than the old 4M value was)
2149 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2150 and *DEBUG-PRINT-LEVEL*
2151 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2152 of CMU CL's support for SPARC, and various endianness and other
2153 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2154 * new syntactic sugar for the Unix command line: --load foo.bar is now
2155 an alternate notation for --eval '(load "foo.bar")'.
2157 ** The system now detects stack overflow and handles it gracefully,
2158 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2159 optimization settings. (This is a good thing in general, and
2160 its introduction in this version should be particularly timely
2161 for anyone whose code fails because of suppression of tail
2163 ** The system now hunts for the C variable "environ" in a more
2164 devious way, to avoid segfaults when the C library version differs
2165 between compile time and run time. (thanks to Christophe Rhodes)
2166 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2167 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2168 ** The compiler no longer issues bogus style warnings for undefined
2169 classes in the same source file as the DEFCLASSes which defined
2170 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2172 ** fixes in CONDITION class precedence list for undefined function
2173 errors (thanks to Alexei Dejneka)
2174 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2175 correctly. (thanks to Dan Barlow)
2176 ** portability fixes aiming at bootstrapping under CLISP (thanks
2177 to Dave McDonald and Christophe Rhodes)
2178 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2179 ** fixes in type translation and and type inference (thanks to
2181 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2182 ** various fixes in the new ports (thanks to Dan Barlow)
2183 * several changes related to debugging:
2184 ** suppression of tail recursion, as noted above
2185 ** stack overflow detection, as noted above
2186 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2187 is now the default. (For some time encapsulation has been more
2188 reliable than the breakpoint-based :ENCAPSULATE NIL
2189 implementation, at least on X86 systems; and I just noticed that
2190 encapsulation also seems closer to the spirit of the ANSI
2193 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2195 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2196 up properly again. (There was a packaging bug in 0.7.0 which
2197 left their definitions in SB-SYS::LOAD-FOREIGN and
2198 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2199 vital for most things which interface to C-level interfaces,
2200 like extensions working with sockets or databases or
2201 Perl-compatible regexes or whatever, and the need to fix
2202 this bug is the main reason that 0.7.1 was released so
2204 ** DEFGENERIC is now choosier about the methods it redefines, so that
2205 reLOADing a previously-LOADed file containing DEFGENERICs does
2206 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2207 cycle now works as it should. (thanks to Alexey Dejneka)
2208 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2209 fixed by Christophe Rhodes. (He actually submitted this patch
2210 months ago, and I delayed until after 0.7.0.)
2211 ** Bug 111 (internal compiler confusion about runtime checks on
2212 FUNCTION types) was fixed by Alexey Dejneka.
2213 * Some internal cleanups (getting rid of variables which aren't
2214 needed now that the byte interpreter is gone) caused the fasl
2215 file format number to change again.
2217 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2218 * major incompatible change: The default fasl file extension, i.e. the
2219 default extension for files produced by COMPILE-FILE, has changed
2220 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2222 ** There are many changes in the implementation of the compiler.
2223 SBCL is now essentially a compiler-only implementation of ANSI
2224 Common Lisp. EVAL still "interprets" a few special cases, but
2225 almost all the interesting cases are handled by creating
2226 a LAMBDA expression, calling COMPILE on it, then calling
2227 FUNCALL on the result.
2228 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2229 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2230 Since the code is newer, there might still be some new bugs
2231 (though not as many as before Martin Atzmueller's fixes:-). But
2232 the new code is substantially simpler and clearer, and hopefully
2233 any remaining bugs will be simpler, less fundamental, and more
2234 fixable then the bugs in the old code.
2235 ** The revised compiler is still a little unsteady on its feet.
2237 *** The debugging information it produces (particularly the names
2238 of FUNCTION objects) is sometimes much less useful than what
2239 the old compiler produced.
2240 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2241 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2242 within a MACROLET) has been temporarily weakened.
2243 ** There are new compiler optimizations for various functions:
2244 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2245 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2246 *** the math functions TRUNCATE, FLOOR, and CEILING
2247 *** the function-of-all-trades COERCE
2248 Mostly these should be transparent, but there's one
2249 potentially-annoying problem (bug 117): when the compiler
2250 inline-expands a function and does type analysis on the result,
2251 it can create control paths which have type mismatches, and
2252 when it can't prove that those control paths aren't taken,
2253 it will issue WARNINGs about the type mismatches. This is
2254 a particular problem in practice for the new sequence functions.
2255 It's not clear how this should be fixed, and for now, a
2256 workaround is given in the entry for 117 in the BUGS file.
2257 ** (Because of the interaction between the two previous items --
2258 occasional inlining problems and new inline expansions -- some
2259 of the new sequence function optimizations won't really kick in
2260 completely until debugging information, and then inlining, are
2261 straightened out in some future version.)
2262 * minor incompatible changes:
2263 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2264 DIRECTORY no longer implicitly promotes NIL slots of its
2265 pathname argument to :WILD. In particular, when you ask for the
2266 contents of a directory (which you used to be able to do without
2267 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2268 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2269 ** changes in behavior that ANSI explicitly defines to be
2270 implementation dependent:
2271 *** The new compiler-only implementation still conforms with ANSI,
2272 but acts a little different than before. Besides the obvious
2273 changes in performance tradeoffs (that the cost per form passed
2274 to EVAL has gone up, and the cost per form executed by EVAL
2275 has gone down), the behavior of the system changes a little
2276 because there are no longer any interpreted function objects.
2277 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2278 e.g. doing COMPILE on the output of interactive DEFUN is
2280 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2282 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2283 has changed from (:CL) to NIL.
2284 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2285 e.g. #\Newline and #\Tab, don't have names specified in the
2286 ANSI Common Lisp standard, is now based on their ASCII symbolic
2287 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2288 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2289 are no longer used for output.
2290 ** changes in internal implementation constants:
2291 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2292 4 million. (If your application spends a lot of time GCing and
2293 you have a lot of RAM, you might want to experiment with
2294 increasing it even more.)
2295 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2296 However, almost all old code should still continue to work without
2297 immediate update, as SB-C-CALL is now a (deprecated) nickname
2299 ** Old operator names in the style DEF-FOO are now deprecated in
2300 favor of new corresponding names DEFINE-FOO, for consistency with
2301 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2302 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2303 internal symbols, but a few supported extensions like
2304 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2305 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2306 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2307 etc. as you get deeper into recursive calls to the debugger
2308 command loop, instead of the old "5]", "5]]", "5]]]"
2309 sequence. (I was motivated to do this when squabbles between
2310 ILISP and SBCL left me very deeply nested in the debugger. In the
2311 short term, this change will probably provoke more ILISP/SBCL
2312 squabbles, but hopefully it will be an improvement in the long run.)
2313 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2314 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2315 function, since declaiming return types involving aliens is
2316 (1) annoyingly messy to do by hand and (2) vital to efficient
2317 compilation of code which calls such functions.
2318 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2319 longer reexported by the SB-EXT package. They're solely useful
2320 for alien code, so it seems more logical that you should get
2321 them from the SB-ALIEN package, not in SB-EXT.
2322 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2323 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2324 features. Instead, the code that they used to control is always
2325 built into the system.
2326 * many other bug fixes
2327 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2328 advantage of the new EVAL-WHEN stuff and to clean them up in
2329 general, and they are now more ANSI-compliant in a number of
2330 ways. Martin Atzmueller is responsible for a lot of this.
2331 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2333 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2334 *** correct ERROR type for various file operations
2335 *** some fixes for Lisp streams
2336 *** DEFMETHOD syntax checking
2337 *** changing old weird representation of debug information as
2338 strings (which, among their other deficiencies, don't transform
2339 correctly when you rename packages, and don't change their
2340 print representation when you change things like *PACKAGE*
2341 and *PRINT-LENGTH*) to symbols and lists of symbols
2342 He also made several improvements and fixed several bugs in DESCRIBE.
2343 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2345 *** misbehavior of WRITE-STRING/WRITE-LINE
2346 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2347 and several other LOOP problems as well
2348 *** DIRECTORY when similar filenames are present
2349 *** DEFGENERIC with :METHOD options
2350 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2351 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2352 *** argument ordering in FIND with :TEST option
2353 *** mishandled package designator argument in APROPOS-LIST
2354 *** various problems in the backquote readmacro
2355 *** a bug in APROPOS
2356 *** probably some others that I'm not describing very well here,
2357 since the CVS log documents them by reference to sbcl-devel
2358 messages, and the SourceForge archives aren't working well.:-(
2359 ** Dan Barlow improved the Alpha port (and is making progress on the
2360 PPC port, for those of you who think different).
2361 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2362 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2363 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2364 continues to work on the SPARC port (for those of you in a position
2365 to look down upon our little PC-compatible boxes from a great height).
2366 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2367 and uses that copy, rather than the current dynamic value, when
2368 it's trying to decide whether to truncate output. Thus e.g.
2369 (let ((*print-lines* 50))
2370 (pprint-logical-block (stream nil)
2372 (let ((*print-lines* 8))
2373 (print (aref possiblybigthings i) stream)))))
2374 should now truncate the logical block only at 50 lines, instead of
2375 often truncating it at 8 lines, as it did before.
2376 * The doc/cmucl/ directory, containing old CMU CL documentation
2377 from the time of the fork, is no longer part of the base system.
2378 SourceForge has shut down its anonymous FTP service, and with it
2379 my original plan for distributing the old CMU CL documentation
2380 there. For now, if you need these files you can download an old
2381 SBCL source release and extract them from it.
2382 * The fasl file version number changed again, for dozens of reasons,
2383 some of which are apparent above.
2385 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2386 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2387 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2388 patch, so that DISASSEMBLE output is much nicer.
2389 * The code in the SB-PROFILE package now seems reasonably stable.
2390 I still haven't decided what the final interface should look like
2391 (I'd like PROFILE to interact cleanly with TRACE, since both
2392 facilities use function encapsulation) but if you have a need
2393 for profiling now, you can probably use it successfully with
2394 the current CMU-CL-style interface.
2395 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2396 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2397 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2398 link now returns the dangling link itself, and for similar
2399 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2400 link itself. (In these cases the old code signalled an error and
2401 looped endlessly, respectively.) Thus, DIRECTORY now works even
2402 in the presence of dangling and cyclic symbolic links.
2403 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2404 is now a supported extension again, since the consensus on
2405 sbcl-devel was that it can be useful for ordinary development
2406 work, not just for debugging SBCL itself.
2407 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2408 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2409 that currently-defined functions might be redefined later with
2410 different return types.
2411 * Hash tables can be printed readably, as inspired by CMU CL code
2412 of Eric Marsden and SBCL code of Martin Atzmueller.
2413 * better error handling in CLOS method combination, thanks to
2414 Martin Atzmueller porting Pierre Mai's CMU CL patches
2415 * more overflow fixes for >16Mbyte I/O buffers
2416 * A bug in READ has been fixed, so that now a single Ctrl-D
2417 character suffices to cause end-of-file on character streams.
2418 In particular, now you only need one Ctrl-D at the command
2419 line (not two) to exit SBCL.
2420 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2422 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2423 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2424 does the right thing.
2425 * The implementation of some type tests, especially for CONDITION
2426 types, is now tidier and maybe faster, due to CMU CL code
2427 originally by Douglas Crosher, ported by Martin Atzmueller.
2428 * Some math functions have been fixed, and there are new
2429 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2430 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2431 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2432 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2433 likely to be built into sbcl-0.7.0.)
2434 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2435 CL, ported to SBCL by Martin Atzmueller)
2436 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2437 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2438 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2439 * There's a new slam.sh hack to shorten the edit/compile/debug
2440 cycle for low-level changes to SBCL itself, and a new
2441 :SB-AFTER-XC-CORE target feature to control the generation of
2442 the after-xc.core file needed by slam.sh.
2443 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2444 COMPILE-FILE is no longer supported, so that now every function
2445 gets an entry point, so that block compilation looks a little
2446 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2447 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2448 returns the number of bytes consed since the system started,
2449 rather than the number consed since the first time the function
2450 was called. (The new definition parallels ANSI functions like
2451 CL:GET-INTERNAL-RUN-TIME.)
2452 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2453 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2454 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2455 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2456 * Fasl file version numbers are now independent of the target CPU,
2457 since historically most system changes which required version
2458 number changes have affected all CPUs equally. Similarly,
2459 the byte fasl file version is now equal to the ordinary
2462 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2463 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2464 is no longer recognized. I apologize for this, because it was
2465 listed in SB-EXT as a supported extension, but I found that
2466 its existing behavior was poorly specified, as well as incorrectly
2467 specified, and it looked like too much of a mess to straighten it
2468 out. I have enough on my hands trying to get ANSI stuff to work..
2469 * many patches ported from CMU CL by Martin Atzmueller, with
2470 half a dozen bug fixes in pretty-printing and the debugger, and
2471 half a dozen others elsewhere
2472 * fixed bug 13: Floating point infinities are now supported again.
2473 They might still be a little bit flaky, but thanks to bug reports
2474 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2475 as flaky as they were.
2476 * The --noprogrammer command line option is now supported. (Its
2477 behavior is slightly different in detail from what the old man
2478 page claimed it would do, but it's still appropriate under the
2479 same circumstances that the man page talks about.)
2480 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2481 are now supported, and enabled by default. Thus, the compiler can
2482 handle many floating point and complex operations much less
2483 inefficiently. (Thus e.g. you can implement a complex FFT
2485 * The compiler now detects type mismatches between DECLAIM FTYPE
2486 and DEFUN better, and implements CHECK-TYPE more correctly, and
2487 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2488 thanks to patches from Martin Atzmueller.
2489 * various fixes to make the cross-compiler more portable to
2490 ANSI-conforming-but-different cross-compilation hosts (notably
2491 Lispworks for Windows, following bug reports from Arthur Lemmens)
2492 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2493 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2494 fixed thanks to Pierre Mai's CMU CL patches.
2495 * fixes to keep the system from overflowing internal counters when
2496 it tries to use i/o buffers larger than 16M bytes
2497 * fixed bug 45a: Various internal functions required to support
2498 complex special functions have been merged from CMU CL sources.
2499 (When I was first setting up SBCL, I misunderstood a compile-time
2500 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2501 * improved support for type intersection and union, fixing bug 12
2502 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2503 more obscure bugs as well
2504 * some steps toward byte-compiling non-performance-critical
2505 parts of the system, courtesy of patches from Martin Atzmueller
2506 * Christophe Rhodes has made some debian packages of sbcl at
2507 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2508 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2509 stable, but are nonetheless usable. When he's ready, I'd be happy
2510 to add them to the SourceForge "File Releases" section. (And if
2511 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2512 * new fasl file format version number (because of changes in
2513 internal representation of (OR ..) types to accommodate the new
2514 support for (AND ..) types, among other things)
2516 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2517 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2519 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2521 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2522 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2523 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2524 * DESCRIBE now works on structure objects again.
2525 * Most function call argument type mismatches are now handled as
2526 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2527 know whether the function will be redefined before the call is
2528 executed. (The compiler could flag local calls with full WARNINGs,
2529 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2530 it doesn't keep track of enough information to know whether calls
2531 are local in this sense.)
2532 * Compiler output is now more verbose, with messages truncated
2533 later than before. (There should be some supported way for users
2534 to override the default verbosity, but I haven't decided how to
2535 provide it yet, so this behavior is still controlled by the internal
2536 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2537 src/compiler/ir1util.lisp.)
2538 * Fasl file format version numbers have increased again, because
2539 support for the Gray streams extension changes the layout of the
2540 system's STREAM objects.
2541 * The Gray subclassable streams extension now works, thanks to a
2542 patch from Martin Atzmueller.
2543 * The full LOAD-FOREIGN extension (not just the primitive
2544 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2545 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2546 but like most other programs, it defaults to copying the Unix
2547 environment from the original process instead of starting the
2548 new process in an empty environment.
2549 * Extensions which manipulate the Unix environment now support
2550 an :ENVIRONMENT keyword option which doesn't smash case or
2551 do other bad things. The CMU-CL-style :ENV option is retained
2552 for porting convenience.
2553 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2554 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2556 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2558 * A patch from Martin Atzmueller seems to have solved the SIGINT
2559 problem, and as far as we know, signal-handling now works cleanly.
2560 (If you find any new bugs, please report them!)
2561 * The system no longer defaults Lisp source file names to types
2562 ".l", ".cl", or ".lsp", but only to ".lisp".
2563 * The compiler no longer uses special default file extensions for
2564 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2565 seems to expect a single default extension for all compiled code,
2566 and there's no compelling reason to try to stretch the standard
2567 to allow two different extensions.) Instead, byte-compiled files
2568 default to the same extension as native-compiled files.
2569 * Fasl file format version numbers have increased again, because
2570 a rearrangement of internal implementation packages made some
2571 dumped symbols in old fasl files unreadable in new cores.
2572 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2573 many fewer weird special cases.
2574 * Bug #17 (differing COMPILE-FILE behavior between logical and
2575 physical pathnames) has been fixed, and some related misbehavior too,
2576 thanks to a patch from Martin Atzmueller.
2577 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2578 by Tim Moore, ported to SBCL by Martin Atzmueller.
2579 * Martin Atzmueller fixed several filesystem-related problems,
2580 including bug #36, in part by porting CMU CL patches, which were
2581 written in part by Paul Werkowski.
2582 * More compiler warnings in src/runtime/ are gone, thanks to
2583 more patches from Martin Atzmueller.
2584 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2587 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2589 * DESCRIBE now works on CONDITION objects.
2590 * The debugger now handles errors which arise when trying to print
2591 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2593 * The build system now uses an additional file, customize-target-features.lisp,
2594 to allow local modifications to the target *FEATURES* list. (The point of
2595 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2596 debugging features enabled, without having to worry about propagating it
2597 into everyone's system when I do a "cvs update".) When no
2598 customize-target-features.lisp file exists, the target *FEATURES* list
2599 should be constructed the same way as before.
2600 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2601 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2602 warn about behavior which is undefined under the ANSI spec. Note: This
2603 is specified by ANSI, but it's not very popular with programmers.
2604 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2605 macro in the SBCL sources for an example of a workaround which you
2606 might use to make portable ANSI-standard code which does what you want.
2607 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2608 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2609 when it's used as a non-toplevel form. (This is required in order
2610 to implement the DEFCONSTANT-EQX macro.)
2611 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2612 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2613 funny things, due to bugs in the implementation of EVAL-WHEN
2614 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2615 non-toplevel DEFCONSTANTs are uncommon.)
2616 * The core file version number and fasl file version number have been
2617 incremented, because the old noncompliant DEFCONSTANT behavior involved
2618 calling functions which no longer exist, and because I also took the
2619 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2620 * fixed bug 1 (error handling before read-eval-print loop starts), and
2621 redid debugger restarts and related debugger commands somewhat while
2623 ** The QUIT debugger command is gone, since it did something
2624 rather different than the SB-EXT:QUIT command, and since it never
2625 worked properly outside the main toplevel read/eval/print loop.
2626 Invoking the new TOPLEVEL restart provides the same functionality.
2627 ** The GO debugger command is also gone, since you can just invoke
2628 the CONTINUE restart directly instead.
2629 ** The TOP debugger command is also gone, since it's redundant with the
2630 FRAME 0 command, and since it interfered with abbreviations for the
2632 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2634 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2635 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2636 some porting work by Martin Atzmueller)
2637 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2638 with a boatload of patches which he ported to SBCL. Now that those
2640 ** The system tries to make sure that its low-priority messages
2641 are prefixed by semicolons, to help people who like to use
2642 syntax highlighting in their ILISP buffer. (This patch
2643 was originally due to Raymond Toy.)
2644 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2645 patches originally written by Raymond Toy.
2646 ** The compiler understands coercion between single-value and
2647 multiple-VALUES type expressions better, getting rid of some very
2648 weird behavior, thanks to patches originally by Robert MacLachlan
2649 and Douglas Crosher.
2650 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2651 lambda lists, thanks to a patch originally by Pierre Mai.
2652 ** The system no longer bogusly warns about "abbreviated type
2654 ** The compiler gets less confused by inlining and RETURN-FROM,
2655 thanks to some patches originally by Tim Moore.
2656 ** The system no longer hangs when dumping circular lists to fasl
2657 files, thanks to a patch originally from Douglas Crosher.
2658 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2659 undefined function error.
2660 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2661 in structure slot initforms) for now, documented workaround instead:-|
2662 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2663 * fixed bug 5: added stubs for various Gray stream functions called
2664 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2665 installed, at least appropriate type errors are generated
2666 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2667 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2668 consistently in DEFMETHOD forms.
2669 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2670 it doesn't seem to affect SBCL after all
2671 * The C runtime system now builds with better optimization and many
2672 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2674 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2676 * The system is now under CVS at SourceForge (instead of the
2677 CVS repository on my home machine).
2678 * The new signal handling code has been tweaked to treat register
2679 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2680 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2681 downstream, e.g. in debug-int.lisp, has implicit dependencies
2682 on the unsignedness of integer representation of machine words,
2683 and that caused the system to bomb out with infinite regress
2684 when trying to recover from type errors involving signed values,
2685 e.g. (BUTLAST '(1 2 3) -1).)
2686 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2687 (This was one of the bugs Peter Van Eynde reported back in July.)
2688 * The system now uses code inspired by Colin Walters' O(N)
2689 implementation of MAP (from the cmucl-imp@cons.org mailing
2690 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2691 inline the MAP operation, and there is more than one
2692 sequence argument to the MAP call (so that it can't just
2693 do ETYPECASE once and for all based on the type of the
2694 single sequence argument). (The old non-inline implementation
2695 of the general M-argument sequence-of-length-N case required
2696 O(M*N*N) time when any of the sequence arguments were LISTs.)
2697 * The QUIT :UNIX-CODE keyword argument has been renamed to
2698 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2699 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2701 * Raymond Wiker's patches to port dynamic loading from Linux to
2702 FreeBSD have been added.
2703 * The BUGS file is now more nearly up to date, thanks in large part
2704 to Martin Atzmueller's review of it.
2705 * The debugger now flushes standard output streams before it begins
2706 its output ("debugger invoked" and so forth).
2707 * The core version number and fasl file version number have both
2708 been incremented, because of incompatible changes in the layout
2710 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2711 used to not be called for a saved Lisp image.)
2712 * Martin Atzmueller's version of a patch to fix a compiler crash,
2713 as posted on sbcl-devel 13 September 2000, has been installed.
2714 * Instead of installing Martin Atzmueller's patch for the
2715 compiler transform for SUBSEQ, I deleted the compiler transform,
2716 and transforms for some similar consing operations.
2717 * A bug in signal handling which kept TRACE from working on OpenBSD
2719 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2720 away by constant folding
2721 * The system now defines its address space constants in one place
2722 (in the Lisp sources), and propagates them automatically elsewhere
2723 (through GENESIS and the sbcl.h file). Therefore, patching the
2724 address map is less unnecessarily tedious and error-prone. The
2725 Lisp names of address space constants have also been systematized.
2726 * CVS tags like dollar-Header-dollar have been removed from
2727 the sources, because they have never saved me trouble and
2728 they've been source of trouble working with patches and other
2729 diff-related operations.
2730 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2731 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2733 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2735 * The system has been ported to OpenBSD.
2736 * The system now compiles with a simple "sh make.sh" on the systems
2737 that it's supported on. I.e., now you no longer need to tweak
2738 text in the target-features.lisp-expr and symlinks in src/runtime/
2739 by hand, the make.sh takes care of it for you.
2740 * The system is no longer so grossly inefficient when compiling code
2741 involving vectors implemented as general (not simple) vectors (VECTOR T),
2742 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2743 which dares to use the various sequence functions on non-simple
2744 vectors, takes less of a performance hit.
2745 * There is now a primitive type predicate VECTOR-T-P
2746 to test for the (VECTOR T) type, so that e.g.
2747 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2748 can now be compiled with some semblance of efficiency. (The old code
2749 turned the type declaration into a full call to %TYPEP at runtime!)
2750 * AREF on (VECTOR T) is still not fast, since it's still compiled
2751 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2752 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2753 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2754 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2755 without making full calls to SUBTYPEP at runtime.
2756 (Some analogous efficiency issues for non-simple vectors specialized to
2757 element types other than T, or for non-simple multidimensional arrays,
2758 have not been addressed. They could almost certainly be handled the
2759 same way if anyone is motivated to do so.)
2760 * The changes in array handling break binary compatibility, so
2761 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2762 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2763 instead of (VALUES T T).
2764 * By following the instructions that Dan Barlow posted to sbcl-devel
2765 on 2 July 2000, I was able to enable primitive dynamic object
2766 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2767 functionality is not implemented (since it calls ld to resolve
2768 library references automatically, requiring RUN-PROGRAM for its
2769 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2770 try to resolve library references) is now supported.
2771 * The system now flushes the standard output streams when it terminates,
2772 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2773 them at several other probably-convenient times, e.g. in each pass of
2774 the toplevel read-eval-print loop, and after evaluating a form given
2775 as an "--eval" command-line option. (These changes were motivated by a
2776 discussion of stream flushing issues on cmucl-imp in August 2000.)
2777 * The source transform for TYPEP of array types no longer assumes
2778 that an array whose element type is a not-yet-defined type
2779 is implemented as an array of T, but instead punts, so that the
2780 type will be interpreted at runtime.
2781 * There is now some support for cross-compiling in make.sh: each of
2782 the phases of make.sh has its own script. (This should be transparent
2783 to people doing ordinary, non-cross-compile builds.)
2784 * Since my laptop doesn't have hundreds of megabytes of memory like
2785 my desktop machine, I became more motivated to do some items on
2786 my to-do list in order to reduce the size of the system a little:
2787 ** Arrange for various needed-only-at-cold-init things to be
2788 uninterned after cold init. To support this, those things have
2789 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2790 symbols with such names are now uninterned after cold init).
2791 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2792 fasl files for cold load.
2793 ** Remove the old compiler structure pooling code (which used to
2794 be conditional on the target feature :SB-ALLOC) completely.
2795 ** Redo the representation of some data in cold init to be more compact.
2796 (I also looked into supporting byte compiled code at bootstrap time,
2797 which would probably reduce the size of the system a lot, but that
2798 looked too complicated, so I punted for now.)
2799 * The maximum signal nesting depth in the src/runtime/ support code has
2800 been reduced from 4096 to 256. (I don't know any reason for the very
2801 large old value. If the new smaller value turns out to break something,
2802 I'll probably just bump it back up.)
2803 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2805 * Many, many bugs reported by Peter Van Eynde have been added to
2806 the BUGS list; some have even been fixed.
2807 * While enabling dynamic object file loading, I tried to make the
2808 code easier to understand, renaming various functions and variables
2809 with less ambiguous names, and changing some function calling
2810 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2811 * While trying to figure out how to do the OpenBSD port, I tried to
2812 clean up some of the code in src/runtime/. In particular, I dropped
2813 support for non-POSIX signal handling, added various comments,
2814 tweaked the code to reduce the number of compilation warnings, and
2815 renamed some files to increase consistency.
2816 * To support the new automatic configuration functionality in make.sh,
2817 the source file target-features.lisp-expr has been replaced with the
2818 source file base-target-features.lisp-expr and the machine-generated
2819 file local-target-features.lisp-expr.
2820 * fixed a stupid quoting error in make.sh so that using CMU CL
2821 "lisp -batch" as cross-compilation host works again
2823 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2825 * DESCRIBE no longer tries to call itself recursively to describe
2826 bound/fbound values, so that it no longer fails on symbols which are
2827 bound to themselves (like keywords, T, and NIL).
2828 * DESCRIBE now works on generic functions.
2829 * The printer now prints less-screwed-up representations of closures
2830 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2831 * A private symbol is used instead of the :EMPTY keyword previously
2832 used to mark empty slots in hash tables. Thus
2833 (DEFVAR *HT* (MAKE-HASH-TABLE))
2834 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2835 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2836 now does what ANSI says that it should. (You can still get
2837 similar noncompliant behavior if bang on the hash table
2838 implementation with all the symbols you get back from
2839 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2840 This breaks binary compatibility, since tests for equality to
2841 :EMPTY are wired into things like the macroexpansion of
2842 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2844 * There's now a minimal placeholder implementation for CL:STEP,
2845 as required by ANSI.
2846 * An obscure bug in the interaction of the normal compiler, the byte
2847 compiler, inlining, and structure predicates has been patched
2848 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2849 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2851 * Missing ordinary arguments in a macro call are now detected even
2852 when the macro lambda list contains &KEY or &REST.
2853 * The debugger no longer complains about encountering the top of the
2854 stack when you type "FRAME 0" to explicitly instruct it to go to
2855 the top of the stack. And it now prints the frame you request even
2856 if it's the current frame (instead of saying "You are here.").
2857 * As specified by ANSI, the system now always prints keywords
2858 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2860 * The default initial SIZE of HASH-TABLEs is now smaller.
2861 * Type information from CLOS class dispatch is now propagated
2862 into DEFMETHOD bodies, so that e.g.
2863 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2865 is now basically equivalent to
2866 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2867 (DECLARE (TYPE SINGLE-FLOAT X))
2869 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2870 operation, without having to do run-time type dispatch.
2871 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2872 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2873 * Since I don't seem to be making big file reorganizations very often
2874 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2875 is overflowing my ability to conveniently back them up), I've finally
2876 checked the system into CVS. (The CVS repository is on my home system,
2877 not at SourceForge -- putting it on SourceForge might come later.)
2878 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2879 high-level GC-NOTIFY-FOO functions send their output. (There's
2880 still very little control of where low-level verbose GC functions
2881 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2882 less than it used to -- the GC-NOTIFY-FOO functions are now under
2883 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2884 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2885 in only one place in the source code, and propagates it automatically
2886 everywhere that it's needed. Thus e.g. when I bump the version from
2887 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2888 * The C source files now include boilerplate legalese and documentation
2889 at the head of each file (just as the Lisp source files already did).
2890 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2891 to his page will be replaced with a link to his new CLiki service.
2893 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2895 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2896 * The build process now looks for GNU make under the default name "gmake",
2897 instead of "make" as it used to. If GNU make is not available as "gmake"
2898 on your system, you can change this default behavior by setting the
2899 GNUMAKE environment variable.
2900 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2901 can build without error under CMU CL.
2903 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2905 * There is now a partial SBCL user manual (with some new text and some
2906 text cribbed from the CMU CL manual).
2907 * The beginnings of a profiler have been added (starting with the
2908 CMU CL profiler and simplifying and cleaning up). Eventually the
2909 main interface should be through the TRACE macro, but for now,
2910 it's still accessed through vaguely CMU-CL-style functions and macros
2911 exported from the package SB-PROFILE.
2912 * Some problems left over from porting CMU CL to the new
2913 cross-compilation bootstrap process have been cleaned up:
2914 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2915 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2916 part when getting rid of the compiler *BACKEND* stuff.)
2917 ** The value of *NULL-TYPE* was screwed up, because it was
2918 being initialized before the type system knew the final
2919 definition of the 'NULL type. This screwed up several key
2920 optimizations in the compiler, causing inefficiency in all sorts
2921 of places. (I found it because I wanted to understand why
2922 GET-INTERNAL-RUN-TIME was consing.)
2923 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2924 PROCLAIM FTYPE information. Unfortunately this broke binary
2925 compatibility again, since now the forms output by DEFGENERIC
2926 to refer to functions which didn't exist in 0.6.3.
2927 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2928 can use the new (as of 0.6.3) transform for SOME into MAP into
2930 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2931 values to (INTEGER 0 1000000), so that the time code will no longer
2932 occasionally get blown up by Linux returning 1000000 microseconds
2933 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2934 its output conform to the ANSI spec. (Alas, this makes its output
2935 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2936 * A full call to MAP NIL with a single sequence argument no longer conses.
2937 * fixes to problems pointed out by Martin Atzmueller:
2938 * The manual page no longer talks about multiprocessing as though
2939 it were currently supported.
2940 * The ILISP support patches have been removed from the distribution,
2941 because as of version 5.10.1, ILISP now supports SBCL without us
2942 having to maintain patches.
2943 * added a modified version of Raymond Toy's recent CMU CL patch for
2944 EQUALP comparison of HASH-TABLE
2946 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2948 * The system still can't cross-compile itself with
2949 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2950 entails), but at least it can compile more of itself that way
2951 than it used to be able to, and various buglets which were uncovered
2952 by trying to cross-compile itself that way have now been fixed.
2953 * This release breaks binary compatibility again. This time
2954 at least I've incremented the FASL file format version to 2, so that the
2955 problem can be detected reliably instead of just causing weird errors.
2956 * various new style warnings:
2957 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2958 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2959 ** using the lexical binding of a variable named in the *FOO* style
2960 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2961 as specified by ANSI.
2962 * *RANDOM-STATE* is no longer automatically initialized from
2963 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2964 default behavior of the system is to repeat its behavior every time
2965 it's run. If you'd like to change this behavior, you can always
2966 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2967 old convention there was no comparably easy way to get the system to
2968 repeat its behavior every time it was run.
2969 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2970 FTYPE declarations has been removed, in favor of their ANSI
2971 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2972 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2973 the types of their sequence arguments can be determined at compile time.
2974 This is done through a new open code expansion for MAP which eliminates
2975 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2976 argument types can be determined at compile time.
2977 * The optimizer now transforms COERCE into an identity operation when it
2978 can prove that the coerced object is already of the correct type. (This
2979 can be a win for machine generated code, including the output of other
2980 optimization transforms, such as the MAP transform above.)
2981 * Credit information has been moved from source file headers into CREDITS.
2982 * Source file headers have been made more standard.
2983 * The CASE macro now compiles without complaining even when it has
2986 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2988 * (Note that the way that the PCL macroexpansions were rewritten
2989 to accommodate the change in DEFGENERIC below breaks binary
2990 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2991 not run under sbcl-0.6.2. Once we get out of alpha releases,
2992 i.e. hit release 1.0.0, we'll probably try to maintain binary
2993 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2994 and sbcl-1.4.4. Until then, however, it might be fairly common
2995 for maintenance releases to break binary compatibility.)
2996 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2998 * The reporting of unhandled signals has been changed to print some
2999 explanatory text as well as the report form. (Previously only
3000 the report form was printed.)
3001 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3002 it defines, so that the compiler no longer issues undefined function
3003 warnings for compiled-but-not-yet-loaded generic functions.
3004 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3005 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3006 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3007 section of the ANSI spec are supported.
3008 * The "" nickname for the "KEYWORD" package has been removed.
3009 The reader still handles symbol tokens which begin with a package marker
3010 as keywords, but it doesn't expose its mechanism for doing so in the
3011 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3012 * The system now issues STYLE-WARNINGs for contradictory TYPE
3013 proclamations. (Warnings for contradictory FTYPE proclamations would
3014 be nice too, but those can't be done usefully unless the type system
3015 is made smarter about FUNCTION types.)
3016 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3017 systematized, so that "*target-*.lisp is supposed to exist only on the
3018 target and imply that there's a related file which exists on the
3019 host, and *host-*.lisp is supposed to exist only on the host and imply
3020 that there's a related file which exists on the target. This involves a
3021 lot of renaming. Hopefully the acute confusion caused by the renaming
3022 will be justified by the reduction in chronic confusion..
3023 ** runtime-type.lisp -> early-target-type.lisp
3024 ** target-type.lisp -> late-target-type.lisp
3025 ** early-host-format.lisp -> early-format.lisp
3026 ** late-host-format.lisp -> late-format.lisp
3027 ** host-error.lisp -> misc-error.lisp
3028 ** early-error.lisp -> early-target-error.lisp
3029 ** late-error.lisp -> late-target-error.lisp
3030 ** host-defboot.lisp -> early-defboot.lisp
3031 ** code/misc.lisp -> code/target-misc.lisp
3032 ** code/host-misc.lisp -> code/misc.lisp
3033 ** code/numbers.lisp -> code/target-numbers.lisp
3034 ** code/early-numbers.lisp -> numbers.lisp
3035 ** early-host-type.lisp -> early-type.lisp
3036 ** late-host-type.lisp -> late-type.lisp
3037 ** host-typep.lisp -> typep.lisp
3038 ** load.lisp -> target-load.lisp
3039 ** host-load.lisp -> load.lisp
3040 ** host-disassem.lisp -> disassem.lisp
3041 ** host-insts.lisp -> insts.lisp
3042 ** byte-comp.lisp -> target-byte-comp.lisp
3043 ** host-byte-comp.lisp -> byte-comp.lisp
3044 ** host-signal.lisp -> signal.lisp
3045 ** host-defstruct.lisp -> defstruct.lisp
3046 ** late-target-type.lisp -> deftypes-for-target.lisp
3047 Furthermore, several other previously target-only files foo.lisp (e.g.
3048 hash-table.lisp and random.lisp) have been split into a target-and-host
3049 foo.lisp file and a target-only target-foo.lisp file, with their key type
3050 definitions in the target-and-host part, so that the cross-compiler will
3051 know more about target types.
3052 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3053 gone away. In their place are various *BACKEND-FOO* variables
3054 corresponding to the slots of the old structure.
3055 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3056 into the target SBCL has been fixed.
3057 * The chill.lisp system for loading cold code into a running SBCL
3059 * Support for the CMU CL "scavenger hook" extension has been removed.
3060 (It was undocumented and unused in the CMU CL sources that SBCL was
3061 derived from, and stale in sbcl-0.6.1.)
3062 * Various errors in the cross-compiler type system were detected
3063 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3064 (enabling various consistency checks). Many of them were fixed,
3065 but some hard problems remain, so the compiler is back to
3066 running without *TYPE-SYSTEM-INITIALIZED* for now.
3067 * As part of the cross-compiler type system cleanup, I implemented
3068 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3069 * I have started adding UNCROSS calls throughout the type system
3070 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3071 on cross-compiler input files will be able to go away, and various
3073 * CONSTANTP now returns true for quoted forms (as explicitly required
3076 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3078 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3079 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3080 (MAKE-LIST -1). In the longer term, it ought to become true
3081 that declarations are assertions even at SAFETY 1. For now, it's not
3082 quite true even at SAFETY 3, but it's at least more nearly true..
3083 (Note that this change seems to increases the size of the system by
3084 O(5%) and to decrease the speed of the compiler by 20% or more.)
3085 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3086 for the problem of printing dozens of lines of distracting information
3087 about low-level system machinery as part of the top stack frame
3088 on entry to the debugger when an undefined function was called.
3089 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3090 is not reset to COMMON-LISP-USER.
3091 * Compilation of stuff related to dyncount.lisp has been made conditional
3092 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3093 smaller. The various dyncount-related symbols have been moved into
3094 a new "SB-DYNCOUNT" package.
3095 * tty-inspect.lisp has been renamed to inspect.lisp.
3096 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3097 feature has gone away. (When we eventually port to other flavors of
3098 libc and/or Unix, we'll try to make the differences between flavors
3099 invisible at the user level.)
3100 * Various other *FEATURES* tags, and/or their associated conditionals,
3101 have been removed if obsolescent, or given better documentation, or
3102 sometimes given more-mnemonic names.
3104 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3106 * tidied up "make.sh" script
3107 * tidied up system directory structure
3108 * better "clean.sh" behavior
3109 * added doc/FOR-CMUCL-DEVELOPERS
3110 * many many small tweaks to output format, e.g. removing possibly-confusing
3111 trailing #\. character in DESCRIBE-INSTANCE
3112 * (EQUALP #\A 'A) no longer signals an error.
3113 * new hashing code, including EQUALP hashing
3114 * tidied up Lisp initialization and toplevel
3115 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3116 * command line argument processing
3117 * added POSIX-GETENV function to deal with Unix-ish environment variables
3118 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3119 terminating SBCL on EOF
3120 * non-verbose GC by default
3121 * There is no more "sbcl" shell script; the sbcl file is now the C
3122 runtime executable (just like CMU CL).
3123 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3125 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3126 symbol and package information as Lisp native symbol and package objects
3127 instead of strings naming symbols and strings naming packages. This way,
3128 whenever packages are renamed (as in warm init), debug information is
3129 transformed along with everything else.
3130 * tweaked the optimization policy declarations which control the building
3131 of SBCL itself. Now, among other things, the system no longer saves
3132 source location debugging information. (This helps two problems at once
3133 by reducing SBCL size and by keeping SBCL from trying to look for its
3134 sources -- which may not exist -- when reporting errors.)
3135 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3136 debugging and testing purposes
3137 * cleaned up printing, making the printer call PRINT-OBJECT for
3138 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3139 methods, giving nearly-ANSI behavior
3140 * converted almost all special variables to use *FOO* naming convention
3141 * deleted PARSE-TIME functionality, since it can be done portably
3142 * moved some files out of cold init into warm init
3143 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3145 * regularized formatting of source files
3146 * added an install.sh script
3147 * fixed ridiculous memory usage of cross-compiler by making
3148 compiler/alloc.lisp not try to do pooling unless it can hook
3149 itself into the GC of the cross-compilation host. Now the system
3150 builds nicely on my old laptop.
3151 * added :SB-ALLOC in target-features.lisp-expr
3152 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3153 was not implemented)
3154 * re-did condition handling and note reporting in the compiler. Notes
3155 are no longer handled by signalling conditions. Style warnings
3156 and warnings are handled more correctly and reported in such a way
3157 that it's easy to find one or the other in your output (so that you
3158 can e.g. figure out which of many problems caused COMPILE-FILE to
3160 * changed the severity of several compiler warnings from full WARNING
3161 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3162 compiler note reporting so that it doesn't use the condition system
3163 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3164 COMPILE-FILE command)
3165 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3166 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3167 no longer supported, since their implementation was deeply intertwingled
3168 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3169 * removed bogus "support" for compiler macros named (SETF FOO), and
3170 removed the compiler macro for SETF INFO (but only after making a fool
3171 of myself on the cmucl-imp mailing list by posting a bogus patch for
3172 DEFINE-COMPILER-MACRO..)
3173 * Compiled files containing forms which have side effects on the Lisp
3174 reader (such as DEFPACKAGE forms) are now handled more correctly.
3175 (Compiler queuing of top level lambdas has been suppressed by setting
3176 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3177 * deleted various currently-unused source files, e.g. gengc.lisp. They
3178 may be added back at some point e.g. when porting to other architectures,
3179 but until they are it's distracting to distribute them and to try to
3181 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3182 there were so many of them they're just distractions, and UNCROSS is
3183 known to be able to handle the current sources
3184 * moved PROFILE functionality into TRACE, so that it will be clear
3185 how the wrapping and unwrapping of functions when you profile them
3186 interacts with the wrapping and unwrapping of functions when you
3187 trace them. (Actually, the functionality isn't there yet, but at least
3188 the interface specification is there. Hopefully, the functionality will
3189 arrive with some future maintenance release.)
3190 * removed host-oops.lisp
3191 * changed signature of QUIT function to allow UNIX-CODE argument
3192 * fixed READ-SEQUENCE bug
3193 * tweaked verbose GC output so that it looks more like the progress
3194 output that ANSI specifies for functions like LOAD
3195 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3196 * added <http://sbcl.sourceforge.com> to the banner information printed by