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 * fixed some bugs revealed by Paul Dietz' test suite:
35 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
37 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
38 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
39 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
41 * fixed loading of multiply forward-referenced layouts.
42 (thanks to Cheuksan Wang)
43 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
44 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
45 (see COMPILE-FILE documentation for details of :PRINT option)
46 * fixed bugs 19 and 317: fixed-format floating point printing is
47 more accurate. This also fixes a bug reported by Adam Warner
48 related to the ~@F format directive.
49 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
51 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
52 dispatch table if the from-char is a dispatch macro character.
53 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
54 by the wordsize no longer ignore the last word. (reported by Lutz
56 * fixed bug in type checking of dynamic-extent variables. (reported
58 * optimization: sequence traversal functions use their freedom to
59 coerce function designators to functions.
60 * optimization: code with many calls to CLOS methods specialized on
61 CLOS classes has had redundant type checks removed.
62 * optimization: added declarations to speed up operations that access
63 the internal character database (for example STRING-UPCASE)
64 * optimization: comparison operations between floats and sufficiently small
65 fixnums no longer create extra rationals
66 * fixed some bugs related to Unicode integration:
67 ** portions of multibyte characters at the end of buffers for
68 character-based file input are correctly transferred to the
69 start of the buffer at the next read.
70 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
71 passing it through to OPEN.
72 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
73 argument given to internal calls to OPEN.
74 ** the built-in comment readers (introduced by character sequences
75 ";" and "#|") are more forgiving to encoding errors; they will
76 STYLE-WARN and then attempt to resync the stream at a character
77 boundary. (thanks to Teemu Kalvas)
78 * fixed some bugs in the x86-64 port:
79 ** Negative short int return values from c-calls are sign-extended
81 ** The stack is aligned to 16-bytes for c-calls, as required by
82 the ABI. (thanks to Cheuksan Wang)
83 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
84 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
85 for enums. (thanks to Vincent Arkesteijn)
86 ** Multiple small optimizations and bugfixes for floating point
88 * fixed some bugs revealed by Paul Dietz' test suite:
89 ** Space, Tab, Linefeed, Return and Page have the invalid
90 secondary constituent character trait.
91 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
93 ** WITH-INPUT-FROM-STRING should only update the index place on
95 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
97 ** Bit-array operations (BIT-AND and similar) worked incorrectly
98 with one-dimensional arrays with fill pointers.
99 ** TYPE-OF failed on a complex with an integer realpart and a
101 ** compiler failure during type inference for the code of form
102 (IF (EQL X (THE ...)) ...) (MISC.535).
104 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
105 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
106 platform. The port supports 61-bit fixnums, large memory spaces
107 and reloading shared object files.
108 * enhancement: saving cores with foreign code loaded is now
109 supported on ppc/Darwin in addition to the previously supported
111 * enhancement: the statistical profiler now walks deeper into the
112 call stack for more meaningful call-graphs and accrued time
113 reports (x86/x86-64 only). It also now reports time spent in
115 * enhancement: it is now possible to trace most individual methods
116 of a generic function in addition to tracing the generic function
118 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
120 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
121 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
122 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
123 produces an error. (thanks to Vincent Arkesteijn)
124 * bug fix: NAMESTRING on pathnames with :WILD components in their
125 directories works correctly. (thanks to Artem V. Andreev)
126 * fixed bug 125: compiler preserves identity of closures. (reported
128 * bug fixed: functions with &REST arguments sometimes failed with
129 "Undefined function" when compiled with (DEBUG 3). (reported by
131 * bug fix: overflow during compiling of setting element of a bit
132 vector with constant index and value. (reported by Timmy Douglas)
133 * build fix: fixed the dependence on *LOAD-PATHNAME* and
134 *COMPILE-FILE-PATHNAME* being absolute pathnames.
135 * on x86 compiler partially supports stack allocation of dynamic-extent
137 * GO and RETURN-FROM do not check the extent of their exit points
138 when compiled with SAFETY 0.
139 * fixed some bugs related to Unicode integration:
140 ** encoding and decoding errors are now much more robustly
141 handled; it should now be possible to recover even from invalid
142 input or output to the terminal. (thanks to Teemu Kalvas)
143 ** provided a first cut at implementing STRING-TO-OCTETS and
144 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
145 ** altered the SB-MD5 contributed module slightly, changing the
146 interface just enough for it to be supportable for builds where
147 lisp characters are not eight bits.
148 * fixed some bugs revealed by Paul Dietz' test suite:
149 ** the FORMATTER-generated functions for ~V[ conditionals require
150 the correct number of arguments.
151 ** READ-FROM-STRING returns the mandated second value when applied
152 to displaced strings.
153 ** the #\Rubout and #\Backspace characters are treated as invalid
154 constituent characters by the tokenizer.
156 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
157 * new feature: reloading changed shared object files with
158 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
159 * new feature: references to foreign variables and functions
160 can now be compiled and loaded before the corresponding shared
161 object file is loaded, as long as the foreign definitions are
162 available at runtime.
163 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
164 supported. (thanks to Dan Debertin)
165 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
166 just keywords. (thanks to Vincent Arkesteijn)
167 * fixed bug #331: structure-class instances corresponding to
168 DEFSTRUCT forms are now created eagerly.
169 * fixed bug #345: backtraces from calls to undefined functions work
170 on x86 as well. Related bug #61 is now also partially fixed on x86
171 (backtraces from throws to unknown catch tags.)
172 * bug fix: lambda-list parsing is now stricter vrt. order and number
173 of lambda-list keywords.
174 * bug fix: as specified by AMOP, an error is signalled if a
175 class-option appears multiple times in a DEFCLASS form. (reported
177 * bug fix: RANDOM can be compiled when the compiler derives the type
178 of its numeric argument as a disjoint set of small integers.
179 (reported by Paul Dietz)
180 * bug fix: starting a core saved with shared objects loaded when
181 those objects are not available no longer causes threaded SBCL to
182 hang. (reported by Sean Ross)
183 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
185 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
186 argument, which must always be NIL. (reported by Kalle Niemitalo)
187 * bug fix: printing 1.0d+23 no longer results in an error.
188 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
189 * bug fix: structure slot setters preserve evaluation order. (thanks
191 * fixed some bugs related to Unicode integration:
192 ** RUN-PROGRAM can allow its child to take input from a Lisp
193 stream. (reported by Stefan Scholl)
194 ** ASDF-INSTALL successfully downloads .tar.gz files in all
195 locales. (reported by Ken Causey)
196 * fixed some bugs revealed by Paul Dietz' test suite:
197 ** INCF, DECF and REMF evaluate their place form as specified in
199 ** FORMATTER expands ~{ ~} iteration directives with V or #
200 parameters correctly.
201 ** FORMATTER deals with the ~@[ ~] conditional directive where the
202 consequent uses no arguments correctly.
203 ** the system has a partial understanding of the (COMPLEX RATIO)
205 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
208 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
209 * new feature: a build-time option (controlled by the :SB-UNICODE
210 keyword feature, enabled by default) for building the system with
211 support for the entire 21-bit character space defined by the
213 * new feature: the system now has rudimentary external-format
214 support; the primary user-visible change at this time is that
215 characters with the high bit set (such as lower-case-e-acute) will
216 print correctly to a terminal in a UTF-8 environment.
217 * minor incompatible change: BASE-CHAR no longer names a class;
218 however, CHARACTER continues to do so, as required by ANSI.
219 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
220 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
221 has been moved to the SB-EXT package (temporarily re-exported from
223 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
224 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
226 * fixed bug #351: better error-handlind and reporting for malformed
228 * fixed bug #350: bignum-printing is now more memory-efficient,
229 allowing printing of very large bignums, eg. (expt 2 10000000).
230 (reported by Bruno Haible)
231 * fixed bug #302: better primitive-type selection for intersection
233 * fixed bug #308: non-graphic characters now all have names, as
234 required. (reported by Bruno Haible)
235 * bug fix: redefining a class with different superclasses now correctly
236 removes it from the direct-subclasses of its previous superclasses.
237 (reported by David Morse)
238 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
239 new value now works. (reported by Bruno Haible)
240 * bug fix: correct canonicalization of multiple non-standard slot
241 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
242 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
243 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
244 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
245 now exists, an signals an error.
246 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
247 during macroexpansion for non-iteration variables as well. (reported
248 by Bruno Haible for CMUCL)
249 * bug fix: Cyclic structures and unprintable objects in compiler
250 messages no longer cause errors. (reported by Bruno Haible)
251 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
252 and READ-FROM-STRING all now return a primary value of NIL if
253 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
254 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
255 (reported by Bruno Haible for CMUCL)
256 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
257 do not designate a condition. (reported by Bruno Haible for
259 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
260 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
261 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
262 permitting resolution in favour of any of the conflicting symbols.
263 (reported by Bruno Haible for CMUCL)
264 * bug fix: EQUAL compiler optimizations is less aggressive on
265 strings which can potentially compare true despite having distinct
266 specialized array element types.
267 * bug fix: unit enumerations can be defined without dividing by
268 zero. (thanks to Vincent Arkesteijn)
269 * FORMAT compile-time argument count checking has been enhanced.
270 (report from Bruno Haible for CMUCL)
271 * a partial workaround for the bug 262: the compiler does not try to
272 inline-expand a local function doing RETURN-FROM from a deleted
273 BLOCK. (thanks to Peter Denno for the bug report and to David
274 Wragg for the simple test case)
275 * fixed some bugs revealed by Paul Dietz' test suite:
276 ** RENAME-PACKAGE allows all package designators as new package
278 ** constraint propagation and lambda variable substitution are
279 more cautious in dealing with partially deleted code.
280 ** compiler sometimes lost reoptimization passes.
281 ** CERROR, given a condition as condition designator, passes its
282 remaining arguments to the continue format control without
284 ** Case-altering FORMAT directives work correctly on non-ASCII
286 ** The REAL type specifier handles bounds outside the floating
287 point ranges without signalling FLOATING-POINT-OVERFLOW.
288 ** Functions with IR1-transformations can create intercomponent
289 references to global functions.
290 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
292 ** FORMAT ~R treats a nil value for its first parameter correctly.
294 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
295 * enhancement: saving cores with foreign code loaded is now
296 supported on x86/NetBSD and sparc/Linux in addition to the previously
298 * bug fix: on some platforms repeated installations caused multiple
299 copies of HTML documentation to be installed -- should not happen
300 any more. (reported by Stefan Scholl)
301 * bug fix: parsing self-recursive alien record types multiple times
302 no longer causes infinite recursion. (reported by Thomas F. Burdick,
303 original patch by Helmut Eller for CMUCL)
304 * bug fix: stack-exhaustion detection works now on NetBSD as well.
305 (thanks to Richard Kreuter)
306 * bug fix: defining classes whose accessors are methods on existing
307 generic functions in other (locked) packages no longer signals
308 bogus package lock violations. (reported by François-René Rideau)
309 * bug fix: special variables as DEFMETHOD parameters no longer have
310 associated bogus type declarations. (reported by David Wragg and
312 * bug fix: read-write consistency on streams of element-type
313 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
314 * bug fix: redefiniton of the only method of a generic function with
315 no DEFGENERIC no longer emits a full WARNING. In addition,
316 redefinition of generic functions with no DEFGENERIC to an
317 incompatible lambda list now signals an error. (thanks to Zach
319 * bug fix: DEFGENERIC now works even when there's a function of the
320 same name in an enclosing lexical environment. (thanks to Zach
322 * fixed compiler failure, caused by instrumenting code during
323 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
324 * optimization: added loop analysis and improved register allocation
325 * fixed some bugs revealed by Paul Dietz' test suite:
326 ** POSITION on displaced vectors with non-zero displacement
327 returns the right answer.
328 ** (SIMPLE-STRING) is a valid type specifier for sequence
330 ** *PRINT-LEVEL* handling for slotless structures is pedantically
332 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
333 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
335 ** The justification version of the FORMAT ~< directive treats
336 non-zero minpad parameter correctly.
338 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
339 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
340 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
341 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
342 the supported interface.
343 * new feature: Single-stepping of code compiled with DEBUG 2 or
344 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
345 * new feature: saving cores with foreign code loaded is now
346 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
347 Timothy Moore's work for CMUCL)
348 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
349 parameters to * instead of NIL if no initform is supplied.
350 (reported by Johan Bockgaard)
351 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
352 correctly when FUNCALL appears as the car of the form. Note:
353 despite this FUNCALL forms are not currently subject to
354 compiler-macro expansion. (port of Raymond Toy's fix for the
355 same from CMUCL, reported by Johan Bockgaard)
356 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
357 (thanks for Teemu Kalvas)
358 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
359 case instead of terminating the iteration (thanks for Julian
360 Squires, Sean Champ and Raymond Toy)
361 * bug fix: incorrect expansion of defgeneric that caused a style
362 warning. (thanks for Zach Beane)
363 * bug fix: slot accessor effective method computation works properly
364 for classes with multiple non-standard applicable methods on
365 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
366 * on x86 compiler supports stack allocation of results of LIST and
367 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
368 CMUCL implementation by Gerd Moellmann)
369 * fixed some bugs revealed by Paul Dietz' test suite:
370 ** FORMAT strings with both the ~<~:;~> form of the justification
371 directive and pretty-printing directives cause an error.
373 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
374 * incompatible change: the internal functions
375 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
376 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
377 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
378 instead of the old functions.
379 * new feature: on platforms where "dladdr" is available foreign
380 function names now appear in backtraces. (based on Helmut Eller's
382 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
383 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
385 * documentation: ASDF manual is now distributed with SBCL.
386 * bug fix: SBCL can now load its contributed modules with REQUIRE
387 even if the system-provided entries have been removed from
388 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
390 * bug fix: backtraces involving undefined functions or assembly
391 routines are more informative. (thanks to Brian Downing)
392 * bug fix: mutually referent alien structures now work correctly.
393 (reported by Rick Taube)
394 * bug fix: structures defined by WITH-ALIEN can be referred to
395 within other definitions in the same WITH-ALIEN.
396 * bug fix: division operators (MOD, TRUNCATE and the like) with
397 constant zero divisors and integer dividends no longer generate
399 * bug fix: provide default methods for INPUT-STREAM-P and
400 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
401 * bug fix: improve the detection and resolution of MOP metacycles
402 (where effective method computation depends on the generic function
403 having its effective method computed). (reported by Bruno Haible)
404 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
405 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
406 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
407 * bug fix: pathnames with relative directory components can be
408 represented relative to default pathnames.
409 * optimization: in taking the GCD of bignums, reduce the two bignums
410 to approximately the same size (using Euclid's algorithm) before
411 applying the more sophisticated binary GCD. (thanks to Juho
413 * optimization: COUNT on bitvectors now operates word-at-a-time.
414 * optimization: ASH with a positive, but not necessarily constant,
415 (leftwards) shift, when in a modular context, is compiled to a
417 * fixed some bugs revealed by Paul Dietz' test suite:
418 ** FORMAT variable parameters ("~V<char>") are defaulted properly
419 if the corresponding argument is NIL.
420 ** FORMAT directives accepting a minpad parameter treat negative
422 ** literal commas as character parameters to FORMAT directives are
424 ** literal spaces directly after ~<Newline> directives within a
425 format-logical-block (~:< ~@:>) do not induce :FILL-style
426 conditional newlines.
427 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
428 as specified: it no longer includes conditional newlines).
429 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
431 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
432 :PER-LINE-PREFIX argument does not evaluate to a string.
434 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
435 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
436 the manual for details; for now, package locks can be disabled by
437 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
438 if no major problems are found then it is likely that they will be
439 compiled in unconditionally.
440 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
441 unsupported operators on all platforms. To load a shared library
442 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
443 object file, link it into a shared library outside of SBCL and
444 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
446 * bug workaround: ROOM T and the GC were not working together
447 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
448 reporting facility aren't preserved. That reporting has been
449 disabled, so now until and unless someone figures out how to make it
450 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
451 * minor incompatible change: as threatened around sbcl-0.8.0, the
452 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
453 The symbols are also exported from SB-PCL for backwards
454 compatibility, but more so than before SB-PCL should be treated as
455 an implementation-internal package.
456 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
458 * the SB-SPROF contrib now works on (most) non-x86 architectures.
459 It is known as of this release not to work on the Alpha, however.
460 * fixed bug #167: errors signalled due to illegal syntax in method
461 bodies are now more legible.
462 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
463 designators and can hence be used with TYPEP.
464 * fixed bug #333: CHECK-TYPE now ensures that the type error
465 signalled, if any, has the right object to be accessed by
466 TYPE-ERROR-DATUM. (reported by Tony Martinez)
467 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
468 5.1.2.3 for multiple-value place subforms. (reported by Kalle
470 * fixed bug #334: programmatic addition of slots using specialized
471 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
472 and :ALLOCATION :CLASS slots.
473 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
474 float exponents. (rereported by Peter Seibel)
475 * fixed a bug: #\Space (and other whitespace characters) are no
476 longer considered to be macro characters in standard syntax by
478 * fixed bug: initialization of condition class metaobjects no longer
479 causes an instance of the condition to be created. (reported by
481 * fixed bug: it is now possible to have slots such that
482 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
483 description is neither :INSTANCE nor :CLASS.
484 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
485 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
486 SLOT-VALUE-USING-CLASS), and no longer causes errors with
487 non-standard SLOT-DEFINITION-ALLOCATION values.
488 * fixed bugs: various race conditions handling exiting threads.
489 CL-PPCRE's thread torture test now passes.
490 * fixed arguably-a-bug: GC time proportional to number of threads in
491 system even when most of them are idle
492 * optimization: improved performance of BIT and SBIT on bit-vectors.
493 * .fasl file incompatibility: The fasl file version number has
494 been incremented because of changes associated with package locks.
496 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
497 * minor incompatible change: the system no longer provides
498 optimization hints (conditions which are TYPEP
499 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
500 mode; these hints are still emitted when higher SPEED optimization
502 * new contrib module: a sampling profiler (profiling by statistical
503 sampling, rather than by instrumenting functions) is available as
504 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
505 * the behaviour of the standard function ED is now customizeable by
506 third parties through a hook variable: see ED's documentation
507 string for information on the protocol.
508 * the compiler no longer emits efficiency notes for (FUNCALL X)
509 when the type of X is uncertain under default optimization
511 * fixed bug 276: mutating a binding of a specialized parameter to a
512 method to something that is not TYPEP the specializer is now
514 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
515 exhibit the required behaviour on the x86 platform. (thanks to
516 Peter van Eynde, Eric Marsden and Bruno Haible)
517 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
518 even for difficult arguments. (reported by Peter Graves)
519 * fixed bug 141a: the backquote printer now descends quoted
521 * fixed another bug in backquote printing: no more destructive
522 modification of the form's list structure. (reported by Brian
524 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
525 move between its address being taken and the call to
526 interrupt_thread, fixing a crashing race condition.
527 * the SB-POSIX contrib implementation has been adjusted so that it
528 no longer exhibits ridiculously poor performance when constructing
529 instances corresponding to C structs.
531 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
532 * minor incompatible change: the sb-grovel contrib now treats C
533 structures as alien (in the sense of SB-ALIEN) objects rather than
534 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
535 has implications for memory management of client code
536 (sb-grovel-returned objects must now be manually managed) and for
537 type safety (alien objects now have full types).
538 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
539 used to control emission of compiler diagnostics, rather than the
540 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
541 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
542 quality should be considered deprecated.
543 * install.sh now installs the user manual as well
544 * (not quite a new documentable feature, but worth considering in
545 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
546 beginnings of a semantically meaningful condition hierarchy is
547 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
549 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
550 designator as the defaults argument.
551 * fixed bug: Displaced arrays whose displaced-to array has become
552 too small now cause ARRAY-DIMENSION to signal an error, providing
553 for safer bounds-checking. (reported by Bruno Haible)
554 * fixed bug: DEFCLASS slot definitions with identical :READER and
555 :WRITER names now signal a reasonable error. (reported by Thomas
557 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
559 * fixed bug: Invalid slot specification errors now print correctly.
560 (thanks to Zach Beane)
561 * fixed bug 320: Shared to local slot value transfers in class
562 redefinitions now happen corrently from superclasses as
563 well. (reported by Bruno Haible)
564 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
566 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
567 improper lists correctly. (reported by Bruno Haible)
568 * fixed bug 313: source-transform for <fun-name> was erroneously
569 applied to a call of a value of a variable with name <fun-name>.
570 (reported by Antonio Menezes Leitao)
571 * fixed bug 307: The obsolete instance protocol ensures that
572 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
573 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
574 remove dead unknown-values globs from the middle of the stack.
575 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
576 easier to resume long computations after using *BREAK-ON-SIGNALS*
577 to diagnose and fix failures. (thanks to Nikodemus Siivola)
578 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
579 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
580 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
581 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
582 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
584 * fixed a (fairly theoretical) bug in string printing: if
585 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
586 does not have array-element-type equal to the most general string
588 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
589 function name. (thanks to Nikodemus Siivola)
590 * fixed bug: compile-time format string checker failed on
591 non-closed ~{. (reported by Thomas F Burdick)
592 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
593 don't warn on legal loop code involving more than one aggregate
594 boolean. (thanks to Nikodemus Siivola)
595 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
596 in keyword position to MAKE-PACKAGE was not being reported as
598 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
599 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
601 * fixed bugs 280 and 312: the checking for multiple definitions in a
602 file is less likely to become confused by uses of inline
604 * fixed bug: the #S reader macro performs the keyword coercion
605 specified for slot names. (reported by Kalle Niemitalo)
606 * fixed bug: lambda lists may contain symbols whose names start with
607 & but are not lambda-list-keywords; their occurrence triggers a
609 * fixed bug 321: define-method-combination argument lambda lists do
610 not cause invalid code to be generated when &optional or &aux
611 variables have default values. (reported by Bruno Haible)
612 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
613 have CLOS classes; this appears also to have fixed reported
614 instability in stack exhaustion detection.
615 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
616 MUFFLE-WARNING when no associated restart is present is now
618 * optimization: rearranged the expansion of various defining macros
619 so that each expands into only one top-level form in a
620 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
622 * optimization: used a previously little-used slot in symbols to
623 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
625 * fixed some bugs revealed by Paul Dietz' test suite:
626 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
628 ** Fixed an optimization bug related to inheritance of initforms
629 from local to shared slots.
630 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
631 ** CAST optimizer forgot to flush argument derived type.
632 ** print/read consistency on floats is now orders of magnitude
633 more likely. (thanks also to Bruno Haible for a similar report
635 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
636 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
637 for objects of type REAL. Make it so.
638 ** SXHASH is specified (infelicitously) to respect similarity,
639 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
640 it so. (thanks to Markus Ziegler)
641 ** on the Alpha, the compiler succeeds in compiling functions
642 returning a known number of arguments greater than 63.
643 ** fixed handling of invalid NIL arguments in keyword position
645 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
647 ** provided a readably-printable representation for RANDOM-STATE
649 ** ensured that pathnames, where they have a namestring, always
650 print using #P"..." syntax.
652 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
653 * Support for the forthcoming 2.0 version of the NetBSD kernel
654 running on x86 hardware has been added. (thanks to Perry
655 E. Metzger most immediately, and others for their past work)
656 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
657 current release version; SBCL's OpenBSD support had been broken
658 since about the time of OpenBSD's switch to ELF binary format.)
659 * [placeholder for DX summary]
660 ** user code with &REST lists declared dynamic-extent, under high
661 speed or space and low safety and debug optimization policy.
662 * The manual has been converted to Texinfo format and the debugger
663 chapter from the cmucl manual has been added.
664 * A facility has been added to extract documentation strings from
665 sbcl and store them as Texinfo-formatted snippets for inclusion in
666 the manual (via Texinfo's @include directive)
667 * bug fix: compiler emitted division in optimized DEREF. (thanks for
668 the test case to Dave Roberts)
669 * bug fix: multidimensional simple arrays loaded from FASLs had fill
670 pointers. (reported by Sean Ross)
671 * bug fix: PROFILE output is printed nicely even for large numerical
672 values. (thanks to Zach Beane)
673 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
674 greater than 32 handle EOF correctly.
675 * bug fix: on X86 an immediate argument of the IMUL instruction is
676 correctly printed in disassembly. (thanks to Lutz Euler)
677 * bug fix: class slots in redefined classes preserve their old
678 values. (thanks to Bruno Haible and Nikodemus Siivola)
679 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
680 succeeds. (reported by Marco Baringer)
681 * fixed some bugs revealed by Paul Dietz' test suite:
682 ** READ-SEQUENCE now works on ECHO-STREAMs.
683 ** RATIONALIZE works more according to its specification. (thanks
686 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
687 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
688 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
689 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
690 should matter to you only if you rebind the printer control
691 variables and then find you want different bindings in the
692 debugger than in the ordinary execution of your program.)
693 * The runtime build system has been tweaked to support building
694 (on SPARC/SunOS) using a C compiler which invokes Sun's own
695 assembler and linker. (thanks to Nikodemus Siivola)
696 * Unbound, undefined, undeclared variables now trigger full
697 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
698 more often programmer error than deliberate exploitation of undefined
700 * optimization: the hash algorithm for strings has changed to one
701 that is less vulnerable to spurious collisions. (thanks to Juho
703 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
704 needless bounds checking. (thanks to Juho Snellman)
705 * optimization: implemented multiplication as a modular
706 (UNSIGNED-BYTE 32) operation on the PPC backend.
707 * fixed some bugs revealed by Paul Dietz' test suite:
708 ** ADJUST-ARRAY now copies the datum in a zero rank array if
710 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
711 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
712 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
714 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
716 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
718 ** ECHO-STREAMs no longer attempt to echo the end of file value to
719 their output stream on EOF from read.
720 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
721 have been read to end-of-file.
722 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
724 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
725 description of determination of which consecutive characters
727 ** Printing the "Space" character with escaping on now yields "#\\ ",
728 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
729 ** Reading floating-point numbers with *READ-BASE* set to a number
730 less than 10 works correctly.
731 ** Reading floating-point numbers with *READ-BASE* set to a number
732 more than 10 works correctly.
733 ** Printing with *PRINT-READABLY* targets the standard readtable, not
734 the readtable currently in effect.
736 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
737 * minor incompatible change: parsing of namestrings on a physical
738 (Unix) host has changed; numbers after the final #\. in a
739 namestring are no longer interpreted as a version field. This is
740 intented to be largely invisible to the user, except that the
741 meaning of the namestring "*.*.*" has changed: it now refers to a
742 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
743 should usually be replaced by
744 (make-pathname :name :wild :type :wild :version :wild)
745 with the added benefit that this is more likely to be portable.
746 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
747 option to OPEN now signals an error if the file being opened
748 exists; this may have an impact on existing code.
749 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
750 signals received, on the PowerPC platforms (both Linux and
751 Darwin). (thanks to Pierre Mai for pointing out the location of
753 * several fixes on OS X: The system now builds and runs cleanly on
754 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
755 (thanks to Brian Mastenbrook)
756 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
757 second-resolution: integer multiples of 1/3600 between -24 and 24.
758 (thanks to Vincent Arkesteijn)
759 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
760 type when called with 1 argument; PEEK-CHAR checked type of
761 PEEK-TYPE only after having read first character from a
762 stream. (reported by Peter Graves)
763 * bug fix: the garbage collector now has much better locality
764 behaviour, and in particular no longer treats all memory as being
765 exhausted when many small objects point to each other in a deeply
767 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
768 recognized as being TYPEP their class.
769 * bug fix: the PUSHNEW documentation string has been corrected.
770 (thanks to Vincent Arkesteijn)
771 * bug fix: defaulting of the value for the last of an atypically
772 large number of multiple values being bound was not being
773 performed correctly on the Alpha or PPC platforms
774 * optimization: implemented multiplication as a modular
775 (UNSIGNED-BYTE 32) operation on the x86 backend.
776 * optimization: SEARCH on simple-base-strings can now be open-coded.
777 (see also contrib/compiler-extras.lisp for inspiration for
778 teaching the compiler about the Boyer-Moore algorithm).
779 * value, returned by MAX (and MIN) called with several EQUALP, but
780 not EQL, arguments now does not depend on compiler settings.
781 * fixed some bugs revealed by Paul Dietz' test suite:
782 ** in stack analysis liveness information is propagated from
783 non-local entry points.
784 ** pathwise CAST removing failed when the CAST node did not start
786 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
787 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
789 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
790 host is already defined.
791 ** RENAME-FILE works on streams instead of signalling an internal
793 ** PEEK-CHAR uses the current readtable when determining whether
794 or not a character is whitespace.
795 ** MERGE-PATHNAMES handles the case when the pathname does not
796 specify a name while the default-pathname specifies a version.
797 ** Pathnames now stand a better chance of respecting print/read
799 ** Attempting to use standardized file system operators with a
800 pathname with invalid :DIRECTORY components signals a
802 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
804 ** DIRECTORY on logical pathnames is more correct.
805 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
806 signal a TYPE-ERROR if their argument is not a stream.
807 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
808 designator argument does not designate a stream.
809 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
810 examining the synonym.
811 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
813 ** OPEN and WITH-OPEN-STREAM allow opening streams with
814 element-type larger than ([UN]SIGNED-BYTE 32).
816 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
817 * When built with the :SB-FUTEX feature, threaded builds now take
818 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
819 for faster/more reliable mutex and condition variable support.
820 * Incompatible change (but one you probably shouldn't have been using
821 anyway): the interface and code for arbitrating between multiple
822 threads in the same user session has been redesigned.
823 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
824 error when called without an explicit environment argument.
825 (thanks to Nikodemus Siivola)
826 * bug fix: buffered :DIRECTION :IO streams are less likely to become
827 confused about their position. (thanks to Adam Warner and Gerd
829 * bug fix: Pretty printing backquoted forms with unquotations in the
830 argument list position of various code constructs such as LAMBDA
831 now works correctly. (reported by Paul Dietz)
832 * bug fix: Pretty printing unquotations no longer loses all
833 stream position information.
834 * optimization: performance of string output streams is now less
835 poor for multiple small sequence writes.
836 * optimization: performance of CSUBTYPEP in the presence of complex
837 expressions involving CONS and NOT many times has been improved.
838 (reported by Paul Dietz)
839 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
841 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
842 streams for the REPL, for the debugger, and so on. (thanks to
844 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
845 * fixed some bugs revealed by Paul Dietz' test suite:
846 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
847 is ignored if it is too small, rather than propagating through
848 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
849 ** extremely complex negations of CONS types were not being
850 sufficiently canonicalized, leading to inconsistencies in
852 ** VALUES tranformer lost derived type.
854 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
855 * fixed a bootstrapping bug: the build process no longer assumes
856 that the various BOOLE-related constants have the same value in
857 host and target lisps. (noted by Paul Dietz' test suite on an
858 SBCL binary built from CLISP)
859 * The system can now be dynamically linked on the MIPS platform,
860 which enables dynamic loading of foreign code from Lisp. (thanks
861 to Ralf Baechle for discussions on the MIPS ABI)
862 * The system now records debugging information for its own source
863 files in a filesystem-position-independent manner, relative to
864 the "SYS" logical host.
865 * fixed a compiler bug: MV-LET convertion did not check references
866 to the "max args" entry point. (reported by Brian Downing)
867 * tweaked disassembly notes to be less confident about proclaiming
868 some instruction as an LRA. (thanks to Brian Downing)
869 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
870 now each have their own history, command character, and other
871 characteristics. (thanks to David Lichteblau)
872 * fixed some bugs revealed by Paul Dietz' test suite:
873 ** compiler failure in compiling LOGAND expressions including a
875 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
876 shift greater than 32.
877 ** FLUSH-DEST did not mark blocks for type check regeneration.
878 ** HANDLER-CASE failed to accept declarations in handler clauses
879 in some circumstances.
881 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
882 * New code in contrib/sb-introspect (still probably not entirely
883 stable yet) provides some support for smart Lisp development
884 environments like SLIME.
885 * The conditions signalled for errors occurring when loading .fasl
886 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
887 in a way which should help ASDF recover gracefully.
888 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
889 stuff has been cleaned up. If you code contrib/ stuff, this might
890 affect you, and you can look at contrib/README, contrib/STANDARDS,
891 and/or the 0.8.4.27 diff to check.
892 * In full calls the compiler now does not generate checks for declared
893 argument types for all arguments.
894 * various threading fixes
895 ** and some experimental patches which didn't make it into
896 the main tree for this release, but which are shipped in
897 contrib/experimental-thread.patch as a possible fix for some
898 failures (deadlock, spinning...) in GC-intensive multithreaded
900 * fixed PPC build problem (source code incompatibility of different
901 library versions): added offsetof() hackery which attempts to divine
902 where glibc maintainers put uc_mcontext today
903 * fixed bug 282: compiler does not trust type assertions while passing
904 arguments to a full call.
905 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
906 &OPTIONAL VALUES type parameter.
907 * fix bug 214: algorithm for noting rejected templates is now more
908 similar to that of template seletion. (also reported by rydis on
910 * fixed bug 141b: printing backquoted information readably and prettily
911 inserts a space where necessary.
912 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
913 (CHARACTER 10) are now reported as errors, rather than propagated
914 as unknown types. (reported by piso on #lisp)
915 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
916 with values NIL and :ERROR. (thanks to Milan Zamazal)
917 * fixed bug 191c: CLOS now does proper keyword argument checking as
918 described in CLHS 7.6.5 and 7.6.5.1.
919 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
920 longer raise an error; further, using a list as a for-as-arithmetic
921 counter now raises a meaningful error.
922 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
923 now understood by sequence creation functions such as MAKE-SEQUENCE
925 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
926 a STRING-INPUT-STREAM.
927 * compiler enhancement: SIGNUM is now better able to derive the type
929 * type declarations inside WITH-SLOTS are checked. (reported by
931 * fixed some bugs revealed by Paul Dietz' test suite:
932 ** incorrect optimization of TRUNCATE for a positive first
933 argument and negative second.
934 ** compiler failure in let-convertion during flushing dead code.
935 ** compiler failure while deriving type of TRUNCATE on an
936 interval, containing 0.
937 ** ASH of a negative bignum by a negative bignum count now returns
939 ** intersection of CONS types now canonicalizes properly, fixing
940 inconsistencies in SUBTYPEP.
942 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
943 * incompatible change: The --disable-debugger command line
944 option now clobbers the debugger at a more fundamental
945 level, by redefining #'INVOKE-DEBUGGER instead of by
946 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
947 is specified by ANSI to ignore *DEBUGGER-HOOK* and
948 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
949 enter the debugger REPL and then suffer recursive errors
950 because *DEBUG-IO* is also messed up in --disable-debugger mode;
951 while under the new system, BREAK in --disable-debugger mode
952 terminates the system just as an unhandled error would.
953 * fixed compiler performance when processing loops with a step >1;
954 * bug fix: DOCUMENTATION now retrieves generic function
955 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
956 support has been systematized, and now supports the methods
957 specified by ANSI, along with a default method and a method for
958 slot documentation. (reported by Nathan Froyd)
959 * bug fix: effective methods associated with a generic function are
960 no longer cached over a change of that generic function's method
961 combination. (reported by Andreas Fuchs)
962 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
963 names a non-existent directory. (thanks to Andreas Fuchs)
964 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
965 the PPC platform, lead to stack corruption; this has been fixed.
966 (reported by Rainer Joswig)
967 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
968 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
969 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
970 arithmetic in general, is now much more reliable on the PPC
972 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
973 platform now returns the right answer.
974 * optimization: restored some effective method precomputation in
975 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
976 precomputation is now tunable.
977 * optimization: compiler-internal data structure use has been
978 reviewed, and changes have been made that should improve the
979 performance of the compiler by about 20%.
980 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
981 simple-base-strings and simple-bit-vectors is improved.
982 * optimization: the optimization of 32-bit logical and arithmetic
983 functions introduced in version 0.8.3 on the x86 has been
984 implemented on the mips, ppc and sparc platforms; an
985 implementation of the same facility, but for 64-bit arithmetic,
986 has been added for the alpha.
987 * microoptimization: the compiler is better able to make use of the
988 x86 LEA instruction for multiplication by constants.
989 * bug fix: in some situations compiler did not report usage of
990 generic arithmetic in (SPEED 3) policy.
991 * bug 145b fix: compiler used wrong type specifier while converting
992 MEMBER-types to numeric.
993 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
995 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
996 index argument now works properly on the Alpha platform.
997 * bug fix: floating point exception treatment on the Alpha platform
999 * bug fix: FILE-POSITION works much better on string input and
1000 output streams. (thanks to Nikodemus Siivola)
1001 * bug fix: many threading/garbage collection symptoms sorted.
1002 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1003 might be pseudo-atomic.
1004 * internal change: Stopping for GC is now done with signals not
1005 ptrace. GC is now done in whichever thread wanted it, instead of
1007 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1009 * bug fix: result form in DO is not contained in the implicit
1011 * incompatible change: ICR structure is changed; the value part of
1012 CONTINUATION is now called LVAR; corresponding functions are
1013 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1014 * added type deriver for ISQRT (thanks to Robert E. Brown).
1015 * bug fix: better support for loading from the command line when an
1016 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1018 * fixed some bugs revealed by Paul Dietz' test suite:
1019 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1020 ** ROUND and FROUND now give the right answer when given very
1021 small float arguments.
1022 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1024 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1025 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1026 ** LCM should return a non-negative integer.
1027 ** PARSE-INTEGER returned the index of a terminator instead of the
1028 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1029 ** PARSE-INTEGER returned an incorrect index being applied to a
1031 ** LCM with two arguments of 0 returns 0 rather than signalling
1033 ** unsigned addition of a 32-bit constant with the high bit set no
1034 longer causes an internal compiler error on the x86.
1035 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1036 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1037 with negative last argument.
1038 ** byte specifiers with zero size and position no longer cause
1039 an error during type derivation.
1040 ** bignum multiplication on the Alpha platform now returns the
1042 * porting: The system now builds on SuSE AMD64, although it still
1043 generates a 32-bit binary.
1044 * .fasl file incompatibility: The fasl file version number has
1045 been incremented (because of the changes to internal compiler
1046 data structures referred to above).
1048 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1049 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1050 more accurately, on the Darwin kernel running on PowerPC hardware.
1051 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1052 * Compiler code deletion notes now signal a condition of type
1053 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1054 an associated MUFFLE-WARNING restart.
1055 * The compiler now performs limited argument count validation of
1056 constant format strings in FORMAT, and where appropriate in ERROR,
1057 CERROR and WARN. (thanks to Gerd Moellmann)
1058 * New ASDF-INSTALL contrib can be used for automatic download and
1059 installation of third-party Lisp code from CCLAN or other sites
1061 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1062 which forces another thread to execute a function supplied by the
1064 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1065 now accept and act upon their :ELEMENT-TYPE keyword argument.
1066 (reported by Martin Atzmueller, Edi Weitz)
1067 * bug fix: FILE-POSITION now accepts position designators up to
1068 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1069 is the greater. (thanks to Patrik Nordebo)
1070 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1071 Kalle Olavi Niemitalo)
1072 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1073 printable. (reported by Eric Marsden)
1074 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1075 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1076 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1077 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1078 (thanks to Stig Sandoe)
1079 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1080 UNREAD-CHAR now correctly includes the unread character in the
1081 target sequence. (thanks to Gerd Moellmann)
1082 * bug fix in threaded builds: the system can now be suspended and
1083 resumed by shell job control with minimal disruption.
1084 * bug fixes in times and timezones >2038AD
1085 * better handling of "where is GNU make?" problem in build scripts
1086 (thanks to Nikodemus Siivola)
1087 * new optimization: inside a named function any reference to a
1088 function with the same name is considered to be a self-reference;
1089 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1090 optimization quality.
1091 * new optimization on x86: logical functions and + now have
1092 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1093 used when the result is truncated to 32 bits.
1094 * VALUES declaration is partially enabled.
1095 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1096 * fixed some bugs revealed by Paul Dietz' test suite:
1097 ** The system now obeys the constraint imposed by
1098 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1099 types form a lattice under type intersection.
1100 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1101 ** ASSOC now ignores NIL elements in an alist.
1102 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1103 and (1+ MOST-POSITIVE-FIXNUM) answers.
1104 ** The addition of a method with invalid qualifiers to a generic
1105 function does not cause an error to be signalled immediately;
1106 a warning is signalled, and the error is generated only on
1107 calling the generic function.
1108 * changed .fasl file version number, in part to add type codes for
1109 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1110 obscure ANSI requirements
1112 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1113 * fixed bug 148: failure to inline-expand a local function left
1114 garbage, confusing the compiler.
1115 * fixed bugs 3cd: structure slot readers perform type check if the
1116 slot can have an invalid value (i.e. it is either not initialized
1117 or can be written with a less specific slot writer).
1118 * bug fix: the compiler now traps array references to elements off
1119 the end of an array; previously, the bounds checking in some
1120 circumstances could go off-by-one.
1121 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1123 * type declarations for array element types now obey the description
1124 on the CLHS page "Declaration TYPE", as per discussions on
1125 sbcl-help around 2003-05-08. This means that a declaration
1126 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1127 declaration, all references to BAR will be asserted or assumed
1128 (with THE, so dependent on compiler policy) to involve objects of
1129 type FOO. Note that no such declaration is implied in
1130 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1131 * declared types of functions from the "Conditions"
1132 chapter. (reported by Paul Dietz)
1133 * bug fix: CERROR accepts a function as its first argument.
1134 * bug fix: NTH an NTHCDR accept a bignum as index
1135 arguments. (reported by Adam Warner)
1136 * optimization: character compare routines now optimize comparing
1137 against a constant character. (reported by Gilbert Baumann)
1138 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1139 argument now works properly on the MIPS platform.
1140 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1141 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1142 declared to have a functional type, does not cause an error
1144 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1145 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1146 * fixed bug: if last continuation of a deleted block has a
1147 destination, this destination should be deleted too. (reported by
1149 * fixed a bug in the bootstrap process: the host compiler's values
1150 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1151 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1152 test case from Patrik Nordebo)
1153 * improved the ability of the disassembler on the PPC platform to
1154 provide helpful disassembly notes.
1155 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1156 the class in more cases than previously.
1157 * bug fix: FILE-POSITION now understands :START and :END for
1158 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1159 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1160 8) data. (thanks to David Lichteblau)
1161 * bug fix: it is possible to add a method to a generic function
1162 without lambda list.
1163 * bug fix: reader failed to signal END-OF-FILE inside an
1164 object representation. (reported by Nikodemus Siivola)
1165 * fixed some bugs revealed by Paul Dietz' test suite:
1166 ** LAST and [N]BUTLAST should accept a bignum.
1167 ** condition slot accessors are methods.
1168 ** (VECTOR NIL) is a subtype of STRING.
1170 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1171 * minor incompatible change: some nonsensical specialized lambda
1172 lists (used in DEFMETHOD) which were previously ignored now signal
1174 * minor incompatible change: the system is now aware of the types of
1175 variables in the COMMON-LISP package, and will signal errors for
1176 most violations of these type constraints (where previously they
1177 were silently accepted).
1178 * minor incompatible change: COMPILE-FILE now uses the freedom
1179 afforded (ANSI 3.2.2.3) to use derived function types for
1180 functions defined in the same file. This also permits the system
1181 to warn on static type mismatches and function
1182 redefinition. (Currently it does not work with high DEBUG level.)
1183 * minor incompatible change: VALUES declaration is disabled.
1184 * When issuing notes, the compiler now signals a condition of type
1185 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1186 restart for use in user handlers. It is expected that the
1187 COMPILER-NOTE condition will eventually become a condition
1188 supertype to a hierarchy of note types, which will then be
1189 handleable in a similar fashion. However, at the moment, no such
1190 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1191 but it's an implementation detail, not a classification for the
1193 * Changes in type checking closed the following bugs:
1194 ** type checking of unused values (192b, 194d, 203);
1195 ** template selection based on unsafe type assertions (192c, 236);
1196 ** type checking in branches (194bc).
1197 * A short form of VALUES type specifier has ANSI meaning (it has
1198 increased the number of situations when SBCL cannot perform type
1200 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1201 keywords or constants is permissible.
1202 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1203 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1204 classes. (thanks to Antonio Martinez)
1205 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1206 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1207 argument to a pretty-print stream. Instead, it leaves any such
1208 operations to DESCRIBE-OBJECT methods.
1209 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1211 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1212 lambda lists. (thanks to David Lichteblau)
1213 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1214 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1215 (thanks to Gerd Moellmann)
1216 * fixed bug 246: increased compilation speed of long
1217 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1219 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1221 * DEFCONSTANT now throws a condition of type
1222 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1223 constant to a non-EQL value; CONTINUE and ABORT restarts
1224 respectively change and preserve the value.
1225 * fixed bug 63: The code walker, part of the implementation of CLOS,
1226 is now better at handling symbol macros.
1227 * bug fix: There is no longer an internal implementation type named
1228 CL:LENGTH. (reported by Raymond Toy)
1229 * bug fix: In macro-like defining macros/special operators the
1230 implicit block does not enclose the lambda list.
1231 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1232 enclosed in parentheses) are not suitable as type specifiers, and
1233 their use properly signals an error now.
1234 * bug fix: An argument count mismatch for a type specifier in code
1235 being compiled no longer causes an unhandled error at compile
1236 time, but signals a compile-time warning.
1237 * fixed simple vector readable printing
1238 * bug fix: DESCRIBE takes more care over whether the class
1239 precedence list slot of a class is bound before accessing it.
1240 (reported by Markus Krummenacker)
1241 * bug fix: FORMATTER can successfully compile pretty-printer format
1242 strings which use variants of the ~* directive inside.
1243 * bug fix: SEARCH now applies its TEST predicate to the elements of
1244 the arguments in the correct order. (thanks to Wolfhard Buss)
1245 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1246 in type inference. (thanks to Robert E. Brown)
1247 * bug fix: Reading in symbols with an explicit package name of ""
1248 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1249 * fixed some bugs revealed by Paul Dietz' test suite:
1250 ** NIL is now allowed as a structure slot name.
1251 ** Arbitrary numbers, not just REALs, are allowed in certain
1252 circumstances in LOOP for-as-arithmetic clauses.
1253 ** Multiple class redefinitions before slot access no longer
1254 causes a type error.
1255 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1256 association between the name and a class.
1257 ** Generic functions with non-standard method-combination and over
1258 five methods all of which return constants no longer return NIL
1259 after the first few invocations. (thanks to Gerd Moellmann)
1260 ** CALL-NEXT-METHOD with no arguments now passes the original
1261 values of the arguments, even in the presence of assignment.
1262 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1264 ** DISASSEMBLE works with closures and funcallable instances.
1265 ** ADD-METHOD now returns the generic function, not the new method.
1266 ** FIND-METHOD signals an error if the lengths of the specializers
1267 is incompatible with the generic function, even if the ERRORP
1269 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1270 which its argument is a member.
1271 ** DEFCLASS only redefines the class named by its class-name
1272 argument if that name is the proper name of the class;
1273 otherwise, it creates a new class.
1274 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1275 of the UNBOUND-SLOT condition to the name of the slot.
1276 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1277 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1278 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1279 SLOT-MAKUNBOUND in the specified fashion.
1281 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1282 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1283 cross-compilation host. As a consequence, we can now bootstrap our
1284 way up to SBCL starting with a bare gcc toolchain and human-readable
1285 source code (first the source to CLISP, then the source to SBCL).
1286 * A contributed module containing a partial implementation of the
1287 simple-streams interface has been included. (thanks to Rudi
1289 * A contributed module implementing the RFC1321 Message Digest
1290 Algorithm, known as MD5, has been included.
1291 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1292 no longer has any effect, as the code controlled by this feature
1293 has been deleted. (As far as we know, no-one has ever built using
1294 this feature, and its semantics were confused in any case).
1295 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1296 work on conditions (as required by the ANSI specification),
1297 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1298 expected behaviour on conditions. Users should note, however,
1299 that such behaviour is not required by the ANSI specification,
1300 and so use of this behaviour may render their code unportable.
1301 * fixed some bugs revealed by Paul Dietz' test suite:
1302 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1304 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1305 ** SXHASH on condition objects no longer returns NIL.
1306 ** :ALLOCATION :CLASS slots are better treated; their values are
1307 updated on class redefinition, and initforms inherited from
1308 superclasses are applied.
1309 ** REMOVE-METHOD returns its generic function argument even when
1310 no method was removed.
1311 ** SHARED-INITIALIZE now initializes the values of the requested
1312 slots, including those with :ALLOCATION :CLASS.
1313 ** ALLOCATE-INSTANCE now works on structure classes defined via
1314 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1316 ** SLOT-EXISTS-P now works on conditions, as well as structures
1318 ** MAKE-LOAD-FORM now has the required methods on
1319 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1320 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1321 keyword, and now implements the SLOT-NAMES argument.
1322 ** methods with &OPTIONAL arguments no longer allow too many
1323 arguments to be passed in the call without error.
1324 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1325 option is consistent with the required arguments of the generic
1326 function lambda list.
1327 * bug fix: REQUIRE accepts a string designator. (Thanks to
1329 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1330 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1331 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1332 * bug fix: The compiler now checks for duplicated variables in macro
1334 * bug fix: SETQ on globals returns the correct value.
1335 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1336 not a COMPILER-ERROR (followed by some other strange error on
1337 choosing the CONTINUE restart).
1338 * bug fix: make.sh and friends are now more consistent in the way that
1339 they look for GNU "make".
1341 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1342 * experimental native threads support (on x86 Linux >=2.4 only).
1343 This is not compiled in by default: you need to add :SB-THREAD to
1344 the target features. See the "Beyond ANSI" chapter of the manual
1346 * fix for longstanding nonANSIism: The old distinction between
1347 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1348 The return value from CL:FIND-CLASS is now a CLOS class, and
1349 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1350 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1351 * An interface to the MetaObject Protocol, as described in Kiczales,
1352 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1353 MIT Press, 1991, is available from the SB-MOP package.
1354 * incompatible change: the SB-PCL package should now be considered
1355 a private implementation detail, and no longer a semi-private MOP
1357 * minor incompatible change: due to rearrangement for threads, the
1358 control stack and binding stack are now allocated at arbitrary
1359 addresses instead of being hardcoded per-port. Users affected by
1360 this probably have to be doing advanced things with shared
1361 libraries, and will know who they are.
1362 * minor incompatible change: Previously, all --eval forms used were
1363 processed with READ before any of them were processed with EVAL.
1364 Now each --eval form is processed with both READ and EVAL before
1365 the next --eval form is processed. (Thus package operations like
1366 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1367 work as the user might reasonably expect.)
1368 * minor incompatible change: *STANDARD-INPUT* is now only an
1369 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1371 * minor incompatible change: Y-OR-N-P is now character-oriented,
1372 not line oriented. Also, YES-OR-NO-P now works without errors.
1373 (thanks to Antonio Martinez)
1374 * sb-aclrepl module improvements: an integrated inspector, added
1375 repl features, and a bug fix to :trace command.
1376 * Known functions, which cannot be open coded by the backend, are
1377 considered to be able to check types of their arguments. (fixing
1378 a bug report by Nathan J. Froyd)
1379 * fixed a bug in computing method discriminating functions: It is
1380 now possible to define methods specialized on classes which have
1381 forward-referenced superclasses. (thanks to Gerd Moellmann)
1382 * fixed evaluation order in optional entries (reported by Gilbert
1384 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1385 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1386 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1387 required-by-AMOP NAME argument, as well as CLASS and
1388 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1389 * fixed bug 20: DEFMETHOD can define methods using names that are
1390 not the proper names of classes to designate class specializers.
1391 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1392 Unix file descriptors, instead of blowing up. (thanks to Antonio
1394 * Garbage collection refactoring: user-visible change is that a
1395 call to the GC function during WITHOUT-GCING will not do garbage
1396 collection until the end of the WITHOUT-GCING. If you were doing
1397 this you were probably losing anyway.
1398 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1399 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1400 TYPEP the latter but not the former.
1401 * The compiler issues a full WARNING for calls to undefined functions
1402 with names from the CL package.
1403 * MAP-INTO for a vector destination is open coded. (reported by
1404 Brian Downing on c.l.l)
1405 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1406 documentation string.
1407 * fixed some bugs revealed by Paul Dietz' test suite:
1408 ** COPY-ALIST now signals an error if its argument is a dotted
1410 ** Condition slots are now accessed more correctly in the presence
1411 of multiple initargs for a given slot.
1412 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1413 exclude restarts of the same name associated with a different
1415 ** DEFCLASS of forward-referenced classes with another
1416 forward-referenced class in the superclasses list no longer
1418 ** Condition slots are now initialized once each, not multiple
1419 times. (thanks to Gerd Moellmann)
1420 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1422 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1423 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1424 arguments contain duplicated elements.
1425 ** RESTART-CASE understands local macros.
1426 ** RESTART-CASE associates exactly its own restarts with a condition.
1427 ** ENDP in safe mode checks its argument to be of type LIST.
1428 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1429 in question is unbound.
1430 ** Optimized MAKE-INSTANCE functions no longer cause internal
1431 assertion failures in the presence of duplicate initargs.
1432 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1433 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1434 propagated even in the ctor optimized implementation of
1436 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1438 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1439 * a better implementation of SXHASH on (simple) bit vectors,
1440 measured both in execution speed and in distribution of results
1441 over the positive fixnums, has been installed. Likewise, a better
1442 implementation of EQUAL for simple bit vectors is now available.
1443 * fixed CEILING optimization for a divisor of form 2^k.
1444 * fixed bug 240 (emitting extra style warnings "using the lexical
1445 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1446 by Antonio Martinez)
1447 * fixed SXHASH, giving different results for NIL depending on type
1448 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1449 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1450 variable. (found by Rolf Wester)
1451 * SBCL does not ignore type declarations for special
1452 variables. (reported by rif on c.l.l 2003-03-05)
1453 * some bug fixes in contrib/sb-aclrepl/
1454 * fixed some bugs revealed by Paul Dietz' test suite:
1455 ** a bug in the CONS type specifier, whereby the CAR and CDR
1456 types got intertwined, has been fixed;
1457 ** the type system is now able to reason about the interaction
1458 between INTEGER and RATIO types more completely;
1459 ** APPEND, [N]REVERSE and NRECONC check that those their
1460 arguments, which must be proper lists, are really so;
1461 ** An array specialized to be unable to hold elements has been
1462 implemented, as required -- yes, really -- by ANSI;
1463 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1464 on malformed property lists;
1466 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1467 * incompatible packaging change: in line with Unix convention,
1468 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1469 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1470 * REQUIRE and PROVIDE are now optionally capable of doing something
1471 useful. See the documentation string for REQUIRE.
1472 * infrastructure for a managed SBCL contrib system: contributed
1473 modules in this release include:
1474 ** the ASDF system definition facility;
1475 ** an interface to the BSD Sockets API;
1476 ** an ACL-like convenience interface to the repl;
1477 (thanks to Kevin Rosenberg)
1478 ** an implementation of ROTATE-BYTE, with efficient implementation
1480 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1481 gives a closer approximation to the right answer than previously.
1482 (thanks to Raymond Toy)
1483 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1484 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1485 optional environment arguments, as required by ANSI.
1486 * fixed bugs in other functions taking environment objects, allowing
1487 calls with an explicit NIL environment argument to be compiled
1489 * fixed bug 228: primary return values from
1490 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1491 COMPILE or FUNCTION.
1492 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1493 :TYPE will no longer signal errors on innocuous objects.
1494 * fixed bug 231b: SETQ is better at respecting type declarations in
1495 the lexical environment.
1496 * fixed a bug in DEFCLASS: classes named by symbols with no or
1497 unprintable packages can now be defined.
1498 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1499 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1500 * fixed compiler failure related to checking types of functions.
1501 (reported by Robert E. Brown)
1502 * the compiler is now much more consistent in its error-checking
1503 treatment of bounding index arguments to sequence functions: in
1504 (SAFETY 3) code, errors will be signalled in almost all cases if
1505 invalid sequence bounding indices are passed to functions defined
1506 by ANSI to operate on sequences.
1507 * fixed a bug in the build procedure: documentation of SBCL-specific
1508 packages is now preserved and available in the final Lisp image.
1509 * lifted FDEFINITION lookup out of loops in the implementation of
1510 many list operations. (thanks to Robert E. Brown)
1511 * fixed a bug in the reader: the #n# reader macro now works for
1512 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1513 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1514 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1515 E. Brown and Tony Martinez respectively)
1516 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1517 SLOT-BOUNDP: in particular, it is now able to optimize them much
1518 better, and is now not vulnerable to having packages renamed.
1519 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1520 always called when a slot is not present in an instance. (thanks
1522 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1523 longer signal an error on unreading a character following EOF on
1524 the previous constituent stream. (thanks to Tony Martinez)
1525 * fixed some bugs revealed by Paul Dietz' test suite:
1526 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1527 not just nonnegative fixnums;
1528 ** the logical bit-array operators such as BIT-AND now accept an
1529 explicit NIL for their "opt-arg" argument (to indicate a
1530 freshly-consed result bit-array);
1531 ** ELT now signals an error on an invalid sequence index in safe
1533 ** the type system is now cleverer about negations of numeric
1534 types, and consequently understands the BIGNUM and RATIO types
1536 ** the type system is now cleverer about the interaction between
1537 INTEGER and RATIO types: while bugs still remain, many more
1538 cases are accurately computed;
1539 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1540 if it is in the last clause;
1541 ** CONSTANTLY now correctly returns a side-effect-free function in
1543 ** DECLARE is no longer treated as a special-operator; in
1544 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1545 * incremented fasl file version number due to the change in the
1546 DEFSTRUCT-SLOT-DESCRIPTION structure.
1548 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1549 * minor incompatible change: code processed by the "interpreter" or
1550 EVAL now has a compilation optimization policy of (DEBUG 2)
1551 (changed from (DEBUG 1)) to improve debuggability of interactive
1552 development, and to allow the use of the debug RETURN command in
1554 * an experimental implementation of the RETURN command for the
1555 debugger has been included. (thanks to Frederik Kuivinen)
1556 * fixed bug 62: constraints were not propagated into a loop.
1557 * fixed bug in embedded calls of SORT (reported and investigated by
1559 * fixed some bugs revealed by Paul Dietz' test suite:
1560 ** printing and reading of arrays with some dimensions having
1561 length 0 (thanks to Gerd Moellmann);
1562 ** BOA constructor with &AUX argument without a default value does
1563 not cause a type error;
1564 ** CONSTANTP now returns true for all self-evaluating objects.
1566 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1567 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1568 accessors that are related by inheritance, as specified in the
1569 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1570 Valtteri Vuorikoski)
1571 * The compiler is now able to inline functions that were defined in
1572 a complex lexical environment (e.g. inside a MACROLET).
1573 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1575 * Support for the upcoming FreeBSD-5.0 release has been included.
1576 (thanks to Dag-Erling Smorgrav)
1577 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1578 effect when it is not in a toplevel context.
1579 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1580 stand a better chance of being correct. (thanks to Gerd
1582 * fixed bug in COERCE, which now signals an error on coercing a
1583 rational to a bounded real type which excludes the expected
1585 * The compiler is now able to derive types more accurately from the
1586 COERCE and COMPILE functions.
1587 * fixed bug 223: functional binding is considered to be constant
1588 only for symbols in the CL package.
1589 * fixed bug 231: SETQ did not check the type of a variable being set
1590 (reported by Robert E. Brown)
1591 * A new optimization for MAKE-INSTANCE has been included, fixing
1592 various bugs (including relating to :ALLOCATION :CLASS slots and
1593 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1595 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1596 ** As required by ANSI, LOOP now disallows anonymous collection
1597 clauses such as COLLECT I in conjunction with aggregate boolean
1598 clauses such as THEREIS (= I 1);
1599 ** LOOP now signals an error when any variable is reused in the
1600 same loop (including the potentially useful construct analogous
1601 to WITH A = 1 WITH A = (1+ A);
1602 ** IT is only a special loop symbol within the first clause of a
1603 conditional loop clause;
1604 ** LOOP with a typed iteration variable over a hashtable now
1605 signals a type error iff it should.
1606 * fixed some other bugs revealed by Paul Dietz' test suite:
1607 ** FILE-STREAM now names the class previously known as FD-STREAM;
1608 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1609 argument) no longer signals an error;
1610 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1611 of the slot symbol, rather than using the current package
1612 ((:CONC-NAME "") continues to intern the slot's name in the
1614 * incremented fasl file version number, because of the incompatible
1615 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1616 of the new implementation of DEFINE-COMPILER-MACRO.
1618 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1619 * Support for building SBCL for MIPS platforms running in
1620 little-endian mode has now been checked in, and basic
1621 functionality on said platforms verified.
1622 * minor incompatible change: PCL now records the pathname of a file
1623 in which methods and the like are defined, rather than its
1625 * minor incompatible change: TRUENAME now considers the truename of
1626 a file naming a directory to be the pathname with :DIRECTORY
1627 component indicating that directory.
1628 * minor incompatible change: a NAMED clause in the extended form of
1629 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1630 reason for the previous behaviour is unclear.
1631 * more systematization and improvement of CLOS and MOP conformance
1632 in PCL (thanks to Gerd Moellman and Pierre Mai):
1633 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1635 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1636 primary methods with no specializers;
1637 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1639 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1640 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1641 FORWARD-REFERENCED-CLASSes; error reporting on
1642 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1644 ** SXHASH on CLOS instances now uses a slot internal to the
1645 instance to return different numbers on distinct instances,
1646 while preserving the same return value through invocations of
1648 ** DEFMETHOD signals errors when methods with longer incongruent
1649 lambda lists are added to generic functions;
1650 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1651 CLASS, as specified in AMOP;
1652 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1653 on the order returned by the primary method for classes of
1654 class STANDARD-CLASS;
1655 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1656 * fixed some bugs shown by Paul Dietz' test suite:
1657 ** DOLIST puts its body in TAGBODY;
1658 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1660 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1661 value producing form;
1662 ** if more variables are given to PROGV than values, extra
1663 variables are bound and made to have no value;
1664 ** NSUBSTITUTE on list arguments gets the right answer with
1666 ** ELT signals an error of type TYPE-ERROR when the index argument
1667 is not a valid sequence index;
1668 ** LOOP signals (at macroexpansion time) an error of type
1669 PROGRAM-ERROR when duplicate variable names are found;
1670 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1671 ** FUNCALL of special-operators now cause an error of type
1673 ** PSETQ now works as required in the presence of side-effecting
1674 symbol-macro places;
1675 ** NCONC accepts any object as its last argument;
1676 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1678 ** loop-for-as-package does not require a package to be explicitely
1680 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1682 * fixed bug 166: compiler preserves "there is a way to go"
1683 invariant when deleting code.
1684 * fixed bug 172: macro lambda lists with required arguments after
1685 &REST arguments now cause an error to be signalled. (thanks to
1687 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1688 bug. (thanks to Matthew Danish)
1689 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1691 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1692 arguments are lexically rebound. (thanks to Gerd Moellmann and
1694 * fixed bug 194: error messages are now more informative when there
1695 is no primary method applicable in a call to a generic
1696 function. (thanks to Gerd Moellmann)
1697 * fixed bug in command line argument checking (thanks to Julian
1699 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1700 (thanks to Matthew Danish)
1701 * incremented fasl file version number, because of the
1702 SXHASH-related changes in the layout of CLOS data structures
1704 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1705 * minor incompatible change: The runtime (the Unix executable named
1706 "sbcl") is now much pickier about the .core files it will load.
1707 Essentially it now requires .core files to descend from the same
1708 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1709 as the runtime does. (The intent is to prevent the crashes which
1710 can occur, and which can even be reported as mysterious failures,
1711 when people patch the sources or change the build parameters
1712 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1713 sbcl and .core files.)
1714 * fixed bug: VALUES-LIST is no longer optimized away.
1715 * fixed bug 142: The FFI conversion of C string values to Lisp
1716 string values no longer conses excessively. (thanks to Nathan
1717 Froyd porting Raymond Toy's fix to CMU CL)
1718 * began to systematize and improve MOP conformance in PCL (thanks to
1719 Nathan Froyd, Gerd Moellman and Pierre Mai):
1720 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1722 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1723 ** FINALIZE-INHERITANCE is now called on class finalization;
1724 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1725 argument precedence order.
1726 * fixed bug 202: The compiler no longer fails on functions whose
1727 derived types contradict their declared type.
1728 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1729 so it can be non-toplevel.
1730 * The fasl file version number has changed (because of the new
1731 implementation of DEFMACRO).
1732 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1733 safe code, that any length requirement by their type-specifier
1734 argument is valid. The exceptions to this are described in bug
1736 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1737 functions now check the types of their inputs as required by ANSI.
1738 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1739 introduced symbol is DECLAREd to be SPECIAL.
1740 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1741 * fixed bug: :COUNT argument to sequence functions may be negative
1742 * fixed bug: body of DO-SYMBOLS may contain declarations
1743 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1744 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1745 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1746 symbol macro only once
1747 * fixed printing of call frame when argument list is unavailable
1748 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1749 * compiler no longer signals WARNING on unknown keyword
1752 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1753 * A beta-quality port to the mips architecture running Linux,
1754 based on the old CMUCL backend, has been made. It has been tested
1755 on a big-endian kernel, and works sufficiently well to be able to
1756 rebuild itself; it has not been tested in little-endian mode.
1757 * fixed an inconsistency between gencgc.c and purify.c which made
1758 dumping/loading .core files unreliable
1759 * fixed bug 120a: The compiler now deals correctly with IFs where
1760 the consequent is the same as the alternative, instead of
1761 misderiving the return type. (thanks to Alexey Dejneka)
1762 * fixed bug 113: Logical pathnames are now dumpable (the logical
1763 host is resolved at load-time, throwing an error if it is not
1765 * fixed bug 174: FORMAT's error message is slightly clearer when a
1766 non-printing character is used in a format directive.
1767 * fixed several bugs in compiler checking of type declarations, i.e.
1768 violations of the Python "declarations are assertions" principle
1769 (thanks to Alexey Dejneka)
1770 * fixed several bugs in PCL's error checking (thanks to Gerd
1772 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1774 * fixed bug in compilation of functions as first class values
1775 (thanks to Antonio Martinez-Shotton)
1776 * The compiler's handling TYPE-ERRORs which it can prove will
1777 inevitably happen at runtime has been cleaned up and corrected
1778 in several ways. (thanks to Alexey Dejneka)
1779 * improved argument type checking for various basic arithmetic
1780 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1781 lavished on them in the past that they can be compiled in many
1782 ways in different special cases
1783 * fixed bug 181: compiler checks validity of user supplied type
1785 * cleaned up code flushing in optimization: Function calls which
1786 should signal errors for safety purposes (e.g. which ANSI says
1787 should signal errors when their arguments are of incorrect type)
1788 are no longer optimized away.
1789 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1790 * incremented fasl file version number, because changes in the
1791 implementation of sequence functions like COERCE caused
1792 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1793 in old inline expansions) to become undefined. (Actually these
1794 changes were later undone, so we might very well be binary
1795 compatible with 0.7.7 after all, but leaving the version number
1796 incremented seemed like the simplest and most conservative
1799 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1800 * An alpha-quality port to the parisc architecture running Linux,
1801 based on the old CMUCL backend, has been made. This, even more so
1802 than the other backends, should be considered still a work in
1803 progress; known problems include that the Linux kernel in 64-bit
1804 mode does not propagate the correct sigcontext structure to
1805 userspace, and consequently SBCL on a parisc64 kernel will not
1807 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1808 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1809 suppressed.) Also now that the compiler is looking at declarations
1810 in the environment, it checks optimization declarations as well,
1811 and suppresses inlining when (> DEBUG SPEED).
1812 * More fixes have been made to treatment of floating point exception
1813 treatment and other Unix signals. In particular, floating point
1814 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1815 * The detection and handling of control stack exhaustion (infinite
1816 or very deeply nested recursion) has changed. Stack exhaustion
1817 detection is now done by write-protecting pages at the OS level
1818 and applies at all optimization settings; when found, a
1819 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1820 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1821 be caught using IGNORE-ERRORS.
1822 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1823 correctly when transforming recursive LABELS functions to LETs.
1824 (thanks to Alexey Dejneka)
1825 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1826 that are names of constants or global variables.
1827 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1828 alien routines with docstrings.
1829 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1830 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1832 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1833 (thanks to Eric Marsden)
1834 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1835 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1836 * Bug fix: The SPARC backend can now compile functions involving
1837 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1838 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1839 to a functions expecting a PATHNAME-DESIGNATOR.
1840 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1841 lambda lists. (thanks to Alexey Dejneka)
1842 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1843 functions so that EXPORTing the name of the class doesn't cause
1844 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1845 to Antonio Martinez for reporting this)
1846 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1847 constant-folded without failing an assertion. (thanks to Einar
1848 Floystad Dorum for reporting this)
1849 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1850 types are now dealt with more correctly by the compiler.
1851 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1852 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1853 behaviour analogous to RENAME-FILE. This puts its behaviour more
1854 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1856 * The fasl file version number has changed again. (because of the
1857 bug fix involving the names of PCL MAKE-INSTANCE functions)
1859 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1860 * bug fix: Floating point exceptions are treated much more
1861 consistently on the x86/Linux and PPC/Linux platforms.
1862 * Array initialization with :INITIAL-ELEMENT is now much faster for
1863 cases when the compiler cannot open code the array creation, but
1864 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1865 array accesses have also seen a speed increase.
1866 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1867 specified. (This was at the root of some bad interactions between
1868 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1870 * bug fix: Internal error arguments for undefined functions are now
1871 computed correctly on the PPC/Linux platform.
1872 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1873 Raymond Toy's patch for CMU CL)
1874 * Support for the Solaris 9 operating environment has been included
1875 (thanks to Daniel Merritt)
1876 * A very ugly but hopefully complete draft of the missing FFI chapter
1877 of the manual has been created by reformatting the corresponding
1878 CMU CL manual chapter into (currently very ugly and incoherent)
1879 DocBook and bringing it up to date for SBCL behavior. Thus, the
1880 manual is now essentially complete, at least by my extreme
1881 once-and-only-once standards, whereby it's acceptable to refer to
1882 the doc strings of SB-EXT functions as the primary documentation.
1883 * The fasl file version number has changed again, due to cleanup of
1884 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1885 is no longer a static symbol.)
1887 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1888 * SBCL now builds with OpenMCL (version 0.12) as the
1889 cross-compilation host; also, more progress has been made toward
1890 bootstrapping under CLISP.
1891 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1893 * bug 158 fixed: The compiler can now deal with integer loop
1894 increments different from 1; fixing this turned out also to fix
1896 * bug 169 fixed: no more bogus warnings about using lexical bindings
1897 despite the presence of perfectly good SPECIAL declarations (thanks
1898 to David Lichteblau)
1899 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1900 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1901 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1903 * bug 180 fixed: Method combination specifications no longer ignore
1904 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1905 * bug fix: Structure type predicate functions now check their argument
1906 count as they should.
1907 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1908 correctly. (thanks to Pierre Mai)
1909 * minor incompatible change: The --noprogrammer option is deprecated
1910 in favor of the new --disable-debugger option, which is very similar.
1911 (The major difference is that it takes effect at a slightly different
1912 time at startup, causing handling of errors in --sysinit and
1913 --userinit files will be affected differently.) The
1914 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1915 been added to allow this functionality to be controlled from ordinary
1916 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1917 maintainers, who might want to run non-interactive scripts to
1918 build SBCL cores which will later be used interactively.)
1919 * minor incompatible change: The LOAD function no longer, when given
1920 a wild pathname to load, loads all files matching that pathname.
1921 Instead, an error of type FILE-ERROR is signalled.
1923 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1924 * bug 147 fixed: The compiler preserves its block link/count
1925 invariants more correctly now so that it doesn't crash. (thanks
1927 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1929 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1930 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1931 no-value-for-this-character as NIL (as specified by ANSI).
1932 * HOST-NAMESTRING on physical pathnames now returns a string that is
1933 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1934 (thanks to Christophe Rhodes)
1935 * The Alpha port handles icache flushing more correctly. (thanks to
1937 * More progress has been made toward bootstrapping under CLISP. (thanks
1938 to Christophe Rhodes)
1939 * The fasl file format has changed again, because dynamic loading
1940 on OpenBSD (which has non-ELF object files) motivated some cleanups
1941 in the way that foreign symbols are transformed and passed around.
1942 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1943 is no longer treated as whitespace by the reader, but instead as
1944 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
1945 |A
\7fB|, instead of A as it used to.
1947 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1948 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1949 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1950 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1951 this as of 0.7.1.45, but was left out of the previous news section
1952 (thanks to Dan Barlow)
1953 * SBCL now runs on the Solaris operating system on SPARC architectures
1954 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1955 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1956 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1957 * SPARC backend cleanups, allowing builds of cores optimized for V8
1958 and V9 SPARCS, and also emission of code targeted to a particular
1959 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1961 * SBCL is closer to bootstrapping under CLISP, thanks to various
1962 fixes by Christophe Rhodes.
1963 * The fasl file format has changed again, to allow the compiler's
1964 INFO database to support symbol macros.
1965 * The user manual (in doc/) is formatted into HTML more nicely.
1966 (thanks to coreythomas)
1967 * The system is smarter about SUBTYPEP relationships, especially
1968 those involving NOT types (including types such as ATOM which are
1969 represented internally using NOT types). Thus SUBTYPEP is less
1970 likely to return (VALUES NIL NIL) in general, and in particular
1971 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1973 * The fasl file format has changed again, because the internal
1974 representation of types now includes a new slot to support the new
1975 SUBTYPEP-of-NOT-types logic.
1976 * (not a change in the main branch of SBCL, but a related prototype
1977 which can hopefully be merged into the main branch of SBCL in the
1978 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1979 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1980 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1981 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1982 dispatch macro character combinations. (thanks to Alexey Dejneka)
1983 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1984 that superficially look like logical namestrings correctly.
1985 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1987 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1988 default supplied-p) for &optional and &key arguments. (thanks to
1991 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1992 * incompatible change: The compiler is now less aggressive about
1993 tail call optimization, doing it only when (> SPACE DEBUG) or
1994 (> SPEED DEBUG). (This is an incompatible change because there are
1995 programs which relied on the old CMU-CL-style behavior to optimize
1996 away their unbounded recursion which will now die of stack overflow.)
1997 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1998 for non-GENCGC systems has been increased to 20M (since that
1999 seems much closer to the likely performance optimum for modern
2000 systems than the old 4M value was)
2001 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2002 and *DEBUG-PRINT-LEVEL*
2003 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2004 of CMU CL's support for SPARC, and various endianness and other
2005 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2006 * new syntactic sugar for the Unix command line: --load foo.bar is now
2007 an alternate notation for --eval '(load "foo.bar")'.
2009 ** The system now detects stack overflow and handles it gracefully,
2010 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2011 optimization settings. (This is a good thing in general, and
2012 its introduction in this version should be particularly timely
2013 for anyone whose code fails because of suppression of tail
2015 ** The system now hunts for the C variable "environ" in a more
2016 devious way, to avoid segfaults when the C library version differs
2017 between compile time and run time. (thanks to Christophe Rhodes)
2018 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2019 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2020 ** The compiler no longer issues bogus style warnings for undefined
2021 classes in the same source file as the DEFCLASSes which defined
2022 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2024 ** fixes in CONDITION class precedence list for undefined function
2025 errors (thanks to Alexei Dejneka)
2026 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2027 correctly. (thanks to Dan Barlow)
2028 ** portability fixes aiming at bootstrapping under CLISP (thanks
2029 to Dave McDonald and Christophe Rhodes)
2030 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2031 ** fixes in type translation and and type inference (thanks to
2033 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2034 ** various fixes in the new ports (thanks to Dan Barlow)
2035 * several changes related to debugging:
2036 ** suppression of tail recursion, as noted above
2037 ** stack overflow detection, as noted above
2038 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2039 is now the default. (For some time encapsulation has been more
2040 reliable than the breakpoint-based :ENCAPSULATE NIL
2041 implementation, at least on X86 systems; and I just noticed that
2042 encapsulation also seems closer to the spirit of the ANSI
2045 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2047 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2048 up properly again. (There was a packaging bug in 0.7.0 which
2049 left their definitions in SB-SYS::LOAD-FOREIGN and
2050 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2051 vital for most things which interface to C-level interfaces,
2052 like extensions working with sockets or databases or
2053 Perl-compatible regexes or whatever, and the need to fix
2054 this bug is the main reason that 0.7.1 was released so
2056 ** DEFGENERIC is now choosier about the methods it redefines, so that
2057 reLOADing a previously-LOADed file containing DEFGENERICs does
2058 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2059 cycle now works as it should. (thanks to Alexey Dejneka)
2060 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2061 fixed by Christophe Rhodes. (He actually submitted this patch
2062 months ago, and I delayed until after 0.7.0.)
2063 ** Bug 111 (internal compiler confusion about runtime checks on
2064 FUNCTION types) was fixed by Alexey Dejneka.
2065 * Some internal cleanups (getting rid of variables which aren't
2066 needed now that the byte interpreter is gone) caused the fasl
2067 file format number to change again.
2069 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2070 * major incompatible change: The default fasl file extension, i.e. the
2071 default extension for files produced by COMPILE-FILE, has changed
2072 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2074 ** There are many changes in the implementation of the compiler.
2075 SBCL is now essentially a compiler-only implementation of ANSI
2076 Common Lisp. EVAL still "interprets" a few special cases, but
2077 almost all the interesting cases are handled by creating
2078 a LAMBDA expression, calling COMPILE on it, then calling
2079 FUNCALL on the result.
2080 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2081 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2082 Since the code is newer, there might still be some new bugs
2083 (though not as many as before Martin Atzmueller's fixes:-). But
2084 the new code is substantially simpler and clearer, and hopefully
2085 any remaining bugs will be simpler, less fundamental, and more
2086 fixable then the bugs in the old code.
2087 ** The revised compiler is still a little unsteady on its feet.
2089 *** The debugging information it produces (particularly the names
2090 of FUNCTION objects) is sometimes much less useful than what
2091 the old compiler produced.
2092 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2093 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2094 within a MACROLET) has been temporarily weakened.
2095 ** There are new compiler optimizations for various functions:
2096 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2097 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2098 *** the math functions TRUNCATE, FLOOR, and CEILING
2099 *** the function-of-all-trades COERCE
2100 Mostly these should be transparent, but there's one
2101 potentially-annoying problem (bug 117): when the compiler
2102 inline-expands a function and does type analysis on the result,
2103 it can create control paths which have type mismatches, and
2104 when it can't prove that those control paths aren't taken,
2105 it will issue WARNINGs about the type mismatches. This is
2106 a particular problem in practice for the new sequence functions.
2107 It's not clear how this should be fixed, and for now, a
2108 workaround is given in the entry for 117 in the BUGS file.
2109 ** (Because of the interaction between the two previous items --
2110 occasional inlining problems and new inline expansions -- some
2111 of the new sequence function optimizations won't really kick in
2112 completely until debugging information, and then inlining, are
2113 straightened out in some future version.)
2114 * minor incompatible changes:
2115 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2116 DIRECTORY no longer implicitly promotes NIL slots of its
2117 pathname argument to :WILD. In particular, when you ask for the
2118 contents of a directory (which you used to be able to do without
2119 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2120 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2121 ** changes in behavior that ANSI explicitly defines to be
2122 implementation dependent:
2123 *** The new compiler-only implementation still conforms with ANSI,
2124 but acts a little different than before. Besides the obvious
2125 changes in performance tradeoffs (that the cost per form passed
2126 to EVAL has gone up, and the cost per form executed by EVAL
2127 has gone down), the behavior of the system changes a little
2128 because there are no longer any interpreted function objects.
2129 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2130 e.g. doing COMPILE on the output of interactive DEFUN is
2132 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2134 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2135 has changed from (:CL) to NIL.
2136 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2137 e.g. #\Newline and #\Tab, don't have names specified in the
2138 ANSI Common Lisp standard, is now based on their ASCII symbolic
2139 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2140 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2141 are no longer used for output.
2142 ** changes in internal implementation constants:
2143 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2144 4 million. (If your application spends a lot of time GCing and
2145 you have a lot of RAM, you might want to experiment with
2146 increasing it even more.)
2147 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2148 However, almost all old code should still continue to work without
2149 immediate update, as SB-C-CALL is now a (deprecated) nickname
2151 ** Old operator names in the style DEF-FOO are now deprecated in
2152 favor of new corresponding names DEFINE-FOO, for consistency with
2153 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2154 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2155 internal symbols, but a few supported extensions like
2156 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2157 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2158 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2159 etc. as you get deeper into recursive calls to the debugger
2160 command loop, instead of the old "5]", "5]]", "5]]]"
2161 sequence. (I was motivated to do this when squabbles between
2162 ILISP and SBCL left me very deeply nested in the debugger. In the
2163 short term, this change will probably provoke more ILISP/SBCL
2164 squabbles, but hopefully it will be an improvement in the long run.)
2165 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2166 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2167 function, since declaiming return types involving aliens is
2168 (1) annoyingly messy to do by hand and (2) vital to efficient
2169 compilation of code which calls such functions.
2170 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2171 longer reexported by the SB-EXT package. They're solely useful
2172 for alien code, so it seems more logical that you should get
2173 them from the SB-ALIEN package, not in SB-EXT.
2174 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2175 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2176 features. Instead, the code that they used to control is always
2177 built into the system.
2178 * many other bug fixes
2179 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2180 advantage of the new EVAL-WHEN stuff and to clean them up in
2181 general, and they are now more ANSI-compliant in a number of
2182 ways. Martin Atzmueller is responsible for a lot of this.
2183 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2185 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2186 *** correct ERROR type for various file operations
2187 *** some fixes for Lisp streams
2188 *** DEFMETHOD syntax checking
2189 *** changing old weird representation of debug information as
2190 strings (which, among their other deficiencies, don't transform
2191 correctly when you rename packages, and don't change their
2192 print representation when you change things like *PACKAGE*
2193 and *PRINT-LENGTH*) to symbols and lists of symbols
2194 He also made several improvements and fixed several bugs in DESCRIBE.
2195 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2197 *** misbehavior of WRITE-STRING/WRITE-LINE
2198 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2199 and several other LOOP problems as well
2200 *** DIRECTORY when similar filenames are present
2201 *** DEFGENERIC with :METHOD options
2202 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2203 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2204 *** argument ordering in FIND with :TEST option
2205 *** mishandled package designator argument in APROPOS-LIST
2206 *** various problems in the backquote readmacro
2207 *** a bug in APROPOS
2208 *** probably some others that I'm not describing very well here,
2209 since the CVS log documents them by reference to sbcl-devel
2210 messages, and the SourceForge archives aren't working well.:-(
2211 ** Dan Barlow improved the Alpha port (and is making progress on the
2212 PPC port, for those of you who think different).
2213 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2214 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2215 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2216 continues to work on the SPARC port (for those of you in a position
2217 to look down upon our little PC-compatible boxes from a great height).
2218 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2219 and uses that copy, rather than the current dynamic value, when
2220 it's trying to decide whether to truncate output. Thus e.g.
2221 (let ((*print-lines* 50))
2222 (pprint-logical-block (stream nil)
2224 (let ((*print-lines* 8))
2225 (print (aref possiblybigthings i) stream)))))
2226 should now truncate the logical block only at 50 lines, instead of
2227 often truncating it at 8 lines, as it did before.
2228 * The doc/cmucl/ directory, containing old CMU CL documentation
2229 from the time of the fork, is no longer part of the base system.
2230 SourceForge has shut down its anonymous FTP service, and with it
2231 my original plan for distributing the old CMU CL documentation
2232 there. For now, if you need these files you can download an old
2233 SBCL source release and extract them from it.
2234 * The fasl file version number changed again, for dozens of reasons,
2235 some of which are apparent above.
2237 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2238 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2239 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2240 patch, so that DISASSEMBLE output is much nicer.
2241 * The code in the SB-PROFILE package now seems reasonably stable.
2242 I still haven't decided what the final interface should look like
2243 (I'd like PROFILE to interact cleanly with TRACE, since both
2244 facilities use function encapsulation) but if you have a need
2245 for profiling now, you can probably use it successfully with
2246 the current CMU-CL-style interface.
2247 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2248 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2249 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2250 link now returns the dangling link itself, and for similar
2251 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2252 link itself. (In these cases the old code signalled an error and
2253 looped endlessly, respectively.) Thus, DIRECTORY now works even
2254 in the presence of dangling and cyclic symbolic links.
2255 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2256 is now a supported extension again, since the consensus on
2257 sbcl-devel was that it can be useful for ordinary development
2258 work, not just for debugging SBCL itself.
2259 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2260 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2261 that currently-defined functions might be redefined later with
2262 different return types.
2263 * Hash tables can be printed readably, as inspired by CMU CL code
2264 of Eric Marsden and SBCL code of Martin Atzmueller.
2265 * better error handling in CLOS method combination, thanks to
2266 Martin Atzmueller porting Pierre Mai's CMU CL patches
2267 * more overflow fixes for >16Mbyte I/O buffers
2268 * A bug in READ has been fixed, so that now a single Ctrl-D
2269 character suffices to cause end-of-file on character streams.
2270 In particular, now you only need one Ctrl-D at the command
2271 line (not two) to exit SBCL.
2272 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2274 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2275 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2276 does the right thing.
2277 * The implementation of some type tests, especially for CONDITION
2278 types, is now tidier and maybe faster, due to CMU CL code
2279 originally by Douglas Crosher, ported by Martin Atzmueller.
2280 * Some math functions have been fixed, and there are new
2281 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2282 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2283 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2284 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2285 likely to be built into sbcl-0.7.0.)
2286 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2287 CL, ported to SBCL by Martin Atzmueller)
2288 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2289 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2290 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2291 * There's a new slam.sh hack to shorten the edit/compile/debug
2292 cycle for low-level changes to SBCL itself, and a new
2293 :SB-AFTER-XC-CORE target feature to control the generation of
2294 the after-xc.core file needed by slam.sh.
2295 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2296 COMPILE-FILE is no longer supported, so that now every function
2297 gets an entry point, so that block compilation looks a little
2298 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2299 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2300 returns the number of bytes consed since the system started,
2301 rather than the number consed since the first time the function
2302 was called. (The new definition parallels ANSI functions like
2303 CL:GET-INTERNAL-RUN-TIME.)
2304 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2305 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2306 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2307 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2308 * Fasl file version numbers are now independent of the target CPU,
2309 since historically most system changes which required version
2310 number changes have affected all CPUs equally. Similarly,
2311 the byte fasl file version is now equal to the ordinary
2314 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2315 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2316 is no longer recognized. I apologize for this, because it was
2317 listed in SB-EXT as a supported extension, but I found that
2318 its existing behavior was poorly specified, as well as incorrectly
2319 specified, and it looked like too much of a mess to straighten it
2320 out. I have enough on my hands trying to get ANSI stuff to work..
2321 * many patches ported from CMU CL by Martin Atzmueller, with
2322 half a dozen bug fixes in pretty-printing and the debugger, and
2323 half a dozen others elsewhere
2324 * fixed bug 13: Floating point infinities are now supported again.
2325 They might still be a little bit flaky, but thanks to bug reports
2326 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2327 as flaky as they were.
2328 * The --noprogrammer command line option is now supported. (Its
2329 behavior is slightly different in detail from what the old man
2330 page claimed it would do, but it's still appropriate under the
2331 same circumstances that the man page talks about.)
2332 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2333 are now supported, and enabled by default. Thus, the compiler can
2334 handle many floating point and complex operations much less
2335 inefficiently. (Thus e.g. you can implement a complex FFT
2337 * The compiler now detects type mismatches between DECLAIM FTYPE
2338 and DEFUN better, and implements CHECK-TYPE more correctly, and
2339 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2340 thanks to patches from Martin Atzmueller.
2341 * various fixes to make the cross-compiler more portable to
2342 ANSI-conforming-but-different cross-compilation hosts (notably
2343 Lispworks for Windows, following bug reports from Arthur Lemmens)
2344 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2345 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2346 fixed thanks to Pierre Mai's CMU CL patches.
2347 * fixes to keep the system from overflowing internal counters when
2348 it tries to use i/o buffers larger than 16M bytes
2349 * fixed bug 45a: Various internal functions required to support
2350 complex special functions have been merged from CMU CL sources.
2351 (When I was first setting up SBCL, I misunderstood a compile-time
2352 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2353 * improved support for type intersection and union, fixing bug 12
2354 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2355 more obscure bugs as well
2356 * some steps toward byte-compiling non-performance-critical
2357 parts of the system, courtesy of patches from Martin Atzmueller
2358 * Christophe Rhodes has made some debian packages of sbcl at
2359 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2360 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2361 stable, but are nonetheless usable. When he's ready, I'd be happy
2362 to add them to the SourceForge "File Releases" section. (And if
2363 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2364 * new fasl file format version number (because of changes in
2365 internal representation of (OR ..) types to accommodate the new
2366 support for (AND ..) types, among other things)
2368 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2369 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2371 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2373 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2374 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2375 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2376 * DESCRIBE now works on structure objects again.
2377 * Most function call argument type mismatches are now handled as
2378 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2379 know whether the function will be redefined before the call is
2380 executed. (The compiler could flag local calls with full WARNINGs,
2381 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2382 it doesn't keep track of enough information to know whether calls
2383 are local in this sense.)
2384 * Compiler output is now more verbose, with messages truncated
2385 later than before. (There should be some supported way for users
2386 to override the default verbosity, but I haven't decided how to
2387 provide it yet, so this behavior is still controlled by the internal
2388 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2389 src/compiler/ir1util.lisp.)
2390 * Fasl file format version numbers have increased again, because
2391 support for the Gray streams extension changes the layout of the
2392 system's STREAM objects.
2393 * The Gray subclassable streams extension now works, thanks to a
2394 patch from Martin Atzmueller.
2395 * The full LOAD-FOREIGN extension (not just the primitive
2396 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2397 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2398 but like most other programs, it defaults to copying the Unix
2399 environment from the original process instead of starting the
2400 new process in an empty environment.
2401 * Extensions which manipulate the Unix environment now support
2402 an :ENVIRONMENT keyword option which doesn't smash case or
2403 do other bad things. The CMU-CL-style :ENV option is retained
2404 for porting convenience.
2405 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2406 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2408 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2410 * A patch from Martin Atzmueller seems to have solved the SIGINT
2411 problem, and as far as we know, signal-handling now works cleanly.
2412 (If you find any new bugs, please report them!)
2413 * The system no longer defaults Lisp source file names to types
2414 ".l", ".cl", or ".lsp", but only to ".lisp".
2415 * The compiler no longer uses special default file extensions for
2416 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2417 seems to expect a single default extension for all compiled code,
2418 and there's no compelling reason to try to stretch the standard
2419 to allow two different extensions.) Instead, byte-compiled files
2420 default to the same extension as native-compiled files.
2421 * Fasl file format version numbers have increased again, because
2422 a rearrangement of internal implementation packages made some
2423 dumped symbols in old fasl files unreadable in new cores.
2424 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2425 many fewer weird special cases.
2426 * Bug #17 (differing COMPILE-FILE behavior between logical and
2427 physical pathnames) has been fixed, and some related misbehavior too,
2428 thanks to a patch from Martin Atzmueller.
2429 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2430 by Tim Moore, ported to SBCL by Martin Atzmueller.
2431 * Martin Atzmueller fixed several filesystem-related problems,
2432 including bug #36, in part by porting CMU CL patches, which were
2433 written in part by Paul Werkowski.
2434 * More compiler warnings in src/runtime/ are gone, thanks to
2435 more patches from Martin Atzmueller.
2436 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2439 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2441 * DESCRIBE now works on CONDITION objects.
2442 * The debugger now handles errors which arise when trying to print
2443 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2445 * The build system now uses an additional file, customize-target-features.lisp,
2446 to allow local modifications to the target *FEATURES* list. (The point of
2447 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2448 debugging features enabled, without having to worry about propagating it
2449 into everyone's system when I do a "cvs update".) When no
2450 customize-target-features.lisp file exists, the target *FEATURES* list
2451 should be constructed the same way as before.
2452 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2453 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2454 warn about behavior which is undefined under the ANSI spec. Note: This
2455 is specified by ANSI, but it's not very popular with programmers.
2456 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2457 macro in the SBCL sources for an example of a workaround which you
2458 might use to make portable ANSI-standard code which does what you want.
2459 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2460 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2461 when it's used as a non-toplevel form. (This is required in order
2462 to implement the DEFCONSTANT-EQX macro.)
2463 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2464 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2465 funny things, due to bugs in the implementation of EVAL-WHEN
2466 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2467 non-toplevel DEFCONSTANTs are uncommon.)
2468 * The core file version number and fasl file version number have been
2469 incremented, because the old noncompliant DEFCONSTANT behavior involved
2470 calling functions which no longer exist, and because I also took the
2471 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2472 * fixed bug 1 (error handling before read-eval-print loop starts), and
2473 redid debugger restarts and related debugger commands somewhat while
2475 ** The QUIT debugger command is gone, since it did something
2476 rather different than the SB-EXT:QUIT command, and since it never
2477 worked properly outside the main toplevel read/eval/print loop.
2478 Invoking the new TOPLEVEL restart provides the same functionality.
2479 ** The GO debugger command is also gone, since you can just invoke
2480 the CONTINUE restart directly instead.
2481 ** The TOP debugger command is also gone, since it's redundant with the
2482 FRAME 0 command, and since it interfered with abbreviations for the
2484 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2486 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2487 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2488 some porting work by Martin Atzmueller)
2489 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2490 with a boatload of patches which he ported to SBCL. Now that those
2492 ** The system tries to make sure that its low-priority messages
2493 are prefixed by semicolons, to help people who like to use
2494 syntax highlighting in their ILISP buffer. (This patch
2495 was originally due to Raymond Toy.)
2496 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2497 patches originally written by Raymond Toy.
2498 ** The compiler understands coercion between single-value and
2499 multiple-VALUES type expressions better, getting rid of some very
2500 weird behavior, thanks to patches originally by Robert MacLachlan
2501 and Douglas Crosher.
2502 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2503 lambda lists, thanks to a patch originally by Pierre Mai.
2504 ** The system no longer bogusly warns about "abbreviated type
2506 ** The compiler gets less confused by inlining and RETURN-FROM,
2507 thanks to some patches originally by Tim Moore.
2508 ** The system no longer hangs when dumping circular lists to fasl
2509 files, thanks to a patch originally from Douglas Crosher.
2510 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2511 undefined function error.
2512 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2513 in structure slot initforms) for now, documented workaround instead:-|
2514 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2515 * fixed bug 5: added stubs for various Gray stream functions called
2516 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2517 installed, at least appropriate type errors are generated
2518 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2519 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2520 consistently in DEFMETHOD forms.
2521 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2522 it doesn't seem to affect SBCL after all
2523 * The C runtime system now builds with better optimization and many
2524 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2526 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2528 * The system is now under CVS at SourceForge (instead of the
2529 CVS repository on my home machine).
2530 * The new signal handling code has been tweaked to treat register
2531 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2532 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2533 downstream, e.g. in debug-int.lisp, has implicit dependencies
2534 on the unsignedness of integer representation of machine words,
2535 and that caused the system to bomb out with infinite regress
2536 when trying to recover from type errors involving signed values,
2537 e.g. (BUTLAST '(1 2 3) -1).)
2538 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2539 (This was one of the bugs Peter Van Eynde reported back in July.)
2540 * The system now uses code inspired by Colin Walters' O(N)
2541 implementation of MAP (from the cmucl-imp@cons.org mailing
2542 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2543 inline the MAP operation, and there is more than one
2544 sequence argument to the MAP call (so that it can't just
2545 do ETYPECASE once and for all based on the type of the
2546 single sequence argument). (The old non-inline implementation
2547 of the general M-argument sequence-of-length-N case required
2548 O(M*N*N) time when any of the sequence arguments were LISTs.)
2549 * The QUIT :UNIX-CODE keyword argument has been renamed to
2550 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2551 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2553 * Raymond Wiker's patches to port dynamic loading from Linux to
2554 FreeBSD have been added.
2555 * The BUGS file is now more nearly up to date, thanks in large part
2556 to Martin Atzmueller's review of it.
2557 * The debugger now flushes standard output streams before it begins
2558 its output ("debugger invoked" and so forth).
2559 * The core version number and fasl file version number have both
2560 been incremented, because of incompatible changes in the layout
2562 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2563 used to not be called for a saved Lisp image.)
2564 * Martin Atzmueller's version of a patch to fix a compiler crash,
2565 as posted on sbcl-devel 13 September 2000, has been installed.
2566 * Instead of installing Martin Atzmueller's patch for the
2567 compiler transform for SUBSEQ, I deleted the compiler transform,
2568 and transforms for some similar consing operations.
2569 * A bug in signal handling which kept TRACE from working on OpenBSD
2571 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2572 away by constant folding
2573 * The system now defines its address space constants in one place
2574 (in the Lisp sources), and propagates them automatically elsewhere
2575 (through GENESIS and the sbcl.h file). Therefore, patching the
2576 address map is less unnecessarily tedious and error-prone. The
2577 Lisp names of address space constants have also been systematized.
2578 * CVS tags like dollar-Header-dollar have been removed from
2579 the sources, because they have never saved me trouble and
2580 they've been source of trouble working with patches and other
2581 diff-related operations.
2582 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2583 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2585 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2587 * The system has been ported to OpenBSD.
2588 * The system now compiles with a simple "sh make.sh" on the systems
2589 that it's supported on. I.e., now you no longer need to tweak
2590 text in the target-features.lisp-expr and symlinks in src/runtime/
2591 by hand, the make.sh takes care of it for you.
2592 * The system is no longer so grossly inefficient when compiling code
2593 involving vectors implemented as general (not simple) vectors (VECTOR T),
2594 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2595 which dares to use the various sequence functions on non-simple
2596 vectors, takes less of a performance hit.
2597 * There is now a primitive type predicate VECTOR-T-P
2598 to test for the (VECTOR T) type, so that e.g.
2599 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2600 can now be compiled with some semblance of efficiency. (The old code
2601 turned the type declaration into a full call to %TYPEP at runtime!)
2602 * AREF on (VECTOR T) is still not fast, since it's still compiled
2603 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2604 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2605 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2606 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2607 without making full calls to SUBTYPEP at runtime.
2608 (Some analogous efficiency issues for non-simple vectors specialized to
2609 element types other than T, or for non-simple multidimensional arrays,
2610 have not been addressed. They could almost certainly be handled the
2611 same way if anyone is motivated to do so.)
2612 * The changes in array handling break binary compatibility, so
2613 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2614 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2615 instead of (VALUES T T).
2616 * By following the instructions that Dan Barlow posted to sbcl-devel
2617 on 2 July 2000, I was able to enable primitive dynamic object
2618 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2619 functionality is not implemented (since it calls ld to resolve
2620 library references automatically, requiring RUN-PROGRAM for its
2621 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2622 try to resolve library references) is now supported.
2623 * The system now flushes the standard output streams when it terminates,
2624 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2625 them at several other probably-convenient times, e.g. in each pass of
2626 the toplevel read-eval-print loop, and after evaluating a form given
2627 as an "--eval" command-line option. (These changes were motivated by a
2628 discussion of stream flushing issues on cmucl-imp in August 2000.)
2629 * The source transform for TYPEP of array types no longer assumes
2630 that an array whose element type is a not-yet-defined type
2631 is implemented as an array of T, but instead punts, so that the
2632 type will be interpreted at runtime.
2633 * There is now some support for cross-compiling in make.sh: each of
2634 the phases of make.sh has its own script. (This should be transparent
2635 to people doing ordinary, non-cross-compile builds.)
2636 * Since my laptop doesn't have hundreds of megabytes of memory like
2637 my desktop machine, I became more motivated to do some items on
2638 my to-do list in order to reduce the size of the system a little:
2639 ** Arrange for various needed-only-at-cold-init things to be
2640 uninterned after cold init. To support this, those things have
2641 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2642 symbols with such names are now uninterned after cold init).
2643 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2644 fasl files for cold load.
2645 ** Remove the old compiler structure pooling code (which used to
2646 be conditional on the target feature :SB-ALLOC) completely.
2647 ** Redo the representation of some data in cold init to be more compact.
2648 (I also looked into supporting byte compiled code at bootstrap time,
2649 which would probably reduce the size of the system a lot, but that
2650 looked too complicated, so I punted for now.)
2651 * The maximum signal nesting depth in the src/runtime/ support code has
2652 been reduced from 4096 to 256. (I don't know any reason for the very
2653 large old value. If the new smaller value turns out to break something,
2654 I'll probably just bump it back up.)
2655 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2657 * Many, many bugs reported by Peter Van Eynde have been added to
2658 the BUGS list; some have even been fixed.
2659 * While enabling dynamic object file loading, I tried to make the
2660 code easier to understand, renaming various functions and variables
2661 with less ambiguous names, and changing some function calling
2662 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2663 * While trying to figure out how to do the OpenBSD port, I tried to
2664 clean up some of the code in src/runtime/. In particular, I dropped
2665 support for non-POSIX signal handling, added various comments,
2666 tweaked the code to reduce the number of compilation warnings, and
2667 renamed some files to increase consistency.
2668 * To support the new automatic configuration functionality in make.sh,
2669 the source file target-features.lisp-expr has been replaced with the
2670 source file base-target-features.lisp-expr and the machine-generated
2671 file local-target-features.lisp-expr.
2672 * fixed a stupid quoting error in make.sh so that using CMU CL
2673 "lisp -batch" as cross-compilation host works again
2675 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2677 * DESCRIBE no longer tries to call itself recursively to describe
2678 bound/fbound values, so that it no longer fails on symbols which are
2679 bound to themselves (like keywords, T, and NIL).
2680 * DESCRIBE now works on generic functions.
2681 * The printer now prints less-screwed-up representations of closures
2682 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2683 * A private symbol is used instead of the :EMPTY keyword previously
2684 used to mark empty slots in hash tables. Thus
2685 (DEFVAR *HT* (MAKE-HASH-TABLE))
2686 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2687 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2688 now does what ANSI says that it should. (You can still get
2689 similar noncompliant behavior if bang on the hash table
2690 implementation with all the symbols you get back from
2691 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2692 This breaks binary compatibility, since tests for equality to
2693 :EMPTY are wired into things like the macroexpansion of
2694 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2696 * There's now a minimal placeholder implementation for CL:STEP,
2697 as required by ANSI.
2698 * An obscure bug in the interaction of the normal compiler, the byte
2699 compiler, inlining, and structure predicates has been patched
2700 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2701 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2703 * Missing ordinary arguments in a macro call are now detected even
2704 when the macro lambda list contains &KEY or &REST.
2705 * The debugger no longer complains about encountering the top of the
2706 stack when you type "FRAME 0" to explicitly instruct it to go to
2707 the top of the stack. And it now prints the frame you request even
2708 if it's the current frame (instead of saying "You are here.").
2709 * As specified by ANSI, the system now always prints keywords
2710 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2712 * The default initial SIZE of HASH-TABLEs is now smaller.
2713 * Type information from CLOS class dispatch is now propagated
2714 into DEFMETHOD bodies, so that e.g.
2715 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2717 is now basically equivalent to
2718 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2719 (DECLARE (TYPE SINGLE-FLOAT X))
2721 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2722 operation, without having to do run-time type dispatch.
2723 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2724 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2725 * Since I don't seem to be making big file reorganizations very often
2726 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2727 is overflowing my ability to conveniently back them up), I've finally
2728 checked the system into CVS. (The CVS repository is on my home system,
2729 not at SourceForge -- putting it on SourceForge might come later.)
2730 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2731 high-level GC-NOTIFY-FOO functions send their output. (There's
2732 still very little control of where low-level verbose GC functions
2733 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2734 less than it used to -- the GC-NOTIFY-FOO functions are now under
2735 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2736 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2737 in only one place in the source code, and propagates it automatically
2738 everywhere that it's needed. Thus e.g. when I bump the version from
2739 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2740 * The C source files now include boilerplate legalese and documentation
2741 at the head of each file (just as the Lisp source files already did).
2742 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2743 to his page will be replaced with a link to his new CLiki service.
2745 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2747 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2748 * The build process now looks for GNU make under the default name "gmake",
2749 instead of "make" as it used to. If GNU make is not available as "gmake"
2750 on your system, you can change this default behavior by setting the
2751 GNUMAKE environment variable.
2752 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2753 can build without error under CMU CL.
2755 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2757 * There is now a partial SBCL user manual (with some new text and some
2758 text cribbed from the CMU CL manual).
2759 * The beginnings of a profiler have been added (starting with the
2760 CMU CL profiler and simplifying and cleaning up). Eventually the
2761 main interface should be through the TRACE macro, but for now,
2762 it's still accessed through vaguely CMU-CL-style functions and macros
2763 exported from the package SB-PROFILE.
2764 * Some problems left over from porting CMU CL to the new
2765 cross-compilation bootstrap process have been cleaned up:
2766 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2767 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2768 part when getting rid of the compiler *BACKEND* stuff.)
2769 ** The value of *NULL-TYPE* was screwed up, because it was
2770 being initialized before the type system knew the final
2771 definition of the 'NULL type. This screwed up several key
2772 optimizations in the compiler, causing inefficiency in all sorts
2773 of places. (I found it because I wanted to understand why
2774 GET-INTERNAL-RUN-TIME was consing.)
2775 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2776 PROCLAIM FTYPE information. Unfortunately this broke binary
2777 compatibility again, since now the forms output by DEFGENERIC
2778 to refer to functions which didn't exist in 0.6.3.
2779 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2780 can use the new (as of 0.6.3) transform for SOME into MAP into
2782 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2783 values to (INTEGER 0 1000000), so that the time code will no longer
2784 occasionally get blown up by Linux returning 1000000 microseconds
2785 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2786 its output conform to the ANSI spec. (Alas, this makes its output
2787 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2788 * A full call to MAP NIL with a single sequence argument no longer conses.
2789 * fixes to problems pointed out by Martin Atzmueller:
2790 * The manual page no longer talks about multiprocessing as though
2791 it were currently supported.
2792 * The ILISP support patches have been removed from the distribution,
2793 because as of version 5.10.1, ILISP now supports SBCL without us
2794 having to maintain patches.
2795 * added a modified version of Raymond Toy's recent CMU CL patch for
2796 EQUALP comparison of HASH-TABLE
2798 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2800 * The system still can't cross-compile itself with
2801 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2802 entails), but at least it can compile more of itself that way
2803 than it used to be able to, and various buglets which were uncovered
2804 by trying to cross-compile itself that way have now been fixed.
2805 * This release breaks binary compatibility again. This time
2806 at least I've incremented the FASL file format version to 2, so that the
2807 problem can be detected reliably instead of just causing weird errors.
2808 * various new style warnings:
2809 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2810 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2811 ** using the lexical binding of a variable named in the *FOO* style
2812 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2813 as specified by ANSI.
2814 * *RANDOM-STATE* is no longer automatically initialized from
2815 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2816 default behavior of the system is to repeat its behavior every time
2817 it's run. If you'd like to change this behavior, you can always
2818 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2819 old convention there was no comparably easy way to get the system to
2820 repeat its behavior every time it was run.
2821 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2822 FTYPE declarations has been removed, in favor of their ANSI
2823 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2824 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2825 the types of their sequence arguments can be determined at compile time.
2826 This is done through a new open code expansion for MAP which eliminates
2827 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2828 argument types can be determined at compile time.
2829 * The optimizer now transforms COERCE into an identity operation when it
2830 can prove that the coerced object is already of the correct type. (This
2831 can be a win for machine generated code, including the output of other
2832 optimization transforms, such as the MAP transform above.)
2833 * Credit information has been moved from source file headers into CREDITS.
2834 * Source file headers have been made more standard.
2835 * The CASE macro now compiles without complaining even when it has
2838 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2840 * (Note that the way that the PCL macroexpansions were rewritten
2841 to accommodate the change in DEFGENERIC below breaks binary
2842 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2843 not run under sbcl-0.6.2. Once we get out of alpha releases,
2844 i.e. hit release 1.0.0, we'll probably try to maintain binary
2845 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2846 and sbcl-1.4.4. Until then, however, it might be fairly common
2847 for maintenance releases to break binary compatibility.)
2848 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2850 * The reporting of unhandled signals has been changed to print some
2851 explanatory text as well as the report form. (Previously only
2852 the report form was printed.)
2853 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2854 it defines, so that the compiler no longer issues undefined function
2855 warnings for compiled-but-not-yet-loaded generic functions.
2856 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2857 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2858 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2859 section of the ANSI spec are supported.
2860 * The "" nickname for the "KEYWORD" package has been removed.
2861 The reader still handles symbol tokens which begin with a package marker
2862 as keywords, but it doesn't expose its mechanism for doing so in the
2863 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2864 * The system now issues STYLE-WARNINGs for contradictory TYPE
2865 proclamations. (Warnings for contradictory FTYPE proclamations would
2866 be nice too, but those can't be done usefully unless the type system
2867 is made smarter about FUNCTION types.)
2868 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2869 systematized, so that "*target-*.lisp is supposed to exist only on the
2870 target and imply that there's a related file which exists on the
2871 host, and *host-*.lisp is supposed to exist only on the host and imply
2872 that there's a related file which exists on the target. This involves a
2873 lot of renaming. Hopefully the acute confusion caused by the renaming
2874 will be justified by the reduction in chronic confusion..
2875 ** runtime-type.lisp -> early-target-type.lisp
2876 ** target-type.lisp -> late-target-type.lisp
2877 ** early-host-format.lisp -> early-format.lisp
2878 ** late-host-format.lisp -> late-format.lisp
2879 ** host-error.lisp -> misc-error.lisp
2880 ** early-error.lisp -> early-target-error.lisp
2881 ** late-error.lisp -> late-target-error.lisp
2882 ** host-defboot.lisp -> early-defboot.lisp
2883 ** code/misc.lisp -> code/target-misc.lisp
2884 ** code/host-misc.lisp -> code/misc.lisp
2885 ** code/numbers.lisp -> code/target-numbers.lisp
2886 ** code/early-numbers.lisp -> numbers.lisp
2887 ** early-host-type.lisp -> early-type.lisp
2888 ** late-host-type.lisp -> late-type.lisp
2889 ** host-typep.lisp -> typep.lisp
2890 ** load.lisp -> target-load.lisp
2891 ** host-load.lisp -> load.lisp
2892 ** host-disassem.lisp -> disassem.lisp
2893 ** host-insts.lisp -> insts.lisp
2894 ** byte-comp.lisp -> target-byte-comp.lisp
2895 ** host-byte-comp.lisp -> byte-comp.lisp
2896 ** host-signal.lisp -> signal.lisp
2897 ** host-defstruct.lisp -> defstruct.lisp
2898 ** late-target-type.lisp -> deftypes-for-target.lisp
2899 Furthermore, several other previously target-only files foo.lisp (e.g.
2900 hash-table.lisp and random.lisp) have been split into a target-and-host
2901 foo.lisp file and a target-only target-foo.lisp file, with their key type
2902 definitions in the target-and-host part, so that the cross-compiler will
2903 know more about target types.
2904 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
2905 gone away. In their place are various *BACKEND-FOO* variables
2906 corresponding to the slots of the old structure.
2907 * A bug which caused the SB-COLD bootstrap-time package to be propagated
2908 into the target SBCL has been fixed.
2909 * The chill.lisp system for loading cold code into a running SBCL
2911 * Support for the CMU CL "scavenger hook" extension has been removed.
2912 (It was undocumented and unused in the CMU CL sources that SBCL was
2913 derived from, and stale in sbcl-0.6.1.)
2914 * Various errors in the cross-compiler type system were detected
2915 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
2916 (enabling various consistency checks). Many of them were fixed,
2917 but some hard problems remain, so the compiler is back to
2918 running without *TYPE-SYSTEM-INITIALIZED* for now.
2919 * As part of the cross-compiler type system cleanup, I implemented
2920 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
2921 * I have started adding UNCROSS calls throughout the type system
2922 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
2923 on cross-compiler input files will be able to go away, and various
2925 * CONSTANTP now returns true for quoted forms (as explicitly required
2928 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
2930 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
2931 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
2932 (MAKE-LIST -1). In the longer term, it ought to become true
2933 that declarations are assertions even at SAFETY 1. For now, it's not
2934 quite true even at SAFETY 3, but it's at least more nearly true..
2935 (Note that this change seems to increases the size of the system by
2936 O(5%) and to decrease the speed of the compiler by 20% or more.)
2937 * changed ALIEN printing to be much more abbreviated, as a short-term fix
2938 for the problem of printing dozens of lines of distracting information
2939 about low-level system machinery as part of the top stack frame
2940 on entry to the debugger when an undefined function was called.
2941 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
2942 is not reset to COMMON-LISP-USER.
2943 * Compilation of stuff related to dyncount.lisp has been made conditional
2944 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
2945 smaller. The various dyncount-related symbols have been moved into
2946 a new "SB-DYNCOUNT" package.
2947 * tty-inspect.lisp has been renamed to inspect.lisp.
2948 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
2949 feature has gone away. (When we eventually port to other flavors of
2950 libc and/or Unix, we'll try to make the differences between flavors
2951 invisible at the user level.)
2952 * Various other *FEATURES* tags, and/or their associated conditionals,
2953 have been removed if obsolescent, or given better documentation, or
2954 sometimes given more-mnemonic names.
2956 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
2958 * tidied up "make.sh" script
2959 * tidied up system directory structure
2960 * better "clean.sh" behavior
2961 * added doc/FOR-CMUCL-DEVELOPERS
2962 * many many small tweaks to output format, e.g. removing possibly-confusing
2963 trailing #\. character in DESCRIBE-INSTANCE
2964 * (EQUALP #\A 'A) no longer signals an error.
2965 * new hashing code, including EQUALP hashing
2966 * tidied up Lisp initialization and toplevel
2967 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
2968 * command line argument processing
2969 * added POSIX-GETENV function to deal with Unix-ish environment variables
2970 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
2971 terminating SBCL on EOF
2972 * non-verbose GC by default
2973 * There is no more "sbcl" shell script; the sbcl file is now the C
2974 runtime executable (just like CMU CL).
2975 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
2977 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
2978 symbol and package information as Lisp native symbol and package objects
2979 instead of strings naming symbols and strings naming packages. This way,
2980 whenever packages are renamed (as in warm init), debug information is
2981 transformed along with everything else.
2982 * tweaked the optimization policy declarations which control the building
2983 of SBCL itself. Now, among other things, the system no longer saves
2984 source location debugging information. (This helps two problems at once
2985 by reducing SBCL size and by keeping SBCL from trying to look for its
2986 sources -- which may not exist -- when reporting errors.)
2987 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
2988 debugging and testing purposes
2989 * cleaned up printing, making the printer call PRINT-OBJECT for
2990 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
2991 methods, giving nearly-ANSI behavior
2992 * converted almost all special variables to use *FOO* naming convention
2993 * deleted PARSE-TIME functionality, since it can be done portably
2994 * moved some files out of cold init into warm init
2995 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
2997 * regularized formatting of source files
2998 * added an install.sh script
2999 * fixed ridiculous memory usage of cross-compiler by making
3000 compiler/alloc.lisp not try to do pooling unless it can hook
3001 itself into the GC of the cross-compilation host. Now the system
3002 builds nicely on my old laptop.
3003 * added :SB-ALLOC in target-features.lisp-expr
3004 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3005 was not implemented)
3006 * re-did condition handling and note reporting in the compiler. Notes
3007 are no longer handled by signalling conditions. Style warnings
3008 and warnings are handled more correctly and reported in such a way
3009 that it's easy to find one or the other in your output (so that you
3010 can e.g. figure out which of many problems caused COMPILE-FILE to
3012 * changed the severity of several compiler warnings from full WARNING
3013 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3014 compiler note reporting so that it doesn't use the condition system
3015 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3016 COMPILE-FILE command)
3017 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3018 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3019 no longer supported, since their implementation was deeply intertwingled
3020 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3021 * removed bogus "support" for compiler macros named (SETF FOO), and
3022 removed the compiler macro for SETF INFO (but only after making a fool
3023 of myself on the cmucl-imp mailing list by posting a bogus patch for
3024 DEFINE-COMPILER-MACRO..)
3025 * Compiled files containing forms which have side effects on the Lisp
3026 reader (such as DEFPACKAGE forms) are now handled more correctly.
3027 (Compiler queuing of top level lambdas has been suppressed by setting
3028 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3029 * deleted various currently-unused source files, e.g. gengc.lisp. They
3030 may be added back at some point e.g. when porting to other architectures,
3031 but until they are it's distracting to distribute them and to try to
3033 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3034 there were so many of them they're just distractions, and UNCROSS is
3035 known to be able to handle the current sources
3036 * moved PROFILE functionality into TRACE, so that it will be clear
3037 how the wrapping and unwrapping of functions when you profile them
3038 interacts with the wrapping and unwrapping of functions when you
3039 trace them. (Actually, the functionality isn't there yet, but at least
3040 the interface specification is there. Hopefully, the functionality will
3041 arrive with some future maintenance release.)
3042 * removed host-oops.lisp
3043 * changed signature of QUIT function to allow UNIX-CODE argument
3044 * fixed READ-SEQUENCE bug
3045 * tweaked verbose GC output so that it looks more like the progress
3046 output that ANSI specifies for functions like LOAD
3047 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3048 * added <http://sbcl.sourceforge.com> to the banner information printed by