1 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2 * incompatible change: the top level REPL now has only an ABORT
3 restart associated with it, not TOPLEVEL and ABORT as it used to.
4 TOP and TOPLEVEL are now available as debugger commands for
5 returning to the top level.
6 * incompatible change: forms evaluated in the REPL now use the
7 global optimization policy.
8 * incompatible change: user- and system-initialization files are
9 no longer processed with LOAD, but by READ and EVAL; hence the
10 global optimization policy, startup package, readtable, etc,
12 * internal entry point details and argument counts no longer appear
13 in backtraces unless explicitly requested by setting
14 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
15 * built-in and standard functions no longer have names like "top
16 level local call to FOO".
17 * fixed bug 32: functions defined in non-null lexical environments
18 now have more legible printed representation
19 * fixed bug 33: functions defined in non-null lexical environemnts
20 are now more amenable to inspection by INSPECT.
21 * workaround for bug 354: XEPs no longer appear in backtraces unless
23 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
24 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
25 (reported by Lutz Euler)
26 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
27 compiled code. (reported by Paul Dietz)
28 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
29 specialized parameter to a method to something that is not TYPEP
30 the specializer is now possible.
31 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
32 face of package deletion.
33 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
34 pathname. (reported by David Tolpin on c.l.l)
35 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
36 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
37 STRING-STREAM and FILE-STREAM, respectively.
38 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
39 * fixed some bugs revealed by Paul Dietz' test suite:
40 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
42 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
43 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
44 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
46 * fixed loading of multiply forward-referenced layouts.
47 (thanks to Cheuksan Wang)
48 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
49 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
50 (see COMPILE-FILE documentation for details of :PRINT option)
51 * fixed bugs 19 and 317: fixed-format floating point printing is
52 more accurate. This also fixes a bug reported by Adam Warner
53 related to the ~@F format directive.
54 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
56 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
57 dispatch table if the from-char is a dispatch macro character.
58 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
59 by the wordsize no longer ignore the last word. (reported by Lutz
61 * fixed bug in type checking of dynamic-extent variables. (reported
63 * optimization: sequence traversal functions use their freedom to
64 coerce function designators to functions.
65 * optimization: code with many calls to CLOS methods specialized on
66 CLOS classes has had redundant type checks removed.
67 * optimization: added declarations to speed up operations that access
68 the internal character database (for example STRING-UPCASE)
69 * optimization: comparison operations between floats and sufficiently small
70 fixnums no longer create extra rationals
71 * fixed some bugs related to Unicode integration:
72 ** portions of multibyte characters at the end of buffers for
73 character-based file input are correctly transferred to the
74 start of the buffer at the next read.
75 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
76 passing it through to OPEN.
77 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
78 argument given to internal calls to OPEN.
79 ** the built-in comment readers (introduced by character sequences
80 ";" and "#|") are more forgiving to encoding errors; they will
81 STYLE-WARN and then attempt to resync the stream at a character
82 boundary. (thanks to Teemu Kalvas)
83 * fixed some bugs in the x86-64 port:
84 ** Negative short int return values from c-calls are sign-extended
86 ** The stack is aligned to 16-bytes for c-calls, as required by
87 the ABI. (thanks to Cheuksan Wang)
88 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
89 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
90 for enums. (thanks to Vincent Arkesteijn)
91 ** Multiple small optimizations and bugfixes for floating point
93 * fixed some bugs revealed by Paul Dietz' test suite:
94 ** Space, Tab, Linefeed, Return and Page have the invalid
95 secondary constituent character trait.
96 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
98 ** WITH-INPUT-FROM-STRING should only update the index place on
100 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
102 ** Bit-array operations (BIT-AND and similar) worked incorrectly
103 with one-dimensional arrays with fill pointers.
104 ** TYPE-OF failed on a complex with an integer realpart and a
106 ** compiler failure during type inference for the code of form
107 (IF (EQL X (THE ...)) ...) (MISC.535).
109 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
110 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
111 platform. The port supports 61-bit fixnums, large memory spaces
112 and reloading shared object files.
113 * enhancement: saving cores with foreign code loaded is now
114 supported on ppc/Darwin in addition to the previously supported
116 * enhancement: the statistical profiler now walks deeper into the
117 call stack for more meaningful call-graphs and accrued time
118 reports (x86/x86-64 only). It also now reports time spent in
120 * enhancement: it is now possible to trace most individual methods
121 of a generic function in addition to tracing the generic function
123 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
125 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
126 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
127 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
128 produces an error. (thanks to Vincent Arkesteijn)
129 * bug fix: NAMESTRING on pathnames with :WILD components in their
130 directories works correctly. (thanks to Artem V. Andreev)
131 * fixed bug 125: compiler preserves identity of closures. (reported
133 * bug fixed: functions with &REST arguments sometimes failed with
134 "Undefined function" when compiled with (DEBUG 3). (reported by
136 * bug fix: overflow during compiling of setting element of a bit
137 vector with constant index and value. (reported by Timmy Douglas)
138 * build fix: fixed the dependence on *LOAD-PATHNAME* and
139 *COMPILE-FILE-PATHNAME* being absolute pathnames.
140 * on x86 compiler partially supports stack allocation of dynamic-extent
142 * GO and RETURN-FROM do not check the extent of their exit points
143 when compiled with SAFETY 0.
144 * fixed some bugs related to Unicode integration:
145 ** encoding and decoding errors are now much more robustly
146 handled; it should now be possible to recover even from invalid
147 input or output to the terminal. (thanks to Teemu Kalvas)
148 ** provided a first cut at implementing STRING-TO-OCTETS and
149 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
150 ** altered the SB-MD5 contributed module slightly, changing the
151 interface just enough for it to be supportable for builds where
152 lisp characters are not eight bits.
153 * fixed some bugs revealed by Paul Dietz' test suite:
154 ** the FORMATTER-generated functions for ~V[ conditionals require
155 the correct number of arguments.
156 ** READ-FROM-STRING returns the mandated second value when applied
157 to displaced strings.
158 ** the #\Rubout and #\Backspace characters are treated as invalid
159 constituent characters by the tokenizer.
161 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
162 * new feature: reloading changed shared object files with
163 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
164 * new feature: references to foreign variables and functions
165 can now be compiled and loaded before the corresponding shared
166 object file is loaded, as long as the foreign definitions are
167 available at runtime.
168 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
169 supported. (thanks to Dan Debertin)
170 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
171 just keywords. (thanks to Vincent Arkesteijn)
172 * fixed bug #331: structure-class instances corresponding to
173 DEFSTRUCT forms are now created eagerly.
174 * fixed bug #345: backtraces from calls to undefined functions work
175 on x86 as well. Related bug #61 is now also partially fixed on x86
176 (backtraces from throws to unknown catch tags.)
177 * bug fix: lambda-list parsing is now stricter vrt. order and number
178 of lambda-list keywords.
179 * bug fix: as specified by AMOP, an error is signalled if a
180 class-option appears multiple times in a DEFCLASS form. (reported
182 * bug fix: RANDOM can be compiled when the compiler derives the type
183 of its numeric argument as a disjoint set of small integers.
184 (reported by Paul Dietz)
185 * bug fix: starting a core saved with shared objects loaded when
186 those objects are not available no longer causes threaded SBCL to
187 hang. (reported by Sean Ross)
188 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
190 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
191 argument, which must always be NIL. (reported by Kalle Niemitalo)
192 * bug fix: printing 1.0d+23 no longer results in an error.
193 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
194 * bug fix: structure slot setters preserve evaluation order. (thanks
196 * fixed some bugs related to Unicode integration:
197 ** RUN-PROGRAM can allow its child to take input from a Lisp
198 stream. (reported by Stefan Scholl)
199 ** ASDF-INSTALL successfully downloads .tar.gz files in all
200 locales. (reported by Ken Causey)
201 * fixed some bugs revealed by Paul Dietz' test suite:
202 ** INCF, DECF and REMF evaluate their place form as specified in
204 ** FORMATTER expands ~{ ~} iteration directives with V or #
205 parameters correctly.
206 ** FORMATTER deals with the ~@[ ~] conditional directive where the
207 consequent uses no arguments correctly.
208 ** the system has a partial understanding of the (COMPLEX RATIO)
210 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
213 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
214 * new feature: a build-time option (controlled by the :SB-UNICODE
215 keyword feature, enabled by default) for building the system with
216 support for the entire 21-bit character space defined by the
218 * new feature: the system now has rudimentary external-format
219 support; the primary user-visible change at this time is that
220 characters with the high bit set (such as lower-case-e-acute) will
221 print correctly to a terminal in a UTF-8 environment.
222 * minor incompatible change: BASE-CHAR no longer names a class;
223 however, CHARACTER continues to do so, as required by ANSI.
224 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
225 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
226 has been moved to the SB-EXT package (temporarily re-exported from
228 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
229 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
231 * fixed bug #351: better error-handlind and reporting for malformed
233 * fixed bug #350: bignum-printing is now more memory-efficient,
234 allowing printing of very large bignums, eg. (expt 2 10000000).
235 (reported by Bruno Haible)
236 * fixed bug #302: better primitive-type selection for intersection
238 * fixed bug #308: non-graphic characters now all have names, as
239 required. (reported by Bruno Haible)
240 * bug fix: redefining a class with different superclasses now correctly
241 removes it from the direct-subclasses of its previous superclasses.
242 (reported by David Morse)
243 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
244 new value now works. (reported by Bruno Haible)
245 * bug fix: correct canonicalization of multiple non-standard slot
246 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
247 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
248 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
249 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
250 now exists, an signals an error.
251 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
252 during macroexpansion for non-iteration variables as well. (reported
253 by Bruno Haible for CMUCL)
254 * bug fix: Cyclic structures and unprintable objects in compiler
255 messages no longer cause errors. (reported by Bruno Haible)
256 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
257 and READ-FROM-STRING all now return a primary value of NIL if
258 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
259 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
260 (reported by Bruno Haible for CMUCL)
261 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
262 do not designate a condition. (reported by Bruno Haible for
264 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
265 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
266 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
267 permitting resolution in favour of any of the conflicting symbols.
268 (reported by Bruno Haible for CMUCL)
269 * bug fix: EQUAL compiler optimizations is less aggressive on
270 strings which can potentially compare true despite having distinct
271 specialized array element types.
272 * bug fix: unit enumerations can be defined without dividing by
273 zero. (thanks to Vincent Arkesteijn)
274 * FORMAT compile-time argument count checking has been enhanced.
275 (report from Bruno Haible for CMUCL)
276 * a partial workaround for the bug 262: the compiler does not try to
277 inline-expand a local function doing RETURN-FROM from a deleted
278 BLOCK. (thanks to Peter Denno for the bug report and to David
279 Wragg for the simple test case)
280 * fixed some bugs revealed by Paul Dietz' test suite:
281 ** RENAME-PACKAGE allows all package designators as new package
283 ** constraint propagation and lambda variable substitution are
284 more cautious in dealing with partially deleted code.
285 ** compiler sometimes lost reoptimization passes.
286 ** CERROR, given a condition as condition designator, passes its
287 remaining arguments to the continue format control without
289 ** Case-altering FORMAT directives work correctly on non-ASCII
291 ** The REAL type specifier handles bounds outside the floating
292 point ranges without signalling FLOATING-POINT-OVERFLOW.
293 ** Functions with IR1-transformations can create intercomponent
294 references to global functions.
295 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
297 ** FORMAT ~R treats a nil value for its first parameter correctly.
299 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
300 * enhancement: saving cores with foreign code loaded is now
301 supported on x86/NetBSD and sparc/Linux in addition to the previously
303 * bug fix: on some platforms repeated installations caused multiple
304 copies of HTML documentation to be installed -- should not happen
305 any more. (reported by Stefan Scholl)
306 * bug fix: parsing self-recursive alien record types multiple times
307 no longer causes infinite recursion. (reported by Thomas F. Burdick,
308 original patch by Helmut Eller for CMUCL)
309 * bug fix: stack-exhaustion detection works now on NetBSD as well.
310 (thanks to Richard Kreuter)
311 * bug fix: defining classes whose accessors are methods on existing
312 generic functions in other (locked) packages no longer signals
313 bogus package lock violations. (reported by François-René Rideau)
314 * bug fix: special variables as DEFMETHOD parameters no longer have
315 associated bogus type declarations. (reported by David Wragg and
317 * bug fix: read-write consistency on streams of element-type
318 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
319 * bug fix: redefiniton of the only method of a generic function with
320 no DEFGENERIC no longer emits a full WARNING. In addition,
321 redefinition of generic functions with no DEFGENERIC to an
322 incompatible lambda list now signals an error. (thanks to Zach
324 * bug fix: DEFGENERIC now works even when there's a function of the
325 same name in an enclosing lexical environment. (thanks to Zach
327 * fixed compiler failure, caused by instrumenting code during
328 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
329 * optimization: added loop analysis and improved register allocation
330 * fixed some bugs revealed by Paul Dietz' test suite:
331 ** POSITION on displaced vectors with non-zero displacement
332 returns the right answer.
333 ** (SIMPLE-STRING) is a valid type specifier for sequence
335 ** *PRINT-LEVEL* handling for slotless structures is pedantically
337 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
338 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
340 ** The justification version of the FORMAT ~< directive treats
341 non-zero minpad parameter correctly.
343 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
344 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
345 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
346 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
347 the supported interface.
348 * new feature: Single-stepping of code compiled with DEBUG 2 or
349 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
350 * new feature: saving cores with foreign code loaded is now
351 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
352 Timothy Moore's work for CMUCL)
353 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
354 parameters to * instead of NIL if no initform is supplied.
355 (reported by Johan Bockgaard)
356 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
357 correctly when FUNCALL appears as the car of the form. Note:
358 despite this FUNCALL forms are not currently subject to
359 compiler-macro expansion. (port of Raymond Toy's fix for the
360 same from CMUCL, reported by Johan Bockgaard)
361 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
362 (thanks for Teemu Kalvas)
363 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
364 case instead of terminating the iteration (thanks for Julian
365 Squires, Sean Champ and Raymond Toy)
366 * bug fix: incorrect expansion of defgeneric that caused a style
367 warning. (thanks for Zach Beane)
368 * bug fix: slot accessor effective method computation works properly
369 for classes with multiple non-standard applicable methods on
370 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
371 * on x86 compiler supports stack allocation of results of LIST and
372 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
373 CMUCL implementation by Gerd Moellmann)
374 * fixed some bugs revealed by Paul Dietz' test suite:
375 ** FORMAT strings with both the ~<~:;~> form of the justification
376 directive and pretty-printing directives cause an error.
378 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
379 * incompatible change: the internal functions
380 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
381 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
382 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
383 instead of the old functions.
384 * new feature: on platforms where "dladdr" is available foreign
385 function names now appear in backtraces. (based on Helmut Eller's
387 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
388 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
390 * documentation: ASDF manual is now distributed with SBCL.
391 * bug fix: SBCL can now load its contributed modules with REQUIRE
392 even if the system-provided entries have been removed from
393 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
395 * bug fix: backtraces involving undefined functions or assembly
396 routines are more informative. (thanks to Brian Downing)
397 * bug fix: mutually referent alien structures now work correctly.
398 (reported by Rick Taube)
399 * bug fix: structures defined by WITH-ALIEN can be referred to
400 within other definitions in the same WITH-ALIEN.
401 * bug fix: division operators (MOD, TRUNCATE and the like) with
402 constant zero divisors and integer dividends no longer generate
404 * bug fix: provide default methods for INPUT-STREAM-P and
405 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
406 * bug fix: improve the detection and resolution of MOP metacycles
407 (where effective method computation depends on the generic function
408 having its effective method computed). (reported by Bruno Haible)
409 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
410 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
411 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
412 * bug fix: pathnames with relative directory components can be
413 represented relative to default pathnames.
414 * optimization: in taking the GCD of bignums, reduce the two bignums
415 to approximately the same size (using Euclid's algorithm) before
416 applying the more sophisticated binary GCD. (thanks to Juho
418 * optimization: COUNT on bitvectors now operates word-at-a-time.
419 * optimization: ASH with a positive, but not necessarily constant,
420 (leftwards) shift, when in a modular context, is compiled to a
422 * fixed some bugs revealed by Paul Dietz' test suite:
423 ** FORMAT variable parameters ("~V<char>") are defaulted properly
424 if the corresponding argument is NIL.
425 ** FORMAT directives accepting a minpad parameter treat negative
427 ** literal commas as character parameters to FORMAT directives are
429 ** literal spaces directly after ~<Newline> directives within a
430 format-logical-block (~:< ~@:>) do not induce :FILL-style
431 conditional newlines.
432 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
433 as specified: it no longer includes conditional newlines).
434 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
436 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
437 :PER-LINE-PREFIX argument does not evaluate to a string.
439 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
440 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
441 the manual for details; for now, package locks can be disabled by
442 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
443 if no major problems are found then it is likely that they will be
444 compiled in unconditionally.
445 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
446 unsupported operators on all platforms. To load a shared library
447 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
448 object file, link it into a shared library outside of SBCL and
449 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
451 * bug workaround: ROOM T and the GC were not working together
452 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
453 reporting facility aren't preserved. That reporting has been
454 disabled, so now until and unless someone figures out how to make it
455 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
456 * minor incompatible change: as threatened around sbcl-0.8.0, the
457 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
458 The symbols are also exported from SB-PCL for backwards
459 compatibility, but more so than before SB-PCL should be treated as
460 an implementation-internal package.
461 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
463 * the SB-SPROF contrib now works on (most) non-x86 architectures.
464 It is known as of this release not to work on the Alpha, however.
465 * fixed bug #167: errors signalled due to illegal syntax in method
466 bodies are now more legible.
467 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
468 designators and can hence be used with TYPEP.
469 * fixed bug #333: CHECK-TYPE now ensures that the type error
470 signalled, if any, has the right object to be accessed by
471 TYPE-ERROR-DATUM. (reported by Tony Martinez)
472 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
473 5.1.2.3 for multiple-value place subforms. (reported by Kalle
475 * fixed bug #334: programmatic addition of slots using specialized
476 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
477 and :ALLOCATION :CLASS slots.
478 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
479 float exponents. (rereported by Peter Seibel)
480 * fixed a bug: #\Space (and other whitespace characters) are no
481 longer considered to be macro characters in standard syntax by
483 * fixed bug: initialization of condition class metaobjects no longer
484 causes an instance of the condition to be created. (reported by
486 * fixed bug: it is now possible to have slots such that
487 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
488 description is neither :INSTANCE nor :CLASS.
489 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
490 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
491 SLOT-VALUE-USING-CLASS), and no longer causes errors with
492 non-standard SLOT-DEFINITION-ALLOCATION values.
493 * fixed bugs: various race conditions handling exiting threads.
494 CL-PPCRE's thread torture test now passes.
495 * fixed arguably-a-bug: GC time proportional to number of threads in
496 system even when most of them are idle
497 * optimization: improved performance of BIT and SBIT on bit-vectors.
498 * .fasl file incompatibility: The fasl file version number has
499 been incremented because of changes associated with package locks.
501 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
502 * minor incompatible change: the system no longer provides
503 optimization hints (conditions which are TYPEP
504 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
505 mode; these hints are still emitted when higher SPEED optimization
507 * new contrib module: a sampling profiler (profiling by statistical
508 sampling, rather than by instrumenting functions) is available as
509 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
510 * the behaviour of the standard function ED is now customizeable by
511 third parties through a hook variable: see ED's documentation
512 string for information on the protocol.
513 * the compiler no longer emits efficiency notes for (FUNCALL X)
514 when the type of X is uncertain under default optimization
516 * fixed bug 276: mutating a binding of a specialized parameter to a
517 method to something that is not TYPEP the specializer is now
519 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
520 exhibit the required behaviour on the x86 platform. (thanks to
521 Peter van Eynde, Eric Marsden and Bruno Haible)
522 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
523 even for difficult arguments. (reported by Peter Graves)
524 * fixed bug 141a: the backquote printer now descends quoted
526 * fixed another bug in backquote printing: no more destructive
527 modification of the form's list structure. (reported by Brian
529 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
530 move between its address being taken and the call to
531 interrupt_thread, fixing a crashing race condition.
532 * the SB-POSIX contrib implementation has been adjusted so that it
533 no longer exhibits ridiculously poor performance when constructing
534 instances corresponding to C structs.
536 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
537 * minor incompatible change: the sb-grovel contrib now treats C
538 structures as alien (in the sense of SB-ALIEN) objects rather than
539 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
540 has implications for memory management of client code
541 (sb-grovel-returned objects must now be manually managed) and for
542 type safety (alien objects now have full types).
543 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
544 used to control emission of compiler diagnostics, rather than the
545 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
546 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
547 quality should be considered deprecated.
548 * install.sh now installs the user manual as well
549 * (not quite a new documentable feature, but worth considering in
550 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
551 beginnings of a semantically meaningful condition hierarchy is
552 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
554 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
555 designator as the defaults argument.
556 * fixed bug: Displaced arrays whose displaced-to array has become
557 too small now cause ARRAY-DIMENSION to signal an error, providing
558 for safer bounds-checking. (reported by Bruno Haible)
559 * fixed bug: DEFCLASS slot definitions with identical :READER and
560 :WRITER names now signal a reasonable error. (reported by Thomas
562 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
564 * fixed bug: Invalid slot specification errors now print correctly.
565 (thanks to Zach Beane)
566 * fixed bug 320: Shared to local slot value transfers in class
567 redefinitions now happen corrently from superclasses as
568 well. (reported by Bruno Haible)
569 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
571 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
572 improper lists correctly. (reported by Bruno Haible)
573 * fixed bug 313: source-transform for <fun-name> was erroneously
574 applied to a call of a value of a variable with name <fun-name>.
575 (reported by Antonio Menezes Leitao)
576 * fixed bug 307: The obsolete instance protocol ensures that
577 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
578 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
579 remove dead unknown-values globs from the middle of the stack.
580 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
581 easier to resume long computations after using *BREAK-ON-SIGNALS*
582 to diagnose and fix failures. (thanks to Nikodemus Siivola)
583 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
584 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
585 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
586 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
587 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
589 * fixed a (fairly theoretical) bug in string printing: if
590 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
591 does not have array-element-type equal to the most general string
593 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
594 function name. (thanks to Nikodemus Siivola)
595 * fixed bug: compile-time format string checker failed on
596 non-closed ~{. (reported by Thomas F Burdick)
597 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
598 don't warn on legal loop code involving more than one aggregate
599 boolean. (thanks to Nikodemus Siivola)
600 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
601 in keyword position to MAKE-PACKAGE was not being reported as
603 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
604 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
606 * fixed bugs 280 and 312: the checking for multiple definitions in a
607 file is less likely to become confused by uses of inline
609 * fixed bug: the #S reader macro performs the keyword coercion
610 specified for slot names. (reported by Kalle Niemitalo)
611 * fixed bug: lambda lists may contain symbols whose names start with
612 & but are not lambda-list-keywords; their occurrence triggers a
614 * fixed bug 321: define-method-combination argument lambda lists do
615 not cause invalid code to be generated when &optional or &aux
616 variables have default values. (reported by Bruno Haible)
617 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
618 have CLOS classes; this appears also to have fixed reported
619 instability in stack exhaustion detection.
620 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
621 MUFFLE-WARNING when no associated restart is present is now
623 * optimization: rearranged the expansion of various defining macros
624 so that each expands into only one top-level form in a
625 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
627 * optimization: used a previously little-used slot in symbols to
628 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
630 * fixed some bugs revealed by Paul Dietz' test suite:
631 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
633 ** Fixed an optimization bug related to inheritance of initforms
634 from local to shared slots.
635 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
636 ** CAST optimizer forgot to flush argument derived type.
637 ** print/read consistency on floats is now orders of magnitude
638 more likely. (thanks also to Bruno Haible for a similar report
640 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
641 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
642 for objects of type REAL. Make it so.
643 ** SXHASH is specified (infelicitously) to respect similarity,
644 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
645 it so. (thanks to Markus Ziegler)
646 ** on the Alpha, the compiler succeeds in compiling functions
647 returning a known number of arguments greater than 63.
648 ** fixed handling of invalid NIL arguments in keyword position
650 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
652 ** provided a readably-printable representation for RANDOM-STATE
654 ** ensured that pathnames, where they have a namestring, always
655 print using #P"..." syntax.
657 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
658 * Support for the forthcoming 2.0 version of the NetBSD kernel
659 running on x86 hardware has been added. (thanks to Perry
660 E. Metzger most immediately, and others for their past work)
661 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
662 current release version; SBCL's OpenBSD support had been broken
663 since about the time of OpenBSD's switch to ELF binary format.)
664 * [placeholder for DX summary]
665 ** user code with &REST lists declared dynamic-extent, under high
666 speed or space and low safety and debug optimization policy.
667 * The manual has been converted to Texinfo format and the debugger
668 chapter from the cmucl manual has been added.
669 * A facility has been added to extract documentation strings from
670 sbcl and store them as Texinfo-formatted snippets for inclusion in
671 the manual (via Texinfo's @include directive)
672 * bug fix: compiler emitted division in optimized DEREF. (thanks for
673 the test case to Dave Roberts)
674 * bug fix: multidimensional simple arrays loaded from FASLs had fill
675 pointers. (reported by Sean Ross)
676 * bug fix: PROFILE output is printed nicely even for large numerical
677 values. (thanks to Zach Beane)
678 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
679 greater than 32 handle EOF correctly.
680 * bug fix: on X86 an immediate argument of the IMUL instruction is
681 correctly printed in disassembly. (thanks to Lutz Euler)
682 * bug fix: class slots in redefined classes preserve their old
683 values. (thanks to Bruno Haible and Nikodemus Siivola)
684 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
685 succeeds. (reported by Marco Baringer)
686 * fixed some bugs revealed by Paul Dietz' test suite:
687 ** READ-SEQUENCE now works on ECHO-STREAMs.
688 ** RATIONALIZE works more according to its specification. (thanks
691 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
692 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
693 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
694 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
695 should matter to you only if you rebind the printer control
696 variables and then find you want different bindings in the
697 debugger than in the ordinary execution of your program.)
698 * The runtime build system has been tweaked to support building
699 (on SPARC/SunOS) using a C compiler which invokes Sun's own
700 assembler and linker. (thanks to Nikodemus Siivola)
701 * Unbound, undefined, undeclared variables now trigger full
702 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
703 more often programmer error than deliberate exploitation of undefined
705 * optimization: the hash algorithm for strings has changed to one
706 that is less vulnerable to spurious collisions. (thanks to Juho
708 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
709 needless bounds checking. (thanks to Juho Snellman)
710 * optimization: implemented multiplication as a modular
711 (UNSIGNED-BYTE 32) operation on the PPC backend.
712 * fixed some bugs revealed by Paul Dietz' test suite:
713 ** ADJUST-ARRAY now copies the datum in a zero rank array if
715 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
716 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
717 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
719 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
721 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
723 ** ECHO-STREAMs no longer attempt to echo the end of file value to
724 their output stream on EOF from read.
725 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
726 have been read to end-of-file.
727 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
729 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
730 description of determination of which consecutive characters
732 ** Printing the "Space" character with escaping on now yields "#\\ ",
733 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
734 ** Reading floating-point numbers with *READ-BASE* set to a number
735 less than 10 works correctly.
736 ** Reading floating-point numbers with *READ-BASE* set to a number
737 more than 10 works correctly.
738 ** Printing with *PRINT-READABLY* targets the standard readtable, not
739 the readtable currently in effect.
741 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
742 * minor incompatible change: parsing of namestrings on a physical
743 (Unix) host has changed; numbers after the final #\. in a
744 namestring are no longer interpreted as a version field. This is
745 intented to be largely invisible to the user, except that the
746 meaning of the namestring "*.*.*" has changed: it now refers to a
747 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
748 should usually be replaced by
749 (make-pathname :name :wild :type :wild :version :wild)
750 with the added benefit that this is more likely to be portable.
751 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
752 option to OPEN now signals an error if the file being opened
753 exists; this may have an impact on existing code.
754 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
755 signals received, on the PowerPC platforms (both Linux and
756 Darwin). (thanks to Pierre Mai for pointing out the location of
758 * several fixes on OS X: The system now builds and runs cleanly on
759 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
760 (thanks to Brian Mastenbrook)
761 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
762 second-resolution: integer multiples of 1/3600 between -24 and 24.
763 (thanks to Vincent Arkesteijn)
764 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
765 type when called with 1 argument; PEEK-CHAR checked type of
766 PEEK-TYPE only after having read first character from a
767 stream. (reported by Peter Graves)
768 * bug fix: the garbage collector now has much better locality
769 behaviour, and in particular no longer treats all memory as being
770 exhausted when many small objects point to each other in a deeply
772 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
773 recognized as being TYPEP their class.
774 * bug fix: the PUSHNEW documentation string has been corrected.
775 (thanks to Vincent Arkesteijn)
776 * bug fix: defaulting of the value for the last of an atypically
777 large number of multiple values being bound was not being
778 performed correctly on the Alpha or PPC platforms
779 * optimization: implemented multiplication as a modular
780 (UNSIGNED-BYTE 32) operation on the x86 backend.
781 * optimization: SEARCH on simple-base-strings can now be open-coded.
782 (see also contrib/compiler-extras.lisp for inspiration for
783 teaching the compiler about the Boyer-Moore algorithm).
784 * value, returned by MAX (and MIN) called with several EQUALP, but
785 not EQL, arguments now does not depend on compiler settings.
786 * fixed some bugs revealed by Paul Dietz' test suite:
787 ** in stack analysis liveness information is propagated from
788 non-local entry points.
789 ** pathwise CAST removing failed when the CAST node did not start
791 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
792 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
794 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
795 host is already defined.
796 ** RENAME-FILE works on streams instead of signalling an internal
798 ** PEEK-CHAR uses the current readtable when determining whether
799 or not a character is whitespace.
800 ** MERGE-PATHNAMES handles the case when the pathname does not
801 specify a name while the default-pathname specifies a version.
802 ** Pathnames now stand a better chance of respecting print/read
804 ** Attempting to use standardized file system operators with a
805 pathname with invalid :DIRECTORY components signals a
807 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
809 ** DIRECTORY on logical pathnames is more correct.
810 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
811 signal a TYPE-ERROR if their argument is not a stream.
812 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
813 designator argument does not designate a stream.
814 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
815 examining the synonym.
816 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
818 ** OPEN and WITH-OPEN-STREAM allow opening streams with
819 element-type larger than ([UN]SIGNED-BYTE 32).
821 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
822 * When built with the :SB-FUTEX feature, threaded builds now take
823 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
824 for faster/more reliable mutex and condition variable support.
825 * Incompatible change (but one you probably shouldn't have been using
826 anyway): the interface and code for arbitrating between multiple
827 threads in the same user session has been redesigned.
828 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
829 error when called without an explicit environment argument.
830 (thanks to Nikodemus Siivola)
831 * bug fix: buffered :DIRECTION :IO streams are less likely to become
832 confused about their position. (thanks to Adam Warner and Gerd
834 * bug fix: Pretty printing backquoted forms with unquotations in the
835 argument list position of various code constructs such as LAMBDA
836 now works correctly. (reported by Paul Dietz)
837 * bug fix: Pretty printing unquotations no longer loses all
838 stream position information.
839 * optimization: performance of string output streams is now less
840 poor for multiple small sequence writes.
841 * optimization: performance of CSUBTYPEP in the presence of complex
842 expressions involving CONS and NOT many times has been improved.
843 (reported by Paul Dietz)
844 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
846 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
847 streams for the REPL, for the debugger, and so on. (thanks to
849 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
850 * fixed some bugs revealed by Paul Dietz' test suite:
851 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
852 is ignored if it is too small, rather than propagating through
853 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
854 ** extremely complex negations of CONS types were not being
855 sufficiently canonicalized, leading to inconsistencies in
857 ** VALUES tranformer lost derived type.
859 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
860 * fixed a bootstrapping bug: the build process no longer assumes
861 that the various BOOLE-related constants have the same value in
862 host and target lisps. (noted by Paul Dietz' test suite on an
863 SBCL binary built from CLISP)
864 * The system can now be dynamically linked on the MIPS platform,
865 which enables dynamic loading of foreign code from Lisp. (thanks
866 to Ralf Baechle for discussions on the MIPS ABI)
867 * The system now records debugging information for its own source
868 files in a filesystem-position-independent manner, relative to
869 the "SYS" logical host.
870 * fixed a compiler bug: MV-LET convertion did not check references
871 to the "max args" entry point. (reported by Brian Downing)
872 * tweaked disassembly notes to be less confident about proclaiming
873 some instruction as an LRA. (thanks to Brian Downing)
874 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
875 now each have their own history, command character, and other
876 characteristics. (thanks to David Lichteblau)
877 * fixed some bugs revealed by Paul Dietz' test suite:
878 ** compiler failure in compiling LOGAND expressions including a
880 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
881 shift greater than 32.
882 ** FLUSH-DEST did not mark blocks for type check regeneration.
883 ** HANDLER-CASE failed to accept declarations in handler clauses
884 in some circumstances.
886 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
887 * New code in contrib/sb-introspect (still probably not entirely
888 stable yet) provides some support for smart Lisp development
889 environments like SLIME.
890 * The conditions signalled for errors occurring when loading .fasl
891 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
892 in a way which should help ASDF recover gracefully.
893 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
894 stuff has been cleaned up. If you code contrib/ stuff, this might
895 affect you, and you can look at contrib/README, contrib/STANDARDS,
896 and/or the 0.8.4.27 diff to check.
897 * In full calls the compiler now does not generate checks for declared
898 argument types for all arguments.
899 * various threading fixes
900 ** and some experimental patches which didn't make it into
901 the main tree for this release, but which are shipped in
902 contrib/experimental-thread.patch as a possible fix for some
903 failures (deadlock, spinning...) in GC-intensive multithreaded
905 * fixed PPC build problem (source code incompatibility of different
906 library versions): added offsetof() hackery which attempts to divine
907 where glibc maintainers put uc_mcontext today
908 * fixed bug 282: compiler does not trust type assertions while passing
909 arguments to a full call.
910 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
911 &OPTIONAL VALUES type parameter.
912 * fix bug 214: algorithm for noting rejected templates is now more
913 similar to that of template seletion. (also reported by rydis on
915 * fixed bug 141b: printing backquoted information readably and prettily
916 inserts a space where necessary.
917 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
918 (CHARACTER 10) are now reported as errors, rather than propagated
919 as unknown types. (reported by piso on #lisp)
920 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
921 with values NIL and :ERROR. (thanks to Milan Zamazal)
922 * fixed bug 191c: CLOS now does proper keyword argument checking as
923 described in CLHS 7.6.5 and 7.6.5.1.
924 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
925 longer raise an error; further, using a list as a for-as-arithmetic
926 counter now raises a meaningful error.
927 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
928 now understood by sequence creation functions such as MAKE-SEQUENCE
930 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
931 a STRING-INPUT-STREAM.
932 * compiler enhancement: SIGNUM is now better able to derive the type
934 * type declarations inside WITH-SLOTS are checked. (reported by
936 * fixed some bugs revealed by Paul Dietz' test suite:
937 ** incorrect optimization of TRUNCATE for a positive first
938 argument and negative second.
939 ** compiler failure in let-convertion during flushing dead code.
940 ** compiler failure while deriving type of TRUNCATE on an
941 interval, containing 0.
942 ** ASH of a negative bignum by a negative bignum count now returns
944 ** intersection of CONS types now canonicalizes properly, fixing
945 inconsistencies in SUBTYPEP.
947 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
948 * incompatible change: The --disable-debugger command line
949 option now clobbers the debugger at a more fundamental
950 level, by redefining #'INVOKE-DEBUGGER instead of by
951 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
952 is specified by ANSI to ignore *DEBUGGER-HOOK* and
953 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
954 enter the debugger REPL and then suffer recursive errors
955 because *DEBUG-IO* is also messed up in --disable-debugger mode;
956 while under the new system, BREAK in --disable-debugger mode
957 terminates the system just as an unhandled error would.
958 * fixed compiler performance when processing loops with a step >1;
959 * bug fix: DOCUMENTATION now retrieves generic function
960 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
961 support has been systematized, and now supports the methods
962 specified by ANSI, along with a default method and a method for
963 slot documentation. (reported by Nathan Froyd)
964 * bug fix: effective methods associated with a generic function are
965 no longer cached over a change of that generic function's method
966 combination. (reported by Andreas Fuchs)
967 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
968 names a non-existent directory. (thanks to Andreas Fuchs)
969 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
970 the PPC platform, lead to stack corruption; this has been fixed.
971 (reported by Rainer Joswig)
972 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
973 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
974 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
975 arithmetic in general, is now much more reliable on the PPC
977 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
978 platform now returns the right answer.
979 * optimization: restored some effective method precomputation in
980 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
981 precomputation is now tunable.
982 * optimization: compiler-internal data structure use has been
983 reviewed, and changes have been made that should improve the
984 performance of the compiler by about 20%.
985 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
986 simple-base-strings and simple-bit-vectors is improved.
987 * optimization: the optimization of 32-bit logical and arithmetic
988 functions introduced in version 0.8.3 on the x86 has been
989 implemented on the mips, ppc and sparc platforms; an
990 implementation of the same facility, but for 64-bit arithmetic,
991 has been added for the alpha.
992 * microoptimization: the compiler is better able to make use of the
993 x86 LEA instruction for multiplication by constants.
994 * bug fix: in some situations compiler did not report usage of
995 generic arithmetic in (SPEED 3) policy.
996 * bug 145b fix: compiler used wrong type specifier while converting
997 MEMBER-types to numeric.
998 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1000 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1001 index argument now works properly on the Alpha platform.
1002 * bug fix: floating point exception treatment on the Alpha platform
1004 * bug fix: FILE-POSITION works much better on string input and
1005 output streams. (thanks to Nikodemus Siivola)
1006 * bug fix: many threading/garbage collection symptoms sorted.
1007 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1008 might be pseudo-atomic.
1009 * internal change: Stopping for GC is now done with signals not
1010 ptrace. GC is now done in whichever thread wanted it, instead of
1012 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1014 * bug fix: result form in DO is not contained in the implicit
1016 * incompatible change: ICR structure is changed; the value part of
1017 CONTINUATION is now called LVAR; corresponding functions are
1018 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1019 * added type deriver for ISQRT (thanks to Robert E. Brown).
1020 * bug fix: better support for loading from the command line when an
1021 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1023 * fixed some bugs revealed by Paul Dietz' test suite:
1024 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1025 ** ROUND and FROUND now give the right answer when given very
1026 small float arguments.
1027 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1029 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1030 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1031 ** LCM should return a non-negative integer.
1032 ** PARSE-INTEGER returned the index of a terminator instead of the
1033 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1034 ** PARSE-INTEGER returned an incorrect index being applied to a
1036 ** LCM with two arguments of 0 returns 0 rather than signalling
1038 ** unsigned addition of a 32-bit constant with the high bit set no
1039 longer causes an internal compiler error on the x86.
1040 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1041 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1042 with negative last argument.
1043 ** byte specifiers with zero size and position no longer cause
1044 an error during type derivation.
1045 ** bignum multiplication on the Alpha platform now returns the
1047 * porting: The system now builds on SuSE AMD64, although it still
1048 generates a 32-bit binary.
1049 * .fasl file incompatibility: The fasl file version number has
1050 been incremented (because of the changes to internal compiler
1051 data structures referred to above).
1053 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1054 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1055 more accurately, on the Darwin kernel running on PowerPC hardware.
1056 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1057 * Compiler code deletion notes now signal a condition of type
1058 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1059 an associated MUFFLE-WARNING restart.
1060 * The compiler now performs limited argument count validation of
1061 constant format strings in FORMAT, and where appropriate in ERROR,
1062 CERROR and WARN. (thanks to Gerd Moellmann)
1063 * New ASDF-INSTALL contrib can be used for automatic download and
1064 installation of third-party Lisp code from CCLAN or other sites
1066 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1067 which forces another thread to execute a function supplied by the
1069 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1070 now accept and act upon their :ELEMENT-TYPE keyword argument.
1071 (reported by Martin Atzmueller, Edi Weitz)
1072 * bug fix: FILE-POSITION now accepts position designators up to
1073 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1074 is the greater. (thanks to Patrik Nordebo)
1075 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1076 Kalle Olavi Niemitalo)
1077 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1078 printable. (reported by Eric Marsden)
1079 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1080 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1081 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1082 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1083 (thanks to Stig Sandoe)
1084 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1085 UNREAD-CHAR now correctly includes the unread character in the
1086 target sequence. (thanks to Gerd Moellmann)
1087 * bug fix in threaded builds: the system can now be suspended and
1088 resumed by shell job control with minimal disruption.
1089 * bug fixes in times and timezones >2038AD
1090 * better handling of "where is GNU make?" problem in build scripts
1091 (thanks to Nikodemus Siivola)
1092 * new optimization: inside a named function any reference to a
1093 function with the same name is considered to be a self-reference;
1094 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1095 optimization quality.
1096 * new optimization on x86: logical functions and + now have
1097 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1098 used when the result is truncated to 32 bits.
1099 * VALUES declaration is partially enabled.
1100 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1101 * fixed some bugs revealed by Paul Dietz' test suite:
1102 ** The system now obeys the constraint imposed by
1103 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1104 types form a lattice under type intersection.
1105 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1106 ** ASSOC now ignores NIL elements in an alist.
1107 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1108 and (1+ MOST-POSITIVE-FIXNUM) answers.
1109 ** The addition of a method with invalid qualifiers to a generic
1110 function does not cause an error to be signalled immediately;
1111 a warning is signalled, and the error is generated only on
1112 calling the generic function.
1113 * changed .fasl file version number, in part to add type codes for
1114 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1115 obscure ANSI requirements
1117 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1118 * fixed bug 148: failure to inline-expand a local function left
1119 garbage, confusing the compiler.
1120 * fixed bugs 3cd: structure slot readers perform type check if the
1121 slot can have an invalid value (i.e. it is either not initialized
1122 or can be written with a less specific slot writer).
1123 * bug fix: the compiler now traps array references to elements off
1124 the end of an array; previously, the bounds checking in some
1125 circumstances could go off-by-one.
1126 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1128 * type declarations for array element types now obey the description
1129 on the CLHS page "Declaration TYPE", as per discussions on
1130 sbcl-help around 2003-05-08. This means that a declaration
1131 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1132 declaration, all references to BAR will be asserted or assumed
1133 (with THE, so dependent on compiler policy) to involve objects of
1134 type FOO. Note that no such declaration is implied in
1135 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1136 * declared types of functions from the "Conditions"
1137 chapter. (reported by Paul Dietz)
1138 * bug fix: CERROR accepts a function as its first argument.
1139 * bug fix: NTH an NTHCDR accept a bignum as index
1140 arguments. (reported by Adam Warner)
1141 * optimization: character compare routines now optimize comparing
1142 against a constant character. (reported by Gilbert Baumann)
1143 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1144 argument now works properly on the MIPS platform.
1145 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1146 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1147 declared to have a functional type, does not cause an error
1149 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1150 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1151 * fixed bug: if last continuation of a deleted block has a
1152 destination, this destination should be deleted too. (reported by
1154 * fixed a bug in the bootstrap process: the host compiler's values
1155 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1156 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1157 test case from Patrik Nordebo)
1158 * improved the ability of the disassembler on the PPC platform to
1159 provide helpful disassembly notes.
1160 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1161 the class in more cases than previously.
1162 * bug fix: FILE-POSITION now understands :START and :END for
1163 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1164 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1165 8) data. (thanks to David Lichteblau)
1166 * bug fix: it is possible to add a method to a generic function
1167 without lambda list.
1168 * bug fix: reader failed to signal END-OF-FILE inside an
1169 object representation. (reported by Nikodemus Siivola)
1170 * fixed some bugs revealed by Paul Dietz' test suite:
1171 ** LAST and [N]BUTLAST should accept a bignum.
1172 ** condition slot accessors are methods.
1173 ** (VECTOR NIL) is a subtype of STRING.
1175 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1176 * minor incompatible change: some nonsensical specialized lambda
1177 lists (used in DEFMETHOD) which were previously ignored now signal
1179 * minor incompatible change: the system is now aware of the types of
1180 variables in the COMMON-LISP package, and will signal errors for
1181 most violations of these type constraints (where previously they
1182 were silently accepted).
1183 * minor incompatible change: COMPILE-FILE now uses the freedom
1184 afforded (ANSI 3.2.2.3) to use derived function types for
1185 functions defined in the same file. This also permits the system
1186 to warn on static type mismatches and function
1187 redefinition. (Currently it does not work with high DEBUG level.)
1188 * minor incompatible change: VALUES declaration is disabled.
1189 * When issuing notes, the compiler now signals a condition of type
1190 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1191 restart for use in user handlers. It is expected that the
1192 COMPILER-NOTE condition will eventually become a condition
1193 supertype to a hierarchy of note types, which will then be
1194 handleable in a similar fashion. However, at the moment, no such
1195 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1196 but it's an implementation detail, not a classification for the
1198 * Changes in type checking closed the following bugs:
1199 ** type checking of unused values (192b, 194d, 203);
1200 ** template selection based on unsafe type assertions (192c, 236);
1201 ** type checking in branches (194bc).
1202 * A short form of VALUES type specifier has ANSI meaning (it has
1203 increased the number of situations when SBCL cannot perform type
1205 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1206 keywords or constants is permissible.
1207 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1208 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1209 classes. (thanks to Antonio Martinez)
1210 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1211 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1212 argument to a pretty-print stream. Instead, it leaves any such
1213 operations to DESCRIBE-OBJECT methods.
1214 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1216 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1217 lambda lists. (thanks to David Lichteblau)
1218 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1219 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1220 (thanks to Gerd Moellmann)
1221 * fixed bug 246: increased compilation speed of long
1222 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1224 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1226 * DEFCONSTANT now throws a condition of type
1227 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1228 constant to a non-EQL value; CONTINUE and ABORT restarts
1229 respectively change and preserve the value.
1230 * fixed bug 63: The code walker, part of the implementation of CLOS,
1231 is now better at handling symbol macros.
1232 * bug fix: There is no longer an internal implementation type named
1233 CL:LENGTH. (reported by Raymond Toy)
1234 * bug fix: In macro-like defining macros/special operators the
1235 implicit block does not enclose the lambda list.
1236 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1237 enclosed in parentheses) are not suitable as type specifiers, and
1238 their use properly signals an error now.
1239 * bug fix: An argument count mismatch for a type specifier in code
1240 being compiled no longer causes an unhandled error at compile
1241 time, but signals a compile-time warning.
1242 * fixed simple vector readable printing
1243 * bug fix: DESCRIBE takes more care over whether the class
1244 precedence list slot of a class is bound before accessing it.
1245 (reported by Markus Krummenacker)
1246 * bug fix: FORMATTER can successfully compile pretty-printer format
1247 strings which use variants of the ~* directive inside.
1248 * bug fix: SEARCH now applies its TEST predicate to the elements of
1249 the arguments in the correct order. (thanks to Wolfhard Buss)
1250 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1251 in type inference. (thanks to Robert E. Brown)
1252 * bug fix: Reading in symbols with an explicit package name of ""
1253 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1254 * fixed some bugs revealed by Paul Dietz' test suite:
1255 ** NIL is now allowed as a structure slot name.
1256 ** Arbitrary numbers, not just REALs, are allowed in certain
1257 circumstances in LOOP for-as-arithmetic clauses.
1258 ** Multiple class redefinitions before slot access no longer
1259 causes a type error.
1260 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1261 association between the name and a class.
1262 ** Generic functions with non-standard method-combination and over
1263 five methods all of which return constants no longer return NIL
1264 after the first few invocations. (thanks to Gerd Moellmann)
1265 ** CALL-NEXT-METHOD with no arguments now passes the original
1266 values of the arguments, even in the presence of assignment.
1267 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1269 ** DISASSEMBLE works with closures and funcallable instances.
1270 ** ADD-METHOD now returns the generic function, not the new method.
1271 ** FIND-METHOD signals an error if the lengths of the specializers
1272 is incompatible with the generic function, even if the ERRORP
1274 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1275 which its argument is a member.
1276 ** DEFCLASS only redefines the class named by its class-name
1277 argument if that name is the proper name of the class;
1278 otherwise, it creates a new class.
1279 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1280 of the UNBOUND-SLOT condition to the name of the slot.
1281 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1282 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1283 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1284 SLOT-MAKUNBOUND in the specified fashion.
1286 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1287 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1288 cross-compilation host. As a consequence, we can now bootstrap our
1289 way up to SBCL starting with a bare gcc toolchain and human-readable
1290 source code (first the source to CLISP, then the source to SBCL).
1291 * A contributed module containing a partial implementation of the
1292 simple-streams interface has been included. (thanks to Rudi
1294 * A contributed module implementing the RFC1321 Message Digest
1295 Algorithm, known as MD5, has been included.
1296 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1297 no longer has any effect, as the code controlled by this feature
1298 has been deleted. (As far as we know, no-one has ever built using
1299 this feature, and its semantics were confused in any case).
1300 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1301 work on conditions (as required by the ANSI specification),
1302 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1303 expected behaviour on conditions. Users should note, however,
1304 that such behaviour is not required by the ANSI specification,
1305 and so use of this behaviour may render their code unportable.
1306 * fixed some bugs revealed by Paul Dietz' test suite:
1307 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1309 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1310 ** SXHASH on condition objects no longer returns NIL.
1311 ** :ALLOCATION :CLASS slots are better treated; their values are
1312 updated on class redefinition, and initforms inherited from
1313 superclasses are applied.
1314 ** REMOVE-METHOD returns its generic function argument even when
1315 no method was removed.
1316 ** SHARED-INITIALIZE now initializes the values of the requested
1317 slots, including those with :ALLOCATION :CLASS.
1318 ** ALLOCATE-INSTANCE now works on structure classes defined via
1319 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1321 ** SLOT-EXISTS-P now works on conditions, as well as structures
1323 ** MAKE-LOAD-FORM now has the required methods on
1324 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1325 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1326 keyword, and now implements the SLOT-NAMES argument.
1327 ** methods with &OPTIONAL arguments no longer allow too many
1328 arguments to be passed in the call without error.
1329 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1330 option is consistent with the required arguments of the generic
1331 function lambda list.
1332 * bug fix: REQUIRE accepts a string designator. (Thanks to
1334 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1335 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1336 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1337 * bug fix: The compiler now checks for duplicated variables in macro
1339 * bug fix: SETQ on globals returns the correct value.
1340 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1341 not a COMPILER-ERROR (followed by some other strange error on
1342 choosing the CONTINUE restart).
1343 * bug fix: make.sh and friends are now more consistent in the way that
1344 they look for GNU "make".
1346 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1347 * experimental native threads support (on x86 Linux >=2.4 only).
1348 This is not compiled in by default: you need to add :SB-THREAD to
1349 the target features. See the "Beyond ANSI" chapter of the manual
1351 * fix for longstanding nonANSIism: The old distinction between
1352 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1353 The return value from CL:FIND-CLASS is now a CLOS class, and
1354 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1355 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1356 * An interface to the MetaObject Protocol, as described in Kiczales,
1357 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1358 MIT Press, 1991, is available from the SB-MOP package.
1359 * incompatible change: the SB-PCL package should now be considered
1360 a private implementation detail, and no longer a semi-private MOP
1362 * minor incompatible change: due to rearrangement for threads, the
1363 control stack and binding stack are now allocated at arbitrary
1364 addresses instead of being hardcoded per-port. Users affected by
1365 this probably have to be doing advanced things with shared
1366 libraries, and will know who they are.
1367 * minor incompatible change: Previously, all --eval forms used were
1368 processed with READ before any of them were processed with EVAL.
1369 Now each --eval form is processed with both READ and EVAL before
1370 the next --eval form is processed. (Thus package operations like
1371 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1372 work as the user might reasonably expect.)
1373 * minor incompatible change: *STANDARD-INPUT* is now only an
1374 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1376 * minor incompatible change: Y-OR-N-P is now character-oriented,
1377 not line oriented. Also, YES-OR-NO-P now works without errors.
1378 (thanks to Antonio Martinez)
1379 * sb-aclrepl module improvements: an integrated inspector, added
1380 repl features, and a bug fix to :trace command.
1381 * Known functions, which cannot be open coded by the backend, are
1382 considered to be able to check types of their arguments. (fixing
1383 a bug report by Nathan J. Froyd)
1384 * fixed a bug in computing method discriminating functions: It is
1385 now possible to define methods specialized on classes which have
1386 forward-referenced superclasses. (thanks to Gerd Moellmann)
1387 * fixed evaluation order in optional entries (reported by Gilbert
1389 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1390 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1391 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1392 required-by-AMOP NAME argument, as well as CLASS and
1393 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1394 * fixed bug 20: DEFMETHOD can define methods using names that are
1395 not the proper names of classes to designate class specializers.
1396 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1397 Unix file descriptors, instead of blowing up. (thanks to Antonio
1399 * Garbage collection refactoring: user-visible change is that a
1400 call to the GC function during WITHOUT-GCING will not do garbage
1401 collection until the end of the WITHOUT-GCING. If you were doing
1402 this you were probably losing anyway.
1403 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1404 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1405 TYPEP the latter but not the former.
1406 * The compiler issues a full WARNING for calls to undefined functions
1407 with names from the CL package.
1408 * MAP-INTO for a vector destination is open coded. (reported by
1409 Brian Downing on c.l.l)
1410 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1411 documentation string.
1412 * fixed some bugs revealed by Paul Dietz' test suite:
1413 ** COPY-ALIST now signals an error if its argument is a dotted
1415 ** Condition slots are now accessed more correctly in the presence
1416 of multiple initargs for a given slot.
1417 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1418 exclude restarts of the same name associated with a different
1420 ** DEFCLASS of forward-referenced classes with another
1421 forward-referenced class in the superclasses list no longer
1423 ** Condition slots are now initialized once each, not multiple
1424 times. (thanks to Gerd Moellmann)
1425 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1427 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1428 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1429 arguments contain duplicated elements.
1430 ** RESTART-CASE understands local macros.
1431 ** RESTART-CASE associates exactly its own restarts with a condition.
1432 ** ENDP in safe mode checks its argument to be of type LIST.
1433 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1434 in question is unbound.
1435 ** Optimized MAKE-INSTANCE functions no longer cause internal
1436 assertion failures in the presence of duplicate initargs.
1437 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1438 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1439 propagated even in the ctor optimized implementation of
1441 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1443 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1444 * a better implementation of SXHASH on (simple) bit vectors,
1445 measured both in execution speed and in distribution of results
1446 over the positive fixnums, has been installed. Likewise, a better
1447 implementation of EQUAL for simple bit vectors is now available.
1448 * fixed CEILING optimization for a divisor of form 2^k.
1449 * fixed bug 240 (emitting extra style warnings "using the lexical
1450 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1451 by Antonio Martinez)
1452 * fixed SXHASH, giving different results for NIL depending on type
1453 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1454 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1455 variable. (found by Rolf Wester)
1456 * SBCL does not ignore type declarations for special
1457 variables. (reported by rif on c.l.l 2003-03-05)
1458 * some bug fixes in contrib/sb-aclrepl/
1459 * fixed some bugs revealed by Paul Dietz' test suite:
1460 ** a bug in the CONS type specifier, whereby the CAR and CDR
1461 types got intertwined, has been fixed;
1462 ** the type system is now able to reason about the interaction
1463 between INTEGER and RATIO types more completely;
1464 ** APPEND, [N]REVERSE and NRECONC check that those their
1465 arguments, which must be proper lists, are really so;
1466 ** An array specialized to be unable to hold elements has been
1467 implemented, as required -- yes, really -- by ANSI;
1468 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1469 on malformed property lists;
1471 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1472 * incompatible packaging change: in line with Unix convention,
1473 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1474 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1475 * REQUIRE and PROVIDE are now optionally capable of doing something
1476 useful. See the documentation string for REQUIRE.
1477 * infrastructure for a managed SBCL contrib system: contributed
1478 modules in this release include:
1479 ** the ASDF system definition facility;
1480 ** an interface to the BSD Sockets API;
1481 ** an ACL-like convenience interface to the repl;
1482 (thanks to Kevin Rosenberg)
1483 ** an implementation of ROTATE-BYTE, with efficient implementation
1485 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1486 gives a closer approximation to the right answer than previously.
1487 (thanks to Raymond Toy)
1488 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1489 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1490 optional environment arguments, as required by ANSI.
1491 * fixed bugs in other functions taking environment objects, allowing
1492 calls with an explicit NIL environment argument to be compiled
1494 * fixed bug 228: primary return values from
1495 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1496 COMPILE or FUNCTION.
1497 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1498 :TYPE will no longer signal errors on innocuous objects.
1499 * fixed bug 231b: SETQ is better at respecting type declarations in
1500 the lexical environment.
1501 * fixed a bug in DEFCLASS: classes named by symbols with no or
1502 unprintable packages can now be defined.
1503 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1504 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1505 * fixed compiler failure related to checking types of functions.
1506 (reported by Robert E. Brown)
1507 * the compiler is now much more consistent in its error-checking
1508 treatment of bounding index arguments to sequence functions: in
1509 (SAFETY 3) code, errors will be signalled in almost all cases if
1510 invalid sequence bounding indices are passed to functions defined
1511 by ANSI to operate on sequences.
1512 * fixed a bug in the build procedure: documentation of SBCL-specific
1513 packages is now preserved and available in the final Lisp image.
1514 * lifted FDEFINITION lookup out of loops in the implementation of
1515 many list operations. (thanks to Robert E. Brown)
1516 * fixed a bug in the reader: the #n# reader macro now works for
1517 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1518 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1519 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1520 E. Brown and Tony Martinez respectively)
1521 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1522 SLOT-BOUNDP: in particular, it is now able to optimize them much
1523 better, and is now not vulnerable to having packages renamed.
1524 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1525 always called when a slot is not present in an instance. (thanks
1527 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1528 longer signal an error on unreading a character following EOF on
1529 the previous constituent stream. (thanks to Tony Martinez)
1530 * fixed some bugs revealed by Paul Dietz' test suite:
1531 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1532 not just nonnegative fixnums;
1533 ** the logical bit-array operators such as BIT-AND now accept an
1534 explicit NIL for their "opt-arg" argument (to indicate a
1535 freshly-consed result bit-array);
1536 ** ELT now signals an error on an invalid sequence index in safe
1538 ** the type system is now cleverer about negations of numeric
1539 types, and consequently understands the BIGNUM and RATIO types
1541 ** the type system is now cleverer about the interaction between
1542 INTEGER and RATIO types: while bugs still remain, many more
1543 cases are accurately computed;
1544 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1545 if it is in the last clause;
1546 ** CONSTANTLY now correctly returns a side-effect-free function in
1548 ** DECLARE is no longer treated as a special-operator; in
1549 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1550 * incremented fasl file version number due to the change in the
1551 DEFSTRUCT-SLOT-DESCRIPTION structure.
1553 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1554 * minor incompatible change: code processed by the "interpreter" or
1555 EVAL now has a compilation optimization policy of (DEBUG 2)
1556 (changed from (DEBUG 1)) to improve debuggability of interactive
1557 development, and to allow the use of the debug RETURN command in
1559 * an experimental implementation of the RETURN command for the
1560 debugger has been included. (thanks to Frederik Kuivinen)
1561 * fixed bug 62: constraints were not propagated into a loop.
1562 * fixed bug in embedded calls of SORT (reported and investigated by
1564 * fixed some bugs revealed by Paul Dietz' test suite:
1565 ** printing and reading of arrays with some dimensions having
1566 length 0 (thanks to Gerd Moellmann);
1567 ** BOA constructor with &AUX argument without a default value does
1568 not cause a type error;
1569 ** CONSTANTP now returns true for all self-evaluating objects.
1571 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1572 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1573 accessors that are related by inheritance, as specified in the
1574 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1575 Valtteri Vuorikoski)
1576 * The compiler is now able to inline functions that were defined in
1577 a complex lexical environment (e.g. inside a MACROLET).
1578 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1580 * Support for the upcoming FreeBSD-5.0 release has been included.
1581 (thanks to Dag-Erling Smorgrav)
1582 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1583 effect when it is not in a toplevel context.
1584 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1585 stand a better chance of being correct. (thanks to Gerd
1587 * fixed bug in COERCE, which now signals an error on coercing a
1588 rational to a bounded real type which excludes the expected
1590 * The compiler is now able to derive types more accurately from the
1591 COERCE and COMPILE functions.
1592 * fixed bug 223: functional binding is considered to be constant
1593 only for symbols in the CL package.
1594 * fixed bug 231: SETQ did not check the type of a variable being set
1595 (reported by Robert E. Brown)
1596 * A new optimization for MAKE-INSTANCE has been included, fixing
1597 various bugs (including relating to :ALLOCATION :CLASS slots and
1598 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1600 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1601 ** As required by ANSI, LOOP now disallows anonymous collection
1602 clauses such as COLLECT I in conjunction with aggregate boolean
1603 clauses such as THEREIS (= I 1);
1604 ** LOOP now signals an error when any variable is reused in the
1605 same loop (including the potentially useful construct analogous
1606 to WITH A = 1 WITH A = (1+ A);
1607 ** IT is only a special loop symbol within the first clause of a
1608 conditional loop clause;
1609 ** LOOP with a typed iteration variable over a hashtable now
1610 signals a type error iff it should.
1611 * fixed some other bugs revealed by Paul Dietz' test suite:
1612 ** FILE-STREAM now names the class previously known as FD-STREAM;
1613 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1614 argument) no longer signals an error;
1615 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1616 of the slot symbol, rather than using the current package
1617 ((:CONC-NAME "") continues to intern the slot's name in the
1619 * incremented fasl file version number, because of the incompatible
1620 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1621 of the new implementation of DEFINE-COMPILER-MACRO.
1623 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1624 * Support for building SBCL for MIPS platforms running in
1625 little-endian mode has now been checked in, and basic
1626 functionality on said platforms verified.
1627 * minor incompatible change: PCL now records the pathname of a file
1628 in which methods and the like are defined, rather than its
1630 * minor incompatible change: TRUENAME now considers the truename of
1631 a file naming a directory to be the pathname with :DIRECTORY
1632 component indicating that directory.
1633 * minor incompatible change: a NAMED clause in the extended form of
1634 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1635 reason for the previous behaviour is unclear.
1636 * more systematization and improvement of CLOS and MOP conformance
1637 in PCL (thanks to Gerd Moellman and Pierre Mai):
1638 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1640 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1641 primary methods with no specializers;
1642 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1644 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1645 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1646 FORWARD-REFERENCED-CLASSes; error reporting on
1647 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1649 ** SXHASH on CLOS instances now uses a slot internal to the
1650 instance to return different numbers on distinct instances,
1651 while preserving the same return value through invocations of
1653 ** DEFMETHOD signals errors when methods with longer incongruent
1654 lambda lists are added to generic functions;
1655 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1656 CLASS, as specified in AMOP;
1657 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1658 on the order returned by the primary method for classes of
1659 class STANDARD-CLASS;
1660 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1661 * fixed some bugs shown by Paul Dietz' test suite:
1662 ** DOLIST puts its body in TAGBODY;
1663 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1665 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1666 value producing form;
1667 ** if more variables are given to PROGV than values, extra
1668 variables are bound and made to have no value;
1669 ** NSUBSTITUTE on list arguments gets the right answer with
1671 ** ELT signals an error of type TYPE-ERROR when the index argument
1672 is not a valid sequence index;
1673 ** LOOP signals (at macroexpansion time) an error of type
1674 PROGRAM-ERROR when duplicate variable names are found;
1675 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1676 ** FUNCALL of special-operators now cause an error of type
1678 ** PSETQ now works as required in the presence of side-effecting
1679 symbol-macro places;
1680 ** NCONC accepts any object as its last argument;
1681 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1683 ** loop-for-as-package does not require a package to be explicitely
1685 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1687 * fixed bug 166: compiler preserves "there is a way to go"
1688 invariant when deleting code.
1689 * fixed bug 172: macro lambda lists with required arguments after
1690 &REST arguments now cause an error to be signalled. (thanks to
1692 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1693 bug. (thanks to Matthew Danish)
1694 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1696 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1697 arguments are lexically rebound. (thanks to Gerd Moellmann and
1699 * fixed bug 194: error messages are now more informative when there
1700 is no primary method applicable in a call to a generic
1701 function. (thanks to Gerd Moellmann)
1702 * fixed bug in command line argument checking (thanks to Julian
1704 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1705 (thanks to Matthew Danish)
1706 * incremented fasl file version number, because of the
1707 SXHASH-related changes in the layout of CLOS data structures
1709 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1710 * minor incompatible change: The runtime (the Unix executable named
1711 "sbcl") is now much pickier about the .core files it will load.
1712 Essentially it now requires .core files to descend from the same
1713 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1714 as the runtime does. (The intent is to prevent the crashes which
1715 can occur, and which can even be reported as mysterious failures,
1716 when people patch the sources or change the build parameters
1717 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1718 sbcl and .core files.)
1719 * fixed bug: VALUES-LIST is no longer optimized away.
1720 * fixed bug 142: The FFI conversion of C string values to Lisp
1721 string values no longer conses excessively. (thanks to Nathan
1722 Froyd porting Raymond Toy's fix to CMU CL)
1723 * began to systematize and improve MOP conformance in PCL (thanks to
1724 Nathan Froyd, Gerd Moellman and Pierre Mai):
1725 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1727 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1728 ** FINALIZE-INHERITANCE is now called on class finalization;
1729 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1730 argument precedence order.
1731 * fixed bug 202: The compiler no longer fails on functions whose
1732 derived types contradict their declared type.
1733 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1734 so it can be non-toplevel.
1735 * The fasl file version number has changed (because of the new
1736 implementation of DEFMACRO).
1737 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1738 safe code, that any length requirement by their type-specifier
1739 argument is valid. The exceptions to this are described in bug
1741 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1742 functions now check the types of their inputs as required by ANSI.
1743 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1744 introduced symbol is DECLAREd to be SPECIAL.
1745 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1746 * fixed bug: :COUNT argument to sequence functions may be negative
1747 * fixed bug: body of DO-SYMBOLS may contain declarations
1748 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1749 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1750 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1751 symbol macro only once
1752 * fixed printing of call frame when argument list is unavailable
1753 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1754 * compiler no longer signals WARNING on unknown keyword
1757 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1758 * A beta-quality port to the mips architecture running Linux,
1759 based on the old CMUCL backend, has been made. It has been tested
1760 on a big-endian kernel, and works sufficiently well to be able to
1761 rebuild itself; it has not been tested in little-endian mode.
1762 * fixed an inconsistency between gencgc.c and purify.c which made
1763 dumping/loading .core files unreliable
1764 * fixed bug 120a: The compiler now deals correctly with IFs where
1765 the consequent is the same as the alternative, instead of
1766 misderiving the return type. (thanks to Alexey Dejneka)
1767 * fixed bug 113: Logical pathnames are now dumpable (the logical
1768 host is resolved at load-time, throwing an error if it is not
1770 * fixed bug 174: FORMAT's error message is slightly clearer when a
1771 non-printing character is used in a format directive.
1772 * fixed several bugs in compiler checking of type declarations, i.e.
1773 violations of the Python "declarations are assertions" principle
1774 (thanks to Alexey Dejneka)
1775 * fixed several bugs in PCL's error checking (thanks to Gerd
1777 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1779 * fixed bug in compilation of functions as first class values
1780 (thanks to Antonio Martinez-Shotton)
1781 * The compiler's handling TYPE-ERRORs which it can prove will
1782 inevitably happen at runtime has been cleaned up and corrected
1783 in several ways. (thanks to Alexey Dejneka)
1784 * improved argument type checking for various basic arithmetic
1785 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1786 lavished on them in the past that they can be compiled in many
1787 ways in different special cases
1788 * fixed bug 181: compiler checks validity of user supplied type
1790 * cleaned up code flushing in optimization: Function calls which
1791 should signal errors for safety purposes (e.g. which ANSI says
1792 should signal errors when their arguments are of incorrect type)
1793 are no longer optimized away.
1794 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1795 * incremented fasl file version number, because changes in the
1796 implementation of sequence functions like COERCE caused
1797 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1798 in old inline expansions) to become undefined. (Actually these
1799 changes were later undone, so we might very well be binary
1800 compatible with 0.7.7 after all, but leaving the version number
1801 incremented seemed like the simplest and most conservative
1804 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1805 * An alpha-quality port to the parisc architecture running Linux,
1806 based on the old CMUCL backend, has been made. This, even more so
1807 than the other backends, should be considered still a work in
1808 progress; known problems include that the Linux kernel in 64-bit
1809 mode does not propagate the correct sigcontext structure to
1810 userspace, and consequently SBCL on a parisc64 kernel will not
1812 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1813 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1814 suppressed.) Also now that the compiler is looking at declarations
1815 in the environment, it checks optimization declarations as well,
1816 and suppresses inlining when (> DEBUG SPEED).
1817 * More fixes have been made to treatment of floating point exception
1818 treatment and other Unix signals. In particular, floating point
1819 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1820 * The detection and handling of control stack exhaustion (infinite
1821 or very deeply nested recursion) has changed. Stack exhaustion
1822 detection is now done by write-protecting pages at the OS level
1823 and applies at all optimization settings; when found, a
1824 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1825 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1826 be caught using IGNORE-ERRORS.
1827 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1828 correctly when transforming recursive LABELS functions to LETs.
1829 (thanks to Alexey Dejneka)
1830 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1831 that are names of constants or global variables.
1832 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1833 alien routines with docstrings.
1834 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1835 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1837 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1838 (thanks to Eric Marsden)
1839 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1840 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1841 * Bug fix: The SPARC backend can now compile functions involving
1842 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1843 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1844 to a functions expecting a PATHNAME-DESIGNATOR.
1845 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1846 lambda lists. (thanks to Alexey Dejneka)
1847 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1848 functions so that EXPORTing the name of the class doesn't cause
1849 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1850 to Antonio Martinez for reporting this)
1851 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1852 constant-folded without failing an assertion. (thanks to Einar
1853 Floystad Dorum for reporting this)
1854 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1855 types are now dealt with more correctly by the compiler.
1856 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1857 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1858 behaviour analogous to RENAME-FILE. This puts its behaviour more
1859 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1861 * The fasl file version number has changed again. (because of the
1862 bug fix involving the names of PCL MAKE-INSTANCE functions)
1864 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1865 * bug fix: Floating point exceptions are treated much more
1866 consistently on the x86/Linux and PPC/Linux platforms.
1867 * Array initialization with :INITIAL-ELEMENT is now much faster for
1868 cases when the compiler cannot open code the array creation, but
1869 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1870 array accesses have also seen a speed increase.
1871 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1872 specified. (This was at the root of some bad interactions between
1873 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1875 * bug fix: Internal error arguments for undefined functions are now
1876 computed correctly on the PPC/Linux platform.
1877 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1878 Raymond Toy's patch for CMU CL)
1879 * Support for the Solaris 9 operating environment has been included
1880 (thanks to Daniel Merritt)
1881 * A very ugly but hopefully complete draft of the missing FFI chapter
1882 of the manual has been created by reformatting the corresponding
1883 CMU CL manual chapter into (currently very ugly and incoherent)
1884 DocBook and bringing it up to date for SBCL behavior. Thus, the
1885 manual is now essentially complete, at least by my extreme
1886 once-and-only-once standards, whereby it's acceptable to refer to
1887 the doc strings of SB-EXT functions as the primary documentation.
1888 * The fasl file version number has changed again, due to cleanup of
1889 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1890 is no longer a static symbol.)
1892 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1893 * SBCL now builds with OpenMCL (version 0.12) as the
1894 cross-compilation host; also, more progress has been made toward
1895 bootstrapping under CLISP.
1896 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1898 * bug 158 fixed: The compiler can now deal with integer loop
1899 increments different from 1; fixing this turned out also to fix
1901 * bug 169 fixed: no more bogus warnings about using lexical bindings
1902 despite the presence of perfectly good SPECIAL declarations (thanks
1903 to David Lichteblau)
1904 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1905 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1906 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1908 * bug 180 fixed: Method combination specifications no longer ignore
1909 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1910 * bug fix: Structure type predicate functions now check their argument
1911 count as they should.
1912 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1913 correctly. (thanks to Pierre Mai)
1914 * minor incompatible change: The --noprogrammer option is deprecated
1915 in favor of the new --disable-debugger option, which is very similar.
1916 (The major difference is that it takes effect at a slightly different
1917 time at startup, causing handling of errors in --sysinit and
1918 --userinit files will be affected differently.) The
1919 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1920 been added to allow this functionality to be controlled from ordinary
1921 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1922 maintainers, who might want to run non-interactive scripts to
1923 build SBCL cores which will later be used interactively.)
1924 * minor incompatible change: The LOAD function no longer, when given
1925 a wild pathname to load, loads all files matching that pathname.
1926 Instead, an error of type FILE-ERROR is signalled.
1928 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1929 * bug 147 fixed: The compiler preserves its block link/count
1930 invariants more correctly now so that it doesn't crash. (thanks
1932 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1934 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1935 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1936 no-value-for-this-character as NIL (as specified by ANSI).
1937 * HOST-NAMESTRING on physical pathnames now returns a string that is
1938 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1939 (thanks to Christophe Rhodes)
1940 * The Alpha port handles icache flushing more correctly. (thanks to
1942 * More progress has been made toward bootstrapping under CLISP. (thanks
1943 to Christophe Rhodes)
1944 * The fasl file format has changed again, because dynamic loading
1945 on OpenBSD (which has non-ELF object files) motivated some cleanups
1946 in the way that foreign symbols are transformed and passed around.
1947 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1948 is no longer treated as whitespace by the reader, but instead as
1949 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1950 |A
\7fB|, instead of A as it used to.
1952 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1953 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1954 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1955 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1956 this as of 0.7.1.45, but was left out of the previous news section
1957 (thanks to Dan Barlow)
1958 * SBCL now runs on the Solaris operating system on SPARC architectures
1959 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1960 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1961 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1962 * SPARC backend cleanups, allowing builds of cores optimized for V8
1963 and V9 SPARCS, and also emission of code targeted to a particular
1964 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1966 * SBCL is closer to bootstrapping under CLISP, thanks to various
1967 fixes by Christophe Rhodes.
1968 * The fasl file format has changed again, to allow the compiler's
1969 INFO database to support symbol macros.
1970 * The user manual (in doc/) is formatted into HTML more nicely.
1971 (thanks to coreythomas)
1972 * The system is smarter about SUBTYPEP relationships, especially
1973 those involving NOT types (including types such as ATOM which are
1974 represented internally using NOT types). Thus SUBTYPEP is less
1975 likely to return (VALUES NIL NIL) in general, and in particular
1976 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1978 * The fasl file format has changed again, because the internal
1979 representation of types now includes a new slot to support the new
1980 SUBTYPEP-of-NOT-types logic.
1981 * (not a change in the main branch of SBCL, but a related prototype
1982 which can hopefully be merged into the main branch of SBCL in the
1983 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1984 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1985 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1986 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1987 dispatch macro character combinations. (thanks to Alexey Dejneka)
1988 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1989 that superficially look like logical namestrings correctly.
1990 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1992 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1993 default supplied-p) for &optional and &key arguments. (thanks to
1996 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1997 * incompatible change: The compiler is now less aggressive about
1998 tail call optimization, doing it only when (> SPACE DEBUG) or
1999 (> SPEED DEBUG). (This is an incompatible change because there are
2000 programs which relied on the old CMU-CL-style behavior to optimize
2001 away their unbounded recursion which will now die of stack overflow.)
2002 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2003 for non-GENCGC systems has been increased to 20M (since that
2004 seems much closer to the likely performance optimum for modern
2005 systems than the old 4M value was)
2006 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2007 and *DEBUG-PRINT-LEVEL*
2008 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2009 of CMU CL's support for SPARC, and various endianness and other
2010 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2011 * new syntactic sugar for the Unix command line: --load foo.bar is now
2012 an alternate notation for --eval '(load "foo.bar")'.
2014 ** The system now detects stack overflow and handles it gracefully,
2015 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2016 optimization settings. (This is a good thing in general, and
2017 its introduction in this version should be particularly timely
2018 for anyone whose code fails because of suppression of tail
2020 ** The system now hunts for the C variable "environ" in a more
2021 devious way, to avoid segfaults when the C library version differs
2022 between compile time and run time. (thanks to Christophe Rhodes)
2023 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2024 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2025 ** The compiler no longer issues bogus style warnings for undefined
2026 classes in the same source file as the DEFCLASSes which defined
2027 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2029 ** fixes in CONDITION class precedence list for undefined function
2030 errors (thanks to Alexei Dejneka)
2031 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2032 correctly. (thanks to Dan Barlow)
2033 ** portability fixes aiming at bootstrapping under CLISP (thanks
2034 to Dave McDonald and Christophe Rhodes)
2035 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2036 ** fixes in type translation and and type inference (thanks to
2038 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2039 ** various fixes in the new ports (thanks to Dan Barlow)
2040 * several changes related to debugging:
2041 ** suppression of tail recursion, as noted above
2042 ** stack overflow detection, as noted above
2043 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2044 is now the default. (For some time encapsulation has been more
2045 reliable than the breakpoint-based :ENCAPSULATE NIL
2046 implementation, at least on X86 systems; and I just noticed that
2047 encapsulation also seems closer to the spirit of the ANSI
2050 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2052 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2053 up properly again. (There was a packaging bug in 0.7.0 which
2054 left their definitions in SB-SYS::LOAD-FOREIGN and
2055 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2056 vital for most things which interface to C-level interfaces,
2057 like extensions working with sockets or databases or
2058 Perl-compatible regexes or whatever, and the need to fix
2059 this bug is the main reason that 0.7.1 was released so
2061 ** DEFGENERIC is now choosier about the methods it redefines, so that
2062 reLOADing a previously-LOADed file containing DEFGENERICs does
2063 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2064 cycle now works as it should. (thanks to Alexey Dejneka)
2065 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2066 fixed by Christophe Rhodes. (He actually submitted this patch
2067 months ago, and I delayed until after 0.7.0.)
2068 ** Bug 111 (internal compiler confusion about runtime checks on
2069 FUNCTION types) was fixed by Alexey Dejneka.
2070 * Some internal cleanups (getting rid of variables which aren't
2071 needed now that the byte interpreter is gone) caused the fasl
2072 file format number to change again.
2074 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2075 * major incompatible change: The default fasl file extension, i.e. the
2076 default extension for files produced by COMPILE-FILE, has changed
2077 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2079 ** There are many changes in the implementation of the compiler.
2080 SBCL is now essentially a compiler-only implementation of ANSI
2081 Common Lisp. EVAL still "interprets" a few special cases, but
2082 almost all the interesting cases are handled by creating
2083 a LAMBDA expression, calling COMPILE on it, then calling
2084 FUNCALL on the result.
2085 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2086 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2087 Since the code is newer, there might still be some new bugs
2088 (though not as many as before Martin Atzmueller's fixes:-). But
2089 the new code is substantially simpler and clearer, and hopefully
2090 any remaining bugs will be simpler, less fundamental, and more
2091 fixable then the bugs in the old code.
2092 ** The revised compiler is still a little unsteady on its feet.
2094 *** The debugging information it produces (particularly the names
2095 of FUNCTION objects) is sometimes much less useful than what
2096 the old compiler produced.
2097 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2098 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2099 within a MACROLET) has been temporarily weakened.
2100 ** There are new compiler optimizations for various functions:
2101 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2102 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2103 *** the math functions TRUNCATE, FLOOR, and CEILING
2104 *** the function-of-all-trades COERCE
2105 Mostly these should be transparent, but there's one
2106 potentially-annoying problem (bug 117): when the compiler
2107 inline-expands a function and does type analysis on the result,
2108 it can create control paths which have type mismatches, and
2109 when it can't prove that those control paths aren't taken,
2110 it will issue WARNINGs about the type mismatches. This is
2111 a particular problem in practice for the new sequence functions.
2112 It's not clear how this should be fixed, and for now, a
2113 workaround is given in the entry for 117 in the BUGS file.
2114 ** (Because of the interaction between the two previous items --
2115 occasional inlining problems and new inline expansions -- some
2116 of the new sequence function optimizations won't really kick in
2117 completely until debugging information, and then inlining, are
2118 straightened out in some future version.)
2119 * minor incompatible changes:
2120 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2121 DIRECTORY no longer implicitly promotes NIL slots of its
2122 pathname argument to :WILD. In particular, when you ask for the
2123 contents of a directory (which you used to be able to do without
2124 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2125 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2126 ** changes in behavior that ANSI explicitly defines to be
2127 implementation dependent:
2128 *** The new compiler-only implementation still conforms with ANSI,
2129 but acts a little different than before. Besides the obvious
2130 changes in performance tradeoffs (that the cost per form passed
2131 to EVAL has gone up, and the cost per form executed by EVAL
2132 has gone down), the behavior of the system changes a little
2133 because there are no longer any interpreted function objects.
2134 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2135 e.g. doing COMPILE on the output of interactive DEFUN is
2137 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2139 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2140 has changed from (:CL) to NIL.
2141 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2142 e.g. #\Newline and #\Tab, don't have names specified in the
2143 ANSI Common Lisp standard, is now based on their ASCII symbolic
2144 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2145 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2146 are no longer used for output.
2147 ** changes in internal implementation constants:
2148 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2149 4 million. (If your application spends a lot of time GCing and
2150 you have a lot of RAM, you might want to experiment with
2151 increasing it even more.)
2152 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2153 However, almost all old code should still continue to work without
2154 immediate update, as SB-C-CALL is now a (deprecated) nickname
2156 ** Old operator names in the style DEF-FOO are now deprecated in
2157 favor of new corresponding names DEFINE-FOO, for consistency with
2158 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2159 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2160 internal symbols, but a few supported extensions like
2161 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2162 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2163 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2164 etc. as you get deeper into recursive calls to the debugger
2165 command loop, instead of the old "5]", "5]]", "5]]]"
2166 sequence. (I was motivated to do this when squabbles between
2167 ILISP and SBCL left me very deeply nested in the debugger. In the
2168 short term, this change will probably provoke more ILISP/SBCL
2169 squabbles, but hopefully it will be an improvement in the long run.)
2170 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2171 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2172 function, since declaiming return types involving aliens is
2173 (1) annoyingly messy to do by hand and (2) vital to efficient
2174 compilation of code which calls such functions.
2175 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2176 longer reexported by the SB-EXT package. They're solely useful
2177 for alien code, so it seems more logical that you should get
2178 them from the SB-ALIEN package, not in SB-EXT.
2179 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2180 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2181 features. Instead, the code that they used to control is always
2182 built into the system.
2183 * many other bug fixes
2184 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2185 advantage of the new EVAL-WHEN stuff and to clean them up in
2186 general, and they are now more ANSI-compliant in a number of
2187 ways. Martin Atzmueller is responsible for a lot of this.
2188 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2190 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2191 *** correct ERROR type for various file operations
2192 *** some fixes for Lisp streams
2193 *** DEFMETHOD syntax checking
2194 *** changing old weird representation of debug information as
2195 strings (which, among their other deficiencies, don't transform
2196 correctly when you rename packages, and don't change their
2197 print representation when you change things like *PACKAGE*
2198 and *PRINT-LENGTH*) to symbols and lists of symbols
2199 He also made several improvements and fixed several bugs in DESCRIBE.
2200 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2202 *** misbehavior of WRITE-STRING/WRITE-LINE
2203 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2204 and several other LOOP problems as well
2205 *** DIRECTORY when similar filenames are present
2206 *** DEFGENERIC with :METHOD options
2207 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2208 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2209 *** argument ordering in FIND with :TEST option
2210 *** mishandled package designator argument in APROPOS-LIST
2211 *** various problems in the backquote readmacro
2212 *** a bug in APROPOS
2213 *** probably some others that I'm not describing very well here,
2214 since the CVS log documents them by reference to sbcl-devel
2215 messages, and the SourceForge archives aren't working well.:-(
2216 ** Dan Barlow improved the Alpha port (and is making progress on the
2217 PPC port, for those of you who think different).
2218 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2219 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2220 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2221 continues to work on the SPARC port (for those of you in a position
2222 to look down upon our little PC-compatible boxes from a great height).
2223 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2224 and uses that copy, rather than the current dynamic value, when
2225 it's trying to decide whether to truncate output. Thus e.g.
2226 (let ((*print-lines* 50))
2227 (pprint-logical-block (stream nil)
2229 (let ((*print-lines* 8))
2230 (print (aref possiblybigthings i) stream)))))
2231 should now truncate the logical block only at 50 lines, instead of
2232 often truncating it at 8 lines, as it did before.
2233 * The doc/cmucl/ directory, containing old CMU CL documentation
2234 from the time of the fork, is no longer part of the base system.
2235 SourceForge has shut down its anonymous FTP service, and with it
2236 my original plan for distributing the old CMU CL documentation
2237 there. For now, if you need these files you can download an old
2238 SBCL source release and extract them from it.
2239 * The fasl file version number changed again, for dozens of reasons,
2240 some of which are apparent above.
2242 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2243 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2244 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2245 patch, so that DISASSEMBLE output is much nicer.
2246 * The code in the SB-PROFILE package now seems reasonably stable.
2247 I still haven't decided what the final interface should look like
2248 (I'd like PROFILE to interact cleanly with TRACE, since both
2249 facilities use function encapsulation) but if you have a need
2250 for profiling now, you can probably use it successfully with
2251 the current CMU-CL-style interface.
2252 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2253 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2254 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2255 link now returns the dangling link itself, and for similar
2256 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2257 link itself. (In these cases the old code signalled an error and
2258 looped endlessly, respectively.) Thus, DIRECTORY now works even
2259 in the presence of dangling and cyclic symbolic links.
2260 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2261 is now a supported extension again, since the consensus on
2262 sbcl-devel was that it can be useful for ordinary development
2263 work, not just for debugging SBCL itself.
2264 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2265 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2266 that currently-defined functions might be redefined later with
2267 different return types.
2268 * Hash tables can be printed readably, as inspired by CMU CL code
2269 of Eric Marsden and SBCL code of Martin Atzmueller.
2270 * better error handling in CLOS method combination, thanks to
2271 Martin Atzmueller porting Pierre Mai's CMU CL patches
2272 * more overflow fixes for >16Mbyte I/O buffers
2273 * A bug in READ has been fixed, so that now a single Ctrl-D
2274 character suffices to cause end-of-file on character streams.
2275 In particular, now you only need one Ctrl-D at the command
2276 line (not two) to exit SBCL.
2277 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2279 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2280 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2281 does the right thing.
2282 * The implementation of some type tests, especially for CONDITION
2283 types, is now tidier and maybe faster, due to CMU CL code
2284 originally by Douglas Crosher, ported by Martin Atzmueller.
2285 * Some math functions have been fixed, and there are new
2286 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2287 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2288 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2289 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2290 likely to be built into sbcl-0.7.0.)
2291 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2292 CL, ported to SBCL by Martin Atzmueller)
2293 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2294 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2295 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2296 * There's a new slam.sh hack to shorten the edit/compile/debug
2297 cycle for low-level changes to SBCL itself, and a new
2298 :SB-AFTER-XC-CORE target feature to control the generation of
2299 the after-xc.core file needed by slam.sh.
2300 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2301 COMPILE-FILE is no longer supported, so that now every function
2302 gets an entry point, so that block compilation looks a little
2303 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2304 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2305 returns the number of bytes consed since the system started,
2306 rather than the number consed since the first time the function
2307 was called. (The new definition parallels ANSI functions like
2308 CL:GET-INTERNAL-RUN-TIME.)
2309 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2310 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2311 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2312 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2313 * Fasl file version numbers are now independent of the target CPU,
2314 since historically most system changes which required version
2315 number changes have affected all CPUs equally. Similarly,
2316 the byte fasl file version is now equal to the ordinary
2319 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2320 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2321 is no longer recognized. I apologize for this, because it was
2322 listed in SB-EXT as a supported extension, but I found that
2323 its existing behavior was poorly specified, as well as incorrectly
2324 specified, and it looked like too much of a mess to straighten it
2325 out. I have enough on my hands trying to get ANSI stuff to work..
2326 * many patches ported from CMU CL by Martin Atzmueller, with
2327 half a dozen bug fixes in pretty-printing and the debugger, and
2328 half a dozen others elsewhere
2329 * fixed bug 13: Floating point infinities are now supported again.
2330 They might still be a little bit flaky, but thanks to bug reports
2331 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2332 as flaky as they were.
2333 * The --noprogrammer command line option is now supported. (Its
2334 behavior is slightly different in detail from what the old man
2335 page claimed it would do, but it's still appropriate under the
2336 same circumstances that the man page talks about.)
2337 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2338 are now supported, and enabled by default. Thus, the compiler can
2339 handle many floating point and complex operations much less
2340 inefficiently. (Thus e.g. you can implement a complex FFT
2342 * The compiler now detects type mismatches between DECLAIM FTYPE
2343 and DEFUN better, and implements CHECK-TYPE more correctly, and
2344 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2345 thanks to patches from Martin Atzmueller.
2346 * various fixes to make the cross-compiler more portable to
2347 ANSI-conforming-but-different cross-compilation hosts (notably
2348 Lispworks for Windows, following bug reports from Arthur Lemmens)
2349 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2350 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2351 fixed thanks to Pierre Mai's CMU CL patches.
2352 * fixes to keep the system from overflowing internal counters when
2353 it tries to use i/o buffers larger than 16M bytes
2354 * fixed bug 45a: Various internal functions required to support
2355 complex special functions have been merged from CMU CL sources.
2356 (When I was first setting up SBCL, I misunderstood a compile-time
2357 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2358 * improved support for type intersection and union, fixing bug 12
2359 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2360 more obscure bugs as well
2361 * some steps toward byte-compiling non-performance-critical
2362 parts of the system, courtesy of patches from Martin Atzmueller
2363 * Christophe Rhodes has made some debian packages of sbcl at
2364 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2365 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2366 stable, but are nonetheless usable. When he's ready, I'd be happy
2367 to add them to the SourceForge "File Releases" section. (And if
2368 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2369 * new fasl file format version number (because of changes in
2370 internal representation of (OR ..) types to accommodate the new
2371 support for (AND ..) types, among other things)
2373 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2374 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2376 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2378 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2379 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2380 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2381 * DESCRIBE now works on structure objects again.
2382 * Most function call argument type mismatches are now handled as
2383 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2384 know whether the function will be redefined before the call is
2385 executed. (The compiler could flag local calls with full WARNINGs,
2386 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2387 it doesn't keep track of enough information to know whether calls
2388 are local in this sense.)
2389 * Compiler output is now more verbose, with messages truncated
2390 later than before. (There should be some supported way for users
2391 to override the default verbosity, but I haven't decided how to
2392 provide it yet, so this behavior is still controlled by the internal
2393 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2394 src/compiler/ir1util.lisp.)
2395 * Fasl file format version numbers have increased again, because
2396 support for the Gray streams extension changes the layout of the
2397 system's STREAM objects.
2398 * The Gray subclassable streams extension now works, thanks to a
2399 patch from Martin Atzmueller.
2400 * The full LOAD-FOREIGN extension (not just the primitive
2401 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2402 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2403 but like most other programs, it defaults to copying the Unix
2404 environment from the original process instead of starting the
2405 new process in an empty environment.
2406 * Extensions which manipulate the Unix environment now support
2407 an :ENVIRONMENT keyword option which doesn't smash case or
2408 do other bad things. The CMU-CL-style :ENV option is retained
2409 for porting convenience.
2410 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2411 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2413 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2415 * A patch from Martin Atzmueller seems to have solved the SIGINT
2416 problem, and as far as we know, signal-handling now works cleanly.
2417 (If you find any new bugs, please report them!)
2418 * The system no longer defaults Lisp source file names to types
2419 ".l", ".cl", or ".lsp", but only to ".lisp".
2420 * The compiler no longer uses special default file extensions for
2421 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2422 seems to expect a single default extension for all compiled code,
2423 and there's no compelling reason to try to stretch the standard
2424 to allow two different extensions.) Instead, byte-compiled files
2425 default to the same extension as native-compiled files.
2426 * Fasl file format version numbers have increased again, because
2427 a rearrangement of internal implementation packages made some
2428 dumped symbols in old fasl files unreadable in new cores.
2429 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2430 many fewer weird special cases.
2431 * Bug #17 (differing COMPILE-FILE behavior between logical and
2432 physical pathnames) has been fixed, and some related misbehavior too,
2433 thanks to a patch from Martin Atzmueller.
2434 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2435 by Tim Moore, ported to SBCL by Martin Atzmueller.
2436 * Martin Atzmueller fixed several filesystem-related problems,
2437 including bug #36, in part by porting CMU CL patches, which were
2438 written in part by Paul Werkowski.
2439 * More compiler warnings in src/runtime/ are gone, thanks to
2440 more patches from Martin Atzmueller.
2441 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2444 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2446 * DESCRIBE now works on CONDITION objects.
2447 * The debugger now handles errors which arise when trying to print
2448 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2450 * The build system now uses an additional file, customize-target-features.lisp,
2451 to allow local modifications to the target *FEATURES* list. (The point of
2452 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2453 debugging features enabled, without having to worry about propagating it
2454 into everyone's system when I do a "cvs update".) When no
2455 customize-target-features.lisp file exists, the target *FEATURES* list
2456 should be constructed the same way as before.
2457 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2458 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2459 warn about behavior which is undefined under the ANSI spec. Note: This
2460 is specified by ANSI, but it's not very popular with programmers.
2461 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2462 macro in the SBCL sources for an example of a workaround which you
2463 might use to make portable ANSI-standard code which does what you want.
2464 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2465 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2466 when it's used as a non-toplevel form. (This is required in order
2467 to implement the DEFCONSTANT-EQX macro.)
2468 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2469 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2470 funny things, due to bugs in the implementation of EVAL-WHEN
2471 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2472 non-toplevel DEFCONSTANTs are uncommon.)
2473 * The core file version number and fasl file version number have been
2474 incremented, because the old noncompliant DEFCONSTANT behavior involved
2475 calling functions which no longer exist, and because I also took the
2476 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2477 * fixed bug 1 (error handling before read-eval-print loop starts), and
2478 redid debugger restarts and related debugger commands somewhat while
2480 ** The QUIT debugger command is gone, since it did something
2481 rather different than the SB-EXT:QUIT command, and since it never
2482 worked properly outside the main toplevel read/eval/print loop.
2483 Invoking the new TOPLEVEL restart provides the same functionality.
2484 ** The GO debugger command is also gone, since you can just invoke
2485 the CONTINUE restart directly instead.
2486 ** The TOP debugger command is also gone, since it's redundant with the
2487 FRAME 0 command, and since it interfered with abbreviations for the
2489 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2491 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2492 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2493 some porting work by Martin Atzmueller)
2494 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2495 with a boatload of patches which he ported to SBCL. Now that those
2497 ** The system tries to make sure that its low-priority messages
2498 are prefixed by semicolons, to help people who like to use
2499 syntax highlighting in their ILISP buffer. (This patch
2500 was originally due to Raymond Toy.)
2501 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2502 patches originally written by Raymond Toy.
2503 ** The compiler understands coercion between single-value and
2504 multiple-VALUES type expressions better, getting rid of some very
2505 weird behavior, thanks to patches originally by Robert MacLachlan
2506 and Douglas Crosher.
2507 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2508 lambda lists, thanks to a patch originally by Pierre Mai.
2509 ** The system no longer bogusly warns about "abbreviated type
2511 ** The compiler gets less confused by inlining and RETURN-FROM,
2512 thanks to some patches originally by Tim Moore.
2513 ** The system no longer hangs when dumping circular lists to fasl
2514 files, thanks to a patch originally from Douglas Crosher.
2515 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2516 undefined function error.
2517 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2518 in structure slot initforms) for now, documented workaround instead:-|
2519 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2520 * fixed bug 5: added stubs for various Gray stream functions called
2521 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2522 installed, at least appropriate type errors are generated
2523 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2524 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2525 consistently in DEFMETHOD forms.
2526 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2527 it doesn't seem to affect SBCL after all
2528 * The C runtime system now builds with better optimization and many
2529 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2531 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2533 * The system is now under CVS at SourceForge (instead of the
2534 CVS repository on my home machine).
2535 * The new signal handling code has been tweaked to treat register
2536 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2537 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2538 downstream, e.g. in debug-int.lisp, has implicit dependencies
2539 on the unsignedness of integer representation of machine words,
2540 and that caused the system to bomb out with infinite regress
2541 when trying to recover from type errors involving signed values,
2542 e.g. (BUTLAST '(1 2 3) -1).)
2543 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2544 (This was one of the bugs Peter Van Eynde reported back in July.)
2545 * The system now uses code inspired by Colin Walters' O(N)
2546 implementation of MAP (from the cmucl-imp@cons.org mailing
2547 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2548 inline the MAP operation, and there is more than one
2549 sequence argument to the MAP call (so that it can't just
2550 do ETYPECASE once and for all based on the type of the
2551 single sequence argument). (The old non-inline implementation
2552 of the general M-argument sequence-of-length-N case required
2553 O(M*N*N) time when any of the sequence arguments were LISTs.)
2554 * The QUIT :UNIX-CODE keyword argument has been renamed to
2555 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2556 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2558 * Raymond Wiker's patches to port dynamic loading from Linux to
2559 FreeBSD have been added.
2560 * The BUGS file is now more nearly up to date, thanks in large part
2561 to Martin Atzmueller's review of it.
2562 * The debugger now flushes standard output streams before it begins
2563 its output ("debugger invoked" and so forth).
2564 * The core version number and fasl file version number have both
2565 been incremented, because of incompatible changes in the layout
2567 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2568 used to not be called for a saved Lisp image.)
2569 * Martin Atzmueller's version of a patch to fix a compiler crash,
2570 as posted on sbcl-devel 13 September 2000, has been installed.
2571 * Instead of installing Martin Atzmueller's patch for the
2572 compiler transform for SUBSEQ, I deleted the compiler transform,
2573 and transforms for some similar consing operations.
2574 * A bug in signal handling which kept TRACE from working on OpenBSD
2576 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2577 away by constant folding
2578 * The system now defines its address space constants in one place
2579 (in the Lisp sources), and propagates them automatically elsewhere
2580 (through GENESIS and the sbcl.h file). Therefore, patching the
2581 address map is less unnecessarily tedious and error-prone. The
2582 Lisp names of address space constants have also been systematized.
2583 * CVS tags like dollar-Header-dollar have been removed from
2584 the sources, because they have never saved me trouble and
2585 they've been source of trouble working with patches and other
2586 diff-related operations.
2587 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2588 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2590 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2592 * The system has been ported to OpenBSD.
2593 * The system now compiles with a simple "sh make.sh" on the systems
2594 that it's supported on. I.e., now you no longer need to tweak
2595 text in the target-features.lisp-expr and symlinks in src/runtime/
2596 by hand, the make.sh takes care of it for you.
2597 * The system is no longer so grossly inefficient when compiling code
2598 involving vectors implemented as general (not simple) vectors (VECTOR T),
2599 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2600 which dares to use the various sequence functions on non-simple
2601 vectors, takes less of a performance hit.
2602 * There is now a primitive type predicate VECTOR-T-P
2603 to test for the (VECTOR T) type, so that e.g.
2604 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2605 can now be compiled with some semblance of efficiency. (The old code
2606 turned the type declaration into a full call to %TYPEP at runtime!)
2607 * AREF on (VECTOR T) is still not fast, since it's still compiled
2608 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2609 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2610 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2611 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2612 without making full calls to SUBTYPEP at runtime.
2613 (Some analogous efficiency issues for non-simple vectors specialized to
2614 element types other than T, or for non-simple multidimensional arrays,
2615 have not been addressed. They could almost certainly be handled the
2616 same way if anyone is motivated to do so.)
2617 * The changes in array handling break binary compatibility, so
2618 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2619 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2620 instead of (VALUES T T).
2621 * By following the instructions that Dan Barlow posted to sbcl-devel
2622 on 2 July 2000, I was able to enable primitive dynamic object
2623 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2624 functionality is not implemented (since it calls ld to resolve
2625 library references automatically, requiring RUN-PROGRAM for its
2626 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2627 try to resolve library references) is now supported.
2628 * The system now flushes the standard output streams when it terminates,
2629 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2630 them at several other probably-convenient times, e.g. in each pass of
2631 the toplevel read-eval-print loop, and after evaluating a form given
2632 as an "--eval" command-line option. (These changes were motivated by a
2633 discussion of stream flushing issues on cmucl-imp in August 2000.)
2634 * The source transform for TYPEP of array types no longer assumes
2635 that an array whose element type is a not-yet-defined type
2636 is implemented as an array of T, but instead punts, so that the
2637 type will be interpreted at runtime.
2638 * There is now some support for cross-compiling in make.sh: each of
2639 the phases of make.sh has its own script. (This should be transparent
2640 to people doing ordinary, non-cross-compile builds.)
2641 * Since my laptop doesn't have hundreds of megabytes of memory like
2642 my desktop machine, I became more motivated to do some items on
2643 my to-do list in order to reduce the size of the system a little:
2644 ** Arrange for various needed-only-at-cold-init things to be
2645 uninterned after cold init. To support this, those things have
2646 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2647 symbols with such names are now uninterned after cold init).
2648 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2649 fasl files for cold load.
2650 ** Remove the old compiler structure pooling code (which used to
2651 be conditional on the target feature :SB-ALLOC) completely.
2652 ** Redo the representation of some data in cold init to be more compact.
2653 (I also looked into supporting byte compiled code at bootstrap time,
2654 which would probably reduce the size of the system a lot, but that
2655 looked too complicated, so I punted for now.)
2656 * The maximum signal nesting depth in the src/runtime/ support code has
2657 been reduced from 4096 to 256. (I don't know any reason for the very
2658 large old value. If the new smaller value turns out to break something,
2659 I'll probably just bump it back up.)
2660 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2662 * Many, many bugs reported by Peter Van Eynde have been added to
2663 the BUGS list; some have even been fixed.
2664 * While enabling dynamic object file loading, I tried to make the
2665 code easier to understand, renaming various functions and variables
2666 with less ambiguous names, and changing some function calling
2667 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2668 * While trying to figure out how to do the OpenBSD port, I tried to
2669 clean up some of the code in src/runtime/. In particular, I dropped
2670 support for non-POSIX signal handling, added various comments,
2671 tweaked the code to reduce the number of compilation warnings, and
2672 renamed some files to increase consistency.
2673 * To support the new automatic configuration functionality in make.sh,
2674 the source file target-features.lisp-expr has been replaced with the
2675 source file base-target-features.lisp-expr and the machine-generated
2676 file local-target-features.lisp-expr.
2677 * fixed a stupid quoting error in make.sh so that using CMU CL
2678 "lisp -batch" as cross-compilation host works again
2680 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2682 * DESCRIBE no longer tries to call itself recursively to describe
2683 bound/fbound values, so that it no longer fails on symbols which are
2684 bound to themselves (like keywords, T, and NIL).
2685 * DESCRIBE now works on generic functions.
2686 * The printer now prints less-screwed-up representations of closures
2687 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2688 * A private symbol is used instead of the :EMPTY keyword previously
2689 used to mark empty slots in hash tables. Thus
2690 (DEFVAR *HT* (MAKE-HASH-TABLE))
2691 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2692 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2693 now does what ANSI says that it should. (You can still get
2694 similar noncompliant behavior if bang on the hash table
2695 implementation with all the symbols you get back from
2696 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2697 This breaks binary compatibility, since tests for equality to
2698 :EMPTY are wired into things like the macroexpansion of
2699 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2701 * There's now a minimal placeholder implementation for CL:STEP,
2702 as required by ANSI.
2703 * An obscure bug in the interaction of the normal compiler, the byte
2704 compiler, inlining, and structure predicates has been patched
2705 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2706 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2708 * Missing ordinary arguments in a macro call are now detected even
2709 when the macro lambda list contains &KEY or &REST.
2710 * The debugger no longer complains about encountering the top of the
2711 stack when you type "FRAME 0" to explicitly instruct it to go to
2712 the top of the stack. And it now prints the frame you request even
2713 if it's the current frame (instead of saying "You are here.").
2714 * As specified by ANSI, the system now always prints keywords
2715 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2717 * The default initial SIZE of HASH-TABLEs is now smaller.
2718 * Type information from CLOS class dispatch is now propagated
2719 into DEFMETHOD bodies, so that e.g.
2720 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2722 is now basically equivalent to
2723 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2724 (DECLARE (TYPE SINGLE-FLOAT X))
2726 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2727 operation, without having to do run-time type dispatch.
2728 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2729 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2730 * Since I don't seem to be making big file reorganizations very often
2731 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2732 is overflowing my ability to conveniently back them up), I've finally
2733 checked the system into CVS. (The CVS repository is on my home system,
2734 not at SourceForge -- putting it on SourceForge might come later.)
2735 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2736 high-level GC-NOTIFY-FOO functions send their output. (There's
2737 still very little control of where low-level verbose GC functions
2738 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2739 less than it used to -- the GC-NOTIFY-FOO functions are now under
2740 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2741 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2742 in only one place in the source code, and propagates it automatically
2743 everywhere that it's needed. Thus e.g. when I bump the version from
2744 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2745 * The C source files now include boilerplate legalese and documentation
2746 at the head of each file (just as the Lisp source files already did).
2747 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2748 to his page will be replaced with a link to his new CLiki service.
2750 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2752 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2753 * The build process now looks for GNU make under the default name "gmake",
2754 instead of "make" as it used to. If GNU make is not available as "gmake"
2755 on your system, you can change this default behavior by setting the
2756 GNUMAKE environment variable.
2757 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2758 can build without error under CMU CL.
2760 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2762 * There is now a partial SBCL user manual (with some new text and some
2763 text cribbed from the CMU CL manual).
2764 * The beginnings of a profiler have been added (starting with the
2765 CMU CL profiler and simplifying and cleaning up). Eventually the
2766 main interface should be through the TRACE macro, but for now,
2767 it's still accessed through vaguely CMU-CL-style functions and macros
2768 exported from the package SB-PROFILE.
2769 * Some problems left over from porting CMU CL to the new
2770 cross-compilation bootstrap process have been cleaned up:
2771 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2772 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2773 part when getting rid of the compiler *BACKEND* stuff.)
2774 ** The value of *NULL-TYPE* was screwed up, because it was
2775 being initialized before the type system knew the final
2776 definition of the 'NULL type. This screwed up several key
2777 optimizations in the compiler, causing inefficiency in all sorts
2778 of places. (I found it because I wanted to understand why
2779 GET-INTERNAL-RUN-TIME was consing.)
2780 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2781 PROCLAIM FTYPE information. Unfortunately this broke binary
2782 compatibility again, since now the forms output by DEFGENERIC
2783 to refer to functions which didn't exist in 0.6.3.
2784 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2785 can use the new (as of 0.6.3) transform for SOME into MAP into
2787 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2788 values to (INTEGER 0 1000000), so that the time code will no longer
2789 occasionally get blown up by Linux returning 1000000 microseconds
2790 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2791 its output conform to the ANSI spec. (Alas, this makes its output
2792 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2793 * A full call to MAP NIL with a single sequence argument no longer conses.
2794 * fixes to problems pointed out by Martin Atzmueller:
2795 * The manual page no longer talks about multiprocessing as though
2796 it were currently supported.
2797 * The ILISP support patches have been removed from the distribution,
2798 because as of version 5.10.1, ILISP now supports SBCL without us
2799 having to maintain patches.
2800 * added a modified version of Raymond Toy's recent CMU CL patch for
2801 EQUALP comparison of HASH-TABLE
2803 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2805 * The system still can't cross-compile itself with
2806 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2807 entails), but at least it can compile more of itself that way
2808 than it used to be able to, and various buglets which were uncovered
2809 by trying to cross-compile itself that way have now been fixed.
2810 * This release breaks binary compatibility again. This time
2811 at least I've incremented the FASL file format version to 2, so that the
2812 problem can be detected reliably instead of just causing weird errors.
2813 * various new style warnings:
2814 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2815 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2816 ** using the lexical binding of a variable named in the *FOO* style
2817 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2818 as specified by ANSI.
2819 * *RANDOM-STATE* is no longer automatically initialized from
2820 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2821 default behavior of the system is to repeat its behavior every time
2822 it's run. If you'd like to change this behavior, you can always
2823 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2824 old convention there was no comparably easy way to get the system to
2825 repeat its behavior every time it was run.
2826 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2827 FTYPE declarations has been removed, in favor of their ANSI
2828 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2829 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2830 the types of their sequence arguments can be determined at compile time.
2831 This is done through a new open code expansion for MAP which eliminates
2832 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2833 argument types can be determined at compile time.
2834 * The optimizer now transforms COERCE into an identity operation when it
2835 can prove that the coerced object is already of the correct type. (This
2836 can be a win for machine generated code, including the output of other
2837 optimization transforms, such as the MAP transform above.)
2838 * Credit information has been moved from source file headers into CREDITS.
2839 * Source file headers have been made more standard.
2840 * The CASE macro now compiles without complaining even when it has
2843 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2845 * (Note that the way that the PCL macroexpansions were rewritten
2846 to accommodate the change in DEFGENERIC below breaks binary
2847 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2848 not run under sbcl-0.6.2. Once we get out of alpha releases,
2849 i.e. hit release 1.0.0, we'll probably try to maintain binary
2850 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2851 and sbcl-1.4.4. Until then, however, it might be fairly common
2852 for maintenance releases to break binary compatibility.)
2853 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2855 * The reporting of unhandled signals has been changed to print some
2856 explanatory text as well as the report form. (Previously only
2857 the report form was printed.)
2858 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2859 it defines, so that the compiler no longer issues undefined function
2860 warnings for compiled-but-not-yet-loaded generic functions.
2861 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2862 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2863 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2864 section of the ANSI spec are supported.
2865 * The "" nickname for the "KEYWORD" package has been removed.
2866 The reader still handles symbol tokens which begin with a package marker
2867 as keywords, but it doesn't expose its mechanism for doing so in the
2868 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2869 * The system now issues STYLE-WARNINGs for contradictory TYPE
2870 proclamations. (Warnings for contradictory FTYPE proclamations would
2871 be nice too, but those can't be done usefully unless the type system
2872 is made smarter about FUNCTION types.)
2873 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2874 systematized, so that "*target-*.lisp is supposed to exist only on the
2875 target and imply that there's a related file which exists on the
2876 host, and *host-*.lisp is supposed to exist only on the host and imply
2877 that there's a related file which exists on the target. This involves a
2878 lot of renaming. Hopefully the acute confusion caused by the renaming
2879 will be justified by the reduction in chronic confusion..
2880 ** runtime-type.lisp -> early-target-type.lisp
2881 ** target-type.lisp -> late-target-type.lisp
2882 ** early-host-format.lisp -> early-format.lisp
2883 ** late-host-format.lisp -> late-format.lisp
2884 ** host-error.lisp -> misc-error.lisp
2885 ** early-error.lisp -> early-target-error.lisp
2886 ** late-error.lisp -> late-target-error.lisp
2887 ** host-defboot.lisp -> early-defboot.lisp
2888 ** code/misc.lisp -> code/target-misc.lisp
2889 ** code/host-misc.lisp -> code/misc.lisp
2890 ** code/numbers.lisp -> code/target-numbers.lisp
2891 ** code/early-numbers.lisp -> numbers.lisp
2892 ** early-host-type.lisp -> early-type.lisp
2893 ** late-host-type.lisp -> late-type.lisp
2894 ** host-typep.lisp -> typep.lisp
2895 ** load.lisp -> target-load.lisp
2896 ** host-load.lisp -> load.lisp
2897 ** host-disassem.lisp -> disassem.lisp
2898 ** host-insts.lisp -> insts.lisp
2899 ** byte-comp.lisp -> target-byte-comp.lisp
2900 ** host-byte-comp.lisp -> byte-comp.lisp
2901 ** host-signal.lisp -> signal.lisp
2902 ** host-defstruct.lisp -> defstruct.lisp
2903 ** late-target-type.lisp -> deftypes-for-target.lisp
2904 Furthermore, several other previously target-only files foo.lisp (e.g.
2905 hash-table.lisp and random.lisp) have been split into a target-and-host
2906 foo.lisp file and a target-only target-foo.lisp file, with their key type
2907 definitions in the target-and-host part, so that the cross-compiler will
2908 know more about target types.
2909 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2910 gone away. In their place are various *BACKEND-FOO* variables
2911 corresponding to the slots of the old structure.
2912 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2913 into the target SBCL has been fixed.
2914 * The chill.lisp system for loading cold code into a running SBCL
2916 * Support for the CMU CL "scavenger hook" extension has been removed.
2917 (It was undocumented and unused in the CMU CL sources that SBCL was
2918 derived from, and stale in sbcl-0.6.1.)
2919 * Various errors in the cross-compiler type system were detected
2920 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2921 (enabling various consistency checks). Many of them were fixed,
2922 but some hard problems remain, so the compiler is back to
2923 running without *TYPE-SYSTEM-INITIALIZED* for now.
2924 * As part of the cross-compiler type system cleanup, I implemented
2925 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2926 * I have started adding UNCROSS calls throughout the type system
2927 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2928 on cross-compiler input files will be able to go away, and various
2930 * CONSTANTP now returns true for quoted forms (as explicitly required
2933 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2935 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2936 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2937 (MAKE-LIST -1). In the longer term, it ought to become true
2938 that declarations are assertions even at SAFETY 1. For now, it's not
2939 quite true even at SAFETY 3, but it's at least more nearly true..
2940 (Note that this change seems to increases the size of the system by
2941 O(5%) and to decrease the speed of the compiler by 20% or more.)
2942 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2943 for the problem of printing dozens of lines of distracting information
2944 about low-level system machinery as part of the top stack frame
2945 on entry to the debugger when an undefined function was called.
2946 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2947 is not reset to COMMON-LISP-USER.
2948 * Compilation of stuff related to dyncount.lisp has been made conditional
2949 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2950 smaller. The various dyncount-related symbols have been moved into
2951 a new "SB-DYNCOUNT" package.
2952 * tty-inspect.lisp has been renamed to inspect.lisp.
2953 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2954 feature has gone away. (When we eventually port to other flavors of
2955 libc and/or Unix, we'll try to make the differences between flavors
2956 invisible at the user level.)
2957 * Various other *FEATURES* tags, and/or their associated conditionals,
2958 have been removed if obsolescent, or given better documentation, or
2959 sometimes given more-mnemonic names.
2961 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2963 * tidied up "make.sh" script
2964 * tidied up system directory structure
2965 * better "clean.sh" behavior
2966 * added doc/FOR-CMUCL-DEVELOPERS
2967 * many many small tweaks to output format, e.g. removing possibly-confusing
2968 trailing #\. character in DESCRIBE-INSTANCE
2969 * (EQUALP #\A 'A) no longer signals an error.
2970 * new hashing code, including EQUALP hashing
2971 * tidied up Lisp initialization and toplevel
2972 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2973 * command line argument processing
2974 * added POSIX-GETENV function to deal with Unix-ish environment variables
2975 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2976 terminating SBCL on EOF
2977 * non-verbose GC by default
2978 * There is no more "sbcl" shell script; the sbcl file is now the C
2979 runtime executable (just like CMU CL).
2980 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2982 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2983 symbol and package information as Lisp native symbol and package objects
2984 instead of strings naming symbols and strings naming packages. This way,
2985 whenever packages are renamed (as in warm init), debug information is
2986 transformed along with everything else.
2987 * tweaked the optimization policy declarations which control the building
2988 of SBCL itself. Now, among other things, the system no longer saves
2989 source location debugging information. (This helps two problems at once
2990 by reducing SBCL size and by keeping SBCL from trying to look for its
2991 sources -- which may not exist -- when reporting errors.)
2992 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2993 debugging and testing purposes
2994 * cleaned up printing, making the printer call PRINT-OBJECT for
2995 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2996 methods, giving nearly-ANSI behavior
2997 * converted almost all special variables to use *FOO* naming convention
2998 * deleted PARSE-TIME functionality, since it can be done portably
2999 * moved some files out of cold init into warm init
3000 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3002 * regularized formatting of source files
3003 * added an install.sh script
3004 * fixed ridiculous memory usage of cross-compiler by making
3005 compiler/alloc.lisp not try to do pooling unless it can hook
3006 itself into the GC of the cross-compilation host. Now the system
3007 builds nicely on my old laptop.
3008 * added :SB-ALLOC in target-features.lisp-expr
3009 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3010 was not implemented)
3011 * re-did condition handling and note reporting in the compiler. Notes
3012 are no longer handled by signalling conditions. Style warnings
3013 and warnings are handled more correctly and reported in such a way
3014 that it's easy to find one or the other in your output (so that you
3015 can e.g. figure out which of many problems caused COMPILE-FILE to
3017 * changed the severity of several compiler warnings from full WARNING
3018 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3019 compiler note reporting so that it doesn't use the condition system
3020 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3021 COMPILE-FILE command)
3022 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3023 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3024 no longer supported, since their implementation was deeply intertwingled
3025 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3026 * removed bogus "support" for compiler macros named (SETF FOO), and
3027 removed the compiler macro for SETF INFO (but only after making a fool
3028 of myself on the cmucl-imp mailing list by posting a bogus patch for
3029 DEFINE-COMPILER-MACRO..)
3030 * Compiled files containing forms which have side effects on the Lisp
3031 reader (such as DEFPACKAGE forms) are now handled more correctly.
3032 (Compiler queuing of top level lambdas has been suppressed by setting
3033 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3034 * deleted various currently-unused source files, e.g. gengc.lisp. They
3035 may be added back at some point e.g. when porting to other architectures,
3036 but until they are it's distracting to distribute them and to try to
3038 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3039 there were so many of them they're just distractions, and UNCROSS is
3040 known to be able to handle the current sources
3041 * moved PROFILE functionality into TRACE, so that it will be clear
3042 how the wrapping and unwrapping of functions when you profile them
3043 interacts with the wrapping and unwrapping of functions when you
3044 trace them. (Actually, the functionality isn't there yet, but at least
3045 the interface specification is there. Hopefully, the functionality will
3046 arrive with some future maintenance release.)
3047 * removed host-oops.lisp
3048 * changed signature of QUIT function to allow UNIX-CODE argument
3049 * fixed READ-SEQUENCE bug
3050 * tweaked verbose GC output so that it looks more like the progress
3051 output that ANSI specifies for functions like LOAD
3052 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3053 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob