1 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2 * incompatible change: forms evaluated in the REPL now use the
3 global optimization policy.
4 * incompatible change: user- and system-initialization files are
5 no longer processed with LOAD, but by READ and EVAL; hence the
6 global optimization policy, startup package, readtable, etc,
8 * internal entry point details and argument counts no longer appear
9 in backtraces unless explicitly requested by setting
10 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
11 * built-in and standard functions no longer have names like "top
12 level local call to FOO".
13 * fixed bug 32: functions defined in non-null lexical environments
14 now have more legible printed representation
15 * fixed bug 33: functions defined in non-null lexical environemnts
16 are now more amenable to inspection by INSPECT.
17 * workaround for bug 354: XEPs no longer appear in backtraces unless
19 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
20 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
21 (reported by Lutz Euler)
22 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
23 compiled code. (reported by Paul Dietz)
24 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
25 specialized parameter to a method to something that is not TYPEP
26 the specializer is now possible.
27 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
28 face of package deletion.
29 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
30 pathname. (reported by David Tolpin on c.l.l)
31 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
32 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
33 STRING-STREAM and FILE-STREAM, respectively.
34 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
35 * fixed some bugs revealed by Paul Dietz' test suite:
36 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
38 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
39 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
40 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
42 * fixed loading of multiply forward-referenced layouts.
43 (thanks to Cheuksan Wang)
44 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
45 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
46 (see COMPILE-FILE documentation for details of :PRINT option)
47 * fixed bugs 19 and 317: fixed-format floating point printing is
48 more accurate. This also fixes a bug reported by Adam Warner
49 related to the ~@F format directive.
50 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
52 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
53 dispatch table if the from-char is a dispatch macro character.
54 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
55 by the wordsize no longer ignore the last word. (reported by Lutz
57 * fixed bug in type checking of dynamic-extent variables. (reported
59 * optimization: sequence traversal functions use their freedom to
60 coerce function designators to functions.
61 * optimization: code with many calls to CLOS methods specialized on
62 CLOS classes has had redundant type checks removed.
63 * optimization: added declarations to speed up operations that access
64 the internal character database (for example STRING-UPCASE)
65 * optimization: comparison operations between floats and sufficiently small
66 fixnums no longer create extra rationals
67 * fixed some bugs related to Unicode integration:
68 ** portions of multibyte characters at the end of buffers for
69 character-based file input are correctly transferred to the
70 start of the buffer at the next read.
71 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
72 passing it through to OPEN.
73 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
74 argument given to internal calls to OPEN.
75 ** the built-in comment readers (introduced by character sequences
76 ";" and "#|") are more forgiving to encoding errors; they will
77 STYLE-WARN and then attempt to resync the stream at a character
78 boundary. (thanks to Teemu Kalvas)
79 * fixed some bugs in the x86-64 port:
80 ** Negative short int return values from c-calls are sign-extended
82 ** The stack is aligned to 16-bytes for c-calls, as required by
83 the ABI. (thanks to Cheuksan Wang)
84 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
85 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
86 for enums. (thanks to Vincent Arkesteijn)
87 ** Multiple small optimizations and bugfixes for floating point
89 * fixed some bugs revealed by Paul Dietz' test suite:
90 ** Space, Tab, Linefeed, Return and Page have the invalid
91 secondary constituent character trait.
92 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
94 ** WITH-INPUT-FROM-STRING should only update the index place on
96 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
98 ** Bit-array operations (BIT-AND and similar) worked incorrectly
99 with one-dimensional arrays with fill pointers.
100 ** TYPE-OF failed on a complex with an integer realpart and a
102 ** compiler failure during type inference for the code of form
103 (IF (EQL X (THE ...)) ...) (MISC.535).
105 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
106 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
107 platform. The port supports 61-bit fixnums, large memory spaces
108 and reloading shared object files.
109 * enhancement: saving cores with foreign code loaded is now
110 supported on ppc/Darwin in addition to the previously supported
112 * enhancement: the statistical profiler now walks deeper into the
113 call stack for more meaningful call-graphs and accrued time
114 reports (x86/x86-64 only). It also now reports time spent in
116 * enhancement: it is now possible to trace most individual methods
117 of a generic function in addition to tracing the generic function
119 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
121 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
122 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
123 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
124 produces an error. (thanks to Vincent Arkesteijn)
125 * bug fix: NAMESTRING on pathnames with :WILD components in their
126 directories works correctly. (thanks to Artem V. Andreev)
127 * fixed bug 125: compiler preserves identity of closures. (reported
129 * bug fixed: functions with &REST arguments sometimes failed with
130 "Undefined function" when compiled with (DEBUG 3). (reported by
132 * bug fix: overflow during compiling of setting element of a bit
133 vector with constant index and value. (reported by Timmy Douglas)
134 * build fix: fixed the dependence on *LOAD-PATHNAME* and
135 *COMPILE-FILE-PATHNAME* being absolute pathnames.
136 * on x86 compiler partially supports stack allocation of dynamic-extent
138 * GO and RETURN-FROM do not check the extent of their exit points
139 when compiled with SAFETY 0.
140 * fixed some bugs related to Unicode integration:
141 ** encoding and decoding errors are now much more robustly
142 handled; it should now be possible to recover even from invalid
143 input or output to the terminal. (thanks to Teemu Kalvas)
144 ** provided a first cut at implementing STRING-TO-OCTETS and
145 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
146 ** altered the SB-MD5 contributed module slightly, changing the
147 interface just enough for it to be supportable for builds where
148 lisp characters are not eight bits.
149 * fixed some bugs revealed by Paul Dietz' test suite:
150 ** the FORMATTER-generated functions for ~V[ conditionals require
151 the correct number of arguments.
152 ** READ-FROM-STRING returns the mandated second value when applied
153 to displaced strings.
154 ** the #\Rubout and #\Backspace characters are treated as invalid
155 constituent characters by the tokenizer.
157 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
158 * new feature: reloading changed shared object files with
159 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
160 * new feature: references to foreign variables and functions
161 can now be compiled and loaded before the corresponding shared
162 object file is loaded, as long as the foreign definitions are
163 available at runtime.
164 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
165 supported. (thanks to Dan Debertin)
166 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
167 just keywords. (thanks to Vincent Arkesteijn)
168 * fixed bug #331: structure-class instances corresponding to
169 DEFSTRUCT forms are now created eagerly.
170 * fixed bug #345: backtraces from calls to undefined functions work
171 on x86 as well. Related bug #61 is now also partially fixed on x86
172 (backtraces from throws to unknown catch tags.)
173 * bug fix: lambda-list parsing is now stricter vrt. order and number
174 of lambda-list keywords.
175 * bug fix: as specified by AMOP, an error is signalled if a
176 class-option appears multiple times in a DEFCLASS form. (reported
178 * bug fix: RANDOM can be compiled when the compiler derives the type
179 of its numeric argument as a disjoint set of small integers.
180 (reported by Paul Dietz)
181 * bug fix: starting a core saved with shared objects loaded when
182 those objects are not available no longer causes threaded SBCL to
183 hang. (reported by Sean Ross)
184 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
186 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
187 argument, which must always be NIL. (reported by Kalle Niemitalo)
188 * bug fix: printing 1.0d+23 no longer results in an error.
189 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
190 * bug fix: structure slot setters preserve evaluation order. (thanks
192 * fixed some bugs related to Unicode integration:
193 ** RUN-PROGRAM can allow its child to take input from a Lisp
194 stream. (reported by Stefan Scholl)
195 ** ASDF-INSTALL successfully downloads .tar.gz files in all
196 locales. (reported by Ken Causey)
197 * fixed some bugs revealed by Paul Dietz' test suite:
198 ** INCF, DECF and REMF evaluate their place form as specified in
200 ** FORMATTER expands ~{ ~} iteration directives with V or #
201 parameters correctly.
202 ** FORMATTER deals with the ~@[ ~] conditional directive where the
203 consequent uses no arguments correctly.
204 ** the system has a partial understanding of the (COMPLEX RATIO)
206 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
209 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
210 * new feature: a build-time option (controlled by the :SB-UNICODE
211 keyword feature, enabled by default) for building the system with
212 support for the entire 21-bit character space defined by the
214 * new feature: the system now has rudimentary external-format
215 support; the primary user-visible change at this time is that
216 characters with the high bit set (such as lower-case-e-acute) will
217 print correctly to a terminal in a UTF-8 environment.
218 * minor incompatible change: BASE-CHAR no longer names a class;
219 however, CHARACTER continues to do so, as required by ANSI.
220 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
221 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
222 has been moved to the SB-EXT package (temporarily re-exported from
224 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
225 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
227 * fixed bug #351: better error-handlind and reporting for malformed
229 * fixed bug #350: bignum-printing is now more memory-efficient,
230 allowing printing of very large bignums, eg. (expt 2 10000000).
231 (reported by Bruno Haible)
232 * fixed bug #302: better primitive-type selection for intersection
234 * fixed bug #308: non-graphic characters now all have names, as
235 required. (reported by Bruno Haible)
236 * bug fix: redefining a class with different superclasses now correctly
237 removes it from the direct-subclasses of its previous superclasses.
238 (reported by David Morse)
239 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
240 new value now works. (reported by Bruno Haible)
241 * bug fix: correct canonicalization of multiple non-standard slot
242 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
243 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
244 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
245 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
246 now exists, an signals an error.
247 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
248 during macroexpansion for non-iteration variables as well. (reported
249 by Bruno Haible for CMUCL)
250 * bug fix: Cyclic structures and unprintable objects in compiler
251 messages no longer cause errors. (reported by Bruno Haible)
252 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
253 and READ-FROM-STRING all now return a primary value of NIL if
254 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
255 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
256 (reported by Bruno Haible for CMUCL)
257 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
258 do not designate a condition. (reported by Bruno Haible for
260 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
261 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
262 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
263 permitting resolution in favour of any of the conflicting symbols.
264 (reported by Bruno Haible for CMUCL)
265 * bug fix: EQUAL compiler optimizations is less aggressive on
266 strings which can potentially compare true despite having distinct
267 specialized array element types.
268 * bug fix: unit enumerations can be defined without dividing by
269 zero. (thanks to Vincent Arkesteijn)
270 * FORMAT compile-time argument count checking has been enhanced.
271 (report from Bruno Haible for CMUCL)
272 * a partial workaround for the bug 262: the compiler does not try to
273 inline-expand a local function doing RETURN-FROM from a deleted
274 BLOCK. (thanks to Peter Denno for the bug report and to David
275 Wragg for the simple test case)
276 * fixed some bugs revealed by Paul Dietz' test suite:
277 ** RENAME-PACKAGE allows all package designators as new package
279 ** constraint propagation and lambda variable substitution are
280 more cautious in dealing with partially deleted code.
281 ** compiler sometimes lost reoptimization passes.
282 ** CERROR, given a condition as condition designator, passes its
283 remaining arguments to the continue format control without
285 ** Case-altering FORMAT directives work correctly on non-ASCII
287 ** The REAL type specifier handles bounds outside the floating
288 point ranges without signalling FLOATING-POINT-OVERFLOW.
289 ** Functions with IR1-transformations can create intercomponent
290 references to global functions.
291 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
293 ** FORMAT ~R treats a nil value for its first parameter correctly.
295 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
296 * enhancement: saving cores with foreign code loaded is now
297 supported on x86/NetBSD and sparc/Linux in addition to the previously
299 * bug fix: on some platforms repeated installations caused multiple
300 copies of HTML documentation to be installed -- should not happen
301 any more. (reported by Stefan Scholl)
302 * bug fix: parsing self-recursive alien record types multiple times
303 no longer causes infinite recursion. (reported by Thomas F. Burdick,
304 original patch by Helmut Eller for CMUCL)
305 * bug fix: stack-exhaustion detection works now on NetBSD as well.
306 (thanks to Richard Kreuter)
307 * bug fix: defining classes whose accessors are methods on existing
308 generic functions in other (locked) packages no longer signals
309 bogus package lock violations. (reported by François-René Rideau)
310 * bug fix: special variables as DEFMETHOD parameters no longer have
311 associated bogus type declarations. (reported by David Wragg and
313 * bug fix: read-write consistency on streams of element-type
314 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
315 * bug fix: redefiniton of the only method of a generic function with
316 no DEFGENERIC no longer emits a full WARNING. In addition,
317 redefinition of generic functions with no DEFGENERIC to an
318 incompatible lambda list now signals an error. (thanks to Zach
320 * bug fix: DEFGENERIC now works even when there's a function of the
321 same name in an enclosing lexical environment. (thanks to Zach
323 * fixed compiler failure, caused by instrumenting code during
324 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
325 * optimization: added loop analysis and improved register allocation
326 * fixed some bugs revealed by Paul Dietz' test suite:
327 ** POSITION on displaced vectors with non-zero displacement
328 returns the right answer.
329 ** (SIMPLE-STRING) is a valid type specifier for sequence
331 ** *PRINT-LEVEL* handling for slotless structures is pedantically
333 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
334 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
336 ** The justification version of the FORMAT ~< directive treats
337 non-zero minpad parameter correctly.
339 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
340 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
341 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
342 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
343 the supported interface.
344 * new feature: Single-stepping of code compiled with DEBUG 2 or
345 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
346 * new feature: saving cores with foreign code loaded is now
347 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
348 Timothy Moore's work for CMUCL)
349 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
350 parameters to * instead of NIL if no initform is supplied.
351 (reported by Johan Bockgaard)
352 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
353 correctly when FUNCALL appears as the car of the form. Note:
354 despite this FUNCALL forms are not currently subject to
355 compiler-macro expansion. (port of Raymond Toy's fix for the
356 same from CMUCL, reported by Johan Bockgaard)
357 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
358 (thanks for Teemu Kalvas)
359 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
360 case instead of terminating the iteration (thanks for Julian
361 Squires, Sean Champ and Raymond Toy)
362 * bug fix: incorrect expansion of defgeneric that caused a style
363 warning. (thanks for Zach Beane)
364 * bug fix: slot accessor effective method computation works properly
365 for classes with multiple non-standard applicable methods on
366 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
367 * on x86 compiler supports stack allocation of results of LIST and
368 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
369 CMUCL implementation by Gerd Moellmann)
370 * fixed some bugs revealed by Paul Dietz' test suite:
371 ** FORMAT strings with both the ~<~:;~> form of the justification
372 directive and pretty-printing directives cause an error.
374 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
375 * incompatible change: the internal functions
376 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
377 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
378 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
379 instead of the old functions.
380 * new feature: on platforms where "dladdr" is available foreign
381 function names now appear in backtraces. (based on Helmut Eller's
383 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
384 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
386 * documentation: ASDF manual is now distributed with SBCL.
387 * bug fix: SBCL can now load its contributed modules with REQUIRE
388 even if the system-provided entries have been removed from
389 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
391 * bug fix: backtraces involving undefined functions or assembly
392 routines are more informative. (thanks to Brian Downing)
393 * bug fix: mutually referent alien structures now work correctly.
394 (reported by Rick Taube)
395 * bug fix: structures defined by WITH-ALIEN can be referred to
396 within other definitions in the same WITH-ALIEN.
397 * bug fix: division operators (MOD, TRUNCATE and the like) with
398 constant zero divisors and integer dividends no longer generate
400 * bug fix: provide default methods for INPUT-STREAM-P and
401 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
402 * bug fix: improve the detection and resolution of MOP metacycles
403 (where effective method computation depends on the generic function
404 having its effective method computed). (reported by Bruno Haible)
405 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
406 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
407 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
408 * bug fix: pathnames with relative directory components can be
409 represented relative to default pathnames.
410 * optimization: in taking the GCD of bignums, reduce the two bignums
411 to approximately the same size (using Euclid's algorithm) before
412 applying the more sophisticated binary GCD. (thanks to Juho
414 * optimization: COUNT on bitvectors now operates word-at-a-time.
415 * optimization: ASH with a positive, but not necessarily constant,
416 (leftwards) shift, when in a modular context, is compiled to a
418 * fixed some bugs revealed by Paul Dietz' test suite:
419 ** FORMAT variable parameters ("~V<char>") are defaulted properly
420 if the corresponding argument is NIL.
421 ** FORMAT directives accepting a minpad parameter treat negative
423 ** literal commas as character parameters to FORMAT directives are
425 ** literal spaces directly after ~<Newline> directives within a
426 format-logical-block (~:< ~@:>) do not induce :FILL-style
427 conditional newlines.
428 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
429 as specified: it no longer includes conditional newlines).
430 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
432 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
433 :PER-LINE-PREFIX argument does not evaluate to a string.
435 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
436 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
437 the manual for details; for now, package locks can be disabled by
438 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
439 if no major problems are found then it is likely that they will be
440 compiled in unconditionally.
441 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
442 unsupported operators on all platforms. To load a shared library
443 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
444 object file, link it into a shared library outside of SBCL and
445 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
447 * bug workaround: ROOM T and the GC were not working together
448 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
449 reporting facility aren't preserved. That reporting has been
450 disabled, so now until and unless someone figures out how to make it
451 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
452 * minor incompatible change: as threatened around sbcl-0.8.0, the
453 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
454 The symbols are also exported from SB-PCL for backwards
455 compatibility, but more so than before SB-PCL should be treated as
456 an implementation-internal package.
457 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
459 * the SB-SPROF contrib now works on (most) non-x86 architectures.
460 It is known as of this release not to work on the Alpha, however.
461 * fixed bug #167: errors signalled due to illegal syntax in method
462 bodies are now more legible.
463 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
464 designators and can hence be used with TYPEP.
465 * fixed bug #333: CHECK-TYPE now ensures that the type error
466 signalled, if any, has the right object to be accessed by
467 TYPE-ERROR-DATUM. (reported by Tony Martinez)
468 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
469 5.1.2.3 for multiple-value place subforms. (reported by Kalle
471 * fixed bug #334: programmatic addition of slots using specialized
472 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
473 and :ALLOCATION :CLASS slots.
474 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
475 float exponents. (rereported by Peter Seibel)
476 * fixed a bug: #\Space (and other whitespace characters) are no
477 longer considered to be macro characters in standard syntax by
479 * fixed bug: initialization of condition class metaobjects no longer
480 causes an instance of the condition to be created. (reported by
482 * fixed bug: it is now possible to have slots such that
483 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
484 description is neither :INSTANCE nor :CLASS.
485 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
486 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
487 SLOT-VALUE-USING-CLASS), and no longer causes errors with
488 non-standard SLOT-DEFINITION-ALLOCATION values.
489 * fixed bugs: various race conditions handling exiting threads.
490 CL-PPCRE's thread torture test now passes.
491 * fixed arguably-a-bug: GC time proportional to number of threads in
492 system even when most of them are idle
493 * optimization: improved performance of BIT and SBIT on bit-vectors.
494 * .fasl file incompatibility: The fasl file version number has
495 been incremented because of changes associated with package locks.
497 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
498 * minor incompatible change: the system no longer provides
499 optimization hints (conditions which are TYPEP
500 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
501 mode; these hints are still emitted when higher SPEED optimization
503 * new contrib module: a sampling profiler (profiling by statistical
504 sampling, rather than by instrumenting functions) is available as
505 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
506 * the behaviour of the standard function ED is now customizeable by
507 third parties through a hook variable: see ED's documentation
508 string for information on the protocol.
509 * the compiler no longer emits efficiency notes for (FUNCALL X)
510 when the type of X is uncertain under default optimization
512 * fixed bug 276: mutating a binding of a specialized parameter to a
513 method to something that is not TYPEP the specializer is now
515 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
516 exhibit the required behaviour on the x86 platform. (thanks to
517 Peter van Eynde, Eric Marsden and Bruno Haible)
518 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
519 even for difficult arguments. (reported by Peter Graves)
520 * fixed bug 141a: the backquote printer now descends quoted
522 * fixed another bug in backquote printing: no more destructive
523 modification of the form's list structure. (reported by Brian
525 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
526 move between its address being taken and the call to
527 interrupt_thread, fixing a crashing race condition.
528 * the SB-POSIX contrib implementation has been adjusted so that it
529 no longer exhibits ridiculously poor performance when constructing
530 instances corresponding to C structs.
532 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
533 * minor incompatible change: the sb-grovel contrib now treats C
534 structures as alien (in the sense of SB-ALIEN) objects rather than
535 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
536 has implications for memory management of client code
537 (sb-grovel-returned objects must now be manually managed) and for
538 type safety (alien objects now have full types).
539 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
540 used to control emission of compiler diagnostics, rather than the
541 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
542 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
543 quality should be considered deprecated.
544 * install.sh now installs the user manual as well
545 * (not quite a new documentable feature, but worth considering in
546 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
547 beginnings of a semantically meaningful condition hierarchy is
548 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
550 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
551 designator as the defaults argument.
552 * fixed bug: Displaced arrays whose displaced-to array has become
553 too small now cause ARRAY-DIMENSION to signal an error, providing
554 for safer bounds-checking. (reported by Bruno Haible)
555 * fixed bug: DEFCLASS slot definitions with identical :READER and
556 :WRITER names now signal a reasonable error. (reported by Thomas
558 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
560 * fixed bug: Invalid slot specification errors now print correctly.
561 (thanks to Zach Beane)
562 * fixed bug 320: Shared to local slot value transfers in class
563 redefinitions now happen corrently from superclasses as
564 well. (reported by Bruno Haible)
565 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
567 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
568 improper lists correctly. (reported by Bruno Haible)
569 * fixed bug 313: source-transform for <fun-name> was erroneously
570 applied to a call of a value of a variable with name <fun-name>.
571 (reported by Antonio Menezes Leitao)
572 * fixed bug 307: The obsolete instance protocol ensures that
573 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
574 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
575 remove dead unknown-values globs from the middle of the stack.
576 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
577 easier to resume long computations after using *BREAK-ON-SIGNALS*
578 to diagnose and fix failures. (thanks to Nikodemus Siivola)
579 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
580 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
581 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
582 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
583 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
585 * fixed a (fairly theoretical) bug in string printing: if
586 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
587 does not have array-element-type equal to the most general string
589 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
590 function name. (thanks to Nikodemus Siivola)
591 * fixed bug: compile-time format string checker failed on
592 non-closed ~{. (reported by Thomas F Burdick)
593 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
594 don't warn on legal loop code involving more than one aggregate
595 boolean. (thanks to Nikodemus Siivola)
596 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
597 in keyword position to MAKE-PACKAGE was not being reported as
599 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
600 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
602 * fixed bugs 280 and 312: the checking for multiple definitions in a
603 file is less likely to become confused by uses of inline
605 * fixed bug: the #S reader macro performs the keyword coercion
606 specified for slot names. (reported by Kalle Niemitalo)
607 * fixed bug: lambda lists may contain symbols whose names start with
608 & but are not lambda-list-keywords; their occurrence triggers a
610 * fixed bug 321: define-method-combination argument lambda lists do
611 not cause invalid code to be generated when &optional or &aux
612 variables have default values. (reported by Bruno Haible)
613 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
614 have CLOS classes; this appears also to have fixed reported
615 instability in stack exhaustion detection.
616 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
617 MUFFLE-WARNING when no associated restart is present is now
619 * optimization: rearranged the expansion of various defining macros
620 so that each expands into only one top-level form in a
621 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
623 * optimization: used a previously little-used slot in symbols to
624 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
626 * fixed some bugs revealed by Paul Dietz' test suite:
627 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
629 ** Fixed an optimization bug related to inheritance of initforms
630 from local to shared slots.
631 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
632 ** CAST optimizer forgot to flush argument derived type.
633 ** print/read consistency on floats is now orders of magnitude
634 more likely. (thanks also to Bruno Haible for a similar report
636 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
637 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
638 for objects of type REAL. Make it so.
639 ** SXHASH is specified (infelicitously) to respect similarity,
640 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
641 it so. (thanks to Markus Ziegler)
642 ** on the Alpha, the compiler succeeds in compiling functions
643 returning a known number of arguments greater than 63.
644 ** fixed handling of invalid NIL arguments in keyword position
646 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
648 ** provided a readably-printable representation for RANDOM-STATE
650 ** ensured that pathnames, where they have a namestring, always
651 print using #P"..." syntax.
653 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
654 * Support for the forthcoming 2.0 version of the NetBSD kernel
655 running on x86 hardware has been added. (thanks to Perry
656 E. Metzger most immediately, and others for their past work)
657 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
658 current release version; SBCL's OpenBSD support had been broken
659 since about the time of OpenBSD's switch to ELF binary format.)
660 * [placeholder for DX summary]
661 ** user code with &REST lists declared dynamic-extent, under high
662 speed or space and low safety and debug optimization policy.
663 * The manual has been converted to Texinfo format and the debugger
664 chapter from the cmucl manual has been added.
665 * A facility has been added to extract documentation strings from
666 sbcl and store them as Texinfo-formatted snippets for inclusion in
667 the manual (via Texinfo's @include directive)
668 * bug fix: compiler emitted division in optimized DEREF. (thanks for
669 the test case to Dave Roberts)
670 * bug fix: multidimensional simple arrays loaded from FASLs had fill
671 pointers. (reported by Sean Ross)
672 * bug fix: PROFILE output is printed nicely even for large numerical
673 values. (thanks to Zach Beane)
674 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
675 greater than 32 handle EOF correctly.
676 * bug fix: on X86 an immediate argument of the IMUL instruction is
677 correctly printed in disassembly. (thanks to Lutz Euler)
678 * bug fix: class slots in redefined classes preserve their old
679 values. (thanks to Bruno Haible and Nikodemus Siivola)
680 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
681 succeeds. (reported by Marco Baringer)
682 * fixed some bugs revealed by Paul Dietz' test suite:
683 ** READ-SEQUENCE now works on ECHO-STREAMs.
684 ** RATIONALIZE works more according to its specification. (thanks
687 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
688 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
689 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
690 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
691 should matter to you only if you rebind the printer control
692 variables and then find you want different bindings in the
693 debugger than in the ordinary execution of your program.)
694 * The runtime build system has been tweaked to support building
695 (on SPARC/SunOS) using a C compiler which invokes Sun's own
696 assembler and linker. (thanks to Nikodemus Siivola)
697 * Unbound, undefined, undeclared variables now trigger full
698 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
699 more often programmer error than deliberate exploitation of undefined
701 * optimization: the hash algorithm for strings has changed to one
702 that is less vulnerable to spurious collisions. (thanks to Juho
704 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
705 needless bounds checking. (thanks to Juho Snellman)
706 * optimization: implemented multiplication as a modular
707 (UNSIGNED-BYTE 32) operation on the PPC backend.
708 * fixed some bugs revealed by Paul Dietz' test suite:
709 ** ADJUST-ARRAY now copies the datum in a zero rank array if
711 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
712 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
713 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
715 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
717 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
719 ** ECHO-STREAMs no longer attempt to echo the end of file value to
720 their output stream on EOF from read.
721 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
722 have been read to end-of-file.
723 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
725 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
726 description of determination of which consecutive characters
728 ** Printing the "Space" character with escaping on now yields "#\\ ",
729 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
730 ** Reading floating-point numbers with *READ-BASE* set to a number
731 less than 10 works correctly.
732 ** Reading floating-point numbers with *READ-BASE* set to a number
733 more than 10 works correctly.
734 ** Printing with *PRINT-READABLY* targets the standard readtable, not
735 the readtable currently in effect.
737 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
738 * minor incompatible change: parsing of namestrings on a physical
739 (Unix) host has changed; numbers after the final #\. in a
740 namestring are no longer interpreted as a version field. This is
741 intented to be largely invisible to the user, except that the
742 meaning of the namestring "*.*.*" has changed: it now refers to a
743 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
744 should usually be replaced by
745 (make-pathname :name :wild :type :wild :version :wild)
746 with the added benefit that this is more likely to be portable.
747 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
748 option to OPEN now signals an error if the file being opened
749 exists; this may have an impact on existing code.
750 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
751 signals received, on the PowerPC platforms (both Linux and
752 Darwin). (thanks to Pierre Mai for pointing out the location of
754 * several fixes on OS X: The system now builds and runs cleanly on
755 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
756 (thanks to Brian Mastenbrook)
757 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
758 second-resolution: integer multiples of 1/3600 between -24 and 24.
759 (thanks to Vincent Arkesteijn)
760 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
761 type when called with 1 argument; PEEK-CHAR checked type of
762 PEEK-TYPE only after having read first character from a
763 stream. (reported by Peter Graves)
764 * bug fix: the garbage collector now has much better locality
765 behaviour, and in particular no longer treats all memory as being
766 exhausted when many small objects point to each other in a deeply
768 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
769 recognized as being TYPEP their class.
770 * bug fix: the PUSHNEW documentation string has been corrected.
771 (thanks to Vincent Arkesteijn)
772 * bug fix: defaulting of the value for the last of an atypically
773 large number of multiple values being bound was not being
774 performed correctly on the Alpha or PPC platforms
775 * optimization: implemented multiplication as a modular
776 (UNSIGNED-BYTE 32) operation on the x86 backend.
777 * optimization: SEARCH on simple-base-strings can now be open-coded.
778 (see also contrib/compiler-extras.lisp for inspiration for
779 teaching the compiler about the Boyer-Moore algorithm).
780 * value, returned by MAX (and MIN) called with several EQUALP, but
781 not EQL, arguments now does not depend on compiler settings.
782 * fixed some bugs revealed by Paul Dietz' test suite:
783 ** in stack analysis liveness information is propagated from
784 non-local entry points.
785 ** pathwise CAST removing failed when the CAST node did not start
787 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
788 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
790 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
791 host is already defined.
792 ** RENAME-FILE works on streams instead of signalling an internal
794 ** PEEK-CHAR uses the current readtable when determining whether
795 or not a character is whitespace.
796 ** MERGE-PATHNAMES handles the case when the pathname does not
797 specify a name while the default-pathname specifies a version.
798 ** Pathnames now stand a better chance of respecting print/read
800 ** Attempting to use standardized file system operators with a
801 pathname with invalid :DIRECTORY components signals a
803 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
805 ** DIRECTORY on logical pathnames is more correct.
806 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
807 signal a TYPE-ERROR if their argument is not a stream.
808 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
809 designator argument does not designate a stream.
810 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
811 examining the synonym.
812 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
814 ** OPEN and WITH-OPEN-STREAM allow opening streams with
815 element-type larger than ([UN]SIGNED-BYTE 32).
817 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
818 * When built with the :SB-FUTEX feature, threaded builds now take
819 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
820 for faster/more reliable mutex and condition variable support.
821 * Incompatible change (but one you probably shouldn't have been using
822 anyway): the interface and code for arbitrating between multiple
823 threads in the same user session has been redesigned.
824 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
825 error when called without an explicit environment argument.
826 (thanks to Nikodemus Siivola)
827 * bug fix: buffered :DIRECTION :IO streams are less likely to become
828 confused about their position. (thanks to Adam Warner and Gerd
830 * bug fix: Pretty printing backquoted forms with unquotations in the
831 argument list position of various code constructs such as LAMBDA
832 now works correctly. (reported by Paul Dietz)
833 * bug fix: Pretty printing unquotations no longer loses all
834 stream position information.
835 * optimization: performance of string output streams is now less
836 poor for multiple small sequence writes.
837 * optimization: performance of CSUBTYPEP in the presence of complex
838 expressions involving CONS and NOT many times has been improved.
839 (reported by Paul Dietz)
840 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
842 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
843 streams for the REPL, for the debugger, and so on. (thanks to
845 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
846 * fixed some bugs revealed by Paul Dietz' test suite:
847 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
848 is ignored if it is too small, rather than propagating through
849 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
850 ** extremely complex negations of CONS types were not being
851 sufficiently canonicalized, leading to inconsistencies in
853 ** VALUES tranformer lost derived type.
855 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
856 * fixed a bootstrapping bug: the build process no longer assumes
857 that the various BOOLE-related constants have the same value in
858 host and target lisps. (noted by Paul Dietz' test suite on an
859 SBCL binary built from CLISP)
860 * The system can now be dynamically linked on the MIPS platform,
861 which enables dynamic loading of foreign code from Lisp. (thanks
862 to Ralf Baechle for discussions on the MIPS ABI)
863 * The system now records debugging information for its own source
864 files in a filesystem-position-independent manner, relative to
865 the "SYS" logical host.
866 * fixed a compiler bug: MV-LET convertion did not check references
867 to the "max args" entry point. (reported by Brian Downing)
868 * tweaked disassembly notes to be less confident about proclaiming
869 some instruction as an LRA. (thanks to Brian Downing)
870 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
871 now each have their own history, command character, and other
872 characteristics. (thanks to David Lichteblau)
873 * fixed some bugs revealed by Paul Dietz' test suite:
874 ** compiler failure in compiling LOGAND expressions including a
876 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
877 shift greater than 32.
878 ** FLUSH-DEST did not mark blocks for type check regeneration.
879 ** HANDLER-CASE failed to accept declarations in handler clauses
880 in some circumstances.
882 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
883 * New code in contrib/sb-introspect (still probably not entirely
884 stable yet) provides some support for smart Lisp development
885 environments like SLIME.
886 * The conditions signalled for errors occurring when loading .fasl
887 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
888 in a way which should help ASDF recover gracefully.
889 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
890 stuff has been cleaned up. If you code contrib/ stuff, this might
891 affect you, and you can look at contrib/README, contrib/STANDARDS,
892 and/or the 0.8.4.27 diff to check.
893 * In full calls the compiler now does not generate checks for declared
894 argument types for all arguments.
895 * various threading fixes
896 ** and some experimental patches which didn't make it into
897 the main tree for this release, but which are shipped in
898 contrib/experimental-thread.patch as a possible fix for some
899 failures (deadlock, spinning...) in GC-intensive multithreaded
901 * fixed PPC build problem (source code incompatibility of different
902 library versions): added offsetof() hackery which attempts to divine
903 where glibc maintainers put uc_mcontext today
904 * fixed bug 282: compiler does not trust type assertions while passing
905 arguments to a full call.
906 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
907 &OPTIONAL VALUES type parameter.
908 * fix bug 214: algorithm for noting rejected templates is now more
909 similar to that of template seletion. (also reported by rydis on
911 * fixed bug 141b: printing backquoted information readably and prettily
912 inserts a space where necessary.
913 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
914 (CHARACTER 10) are now reported as errors, rather than propagated
915 as unknown types. (reported by piso on #lisp)
916 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
917 with values NIL and :ERROR. (thanks to Milan Zamazal)
918 * fixed bug 191c: CLOS now does proper keyword argument checking as
919 described in CLHS 7.6.5 and 7.6.5.1.
920 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
921 longer raise an error; further, using a list as a for-as-arithmetic
922 counter now raises a meaningful error.
923 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
924 now understood by sequence creation functions such as MAKE-SEQUENCE
926 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
927 a STRING-INPUT-STREAM.
928 * compiler enhancement: SIGNUM is now better able to derive the type
930 * type declarations inside WITH-SLOTS are checked. (reported by
932 * fixed some bugs revealed by Paul Dietz' test suite:
933 ** incorrect optimization of TRUNCATE for a positive first
934 argument and negative second.
935 ** compiler failure in let-convertion during flushing dead code.
936 ** compiler failure while deriving type of TRUNCATE on an
937 interval, containing 0.
938 ** ASH of a negative bignum by a negative bignum count now returns
940 ** intersection of CONS types now canonicalizes properly, fixing
941 inconsistencies in SUBTYPEP.
943 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
944 * incompatible change: The --disable-debugger command line
945 option now clobbers the debugger at a more fundamental
946 level, by redefining #'INVOKE-DEBUGGER instead of by
947 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
948 is specified by ANSI to ignore *DEBUGGER-HOOK* and
949 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
950 enter the debugger REPL and then suffer recursive errors
951 because *DEBUG-IO* is also messed up in --disable-debugger mode;
952 while under the new system, BREAK in --disable-debugger mode
953 terminates the system just as an unhandled error would.
954 * fixed compiler performance when processing loops with a step >1;
955 * bug fix: DOCUMENTATION now retrieves generic function
956 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
957 support has been systematized, and now supports the methods
958 specified by ANSI, along with a default method and a method for
959 slot documentation. (reported by Nathan Froyd)
960 * bug fix: effective methods associated with a generic function are
961 no longer cached over a change of that generic function's method
962 combination. (reported by Andreas Fuchs)
963 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
964 names a non-existent directory. (thanks to Andreas Fuchs)
965 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
966 the PPC platform, lead to stack corruption; this has been fixed.
967 (reported by Rainer Joswig)
968 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
969 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
970 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
971 arithmetic in general, is now much more reliable on the PPC
973 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
974 platform now returns the right answer.
975 * optimization: restored some effective method precomputation in
976 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
977 precomputation is now tunable.
978 * optimization: compiler-internal data structure use has been
979 reviewed, and changes have been made that should improve the
980 performance of the compiler by about 20%.
981 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
982 simple-base-strings and simple-bit-vectors is improved.
983 * optimization: the optimization of 32-bit logical and arithmetic
984 functions introduced in version 0.8.3 on the x86 has been
985 implemented on the mips, ppc and sparc platforms; an
986 implementation of the same facility, but for 64-bit arithmetic,
987 has been added for the alpha.
988 * microoptimization: the compiler is better able to make use of the
989 x86 LEA instruction for multiplication by constants.
990 * bug fix: in some situations compiler did not report usage of
991 generic arithmetic in (SPEED 3) policy.
992 * bug 145b fix: compiler used wrong type specifier while converting
993 MEMBER-types to numeric.
994 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
996 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
997 index argument now works properly on the Alpha platform.
998 * bug fix: floating point exception treatment on the Alpha platform
1000 * bug fix: FILE-POSITION works much better on string input and
1001 output streams. (thanks to Nikodemus Siivola)
1002 * bug fix: many threading/garbage collection symptoms sorted.
1003 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1004 might be pseudo-atomic.
1005 * internal change: Stopping for GC is now done with signals not
1006 ptrace. GC is now done in whichever thread wanted it, instead of
1008 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1010 * bug fix: result form in DO is not contained in the implicit
1012 * incompatible change: ICR structure is changed; the value part of
1013 CONTINUATION is now called LVAR; corresponding functions are
1014 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1015 * added type deriver for ISQRT (thanks to Robert E. Brown).
1016 * bug fix: better support for loading from the command line when an
1017 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1019 * fixed some bugs revealed by Paul Dietz' test suite:
1020 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1021 ** ROUND and FROUND now give the right answer when given very
1022 small float arguments.
1023 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1025 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1026 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1027 ** LCM should return a non-negative integer.
1028 ** PARSE-INTEGER returned the index of a terminator instead of the
1029 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1030 ** PARSE-INTEGER returned an incorrect index being applied to a
1032 ** LCM with two arguments of 0 returns 0 rather than signalling
1034 ** unsigned addition of a 32-bit constant with the high bit set no
1035 longer causes an internal compiler error on the x86.
1036 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1037 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1038 with negative last argument.
1039 ** byte specifiers with zero size and position no longer cause
1040 an error during type derivation.
1041 ** bignum multiplication on the Alpha platform now returns the
1043 * porting: The system now builds on SuSE AMD64, although it still
1044 generates a 32-bit binary.
1045 * .fasl file incompatibility: The fasl file version number has
1046 been incremented (because of the changes to internal compiler
1047 data structures referred to above).
1049 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1050 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1051 more accurately, on the Darwin kernel running on PowerPC hardware.
1052 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1053 * Compiler code deletion notes now signal a condition of type
1054 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1055 an associated MUFFLE-WARNING restart.
1056 * The compiler now performs limited argument count validation of
1057 constant format strings in FORMAT, and where appropriate in ERROR,
1058 CERROR and WARN. (thanks to Gerd Moellmann)
1059 * New ASDF-INSTALL contrib can be used for automatic download and
1060 installation of third-party Lisp code from CCLAN or other sites
1062 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1063 which forces another thread to execute a function supplied by the
1065 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1066 now accept and act upon their :ELEMENT-TYPE keyword argument.
1067 (reported by Martin Atzmueller, Edi Weitz)
1068 * bug fix: FILE-POSITION now accepts position designators up to
1069 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1070 is the greater. (thanks to Patrik Nordebo)
1071 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1072 Kalle Olavi Niemitalo)
1073 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1074 printable. (reported by Eric Marsden)
1075 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1076 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1077 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1078 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1079 (thanks to Stig Sandoe)
1080 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1081 UNREAD-CHAR now correctly includes the unread character in the
1082 target sequence. (thanks to Gerd Moellmann)
1083 * bug fix in threaded builds: the system can now be suspended and
1084 resumed by shell job control with minimal disruption.
1085 * bug fixes in times and timezones >2038AD
1086 * better handling of "where is GNU make?" problem in build scripts
1087 (thanks to Nikodemus Siivola)
1088 * new optimization: inside a named function any reference to a
1089 function with the same name is considered to be a self-reference;
1090 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1091 optimization quality.
1092 * new optimization on x86: logical functions and + now have
1093 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1094 used when the result is truncated to 32 bits.
1095 * VALUES declaration is partially enabled.
1096 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1097 * fixed some bugs revealed by Paul Dietz' test suite:
1098 ** The system now obeys the constraint imposed by
1099 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1100 types form a lattice under type intersection.
1101 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1102 ** ASSOC now ignores NIL elements in an alist.
1103 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1104 and (1+ MOST-POSITIVE-FIXNUM) answers.
1105 ** The addition of a method with invalid qualifiers to a generic
1106 function does not cause an error to be signalled immediately;
1107 a warning is signalled, and the error is generated only on
1108 calling the generic function.
1109 * changed .fasl file version number, in part to add type codes for
1110 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1111 obscure ANSI requirements
1113 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1114 * fixed bug 148: failure to inline-expand a local function left
1115 garbage, confusing the compiler.
1116 * fixed bugs 3cd: structure slot readers perform type check if the
1117 slot can have an invalid value (i.e. it is either not initialized
1118 or can be written with a less specific slot writer).
1119 * bug fix: the compiler now traps array references to elements off
1120 the end of an array; previously, the bounds checking in some
1121 circumstances could go off-by-one.
1122 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1124 * type declarations for array element types now obey the description
1125 on the CLHS page "Declaration TYPE", as per discussions on
1126 sbcl-help around 2003-05-08. This means that a declaration
1127 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1128 declaration, all references to BAR will be asserted or assumed
1129 (with THE, so dependent on compiler policy) to involve objects of
1130 type FOO. Note that no such declaration is implied in
1131 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1132 * declared types of functions from the "Conditions"
1133 chapter. (reported by Paul Dietz)
1134 * bug fix: CERROR accepts a function as its first argument.
1135 * bug fix: NTH an NTHCDR accept a bignum as index
1136 arguments. (reported by Adam Warner)
1137 * optimization: character compare routines now optimize comparing
1138 against a constant character. (reported by Gilbert Baumann)
1139 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1140 argument now works properly on the MIPS platform.
1141 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1142 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1143 declared to have a functional type, does not cause an error
1145 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1146 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1147 * fixed bug: if last continuation of a deleted block has a
1148 destination, this destination should be deleted too. (reported by
1150 * fixed a bug in the bootstrap process: the host compiler's values
1151 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1152 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1153 test case from Patrik Nordebo)
1154 * improved the ability of the disassembler on the PPC platform to
1155 provide helpful disassembly notes.
1156 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1157 the class in more cases than previously.
1158 * bug fix: FILE-POSITION now understands :START and :END for
1159 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1160 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1161 8) data. (thanks to David Lichteblau)
1162 * bug fix: it is possible to add a method to a generic function
1163 without lambda list.
1164 * bug fix: reader failed to signal END-OF-FILE inside an
1165 object representation. (reported by Nikodemus Siivola)
1166 * fixed some bugs revealed by Paul Dietz' test suite:
1167 ** LAST and [N]BUTLAST should accept a bignum.
1168 ** condition slot accessors are methods.
1169 ** (VECTOR NIL) is a subtype of STRING.
1171 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1172 * minor incompatible change: some nonsensical specialized lambda
1173 lists (used in DEFMETHOD) which were previously ignored now signal
1175 * minor incompatible change: the system is now aware of the types of
1176 variables in the COMMON-LISP package, and will signal errors for
1177 most violations of these type constraints (where previously they
1178 were silently accepted).
1179 * minor incompatible change: COMPILE-FILE now uses the freedom
1180 afforded (ANSI 3.2.2.3) to use derived function types for
1181 functions defined in the same file. This also permits the system
1182 to warn on static type mismatches and function
1183 redefinition. (Currently it does not work with high DEBUG level.)
1184 * minor incompatible change: VALUES declaration is disabled.
1185 * When issuing notes, the compiler now signals a condition of type
1186 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1187 restart for use in user handlers. It is expected that the
1188 COMPILER-NOTE condition will eventually become a condition
1189 supertype to a hierarchy of note types, which will then be
1190 handleable in a similar fashion. However, at the moment, no such
1191 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1192 but it's an implementation detail, not a classification for the
1194 * Changes in type checking closed the following bugs:
1195 ** type checking of unused values (192b, 194d, 203);
1196 ** template selection based on unsafe type assertions (192c, 236);
1197 ** type checking in branches (194bc).
1198 * A short form of VALUES type specifier has ANSI meaning (it has
1199 increased the number of situations when SBCL cannot perform type
1201 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1202 keywords or constants is permissible.
1203 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1204 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1205 classes. (thanks to Antonio Martinez)
1206 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1207 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1208 argument to a pretty-print stream. Instead, it leaves any such
1209 operations to DESCRIBE-OBJECT methods.
1210 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1212 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1213 lambda lists. (thanks to David Lichteblau)
1214 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1215 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1216 (thanks to Gerd Moellmann)
1217 * fixed bug 246: increased compilation speed of long
1218 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1220 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1222 * DEFCONSTANT now throws a condition of type
1223 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1224 constant to a non-EQL value; CONTINUE and ABORT restarts
1225 respectively change and preserve the value.
1226 * fixed bug 63: The code walker, part of the implementation of CLOS,
1227 is now better at handling symbol macros.
1228 * bug fix: There is no longer an internal implementation type named
1229 CL:LENGTH. (reported by Raymond Toy)
1230 * bug fix: In macro-like defining macros/special operators the
1231 implicit block does not enclose the lambda list.
1232 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1233 enclosed in parentheses) are not suitable as type specifiers, and
1234 their use properly signals an error now.
1235 * bug fix: An argument count mismatch for a type specifier in code
1236 being compiled no longer causes an unhandled error at compile
1237 time, but signals a compile-time warning.
1238 * fixed simple vector readable printing
1239 * bug fix: DESCRIBE takes more care over whether the class
1240 precedence list slot of a class is bound before accessing it.
1241 (reported by Markus Krummenacker)
1242 * bug fix: FORMATTER can successfully compile pretty-printer format
1243 strings which use variants of the ~* directive inside.
1244 * bug fix: SEARCH now applies its TEST predicate to the elements of
1245 the arguments in the correct order. (thanks to Wolfhard Buss)
1246 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1247 in type inference. (thanks to Robert E. Brown)
1248 * bug fix: Reading in symbols with an explicit package name of ""
1249 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1250 * fixed some bugs revealed by Paul Dietz' test suite:
1251 ** NIL is now allowed as a structure slot name.
1252 ** Arbitrary numbers, not just REALs, are allowed in certain
1253 circumstances in LOOP for-as-arithmetic clauses.
1254 ** Multiple class redefinitions before slot access no longer
1255 causes a type error.
1256 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1257 association between the name and a class.
1258 ** Generic functions with non-standard method-combination and over
1259 five methods all of which return constants no longer return NIL
1260 after the first few invocations. (thanks to Gerd Moellmann)
1261 ** CALL-NEXT-METHOD with no arguments now passes the original
1262 values of the arguments, even in the presence of assignment.
1263 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1265 ** DISASSEMBLE works with closures and funcallable instances.
1266 ** ADD-METHOD now returns the generic function, not the new method.
1267 ** FIND-METHOD signals an error if the lengths of the specializers
1268 is incompatible with the generic function, even if the ERRORP
1270 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1271 which its argument is a member.
1272 ** DEFCLASS only redefines the class named by its class-name
1273 argument if that name is the proper name of the class;
1274 otherwise, it creates a new class.
1275 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1276 of the UNBOUND-SLOT condition to the name of the slot.
1277 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1278 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1279 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1280 SLOT-MAKUNBOUND in the specified fashion.
1282 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1283 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1284 cross-compilation host. As a consequence, we can now bootstrap our
1285 way up to SBCL starting with a bare gcc toolchain and human-readable
1286 source code (first the source to CLISP, then the source to SBCL).
1287 * A contributed module containing a partial implementation of the
1288 simple-streams interface has been included. (thanks to Rudi
1290 * A contributed module implementing the RFC1321 Message Digest
1291 Algorithm, known as MD5, has been included.
1292 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1293 no longer has any effect, as the code controlled by this feature
1294 has been deleted. (As far as we know, no-one has ever built using
1295 this feature, and its semantics were confused in any case).
1296 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1297 work on conditions (as required by the ANSI specification),
1298 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1299 expected behaviour on conditions. Users should note, however,
1300 that such behaviour is not required by the ANSI specification,
1301 and so use of this behaviour may render their code unportable.
1302 * fixed some bugs revealed by Paul Dietz' test suite:
1303 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1305 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1306 ** SXHASH on condition objects no longer returns NIL.
1307 ** :ALLOCATION :CLASS slots are better treated; their values are
1308 updated on class redefinition, and initforms inherited from
1309 superclasses are applied.
1310 ** REMOVE-METHOD returns its generic function argument even when
1311 no method was removed.
1312 ** SHARED-INITIALIZE now initializes the values of the requested
1313 slots, including those with :ALLOCATION :CLASS.
1314 ** ALLOCATE-INSTANCE now works on structure classes defined via
1315 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1317 ** SLOT-EXISTS-P now works on conditions, as well as structures
1319 ** MAKE-LOAD-FORM now has the required methods on
1320 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1321 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1322 keyword, and now implements the SLOT-NAMES argument.
1323 ** methods with &OPTIONAL arguments no longer allow too many
1324 arguments to be passed in the call without error.
1325 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1326 option is consistent with the required arguments of the generic
1327 function lambda list.
1328 * bug fix: REQUIRE accepts a string designator. (Thanks to
1330 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1331 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1332 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1333 * bug fix: The compiler now checks for duplicated variables in macro
1335 * bug fix: SETQ on globals returns the correct value.
1336 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1337 not a COMPILER-ERROR (followed by some other strange error on
1338 choosing the CONTINUE restart).
1339 * bug fix: make.sh and friends are now more consistent in the way that
1340 they look for GNU "make".
1342 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1343 * experimental native threads support (on x86 Linux >=2.4 only).
1344 This is not compiled in by default: you need to add :SB-THREAD to
1345 the target features. See the "Beyond ANSI" chapter of the manual
1347 * fix for longstanding nonANSIism: The old distinction between
1348 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1349 The return value from CL:FIND-CLASS is now a CLOS class, and
1350 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1351 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1352 * An interface to the MetaObject Protocol, as described in Kiczales,
1353 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1354 MIT Press, 1991, is available from the SB-MOP package.
1355 * incompatible change: the SB-PCL package should now be considered
1356 a private implementation detail, and no longer a semi-private MOP
1358 * minor incompatible change: due to rearrangement for threads, the
1359 control stack and binding stack are now allocated at arbitrary
1360 addresses instead of being hardcoded per-port. Users affected by
1361 this probably have to be doing advanced things with shared
1362 libraries, and will know who they are.
1363 * minor incompatible change: Previously, all --eval forms used were
1364 processed with READ before any of them were processed with EVAL.
1365 Now each --eval form is processed with both READ and EVAL before
1366 the next --eval form is processed. (Thus package operations like
1367 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1368 work as the user might reasonably expect.)
1369 * minor incompatible change: *STANDARD-INPUT* is now only an
1370 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1372 * minor incompatible change: Y-OR-N-P is now character-oriented,
1373 not line oriented. Also, YES-OR-NO-P now works without errors.
1374 (thanks to Antonio Martinez)
1375 * sb-aclrepl module improvements: an integrated inspector, added
1376 repl features, and a bug fix to :trace command.
1377 * Known functions, which cannot be open coded by the backend, are
1378 considered to be able to check types of their arguments. (fixing
1379 a bug report by Nathan J. Froyd)
1380 * fixed a bug in computing method discriminating functions: It is
1381 now possible to define methods specialized on classes which have
1382 forward-referenced superclasses. (thanks to Gerd Moellmann)
1383 * fixed evaluation order in optional entries (reported by Gilbert
1385 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1386 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1387 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1388 required-by-AMOP NAME argument, as well as CLASS and
1389 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1390 * fixed bug 20: DEFMETHOD can define methods using names that are
1391 not the proper names of classes to designate class specializers.
1392 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1393 Unix file descriptors, instead of blowing up. (thanks to Antonio
1395 * Garbage collection refactoring: user-visible change is that a
1396 call to the GC function during WITHOUT-GCING will not do garbage
1397 collection until the end of the WITHOUT-GCING. If you were doing
1398 this you were probably losing anyway.
1399 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1400 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1401 TYPEP the latter but not the former.
1402 * The compiler issues a full WARNING for calls to undefined functions
1403 with names from the CL package.
1404 * MAP-INTO for a vector destination is open coded. (reported by
1405 Brian Downing on c.l.l)
1406 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1407 documentation string.
1408 * fixed some bugs revealed by Paul Dietz' test suite:
1409 ** COPY-ALIST now signals an error if its argument is a dotted
1411 ** Condition slots are now accessed more correctly in the presence
1412 of multiple initargs for a given slot.
1413 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1414 exclude restarts of the same name associated with a different
1416 ** DEFCLASS of forward-referenced classes with another
1417 forward-referenced class in the superclasses list no longer
1419 ** Condition slots are now initialized once each, not multiple
1420 times. (thanks to Gerd Moellmann)
1421 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1423 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1424 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1425 arguments contain duplicated elements.
1426 ** RESTART-CASE understands local macros.
1427 ** RESTART-CASE associates exactly its own restarts with a condition.
1428 ** ENDP in safe mode checks its argument to be of type LIST.
1429 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1430 in question is unbound.
1431 ** Optimized MAKE-INSTANCE functions no longer cause internal
1432 assertion failures in the presence of duplicate initargs.
1433 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1434 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1435 propagated even in the ctor optimized implementation of
1437 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1439 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1440 * a better implementation of SXHASH on (simple) bit vectors,
1441 measured both in execution speed and in distribution of results
1442 over the positive fixnums, has been installed. Likewise, a better
1443 implementation of EQUAL for simple bit vectors is now available.
1444 * fixed CEILING optimization for a divisor of form 2^k.
1445 * fixed bug 240 (emitting extra style warnings "using the lexical
1446 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1447 by Antonio Martinez)
1448 * fixed SXHASH, giving different results for NIL depending on type
1449 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1450 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1451 variable. (found by Rolf Wester)
1452 * SBCL does not ignore type declarations for special
1453 variables. (reported by rif on c.l.l 2003-03-05)
1454 * some bug fixes in contrib/sb-aclrepl/
1455 * fixed some bugs revealed by Paul Dietz' test suite:
1456 ** a bug in the CONS type specifier, whereby the CAR and CDR
1457 types got intertwined, has been fixed;
1458 ** the type system is now able to reason about the interaction
1459 between INTEGER and RATIO types more completely;
1460 ** APPEND, [N]REVERSE and NRECONC check that those their
1461 arguments, which must be proper lists, are really so;
1462 ** An array specialized to be unable to hold elements has been
1463 implemented, as required -- yes, really -- by ANSI;
1464 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1465 on malformed property lists;
1467 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1468 * incompatible packaging change: in line with Unix convention,
1469 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1470 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1471 * REQUIRE and PROVIDE are now optionally capable of doing something
1472 useful. See the documentation string for REQUIRE.
1473 * infrastructure for a managed SBCL contrib system: contributed
1474 modules in this release include:
1475 ** the ASDF system definition facility;
1476 ** an interface to the BSD Sockets API;
1477 ** an ACL-like convenience interface to the repl;
1478 (thanks to Kevin Rosenberg)
1479 ** an implementation of ROTATE-BYTE, with efficient implementation
1481 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1482 gives a closer approximation to the right answer than previously.
1483 (thanks to Raymond Toy)
1484 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1485 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1486 optional environment arguments, as required by ANSI.
1487 * fixed bugs in other functions taking environment objects, allowing
1488 calls with an explicit NIL environment argument to be compiled
1490 * fixed bug 228: primary return values from
1491 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1492 COMPILE or FUNCTION.
1493 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1494 :TYPE will no longer signal errors on innocuous objects.
1495 * fixed bug 231b: SETQ is better at respecting type declarations in
1496 the lexical environment.
1497 * fixed a bug in DEFCLASS: classes named by symbols with no or
1498 unprintable packages can now be defined.
1499 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1500 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1501 * fixed compiler failure related to checking types of functions.
1502 (reported by Robert E. Brown)
1503 * the compiler is now much more consistent in its error-checking
1504 treatment of bounding index arguments to sequence functions: in
1505 (SAFETY 3) code, errors will be signalled in almost all cases if
1506 invalid sequence bounding indices are passed to functions defined
1507 by ANSI to operate on sequences.
1508 * fixed a bug in the build procedure: documentation of SBCL-specific
1509 packages is now preserved and available in the final Lisp image.
1510 * lifted FDEFINITION lookup out of loops in the implementation of
1511 many list operations. (thanks to Robert E. Brown)
1512 * fixed a bug in the reader: the #n# reader macro now works for
1513 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1514 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1515 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1516 E. Brown and Tony Martinez respectively)
1517 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1518 SLOT-BOUNDP: in particular, it is now able to optimize them much
1519 better, and is now not vulnerable to having packages renamed.
1520 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1521 always called when a slot is not present in an instance. (thanks
1523 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1524 longer signal an error on unreading a character following EOF on
1525 the previous constituent stream. (thanks to Tony Martinez)
1526 * fixed some bugs revealed by Paul Dietz' test suite:
1527 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1528 not just nonnegative fixnums;
1529 ** the logical bit-array operators such as BIT-AND now accept an
1530 explicit NIL for their "opt-arg" argument (to indicate a
1531 freshly-consed result bit-array);
1532 ** ELT now signals an error on an invalid sequence index in safe
1534 ** the type system is now cleverer about negations of numeric
1535 types, and consequently understands the BIGNUM and RATIO types
1537 ** the type system is now cleverer about the interaction between
1538 INTEGER and RATIO types: while bugs still remain, many more
1539 cases are accurately computed;
1540 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1541 if it is in the last clause;
1542 ** CONSTANTLY now correctly returns a side-effect-free function in
1544 ** DECLARE is no longer treated as a special-operator; in
1545 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1546 * incremented fasl file version number due to the change in the
1547 DEFSTRUCT-SLOT-DESCRIPTION structure.
1549 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1550 * minor incompatible change: code processed by the "interpreter" or
1551 EVAL now has a compilation optimization policy of (DEBUG 2)
1552 (changed from (DEBUG 1)) to improve debuggability of interactive
1553 development, and to allow the use of the debug RETURN command in
1555 * an experimental implementation of the RETURN command for the
1556 debugger has been included. (thanks to Frederik Kuivinen)
1557 * fixed bug 62: constraints were not propagated into a loop.
1558 * fixed bug in embedded calls of SORT (reported and investigated by
1560 * fixed some bugs revealed by Paul Dietz' test suite:
1561 ** printing and reading of arrays with some dimensions having
1562 length 0 (thanks to Gerd Moellmann);
1563 ** BOA constructor with &AUX argument without a default value does
1564 not cause a type error;
1565 ** CONSTANTP now returns true for all self-evaluating objects.
1567 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1568 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1569 accessors that are related by inheritance, as specified in the
1570 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1571 Valtteri Vuorikoski)
1572 * The compiler is now able to inline functions that were defined in
1573 a complex lexical environment (e.g. inside a MACROLET).
1574 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1576 * Support for the upcoming FreeBSD-5.0 release has been included.
1577 (thanks to Dag-Erling Smorgrav)
1578 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1579 effect when it is not in a toplevel context.
1580 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1581 stand a better chance of being correct. (thanks to Gerd
1583 * fixed bug in COERCE, which now signals an error on coercing a
1584 rational to a bounded real type which excludes the expected
1586 * The compiler is now able to derive types more accurately from the
1587 COERCE and COMPILE functions.
1588 * fixed bug 223: functional binding is considered to be constant
1589 only for symbols in the CL package.
1590 * fixed bug 231: SETQ did not check the type of a variable being set
1591 (reported by Robert E. Brown)
1592 * A new optimization for MAKE-INSTANCE has been included, fixing
1593 various bugs (including relating to :ALLOCATION :CLASS slots and
1594 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1596 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1597 ** As required by ANSI, LOOP now disallows anonymous collection
1598 clauses such as COLLECT I in conjunction with aggregate boolean
1599 clauses such as THEREIS (= I 1);
1600 ** LOOP now signals an error when any variable is reused in the
1601 same loop (including the potentially useful construct analogous
1602 to WITH A = 1 WITH A = (1+ A);
1603 ** IT is only a special loop symbol within the first clause of a
1604 conditional loop clause;
1605 ** LOOP with a typed iteration variable over a hashtable now
1606 signals a type error iff it should.
1607 * fixed some other bugs revealed by Paul Dietz' test suite:
1608 ** FILE-STREAM now names the class previously known as FD-STREAM;
1609 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1610 argument) no longer signals an error;
1611 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1612 of the slot symbol, rather than using the current package
1613 ((:CONC-NAME "") continues to intern the slot's name in the
1615 * incremented fasl file version number, because of the incompatible
1616 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1617 of the new implementation of DEFINE-COMPILER-MACRO.
1619 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1620 * Support for building SBCL for MIPS platforms running in
1621 little-endian mode has now been checked in, and basic
1622 functionality on said platforms verified.
1623 * minor incompatible change: PCL now records the pathname of a file
1624 in which methods and the like are defined, rather than its
1626 * minor incompatible change: TRUENAME now considers the truename of
1627 a file naming a directory to be the pathname with :DIRECTORY
1628 component indicating that directory.
1629 * minor incompatible change: a NAMED clause in the extended form of
1630 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1631 reason for the previous behaviour is unclear.
1632 * more systematization and improvement of CLOS and MOP conformance
1633 in PCL (thanks to Gerd Moellman and Pierre Mai):
1634 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1636 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1637 primary methods with no specializers;
1638 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1640 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1641 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1642 FORWARD-REFERENCED-CLASSes; error reporting on
1643 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1645 ** SXHASH on CLOS instances now uses a slot internal to the
1646 instance to return different numbers on distinct instances,
1647 while preserving the same return value through invocations of
1649 ** DEFMETHOD signals errors when methods with longer incongruent
1650 lambda lists are added to generic functions;
1651 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1652 CLASS, as specified in AMOP;
1653 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1654 on the order returned by the primary method for classes of
1655 class STANDARD-CLASS;
1656 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1657 * fixed some bugs shown by Paul Dietz' test suite:
1658 ** DOLIST puts its body in TAGBODY;
1659 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1661 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1662 value producing form;
1663 ** if more variables are given to PROGV than values, extra
1664 variables are bound and made to have no value;
1665 ** NSUBSTITUTE on list arguments gets the right answer with
1667 ** ELT signals an error of type TYPE-ERROR when the index argument
1668 is not a valid sequence index;
1669 ** LOOP signals (at macroexpansion time) an error of type
1670 PROGRAM-ERROR when duplicate variable names are found;
1671 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1672 ** FUNCALL of special-operators now cause an error of type
1674 ** PSETQ now works as required in the presence of side-effecting
1675 symbol-macro places;
1676 ** NCONC accepts any object as its last argument;
1677 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1679 ** loop-for-as-package does not require a package to be explicitely
1681 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1683 * fixed bug 166: compiler preserves "there is a way to go"
1684 invariant when deleting code.
1685 * fixed bug 172: macro lambda lists with required arguments after
1686 &REST arguments now cause an error to be signalled. (thanks to
1688 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1689 bug. (thanks to Matthew Danish)
1690 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1692 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1693 arguments are lexically rebound. (thanks to Gerd Moellmann and
1695 * fixed bug 194: error messages are now more informative when there
1696 is no primary method applicable in a call to a generic
1697 function. (thanks to Gerd Moellmann)
1698 * fixed bug in command line argument checking (thanks to Julian
1700 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1701 (thanks to Matthew Danish)
1702 * incremented fasl file version number, because of the
1703 SXHASH-related changes in the layout of CLOS data structures
1705 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1706 * minor incompatible change: The runtime (the Unix executable named
1707 "sbcl") is now much pickier about the .core files it will load.
1708 Essentially it now requires .core files to descend from the same
1709 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1710 as the runtime does. (The intent is to prevent the crashes which
1711 can occur, and which can even be reported as mysterious failures,
1712 when people patch the sources or change the build parameters
1713 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1714 sbcl and .core files.)
1715 * fixed bug: VALUES-LIST is no longer optimized away.
1716 * fixed bug 142: The FFI conversion of C string values to Lisp
1717 string values no longer conses excessively. (thanks to Nathan
1718 Froyd porting Raymond Toy's fix to CMU CL)
1719 * began to systematize and improve MOP conformance in PCL (thanks to
1720 Nathan Froyd, Gerd Moellman and Pierre Mai):
1721 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1723 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1724 ** FINALIZE-INHERITANCE is now called on class finalization;
1725 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1726 argument precedence order.
1727 * fixed bug 202: The compiler no longer fails on functions whose
1728 derived types contradict their declared type.
1729 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1730 so it can be non-toplevel.
1731 * The fasl file version number has changed (because of the new
1732 implementation of DEFMACRO).
1733 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1734 safe code, that any length requirement by their type-specifier
1735 argument is valid. The exceptions to this are described in bug
1737 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1738 functions now check the types of their inputs as required by ANSI.
1739 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1740 introduced symbol is DECLAREd to be SPECIAL.
1741 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1742 * fixed bug: :COUNT argument to sequence functions may be negative
1743 * fixed bug: body of DO-SYMBOLS may contain declarations
1744 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1745 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1746 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1747 symbol macro only once
1748 * fixed printing of call frame when argument list is unavailable
1749 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1750 * compiler no longer signals WARNING on unknown keyword
1753 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1754 * A beta-quality port to the mips architecture running Linux,
1755 based on the old CMUCL backend, has been made. It has been tested
1756 on a big-endian kernel, and works sufficiently well to be able to
1757 rebuild itself; it has not been tested in little-endian mode.
1758 * fixed an inconsistency between gencgc.c and purify.c which made
1759 dumping/loading .core files unreliable
1760 * fixed bug 120a: The compiler now deals correctly with IFs where
1761 the consequent is the same as the alternative, instead of
1762 misderiving the return type. (thanks to Alexey Dejneka)
1763 * fixed bug 113: Logical pathnames are now dumpable (the logical
1764 host is resolved at load-time, throwing an error if it is not
1766 * fixed bug 174: FORMAT's error message is slightly clearer when a
1767 non-printing character is used in a format directive.
1768 * fixed several bugs in compiler checking of type declarations, i.e.
1769 violations of the Python "declarations are assertions" principle
1770 (thanks to Alexey Dejneka)
1771 * fixed several bugs in PCL's error checking (thanks to Gerd
1773 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1775 * fixed bug in compilation of functions as first class values
1776 (thanks to Antonio Martinez-Shotton)
1777 * The compiler's handling TYPE-ERRORs which it can prove will
1778 inevitably happen at runtime has been cleaned up and corrected
1779 in several ways. (thanks to Alexey Dejneka)
1780 * improved argument type checking for various basic arithmetic
1781 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1782 lavished on them in the past that they can be compiled in many
1783 ways in different special cases
1784 * fixed bug 181: compiler checks validity of user supplied type
1786 * cleaned up code flushing in optimization: Function calls which
1787 should signal errors for safety purposes (e.g. which ANSI says
1788 should signal errors when their arguments are of incorrect type)
1789 are no longer optimized away.
1790 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1791 * incremented fasl file version number, because changes in the
1792 implementation of sequence functions like COERCE caused
1793 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1794 in old inline expansions) to become undefined. (Actually these
1795 changes were later undone, so we might very well be binary
1796 compatible with 0.7.7 after all, but leaving the version number
1797 incremented seemed like the simplest and most conservative
1800 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1801 * An alpha-quality port to the parisc architecture running Linux,
1802 based on the old CMUCL backend, has been made. This, even more so
1803 than the other backends, should be considered still a work in
1804 progress; known problems include that the Linux kernel in 64-bit
1805 mode does not propagate the correct sigcontext structure to
1806 userspace, and consequently SBCL on a parisc64 kernel will not
1808 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1809 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1810 suppressed.) Also now that the compiler is looking at declarations
1811 in the environment, it checks optimization declarations as well,
1812 and suppresses inlining when (> DEBUG SPEED).
1813 * More fixes have been made to treatment of floating point exception
1814 treatment and other Unix signals. In particular, floating point
1815 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1816 * The detection and handling of control stack exhaustion (infinite
1817 or very deeply nested recursion) has changed. Stack exhaustion
1818 detection is now done by write-protecting pages at the OS level
1819 and applies at all optimization settings; when found, a
1820 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1821 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1822 be caught using IGNORE-ERRORS.
1823 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1824 correctly when transforming recursive LABELS functions to LETs.
1825 (thanks to Alexey Dejneka)
1826 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1827 that are names of constants or global variables.
1828 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1829 alien routines with docstrings.
1830 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1831 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1833 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1834 (thanks to Eric Marsden)
1835 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1836 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1837 * Bug fix: The SPARC backend can now compile functions involving
1838 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1839 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1840 to a functions expecting a PATHNAME-DESIGNATOR.
1841 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1842 lambda lists. (thanks to Alexey Dejneka)
1843 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1844 functions so that EXPORTing the name of the class doesn't cause
1845 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1846 to Antonio Martinez for reporting this)
1847 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1848 constant-folded without failing an assertion. (thanks to Einar
1849 Floystad Dorum for reporting this)
1850 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1851 types are now dealt with more correctly by the compiler.
1852 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1853 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1854 behaviour analogous to RENAME-FILE. This puts its behaviour more
1855 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1857 * The fasl file version number has changed again. (because of the
1858 bug fix involving the names of PCL MAKE-INSTANCE functions)
1860 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1861 * bug fix: Floating point exceptions are treated much more
1862 consistently on the x86/Linux and PPC/Linux platforms.
1863 * Array initialization with :INITIAL-ELEMENT is now much faster for
1864 cases when the compiler cannot open code the array creation, but
1865 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1866 array accesses have also seen a speed increase.
1867 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1868 specified. (This was at the root of some bad interactions between
1869 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1871 * bug fix: Internal error arguments for undefined functions are now
1872 computed correctly on the PPC/Linux platform.
1873 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1874 Raymond Toy's patch for CMU CL)
1875 * Support for the Solaris 9 operating environment has been included
1876 (thanks to Daniel Merritt)
1877 * A very ugly but hopefully complete draft of the missing FFI chapter
1878 of the manual has been created by reformatting the corresponding
1879 CMU CL manual chapter into (currently very ugly and incoherent)
1880 DocBook and bringing it up to date for SBCL behavior. Thus, the
1881 manual is now essentially complete, at least by my extreme
1882 once-and-only-once standards, whereby it's acceptable to refer to
1883 the doc strings of SB-EXT functions as the primary documentation.
1884 * The fasl file version number has changed again, due to cleanup of
1885 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1886 is no longer a static symbol.)
1888 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1889 * SBCL now builds with OpenMCL (version 0.12) as the
1890 cross-compilation host; also, more progress has been made toward
1891 bootstrapping under CLISP.
1892 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1894 * bug 158 fixed: The compiler can now deal with integer loop
1895 increments different from 1; fixing this turned out also to fix
1897 * bug 169 fixed: no more bogus warnings about using lexical bindings
1898 despite the presence of perfectly good SPECIAL declarations (thanks
1899 to David Lichteblau)
1900 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1901 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1902 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1904 * bug 180 fixed: Method combination specifications no longer ignore
1905 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1906 * bug fix: Structure type predicate functions now check their argument
1907 count as they should.
1908 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1909 correctly. (thanks to Pierre Mai)
1910 * minor incompatible change: The --noprogrammer option is deprecated
1911 in favor of the new --disable-debugger option, which is very similar.
1912 (The major difference is that it takes effect at a slightly different
1913 time at startup, causing handling of errors in --sysinit and
1914 --userinit files will be affected differently.) The
1915 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1916 been added to allow this functionality to be controlled from ordinary
1917 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1918 maintainers, who might want to run non-interactive scripts to
1919 build SBCL cores which will later be used interactively.)
1920 * minor incompatible change: The LOAD function no longer, when given
1921 a wild pathname to load, loads all files matching that pathname.
1922 Instead, an error of type FILE-ERROR is signalled.
1924 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1925 * bug 147 fixed: The compiler preserves its block link/count
1926 invariants more correctly now so that it doesn't crash. (thanks
1928 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1930 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1931 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1932 no-value-for-this-character as NIL (as specified by ANSI).
1933 * HOST-NAMESTRING on physical pathnames now returns a string that is
1934 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1935 (thanks to Christophe Rhodes)
1936 * The Alpha port handles icache flushing more correctly. (thanks to
1938 * More progress has been made toward bootstrapping under CLISP. (thanks
1939 to Christophe Rhodes)
1940 * The fasl file format has changed again, because dynamic loading
1941 on OpenBSD (which has non-ELF object files) motivated some cleanups
1942 in the way that foreign symbols are transformed and passed around.
1943 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1944 is no longer treated as whitespace by the reader, but instead as
1945 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1946 |A
\7fB|, instead of A as it used to.
1948 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1949 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1950 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1951 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1952 this as of 0.7.1.45, but was left out of the previous news section
1953 (thanks to Dan Barlow)
1954 * SBCL now runs on the Solaris operating system on SPARC architectures
1955 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1956 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1957 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1958 * SPARC backend cleanups, allowing builds of cores optimized for V8
1959 and V9 SPARCS, and also emission of code targeted to a particular
1960 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1962 * SBCL is closer to bootstrapping under CLISP, thanks to various
1963 fixes by Christophe Rhodes.
1964 * The fasl file format has changed again, to allow the compiler's
1965 INFO database to support symbol macros.
1966 * The user manual (in doc/) is formatted into HTML more nicely.
1967 (thanks to coreythomas)
1968 * The system is smarter about SUBTYPEP relationships, especially
1969 those involving NOT types (including types such as ATOM which are
1970 represented internally using NOT types). Thus SUBTYPEP is less
1971 likely to return (VALUES NIL NIL) in general, and in particular
1972 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1974 * The fasl file format has changed again, because the internal
1975 representation of types now includes a new slot to support the new
1976 SUBTYPEP-of-NOT-types logic.
1977 * (not a change in the main branch of SBCL, but a related prototype
1978 which can hopefully be merged into the main branch of SBCL in the
1979 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1980 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1981 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1982 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1983 dispatch macro character combinations. (thanks to Alexey Dejneka)
1984 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1985 that superficially look like logical namestrings correctly.
1986 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1988 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1989 default supplied-p) for &optional and &key arguments. (thanks to
1992 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1993 * incompatible change: The compiler is now less aggressive about
1994 tail call optimization, doing it only when (> SPACE DEBUG) or
1995 (> SPEED DEBUG). (This is an incompatible change because there are
1996 programs which relied on the old CMU-CL-style behavior to optimize
1997 away their unbounded recursion which will now die of stack overflow.)
1998 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1999 for non-GENCGC systems has been increased to 20M (since that
2000 seems much closer to the likely performance optimum for modern
2001 systems than the old 4M value was)
2002 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2003 and *DEBUG-PRINT-LEVEL*
2004 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2005 of CMU CL's support for SPARC, and various endianness and other
2006 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2007 * new syntactic sugar for the Unix command line: --load foo.bar is now
2008 an alternate notation for --eval '(load "foo.bar")'.
2010 ** The system now detects stack overflow and handles it gracefully,
2011 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2012 optimization settings. (This is a good thing in general, and
2013 its introduction in this version should be particularly timely
2014 for anyone whose code fails because of suppression of tail
2016 ** The system now hunts for the C variable "environ" in a more
2017 devious way, to avoid segfaults when the C library version differs
2018 between compile time and run time. (thanks to Christophe Rhodes)
2019 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2020 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2021 ** The compiler no longer issues bogus style warnings for undefined
2022 classes in the same source file as the DEFCLASSes which defined
2023 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2025 ** fixes in CONDITION class precedence list for undefined function
2026 errors (thanks to Alexei Dejneka)
2027 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2028 correctly. (thanks to Dan Barlow)
2029 ** portability fixes aiming at bootstrapping under CLISP (thanks
2030 to Dave McDonald and Christophe Rhodes)
2031 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2032 ** fixes in type translation and and type inference (thanks to
2034 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2035 ** various fixes in the new ports (thanks to Dan Barlow)
2036 * several changes related to debugging:
2037 ** suppression of tail recursion, as noted above
2038 ** stack overflow detection, as noted above
2039 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2040 is now the default. (For some time encapsulation has been more
2041 reliable than the breakpoint-based :ENCAPSULATE NIL
2042 implementation, at least on X86 systems; and I just noticed that
2043 encapsulation also seems closer to the spirit of the ANSI
2046 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2048 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2049 up properly again. (There was a packaging bug in 0.7.0 which
2050 left their definitions in SB-SYS::LOAD-FOREIGN and
2051 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2052 vital for most things which interface to C-level interfaces,
2053 like extensions working with sockets or databases or
2054 Perl-compatible regexes or whatever, and the need to fix
2055 this bug is the main reason that 0.7.1 was released so
2057 ** DEFGENERIC is now choosier about the methods it redefines, so that
2058 reLOADing a previously-LOADed file containing DEFGENERICs does
2059 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2060 cycle now works as it should. (thanks to Alexey Dejneka)
2061 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2062 fixed by Christophe Rhodes. (He actually submitted this patch
2063 months ago, and I delayed until after 0.7.0.)
2064 ** Bug 111 (internal compiler confusion about runtime checks on
2065 FUNCTION types) was fixed by Alexey Dejneka.
2066 * Some internal cleanups (getting rid of variables which aren't
2067 needed now that the byte interpreter is gone) caused the fasl
2068 file format number to change again.
2070 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2071 * major incompatible change: The default fasl file extension, i.e. the
2072 default extension for files produced by COMPILE-FILE, has changed
2073 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2075 ** There are many changes in the implementation of the compiler.
2076 SBCL is now essentially a compiler-only implementation of ANSI
2077 Common Lisp. EVAL still "interprets" a few special cases, but
2078 almost all the interesting cases are handled by creating
2079 a LAMBDA expression, calling COMPILE on it, then calling
2080 FUNCALL on the result.
2081 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2082 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2083 Since the code is newer, there might still be some new bugs
2084 (though not as many as before Martin Atzmueller's fixes:-). But
2085 the new code is substantially simpler and clearer, and hopefully
2086 any remaining bugs will be simpler, less fundamental, and more
2087 fixable then the bugs in the old code.
2088 ** The revised compiler is still a little unsteady on its feet.
2090 *** The debugging information it produces (particularly the names
2091 of FUNCTION objects) is sometimes much less useful than what
2092 the old compiler produced.
2093 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2094 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2095 within a MACROLET) has been temporarily weakened.
2096 ** There are new compiler optimizations for various functions:
2097 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2098 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2099 *** the math functions TRUNCATE, FLOOR, and CEILING
2100 *** the function-of-all-trades COERCE
2101 Mostly these should be transparent, but there's one
2102 potentially-annoying problem (bug 117): when the compiler
2103 inline-expands a function and does type analysis on the result,
2104 it can create control paths which have type mismatches, and
2105 when it can't prove that those control paths aren't taken,
2106 it will issue WARNINGs about the type mismatches. This is
2107 a particular problem in practice for the new sequence functions.
2108 It's not clear how this should be fixed, and for now, a
2109 workaround is given in the entry for 117 in the BUGS file.
2110 ** (Because of the interaction between the two previous items --
2111 occasional inlining problems and new inline expansions -- some
2112 of the new sequence function optimizations won't really kick in
2113 completely until debugging information, and then inlining, are
2114 straightened out in some future version.)
2115 * minor incompatible changes:
2116 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2117 DIRECTORY no longer implicitly promotes NIL slots of its
2118 pathname argument to :WILD. In particular, when you ask for the
2119 contents of a directory (which you used to be able to do without
2120 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2121 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2122 ** changes in behavior that ANSI explicitly defines to be
2123 implementation dependent:
2124 *** The new compiler-only implementation still conforms with ANSI,
2125 but acts a little different than before. Besides the obvious
2126 changes in performance tradeoffs (that the cost per form passed
2127 to EVAL has gone up, and the cost per form executed by EVAL
2128 has gone down), the behavior of the system changes a little
2129 because there are no longer any interpreted function objects.
2130 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2131 e.g. doing COMPILE on the output of interactive DEFUN is
2133 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2135 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2136 has changed from (:CL) to NIL.
2137 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2138 e.g. #\Newline and #\Tab, don't have names specified in the
2139 ANSI Common Lisp standard, is now based on their ASCII symbolic
2140 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2141 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2142 are no longer used for output.
2143 ** changes in internal implementation constants:
2144 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2145 4 million. (If your application spends a lot of time GCing and
2146 you have a lot of RAM, you might want to experiment with
2147 increasing it even more.)
2148 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2149 However, almost all old code should still continue to work without
2150 immediate update, as SB-C-CALL is now a (deprecated) nickname
2152 ** Old operator names in the style DEF-FOO are now deprecated in
2153 favor of new corresponding names DEFINE-FOO, for consistency with
2154 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2155 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2156 internal symbols, but a few supported extensions like
2157 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2158 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2159 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2160 etc. as you get deeper into recursive calls to the debugger
2161 command loop, instead of the old "5]", "5]]", "5]]]"
2162 sequence. (I was motivated to do this when squabbles between
2163 ILISP and SBCL left me very deeply nested in the debugger. In the
2164 short term, this change will probably provoke more ILISP/SBCL
2165 squabbles, but hopefully it will be an improvement in the long run.)
2166 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2167 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2168 function, since declaiming return types involving aliens is
2169 (1) annoyingly messy to do by hand and (2) vital to efficient
2170 compilation of code which calls such functions.
2171 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2172 longer reexported by the SB-EXT package. They're solely useful
2173 for alien code, so it seems more logical that you should get
2174 them from the SB-ALIEN package, not in SB-EXT.
2175 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2176 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2177 features. Instead, the code that they used to control is always
2178 built into the system.
2179 * many other bug fixes
2180 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2181 advantage of the new EVAL-WHEN stuff and to clean them up in
2182 general, and they are now more ANSI-compliant in a number of
2183 ways. Martin Atzmueller is responsible for a lot of this.
2184 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2186 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2187 *** correct ERROR type for various file operations
2188 *** some fixes for Lisp streams
2189 *** DEFMETHOD syntax checking
2190 *** changing old weird representation of debug information as
2191 strings (which, among their other deficiencies, don't transform
2192 correctly when you rename packages, and don't change their
2193 print representation when you change things like *PACKAGE*
2194 and *PRINT-LENGTH*) to symbols and lists of symbols
2195 He also made several improvements and fixed several bugs in DESCRIBE.
2196 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2198 *** misbehavior of WRITE-STRING/WRITE-LINE
2199 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2200 and several other LOOP problems as well
2201 *** DIRECTORY when similar filenames are present
2202 *** DEFGENERIC with :METHOD options
2203 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2204 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2205 *** argument ordering in FIND with :TEST option
2206 *** mishandled package designator argument in APROPOS-LIST
2207 *** various problems in the backquote readmacro
2208 *** a bug in APROPOS
2209 *** probably some others that I'm not describing very well here,
2210 since the CVS log documents them by reference to sbcl-devel
2211 messages, and the SourceForge archives aren't working well.:-(
2212 ** Dan Barlow improved the Alpha port (and is making progress on the
2213 PPC port, for those of you who think different).
2214 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2215 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2216 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2217 continues to work on the SPARC port (for those of you in a position
2218 to look down upon our little PC-compatible boxes from a great height).
2219 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2220 and uses that copy, rather than the current dynamic value, when
2221 it's trying to decide whether to truncate output. Thus e.g.
2222 (let ((*print-lines* 50))
2223 (pprint-logical-block (stream nil)
2225 (let ((*print-lines* 8))
2226 (print (aref possiblybigthings i) stream)))))
2227 should now truncate the logical block only at 50 lines, instead of
2228 often truncating it at 8 lines, as it did before.
2229 * The doc/cmucl/ directory, containing old CMU CL documentation
2230 from the time of the fork, is no longer part of the base system.
2231 SourceForge has shut down its anonymous FTP service, and with it
2232 my original plan for distributing the old CMU CL documentation
2233 there. For now, if you need these files you can download an old
2234 SBCL source release and extract them from it.
2235 * The fasl file version number changed again, for dozens of reasons,
2236 some of which are apparent above.
2238 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2239 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2240 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2241 patch, so that DISASSEMBLE output is much nicer.
2242 * The code in the SB-PROFILE package now seems reasonably stable.
2243 I still haven't decided what the final interface should look like
2244 (I'd like PROFILE to interact cleanly with TRACE, since both
2245 facilities use function encapsulation) but if you have a need
2246 for profiling now, you can probably use it successfully with
2247 the current CMU-CL-style interface.
2248 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2249 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2250 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2251 link now returns the dangling link itself, and for similar
2252 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2253 link itself. (In these cases the old code signalled an error and
2254 looped endlessly, respectively.) Thus, DIRECTORY now works even
2255 in the presence of dangling and cyclic symbolic links.
2256 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2257 is now a supported extension again, since the consensus on
2258 sbcl-devel was that it can be useful for ordinary development
2259 work, not just for debugging SBCL itself.
2260 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2261 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2262 that currently-defined functions might be redefined later with
2263 different return types.
2264 * Hash tables can be printed readably, as inspired by CMU CL code
2265 of Eric Marsden and SBCL code of Martin Atzmueller.
2266 * better error handling in CLOS method combination, thanks to
2267 Martin Atzmueller porting Pierre Mai's CMU CL patches
2268 * more overflow fixes for >16Mbyte I/O buffers
2269 * A bug in READ has been fixed, so that now a single Ctrl-D
2270 character suffices to cause end-of-file on character streams.
2271 In particular, now you only need one Ctrl-D at the command
2272 line (not two) to exit SBCL.
2273 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2275 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2276 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2277 does the right thing.
2278 * The implementation of some type tests, especially for CONDITION
2279 types, is now tidier and maybe faster, due to CMU CL code
2280 originally by Douglas Crosher, ported by Martin Atzmueller.
2281 * Some math functions have been fixed, and there are new
2282 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2283 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2284 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2285 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2286 likely to be built into sbcl-0.7.0.)
2287 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2288 CL, ported to SBCL by Martin Atzmueller)
2289 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2290 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2291 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2292 * There's a new slam.sh hack to shorten the edit/compile/debug
2293 cycle for low-level changes to SBCL itself, and a new
2294 :SB-AFTER-XC-CORE target feature to control the generation of
2295 the after-xc.core file needed by slam.sh.
2296 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2297 COMPILE-FILE is no longer supported, so that now every function
2298 gets an entry point, so that block compilation looks a little
2299 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2300 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2301 returns the number of bytes consed since the system started,
2302 rather than the number consed since the first time the function
2303 was called. (The new definition parallels ANSI functions like
2304 CL:GET-INTERNAL-RUN-TIME.)
2305 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2306 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2307 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2308 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2309 * Fasl file version numbers are now independent of the target CPU,
2310 since historically most system changes which required version
2311 number changes have affected all CPUs equally. Similarly,
2312 the byte fasl file version is now equal to the ordinary
2315 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2316 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2317 is no longer recognized. I apologize for this, because it was
2318 listed in SB-EXT as a supported extension, but I found that
2319 its existing behavior was poorly specified, as well as incorrectly
2320 specified, and it looked like too much of a mess to straighten it
2321 out. I have enough on my hands trying to get ANSI stuff to work..
2322 * many patches ported from CMU CL by Martin Atzmueller, with
2323 half a dozen bug fixes in pretty-printing and the debugger, and
2324 half a dozen others elsewhere
2325 * fixed bug 13: Floating point infinities are now supported again.
2326 They might still be a little bit flaky, but thanks to bug reports
2327 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2328 as flaky as they were.
2329 * The --noprogrammer command line option is now supported. (Its
2330 behavior is slightly different in detail from what the old man
2331 page claimed it would do, but it's still appropriate under the
2332 same circumstances that the man page talks about.)
2333 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2334 are now supported, and enabled by default. Thus, the compiler can
2335 handle many floating point and complex operations much less
2336 inefficiently. (Thus e.g. you can implement a complex FFT
2338 * The compiler now detects type mismatches between DECLAIM FTYPE
2339 and DEFUN better, and implements CHECK-TYPE more correctly, and
2340 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2341 thanks to patches from Martin Atzmueller.
2342 * various fixes to make the cross-compiler more portable to
2343 ANSI-conforming-but-different cross-compilation hosts (notably
2344 Lispworks for Windows, following bug reports from Arthur Lemmens)
2345 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2346 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2347 fixed thanks to Pierre Mai's CMU CL patches.
2348 * fixes to keep the system from overflowing internal counters when
2349 it tries to use i/o buffers larger than 16M bytes
2350 * fixed bug 45a: Various internal functions required to support
2351 complex special functions have been merged from CMU CL sources.
2352 (When I was first setting up SBCL, I misunderstood a compile-time
2353 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2354 * improved support for type intersection and union, fixing bug 12
2355 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2356 more obscure bugs as well
2357 * some steps toward byte-compiling non-performance-critical
2358 parts of the system, courtesy of patches from Martin Atzmueller
2359 * Christophe Rhodes has made some debian packages of sbcl at
2360 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2361 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2362 stable, but are nonetheless usable. When he's ready, I'd be happy
2363 to add them to the SourceForge "File Releases" section. (And if
2364 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2365 * new fasl file format version number (because of changes in
2366 internal representation of (OR ..) types to accommodate the new
2367 support for (AND ..) types, among other things)
2369 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2370 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2372 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2374 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2375 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2376 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2377 * DESCRIBE now works on structure objects again.
2378 * Most function call argument type mismatches are now handled as
2379 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2380 know whether the function will be redefined before the call is
2381 executed. (The compiler could flag local calls with full WARNINGs,
2382 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2383 it doesn't keep track of enough information to know whether calls
2384 are local in this sense.)
2385 * Compiler output is now more verbose, with messages truncated
2386 later than before. (There should be some supported way for users
2387 to override the default verbosity, but I haven't decided how to
2388 provide it yet, so this behavior is still controlled by the internal
2389 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2390 src/compiler/ir1util.lisp.)
2391 * Fasl file format version numbers have increased again, because
2392 support for the Gray streams extension changes the layout of the
2393 system's STREAM objects.
2394 * The Gray subclassable streams extension now works, thanks to a
2395 patch from Martin Atzmueller.
2396 * The full LOAD-FOREIGN extension (not just the primitive
2397 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2398 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2399 but like most other programs, it defaults to copying the Unix
2400 environment from the original process instead of starting the
2401 new process in an empty environment.
2402 * Extensions which manipulate the Unix environment now support
2403 an :ENVIRONMENT keyword option which doesn't smash case or
2404 do other bad things. The CMU-CL-style :ENV option is retained
2405 for porting convenience.
2406 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2407 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2409 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2411 * A patch from Martin Atzmueller seems to have solved the SIGINT
2412 problem, and as far as we know, signal-handling now works cleanly.
2413 (If you find any new bugs, please report them!)
2414 * The system no longer defaults Lisp source file names to types
2415 ".l", ".cl", or ".lsp", but only to ".lisp".
2416 * The compiler no longer uses special default file extensions for
2417 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2418 seems to expect a single default extension for all compiled code,
2419 and there's no compelling reason to try to stretch the standard
2420 to allow two different extensions.) Instead, byte-compiled files
2421 default to the same extension as native-compiled files.
2422 * Fasl file format version numbers have increased again, because
2423 a rearrangement of internal implementation packages made some
2424 dumped symbols in old fasl files unreadable in new cores.
2425 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2426 many fewer weird special cases.
2427 * Bug #17 (differing COMPILE-FILE behavior between logical and
2428 physical pathnames) has been fixed, and some related misbehavior too,
2429 thanks to a patch from Martin Atzmueller.
2430 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2431 by Tim Moore, ported to SBCL by Martin Atzmueller.
2432 * Martin Atzmueller fixed several filesystem-related problems,
2433 including bug #36, in part by porting CMU CL patches, which were
2434 written in part by Paul Werkowski.
2435 * More compiler warnings in src/runtime/ are gone, thanks to
2436 more patches from Martin Atzmueller.
2437 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2440 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2442 * DESCRIBE now works on CONDITION objects.
2443 * The debugger now handles errors which arise when trying to print
2444 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2446 * The build system now uses an additional file, customize-target-features.lisp,
2447 to allow local modifications to the target *FEATURES* list. (The point of
2448 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2449 debugging features enabled, without having to worry about propagating it
2450 into everyone's system when I do a "cvs update".) When no
2451 customize-target-features.lisp file exists, the target *FEATURES* list
2452 should be constructed the same way as before.
2453 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2454 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2455 warn about behavior which is undefined under the ANSI spec. Note: This
2456 is specified by ANSI, but it's not very popular with programmers.
2457 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2458 macro in the SBCL sources for an example of a workaround which you
2459 might use to make portable ANSI-standard code which does what you want.
2460 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2461 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2462 when it's used as a non-toplevel form. (This is required in order
2463 to implement the DEFCONSTANT-EQX macro.)
2464 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2465 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2466 funny things, due to bugs in the implementation of EVAL-WHEN
2467 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2468 non-toplevel DEFCONSTANTs are uncommon.)
2469 * The core file version number and fasl file version number have been
2470 incremented, because the old noncompliant DEFCONSTANT behavior involved
2471 calling functions which no longer exist, and because I also took the
2472 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2473 * fixed bug 1 (error handling before read-eval-print loop starts), and
2474 redid debugger restarts and related debugger commands somewhat while
2476 ** The QUIT debugger command is gone, since it did something
2477 rather different than the SB-EXT:QUIT command, and since it never
2478 worked properly outside the main toplevel read/eval/print loop.
2479 Invoking the new TOPLEVEL restart provides the same functionality.
2480 ** The GO debugger command is also gone, since you can just invoke
2481 the CONTINUE restart directly instead.
2482 ** The TOP debugger command is also gone, since it's redundant with the
2483 FRAME 0 command, and since it interfered with abbreviations for the
2485 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2487 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2488 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2489 some porting work by Martin Atzmueller)
2490 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2491 with a boatload of patches which he ported to SBCL. Now that those
2493 ** The system tries to make sure that its low-priority messages
2494 are prefixed by semicolons, to help people who like to use
2495 syntax highlighting in their ILISP buffer. (This patch
2496 was originally due to Raymond Toy.)
2497 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2498 patches originally written by Raymond Toy.
2499 ** The compiler understands coercion between single-value and
2500 multiple-VALUES type expressions better, getting rid of some very
2501 weird behavior, thanks to patches originally by Robert MacLachlan
2502 and Douglas Crosher.
2503 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2504 lambda lists, thanks to a patch originally by Pierre Mai.
2505 ** The system no longer bogusly warns about "abbreviated type
2507 ** The compiler gets less confused by inlining and RETURN-FROM,
2508 thanks to some patches originally by Tim Moore.
2509 ** The system no longer hangs when dumping circular lists to fasl
2510 files, thanks to a patch originally from Douglas Crosher.
2511 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2512 undefined function error.
2513 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2514 in structure slot initforms) for now, documented workaround instead:-|
2515 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2516 * fixed bug 5: added stubs for various Gray stream functions called
2517 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2518 installed, at least appropriate type errors are generated
2519 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2520 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2521 consistently in DEFMETHOD forms.
2522 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2523 it doesn't seem to affect SBCL after all
2524 * The C runtime system now builds with better optimization and many
2525 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2527 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2529 * The system is now under CVS at SourceForge (instead of the
2530 CVS repository on my home machine).
2531 * The new signal handling code has been tweaked to treat register
2532 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2533 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2534 downstream, e.g. in debug-int.lisp, has implicit dependencies
2535 on the unsignedness of integer representation of machine words,
2536 and that caused the system to bomb out with infinite regress
2537 when trying to recover from type errors involving signed values,
2538 e.g. (BUTLAST '(1 2 3) -1).)
2539 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2540 (This was one of the bugs Peter Van Eynde reported back in July.)
2541 * The system now uses code inspired by Colin Walters' O(N)
2542 implementation of MAP (from the cmucl-imp@cons.org mailing
2543 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2544 inline the MAP operation, and there is more than one
2545 sequence argument to the MAP call (so that it can't just
2546 do ETYPECASE once and for all based on the type of the
2547 single sequence argument). (The old non-inline implementation
2548 of the general M-argument sequence-of-length-N case required
2549 O(M*N*N) time when any of the sequence arguments were LISTs.)
2550 * The QUIT :UNIX-CODE keyword argument has been renamed to
2551 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2552 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2554 * Raymond Wiker's patches to port dynamic loading from Linux to
2555 FreeBSD have been added.
2556 * The BUGS file is now more nearly up to date, thanks in large part
2557 to Martin Atzmueller's review of it.
2558 * The debugger now flushes standard output streams before it begins
2559 its output ("debugger invoked" and so forth).
2560 * The core version number and fasl file version number have both
2561 been incremented, because of incompatible changes in the layout
2563 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2564 used to not be called for a saved Lisp image.)
2565 * Martin Atzmueller's version of a patch to fix a compiler crash,
2566 as posted on sbcl-devel 13 September 2000, has been installed.
2567 * Instead of installing Martin Atzmueller's patch for the
2568 compiler transform for SUBSEQ, I deleted the compiler transform,
2569 and transforms for some similar consing operations.
2570 * A bug in signal handling which kept TRACE from working on OpenBSD
2572 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2573 away by constant folding
2574 * The system now defines its address space constants in one place
2575 (in the Lisp sources), and propagates them automatically elsewhere
2576 (through GENESIS and the sbcl.h file). Therefore, patching the
2577 address map is less unnecessarily tedious and error-prone. The
2578 Lisp names of address space constants have also been systematized.
2579 * CVS tags like dollar-Header-dollar have been removed from
2580 the sources, because they have never saved me trouble and
2581 they've been source of trouble working with patches and other
2582 diff-related operations.
2583 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2584 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2586 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2588 * The system has been ported to OpenBSD.
2589 * The system now compiles with a simple "sh make.sh" on the systems
2590 that it's supported on. I.e., now you no longer need to tweak
2591 text in the target-features.lisp-expr and symlinks in src/runtime/
2592 by hand, the make.sh takes care of it for you.
2593 * The system is no longer so grossly inefficient when compiling code
2594 involving vectors implemented as general (not simple) vectors (VECTOR T),
2595 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2596 which dares to use the various sequence functions on non-simple
2597 vectors, takes less of a performance hit.
2598 * There is now a primitive type predicate VECTOR-T-P
2599 to test for the (VECTOR T) type, so that e.g.
2600 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2601 can now be compiled with some semblance of efficiency. (The old code
2602 turned the type declaration into a full call to %TYPEP at runtime!)
2603 * AREF on (VECTOR T) is still not fast, since it's still compiled
2604 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2605 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2606 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2607 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2608 without making full calls to SUBTYPEP at runtime.
2609 (Some analogous efficiency issues for non-simple vectors specialized to
2610 element types other than T, or for non-simple multidimensional arrays,
2611 have not been addressed. They could almost certainly be handled the
2612 same way if anyone is motivated to do so.)
2613 * The changes in array handling break binary compatibility, so
2614 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2615 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2616 instead of (VALUES T T).
2617 * By following the instructions that Dan Barlow posted to sbcl-devel
2618 on 2 July 2000, I was able to enable primitive dynamic object
2619 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2620 functionality is not implemented (since it calls ld to resolve
2621 library references automatically, requiring RUN-PROGRAM for its
2622 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2623 try to resolve library references) is now supported.
2624 * The system now flushes the standard output streams when it terminates,
2625 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2626 them at several other probably-convenient times, e.g. in each pass of
2627 the toplevel read-eval-print loop, and after evaluating a form given
2628 as an "--eval" command-line option. (These changes were motivated by a
2629 discussion of stream flushing issues on cmucl-imp in August 2000.)
2630 * The source transform for TYPEP of array types no longer assumes
2631 that an array whose element type is a not-yet-defined type
2632 is implemented as an array of T, but instead punts, so that the
2633 type will be interpreted at runtime.
2634 * There is now some support for cross-compiling in make.sh: each of
2635 the phases of make.sh has its own script. (This should be transparent
2636 to people doing ordinary, non-cross-compile builds.)
2637 * Since my laptop doesn't have hundreds of megabytes of memory like
2638 my desktop machine, I became more motivated to do some items on
2639 my to-do list in order to reduce the size of the system a little:
2640 ** Arrange for various needed-only-at-cold-init things to be
2641 uninterned after cold init. To support this, those things have
2642 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2643 symbols with such names are now uninterned after cold init).
2644 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2645 fasl files for cold load.
2646 ** Remove the old compiler structure pooling code (which used to
2647 be conditional on the target feature :SB-ALLOC) completely.
2648 ** Redo the representation of some data in cold init to be more compact.
2649 (I also looked into supporting byte compiled code at bootstrap time,
2650 which would probably reduce the size of the system a lot, but that
2651 looked too complicated, so I punted for now.)
2652 * The maximum signal nesting depth in the src/runtime/ support code has
2653 been reduced from 4096 to 256. (I don't know any reason for the very
2654 large old value. If the new smaller value turns out to break something,
2655 I'll probably just bump it back up.)
2656 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2658 * Many, many bugs reported by Peter Van Eynde have been added to
2659 the BUGS list; some have even been fixed.
2660 * While enabling dynamic object file loading, I tried to make the
2661 code easier to understand, renaming various functions and variables
2662 with less ambiguous names, and changing some function calling
2663 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2664 * While trying to figure out how to do the OpenBSD port, I tried to
2665 clean up some of the code in src/runtime/. In particular, I dropped
2666 support for non-POSIX signal handling, added various comments,
2667 tweaked the code to reduce the number of compilation warnings, and
2668 renamed some files to increase consistency.
2669 * To support the new automatic configuration functionality in make.sh,
2670 the source file target-features.lisp-expr has been replaced with the
2671 source file base-target-features.lisp-expr and the machine-generated
2672 file local-target-features.lisp-expr.
2673 * fixed a stupid quoting error in make.sh so that using CMU CL
2674 "lisp -batch" as cross-compilation host works again
2676 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2678 * DESCRIBE no longer tries to call itself recursively to describe
2679 bound/fbound values, so that it no longer fails on symbols which are
2680 bound to themselves (like keywords, T, and NIL).
2681 * DESCRIBE now works on generic functions.
2682 * The printer now prints less-screwed-up representations of closures
2683 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2684 * A private symbol is used instead of the :EMPTY keyword previously
2685 used to mark empty slots in hash tables. Thus
2686 (DEFVAR *HT* (MAKE-HASH-TABLE))
2687 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2688 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2689 now does what ANSI says that it should. (You can still get
2690 similar noncompliant behavior if bang on the hash table
2691 implementation with all the symbols you get back from
2692 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2693 This breaks binary compatibility, since tests for equality to
2694 :EMPTY are wired into things like the macroexpansion of
2695 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2697 * There's now a minimal placeholder implementation for CL:STEP,
2698 as required by ANSI.
2699 * An obscure bug in the interaction of the normal compiler, the byte
2700 compiler, inlining, and structure predicates has been patched
2701 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2702 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2704 * Missing ordinary arguments in a macro call are now detected even
2705 when the macro lambda list contains &KEY or &REST.
2706 * The debugger no longer complains about encountering the top of the
2707 stack when you type "FRAME 0" to explicitly instruct it to go to
2708 the top of the stack. And it now prints the frame you request even
2709 if it's the current frame (instead of saying "You are here.").
2710 * As specified by ANSI, the system now always prints keywords
2711 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2713 * The default initial SIZE of HASH-TABLEs is now smaller.
2714 * Type information from CLOS class dispatch is now propagated
2715 into DEFMETHOD bodies, so that e.g.
2716 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2718 is now basically equivalent to
2719 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2720 (DECLARE (TYPE SINGLE-FLOAT X))
2722 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2723 operation, without having to do run-time type dispatch.
2724 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2725 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2726 * Since I don't seem to be making big file reorganizations very often
2727 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2728 is overflowing my ability to conveniently back them up), I've finally
2729 checked the system into CVS. (The CVS repository is on my home system,
2730 not at SourceForge -- putting it on SourceForge might come later.)
2731 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2732 high-level GC-NOTIFY-FOO functions send their output. (There's
2733 still very little control of where low-level verbose GC functions
2734 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2735 less than it used to -- the GC-NOTIFY-FOO functions are now under
2736 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2737 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2738 in only one place in the source code, and propagates it automatically
2739 everywhere that it's needed. Thus e.g. when I bump the version from
2740 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2741 * The C source files now include boilerplate legalese and documentation
2742 at the head of each file (just as the Lisp source files already did).
2743 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2744 to his page will be replaced with a link to his new CLiki service.
2746 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2748 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2749 * The build process now looks for GNU make under the default name "gmake",
2750 instead of "make" as it used to. If GNU make is not available as "gmake"
2751 on your system, you can change this default behavior by setting the
2752 GNUMAKE environment variable.
2753 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2754 can build without error under CMU CL.
2756 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2758 * There is now a partial SBCL user manual (with some new text and some
2759 text cribbed from the CMU CL manual).
2760 * The beginnings of a profiler have been added (starting with the
2761 CMU CL profiler and simplifying and cleaning up). Eventually the
2762 main interface should be through the TRACE macro, but for now,
2763 it's still accessed through vaguely CMU-CL-style functions and macros
2764 exported from the package SB-PROFILE.
2765 * Some problems left over from porting CMU CL to the new
2766 cross-compilation bootstrap process have been cleaned up:
2767 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2768 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2769 part when getting rid of the compiler *BACKEND* stuff.)
2770 ** The value of *NULL-TYPE* was screwed up, because it was
2771 being initialized before the type system knew the final
2772 definition of the 'NULL type. This screwed up several key
2773 optimizations in the compiler, causing inefficiency in all sorts
2774 of places. (I found it because I wanted to understand why
2775 GET-INTERNAL-RUN-TIME was consing.)
2776 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2777 PROCLAIM FTYPE information. Unfortunately this broke binary
2778 compatibility again, since now the forms output by DEFGENERIC
2779 to refer to functions which didn't exist in 0.6.3.
2780 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2781 can use the new (as of 0.6.3) transform for SOME into MAP into
2783 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2784 values to (INTEGER 0 1000000), so that the time code will no longer
2785 occasionally get blown up by Linux returning 1000000 microseconds
2786 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2787 its output conform to the ANSI spec. (Alas, this makes its output
2788 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2789 * A full call to MAP NIL with a single sequence argument no longer conses.
2790 * fixes to problems pointed out by Martin Atzmueller:
2791 * The manual page no longer talks about multiprocessing as though
2792 it were currently supported.
2793 * The ILISP support patches have been removed from the distribution,
2794 because as of version 5.10.1, ILISP now supports SBCL without us
2795 having to maintain patches.
2796 * added a modified version of Raymond Toy's recent CMU CL patch for
2797 EQUALP comparison of HASH-TABLE
2799 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2801 * The system still can't cross-compile itself with
2802 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2803 entails), but at least it can compile more of itself that way
2804 than it used to be able to, and various buglets which were uncovered
2805 by trying to cross-compile itself that way have now been fixed.
2806 * This release breaks binary compatibility again. This time
2807 at least I've incremented the FASL file format version to 2, so that the
2808 problem can be detected reliably instead of just causing weird errors.
2809 * various new style warnings:
2810 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2811 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2812 ** using the lexical binding of a variable named in the *FOO* style
2813 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2814 as specified by ANSI.
2815 * *RANDOM-STATE* is no longer automatically initialized from
2816 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2817 default behavior of the system is to repeat its behavior every time
2818 it's run. If you'd like to change this behavior, you can always
2819 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2820 old convention there was no comparably easy way to get the system to
2821 repeat its behavior every time it was run.
2822 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2823 FTYPE declarations has been removed, in favor of their ANSI
2824 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2825 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2826 the types of their sequence arguments can be determined at compile time.
2827 This is done through a new open code expansion for MAP which eliminates
2828 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2829 argument types can be determined at compile time.
2830 * The optimizer now transforms COERCE into an identity operation when it
2831 can prove that the coerced object is already of the correct type. (This
2832 can be a win for machine generated code, including the output of other
2833 optimization transforms, such as the MAP transform above.)
2834 * Credit information has been moved from source file headers into CREDITS.
2835 * Source file headers have been made more standard.
2836 * The CASE macro now compiles without complaining even when it has
2839 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2841 * (Note that the way that the PCL macroexpansions were rewritten
2842 to accommodate the change in DEFGENERIC below breaks binary
2843 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2844 not run under sbcl-0.6.2. Once we get out of alpha releases,
2845 i.e. hit release 1.0.0, we'll probably try to maintain binary
2846 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2847 and sbcl-1.4.4. Until then, however, it might be fairly common
2848 for maintenance releases to break binary compatibility.)
2849 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2851 * The reporting of unhandled signals has been changed to print some
2852 explanatory text as well as the report form. (Previously only
2853 the report form was printed.)
2854 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2855 it defines, so that the compiler no longer issues undefined function
2856 warnings for compiled-but-not-yet-loaded generic functions.
2857 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2858 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2859 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2860 section of the ANSI spec are supported.
2861 * The "" nickname for the "KEYWORD" package has been removed.
2862 The reader still handles symbol tokens which begin with a package marker
2863 as keywords, but it doesn't expose its mechanism for doing so in the
2864 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2865 * The system now issues STYLE-WARNINGs for contradictory TYPE
2866 proclamations. (Warnings for contradictory FTYPE proclamations would
2867 be nice too, but those can't be done usefully unless the type system
2868 is made smarter about FUNCTION types.)
2869 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2870 systematized, so that "*target-*.lisp is supposed to exist only on the
2871 target and imply that there's a related file which exists on the
2872 host, and *host-*.lisp is supposed to exist only on the host and imply
2873 that there's a related file which exists on the target. This involves a
2874 lot of renaming. Hopefully the acute confusion caused by the renaming
2875 will be justified by the reduction in chronic confusion..
2876 ** runtime-type.lisp -> early-target-type.lisp
2877 ** target-type.lisp -> late-target-type.lisp
2878 ** early-host-format.lisp -> early-format.lisp
2879 ** late-host-format.lisp -> late-format.lisp
2880 ** host-error.lisp -> misc-error.lisp
2881 ** early-error.lisp -> early-target-error.lisp
2882 ** late-error.lisp -> late-target-error.lisp
2883 ** host-defboot.lisp -> early-defboot.lisp
2884 ** code/misc.lisp -> code/target-misc.lisp
2885 ** code/host-misc.lisp -> code/misc.lisp
2886 ** code/numbers.lisp -> code/target-numbers.lisp
2887 ** code/early-numbers.lisp -> numbers.lisp
2888 ** early-host-type.lisp -> early-type.lisp
2889 ** late-host-type.lisp -> late-type.lisp
2890 ** host-typep.lisp -> typep.lisp
2891 ** load.lisp -> target-load.lisp
2892 ** host-load.lisp -> load.lisp
2893 ** host-disassem.lisp -> disassem.lisp
2894 ** host-insts.lisp -> insts.lisp
2895 ** byte-comp.lisp -> target-byte-comp.lisp
2896 ** host-byte-comp.lisp -> byte-comp.lisp
2897 ** host-signal.lisp -> signal.lisp
2898 ** host-defstruct.lisp -> defstruct.lisp
2899 ** late-target-type.lisp -> deftypes-for-target.lisp
2900 Furthermore, several other previously target-only files foo.lisp (e.g.
2901 hash-table.lisp and random.lisp) have been split into a target-and-host
2902 foo.lisp file and a target-only target-foo.lisp file, with their key type
2903 definitions in the target-and-host part, so that the cross-compiler will
2904 know more about target types.
2905 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2906 gone away. In their place are various *BACKEND-FOO* variables
2907 corresponding to the slots of the old structure.
2908 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2909 into the target SBCL has been fixed.
2910 * The chill.lisp system for loading cold code into a running SBCL
2912 * Support for the CMU CL "scavenger hook" extension has been removed.
2913 (It was undocumented and unused in the CMU CL sources that SBCL was
2914 derived from, and stale in sbcl-0.6.1.)
2915 * Various errors in the cross-compiler type system were detected
2916 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2917 (enabling various consistency checks). Many of them were fixed,
2918 but some hard problems remain, so the compiler is back to
2919 running without *TYPE-SYSTEM-INITIALIZED* for now.
2920 * As part of the cross-compiler type system cleanup, I implemented
2921 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2922 * I have started adding UNCROSS calls throughout the type system
2923 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2924 on cross-compiler input files will be able to go away, and various
2926 * CONSTANTP now returns true for quoted forms (as explicitly required
2929 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2931 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2932 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2933 (MAKE-LIST -1). In the longer term, it ought to become true
2934 that declarations are assertions even at SAFETY 1. For now, it's not
2935 quite true even at SAFETY 3, but it's at least more nearly true..
2936 (Note that this change seems to increases the size of the system by
2937 O(5%) and to decrease the speed of the compiler by 20% or more.)
2938 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2939 for the problem of printing dozens of lines of distracting information
2940 about low-level system machinery as part of the top stack frame
2941 on entry to the debugger when an undefined function was called.
2942 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2943 is not reset to COMMON-LISP-USER.
2944 * Compilation of stuff related to dyncount.lisp has been made conditional
2945 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2946 smaller. The various dyncount-related symbols have been moved into
2947 a new "SB-DYNCOUNT" package.
2948 * tty-inspect.lisp has been renamed to inspect.lisp.
2949 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2950 feature has gone away. (When we eventually port to other flavors of
2951 libc and/or Unix, we'll try to make the differences between flavors
2952 invisible at the user level.)
2953 * Various other *FEATURES* tags, and/or their associated conditionals,
2954 have been removed if obsolescent, or given better documentation, or
2955 sometimes given more-mnemonic names.
2957 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2959 * tidied up "make.sh" script
2960 * tidied up system directory structure
2961 * better "clean.sh" behavior
2962 * added doc/FOR-CMUCL-DEVELOPERS
2963 * many many small tweaks to output format, e.g. removing possibly-confusing
2964 trailing #\. character in DESCRIBE-INSTANCE
2965 * (EQUALP #\A 'A) no longer signals an error.
2966 * new hashing code, including EQUALP hashing
2967 * tidied up Lisp initialization and toplevel
2968 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2969 * command line argument processing
2970 * added POSIX-GETENV function to deal with Unix-ish environment variables
2971 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2972 terminating SBCL on EOF
2973 * non-verbose GC by default
2974 * There is no more "sbcl" shell script; the sbcl file is now the C
2975 runtime executable (just like CMU CL).
2976 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2978 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2979 symbol and package information as Lisp native symbol and package objects
2980 instead of strings naming symbols and strings naming packages. This way,
2981 whenever packages are renamed (as in warm init), debug information is
2982 transformed along with everything else.
2983 * tweaked the optimization policy declarations which control the building
2984 of SBCL itself. Now, among other things, the system no longer saves
2985 source location debugging information. (This helps two problems at once
2986 by reducing SBCL size and by keeping SBCL from trying to look for its
2987 sources -- which may not exist -- when reporting errors.)
2988 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2989 debugging and testing purposes
2990 * cleaned up printing, making the printer call PRINT-OBJECT for
2991 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2992 methods, giving nearly-ANSI behavior
2993 * converted almost all special variables to use *FOO* naming convention
2994 * deleted PARSE-TIME functionality, since it can be done portably
2995 * moved some files out of cold init into warm init
2996 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2998 * regularized formatting of source files
2999 * added an install.sh script
3000 * fixed ridiculous memory usage of cross-compiler by making
3001 compiler/alloc.lisp not try to do pooling unless it can hook
3002 itself into the GC of the cross-compilation host. Now the system
3003 builds nicely on my old laptop.
3004 * added :SB-ALLOC in target-features.lisp-expr
3005 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3006 was not implemented)
3007 * re-did condition handling and note reporting in the compiler. Notes
3008 are no longer handled by signalling conditions. Style warnings
3009 and warnings are handled more correctly and reported in such a way
3010 that it's easy to find one or the other in your output (so that you
3011 can e.g. figure out which of many problems caused COMPILE-FILE to
3013 * changed the severity of several compiler warnings from full WARNING
3014 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3015 compiler note reporting so that it doesn't use the condition system
3016 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3017 COMPILE-FILE command)
3018 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3019 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3020 no longer supported, since their implementation was deeply intertwingled
3021 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3022 * removed bogus "support" for compiler macros named (SETF FOO), and
3023 removed the compiler macro for SETF INFO (but only after making a fool
3024 of myself on the cmucl-imp mailing list by posting a bogus patch for
3025 DEFINE-COMPILER-MACRO..)
3026 * Compiled files containing forms which have side effects on the Lisp
3027 reader (such as DEFPACKAGE forms) are now handled more correctly.
3028 (Compiler queuing of top level lambdas has been suppressed by setting
3029 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3030 * deleted various currently-unused source files, e.g. gengc.lisp. They
3031 may be added back at some point e.g. when porting to other architectures,
3032 but until they are it's distracting to distribute them and to try to
3034 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3035 there were so many of them they're just distractions, and UNCROSS is
3036 known to be able to handle the current sources
3037 * moved PROFILE functionality into TRACE, so that it will be clear
3038 how the wrapping and unwrapping of functions when you profile them
3039 interacts with the wrapping and unwrapping of functions when you
3040 trace them. (Actually, the functionality isn't there yet, but at least
3041 the interface specification is there. Hopefully, the functionality will
3042 arrive with some future maintenance release.)
3043 * removed host-oops.lisp
3044 * changed signature of QUIT function to allow UNIX-CODE argument
3045 * fixed READ-SEQUENCE bug
3046 * tweaked verbose GC output so that it looks more like the progress
3047 output that ANSI specifies for functions like LOAD
3048 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3049 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob