1 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2 * fixed cross-compiler leakages that prevented building a 32-bit
3 target with a 64-bit host compiler.
4 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
5 * optimization: There's now a fast-path for fixnum arguments in the
6 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
7 * fixed some bugs revealed by Paul Dietz' test suite:
8 ** the type-error signalled from WARN has a filled-in DATUM slot.
9 ** the type-error required when a stream is not associated with
10 a file has the stream as its datum.
11 ** type-errors on single-floats on x86-64 no longer have
12 :INVALID-OBJECT as the datum
14 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
15 * incompatible change: the --noprogrammer option, deprecated since
16 version 0.7.5, has been removed. Please use the equivalent
17 --disable-debugger option instead.
18 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
20 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
21 inoperational for a while now) has been completely removed.
22 * null lexical environments are now printed as #<NULL-LEXENV>,
23 significantly reducing the amount of clutter in typical backtraces.
24 * documentation on weak pointers, finalization, and after GC hooks
25 has been added to the manual.
26 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
27 as other specialized array types, is much faster. SUBSEQ and
28 COPY-SEQ on such arrays have also been sped up.
29 * optimization: EQL is now more efficient when at least other argument
30 is known to be of type (OR FIXNUM (NOT NUMBER)).
31 * fixed bug: the runtime is now less vulnerable to changes in the
32 size of the SBCL object on OS X, and virtual memory is reserved for
33 all spaces that need to be at a fixed address.
34 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
35 * fixed bug: finalizers and after GC hooks that cause consing are now
37 * fixed bug: compiler error messages and summaries are now printed to
38 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
39 * fixed inference of the upper bound of an iteration variable.
40 (reported by Rajat Datta).
41 * fixed bug 211e: calling local functions with duplicated constant
42 keyword argument no longer causes a bogus style warning about an
44 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
45 ftype declaration to be disregarded. (reported by Dave Roberts)
46 * fixed bug 373: caused by erronous compilation of references to alien
47 variables in the runtime on ppc/darwin.
48 * fixed bug 376: CONJUGATE type deriver.
49 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
50 (reported by Baughn on #lisp)
51 * a cross-compiler bug on non-x86 platforms has been identified and
52 fixed. (thanks to Bruno Haible)
53 * improvements to the MIPS runtime code for increased stability.
54 (thanks to Thiemo Seufer)
55 * increased the maximimum compact environment size to allow
56 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
57 * partial workaround for bug 135: don't name the function we're
58 calling for hairy cases of EVAL, so as not to accumulate one environment
59 entry per call to EVAL. (reported by Kevin Reid)
60 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
61 * optimization: type testing for non-vector arrays should be faster.
62 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
64 * bug fix: setting 31st element of a bit vector to zero did not work
66 * bug fix: redefining a class definition which failed due to a
67 previous accessor / function clash now works (but see BUGS entry
68 #380 for more problems in this area). (thanks to Zach Beane)
69 * the long form of DEFINE-METHOD-COMBINATION disables method group
70 checking when given a single method group with pattern *. (thanks
72 * on x86 compiler supports stack allocation of results of simple
73 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
74 * support for building and running on Mac OS X 10.4 "Tiger" has been added
75 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
76 * fixed some bugs related to Unicode integration:
77 ** the restarts for recovering from input and output encoding
78 errors only appear when there is in fact such an error to
80 * fixed some bugs revealed by Paul Dietz' test suite:
81 ** MISC.549 and similar: late transformation of unsafe type
82 assertions into derived types caused unexpected code
84 ** SCALE-FLOAT type deriver is less wrong.
85 ** type derivers for EXP, LOG and similar functions compute result
86 types for complex arguments better.
87 ** (MISC.563) CONJUGATE type deriver works for very restricted
89 ** out-of-line type testers for character strings are available.
90 ** EQUAL compiler transform understands specialness of objects
92 ** accessing double-floats stored on the stack now works on x86-64.
93 ** debugger internals could sometimes create invalid lispobjs,
94 resulting in GC crashes.
95 ** MISC.548: type check weakening can convert required type into
97 ** initialization forms for bindings are not in scope of free special
100 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
101 * incompatible change: thread support for non-NPTL systems has
102 been removed - locking is buggy and unreliable. A threaded
103 SBCL build will now warn at startup and refuse to create
104 new threads, unless futex support is detected in the kernel
105 * incompatible change: the top level REPL now has only an ABORT
106 restart associated with it, not TOPLEVEL and ABORT as it used to.
107 TOP and TOPLEVEL are now available as debugger commands for
108 returning to the top level.
109 * incompatible change: forms evaluated in the REPL now use the
110 global optimization policy.
111 * incompatible change: user- and system-initialization files are
112 no longer processed with LOAD, but by READ and EVAL; hence the
113 global optimization policy, startup package, readtable, etc,
115 * The .fasl file format number has been incremented because of
116 various incompatible changes.
117 * internal entry point details and argument counts no longer appear
118 in backtraces unless explicitly requested by setting
119 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
120 * built-in and standard functions no longer have names like "top
121 level local call to FOO".
122 * fixed bug 32: functions defined in non-null lexical environments
123 now have more legible printed representation
124 * fixed bug 33: functions defined in non-null lexical environemnts
125 are now more amenable to inspection by INSPECT.
126 * workaround for bug 354: XEPs no longer appear in backtraces unless
127 explicitly requested.
128 * fixed bug: receiving the signal which results from attempting to
129 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
130 SIGBUS on other *BSDs) on architectures where the C stack is also
131 the Lisp stack (x86 and x86-64) from foreign code no longer leads
132 to debugger confusion or wild instability. (reported by Cheuksan
134 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
135 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
136 (reported by Lutz Euler)
137 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
138 compiled code. (reported by Paul Dietz)
139 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
140 specialized parameter to a method to something that is not TYPEP
141 the specializer is now possible.
142 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
143 face of package deletion.
144 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
145 pathname. (reported by David Tolpin on c.l.l)
146 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
147 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
148 STRING-STREAM and FILE-STREAM, respectively.
149 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
150 than previously; in addition, its test suite is now run on build.
151 (thanks to Luke Gorrie)
152 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
153 * optimization: added a immediate representation for single-floats
155 * fixed some bugs revealed by Paul Dietz' test suite:
156 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
157 ** attempting to create a package with a colliding nickname causes
158 correctable errors to be signalled.
159 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
160 ** several x86-64 backend bugs related to sign-extension of immediate
163 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
164 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
165 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
167 * fixed loading of multiply forward-referenced layouts.
168 (thanks to Cheuksan Wang)
169 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
170 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
171 (see COMPILE-FILE documentation for details of :PRINT option)
172 * fixed bugs 19 and 317: fixed-format floating point printing is
173 more accurate. This also fixes a bug reported by Adam Warner
174 related to the ~@F format directive.
175 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
177 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
178 dispatch table if the from-char is a dispatch macro character.
179 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
180 by the wordsize no longer ignore the last word. (reported by Lutz
182 * fixed bug in type checking of dynamic-extent variables. (reported
184 * optimization: sequence traversal functions use their freedom to
185 coerce function designators to functions.
186 * optimization: code with many calls to CLOS methods specialized on
187 CLOS classes has had redundant type checks removed.
188 * optimization: added declarations to speed up operations that access
189 the internal character database (for example STRING-UPCASE)
190 * optimization: comparison operations between floats and sufficiently small
191 fixnums no longer create extra rationals
192 * fixed some bugs related to Unicode integration:
193 ** portions of multibyte characters at the end of buffers for
194 character-based file input are correctly transferred to the
195 start of the buffer at the next read.
196 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
197 passing it through to OPEN.
198 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
199 argument given to internal calls to OPEN.
200 ** the built-in comment readers (introduced by character sequences
201 ";" and "#|") are more forgiving to encoding errors; they will
202 STYLE-WARN and then attempt to resync the stream at a character
203 boundary. (thanks to Teemu Kalvas)
204 * fixed some bugs in the x86-64 port:
205 ** Negative short int return values from c-calls are sign-extended
207 ** The stack is aligned to 16-bytes for c-calls, as required by
208 the ABI. (thanks to Cheuksan Wang)
209 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
210 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
211 for enums. (thanks to Vincent Arkesteijn)
212 ** Multiple small optimizations and bugfixes for floating point
214 * fixed some bugs revealed by Paul Dietz' test suite:
215 ** Space, Tab, Linefeed, Return and Page have the invalid
216 secondary constituent character trait.
217 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
219 ** WITH-INPUT-FROM-STRING should only update the index place on
221 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
223 ** Bit-array operations (BIT-AND and similar) worked incorrectly
224 with one-dimensional arrays with fill pointers.
225 ** TYPE-OF failed on a complex with an integer realpart and a
227 ** compiler failure during type inference for the code of form
228 (IF (EQL X (THE ...)) ...) (MISC.535).
230 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
231 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
232 platform. The port supports 61-bit fixnums, large memory spaces
233 and reloading shared object files.
234 * enhancement: saving cores with foreign code loaded is now
235 supported on ppc/Darwin in addition to the previously supported
237 * enhancement: the statistical profiler now walks deeper into the
238 call stack for more meaningful call-graphs and accrued time
239 reports (x86/x86-64 only). It also now reports time spent in
241 * enhancement: it is now possible to trace most individual methods
242 of a generic function in addition to tracing the generic function
244 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
246 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
247 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
248 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
249 produces an error. (thanks to Vincent Arkesteijn)
250 * bug fix: NAMESTRING on pathnames with :WILD components in their
251 directories works correctly. (thanks to Artem V. Andreev)
252 * fixed bug 125: compiler preserves identity of closures. (reported
254 * bug fixed: functions with &REST arguments sometimes failed with
255 "Undefined function" when compiled with (DEBUG 3). (reported by
257 * bug fix: overflow during compiling of setting element of a bit
258 vector with constant index and value. (reported by Timmy Douglas)
259 * build fix: fixed the dependence on *LOAD-PATHNAME* and
260 *COMPILE-FILE-PATHNAME* being absolute pathnames.
261 * on x86 compiler partially supports stack allocation of dynamic-extent
263 * GO and RETURN-FROM do not check the extent of their exit points
264 when compiled with SAFETY 0.
265 * fixed some bugs related to Unicode integration:
266 ** encoding and decoding errors are now much more robustly
267 handled; it should now be possible to recover even from invalid
268 input or output to the terminal. (thanks to Teemu Kalvas)
269 ** provided a first cut at implementing STRING-TO-OCTETS and
270 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
271 ** altered the SB-MD5 contributed module slightly, changing the
272 interface just enough for it to be supportable for builds where
273 lisp characters are not eight bits.
274 * fixed some bugs revealed by Paul Dietz' test suite:
275 ** the FORMATTER-generated functions for ~V[ conditionals require
276 the correct number of arguments.
277 ** READ-FROM-STRING returns the mandated second value when applied
278 to displaced strings.
279 ** the #\Rubout and #\Backspace characters are treated as invalid
280 constituent characters by the tokenizer.
282 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
283 * new feature: reloading changed shared object files with
284 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
285 * new feature: references to foreign variables and functions
286 can now be compiled and loaded before the corresponding shared
287 object file is loaded, as long as the foreign definitions are
288 available at runtime.
289 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
290 supported. (thanks to Dan Debertin)
291 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
292 just keywords. (thanks to Vincent Arkesteijn)
293 * fixed bug #331: structure-class instances corresponding to
294 DEFSTRUCT forms are now created eagerly.
295 * fixed bug #345: backtraces from calls to undefined functions work
296 on x86 as well. Related bug #61 is now also partially fixed on x86
297 (backtraces from throws to unknown catch tags.)
298 * bug fix: lambda-list parsing is now stricter vrt. order and number
299 of lambda-list keywords.
300 * bug fix: as specified by AMOP, an error is signalled if a
301 class-option appears multiple times in a DEFCLASS form. (reported
303 * bug fix: RANDOM can be compiled when the compiler derives the type
304 of its numeric argument as a disjoint set of small integers.
305 (reported by Paul Dietz)
306 * bug fix: starting a core saved with shared objects loaded when
307 those objects are not available no longer causes threaded SBCL to
308 hang. (reported by Sean Ross)
309 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
311 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
312 argument, which must always be NIL. (reported by Kalle Niemitalo)
313 * bug fix: printing 1.0d+23 no longer results in an error.
314 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
315 * bug fix: structure slot setters preserve evaluation order. (thanks
317 * fixed some bugs related to Unicode integration:
318 ** RUN-PROGRAM can allow its child to take input from a Lisp
319 stream. (reported by Stefan Scholl)
320 ** ASDF-INSTALL successfully downloads .tar.gz files in all
321 locales. (reported by Ken Causey)
322 * fixed some bugs revealed by Paul Dietz' test suite:
323 ** INCF, DECF and REMF evaluate their place form as specified in
325 ** FORMATTER expands ~{ ~} iteration directives with V or #
326 parameters correctly.
327 ** FORMATTER deals with the ~@[ ~] conditional directive where the
328 consequent uses no arguments correctly.
329 ** the system has a partial understanding of the (COMPLEX RATIO)
331 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
334 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
335 * new feature: a build-time option (controlled by the :SB-UNICODE
336 keyword feature, enabled by default) for building the system with
337 support for the entire 21-bit character space defined by the
339 * new feature: the system now has rudimentary external-format
340 support; the primary user-visible change at this time is that
341 characters with the high bit set (such as lower-case-e-acute) will
342 print correctly to a terminal in a UTF-8 environment.
343 * minor incompatible change: BASE-CHAR no longer names a class;
344 however, CHARACTER continues to do so, as required by ANSI.
345 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
346 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
347 has been moved to the SB-EXT package (temporarily re-exported from
349 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
350 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
352 * fixed bug #351: better error-handlind and reporting for malformed
354 * fixed bug #350: bignum-printing is now more memory-efficient,
355 allowing printing of very large bignums, eg. (expt 2 10000000).
356 (reported by Bruno Haible)
357 * fixed bug #302: better primitive-type selection for intersection
359 * fixed bug #308: non-graphic characters now all have names, as
360 required. (reported by Bruno Haible)
361 * bug fix: redefining a class with different superclasses now correctly
362 removes it from the direct-subclasses of its previous superclasses.
363 (reported by David Morse)
364 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
365 new value now works. (reported by Bruno Haible)
366 * bug fix: correct canonicalization of multiple non-standard slot
367 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
368 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
369 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
370 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
371 now exists, an signals an error.
372 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
373 during macroexpansion for non-iteration variables as well. (reported
374 by Bruno Haible for CMUCL)
375 * bug fix: Cyclic structures and unprintable objects in compiler
376 messages no longer cause errors. (reported by Bruno Haible)
377 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
378 and READ-FROM-STRING all now return a primary value of NIL if
379 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
380 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
381 (reported by Bruno Haible for CMUCL)
382 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
383 do not designate a condition. (reported by Bruno Haible for
385 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
386 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
387 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
388 permitting resolution in favour of any of the conflicting symbols.
389 (reported by Bruno Haible for CMUCL)
390 * bug fix: EQUAL compiler optimizations is less aggressive on
391 strings which can potentially compare true despite having distinct
392 specialized array element types.
393 * bug fix: unit enumerations can be defined without dividing by
394 zero. (thanks to Vincent Arkesteijn)
395 * FORMAT compile-time argument count checking has been enhanced.
396 (report from Bruno Haible for CMUCL)
397 * a partial workaround for the bug 262: the compiler does not try to
398 inline-expand a local function doing RETURN-FROM from a deleted
399 BLOCK. (thanks to Peter Denno for the bug report and to David
400 Wragg for the simple test case)
401 * fixed some bugs revealed by Paul Dietz' test suite:
402 ** RENAME-PACKAGE allows all package designators as new package
404 ** constraint propagation and lambda variable substitution are
405 more cautious in dealing with partially deleted code.
406 ** compiler sometimes lost reoptimization passes.
407 ** CERROR, given a condition as condition designator, passes its
408 remaining arguments to the continue format control without
410 ** Case-altering FORMAT directives work correctly on non-ASCII
412 ** The REAL type specifier handles bounds outside the floating
413 point ranges without signalling FLOATING-POINT-OVERFLOW.
414 ** Functions with IR1-transformations can create intercomponent
415 references to global functions.
416 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
418 ** FORMAT ~R treats a nil value for its first parameter correctly.
420 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
421 * enhancement: saving cores with foreign code loaded is now
422 supported on x86/NetBSD and sparc/Linux in addition to the previously
424 * bug fix: on some platforms repeated installations caused multiple
425 copies of HTML documentation to be installed -- should not happen
426 any more. (reported by Stefan Scholl)
427 * bug fix: parsing self-recursive alien record types multiple times
428 no longer causes infinite recursion. (reported by Thomas F. Burdick,
429 original patch by Helmut Eller for CMUCL)
430 * bug fix: stack-exhaustion detection works now on NetBSD as well.
431 (thanks to Richard Kreuter)
432 * bug fix: defining classes whose accessors are methods on existing
433 generic functions in other (locked) packages no longer signals
434 bogus package lock violations. (reported by François-René Rideau)
435 * bug fix: special variables as DEFMETHOD parameters no longer have
436 associated bogus type declarations. (reported by David Wragg and
438 * bug fix: read-write consistency on streams of element-type
439 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
440 * bug fix: redefiniton of the only method of a generic function with
441 no DEFGENERIC no longer emits a full WARNING. In addition,
442 redefinition of generic functions with no DEFGENERIC to an
443 incompatible lambda list now signals an error. (thanks to Zach
445 * bug fix: DEFGENERIC now works even when there's a function of the
446 same name in an enclosing lexical environment. (thanks to Zach
448 * fixed compiler failure, caused by instrumenting code during
449 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
450 * optimization: added loop analysis and improved register allocation
451 * fixed some bugs revealed by Paul Dietz' test suite:
452 ** POSITION on displaced vectors with non-zero displacement
453 returns the right answer.
454 ** (SIMPLE-STRING) is a valid type specifier for sequence
456 ** *PRINT-LEVEL* handling for slotless structures is pedantically
458 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
459 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
461 ** The justification version of the FORMAT ~< directive treats
462 non-zero minpad parameter correctly.
464 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
465 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
466 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
467 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
468 the supported interface.
469 * new feature: Single-stepping of code compiled with DEBUG 2 or
470 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
471 * new feature: saving cores with foreign code loaded is now
472 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
473 Timothy Moore's work for CMUCL)
474 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
475 parameters to * instead of NIL if no initform is supplied.
476 (reported by Johan Bockgaard)
477 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
478 correctly when FUNCALL appears as the car of the form. Note:
479 despite this FUNCALL forms are not currently subject to
480 compiler-macro expansion. (port of Raymond Toy's fix for the
481 same from CMUCL, reported by Johan Bockgaard)
482 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
483 (thanks for Teemu Kalvas)
484 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
485 case instead of terminating the iteration (thanks for Julian
486 Squires, Sean Champ and Raymond Toy)
487 * bug fix: incorrect expansion of defgeneric that caused a style
488 warning. (thanks for Zach Beane)
489 * bug fix: slot accessor effective method computation works properly
490 for classes with multiple non-standard applicable methods on
491 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
492 * on x86 compiler supports stack allocation of results of LIST and
493 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
494 CMUCL implementation by Gerd Moellmann)
495 * fixed some bugs revealed by Paul Dietz' test suite:
496 ** FORMAT strings with both the ~<~:;~> form of the justification
497 directive and pretty-printing directives cause an error.
499 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
500 * incompatible change: the internal functions
501 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
502 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
503 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
504 instead of the old functions.
505 * new feature: on platforms where "dladdr" is available foreign
506 function names now appear in backtraces. (based on Helmut Eller's
508 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
509 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
511 * documentation: ASDF manual is now distributed with SBCL.
512 * bug fix: SBCL can now load its contributed modules with REQUIRE
513 even if the system-provided entries have been removed from
514 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
516 * bug fix: backtraces involving undefined functions or assembly
517 routines are more informative. (thanks to Brian Downing)
518 * bug fix: mutually referent alien structures now work correctly.
519 (reported by Rick Taube)
520 * bug fix: structures defined by WITH-ALIEN can be referred to
521 within other definitions in the same WITH-ALIEN.
522 * bug fix: division operators (MOD, TRUNCATE and the like) with
523 constant zero divisors and integer dividends no longer generate
525 * bug fix: provide default methods for INPUT-STREAM-P and
526 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
527 * bug fix: improve the detection and resolution of MOP metacycles
528 (where effective method computation depends on the generic function
529 having its effective method computed). (reported by Bruno Haible)
530 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
531 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
532 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
533 * bug fix: pathnames with relative directory components can be
534 represented relative to default pathnames.
535 * optimization: in taking the GCD of bignums, reduce the two bignums
536 to approximately the same size (using Euclid's algorithm) before
537 applying the more sophisticated binary GCD. (thanks to Juho
539 * optimization: COUNT on bitvectors now operates word-at-a-time.
540 * optimization: ASH with a positive, but not necessarily constant,
541 (leftwards) shift, when in a modular context, is compiled to a
543 * fixed some bugs revealed by Paul Dietz' test suite:
544 ** FORMAT variable parameters ("~V<char>") are defaulted properly
545 if the corresponding argument is NIL.
546 ** FORMAT directives accepting a minpad parameter treat negative
548 ** literal commas as character parameters to FORMAT directives are
550 ** literal spaces directly after ~<Newline> directives within a
551 format-logical-block (~:< ~@:>) do not induce :FILL-style
552 conditional newlines.
553 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
554 as specified: it no longer includes conditional newlines).
555 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
557 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
558 :PER-LINE-PREFIX argument does not evaluate to a string.
560 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
561 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
562 the manual for details; for now, package locks can be disabled by
563 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
564 if no major problems are found then it is likely that they will be
565 compiled in unconditionally.
566 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
567 unsupported operators on all platforms. To load a shared library
568 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
569 object file, link it into a shared library outside of SBCL and
570 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
572 * bug workaround: ROOM T and the GC were not working together
573 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
574 reporting facility aren't preserved. That reporting has been
575 disabled, so now until and unless someone figures out how to make it
576 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
577 * minor incompatible change: as threatened around sbcl-0.8.0, the
578 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
579 The symbols are also exported from SB-PCL for backwards
580 compatibility, but more so than before SB-PCL should be treated as
581 an implementation-internal package.
582 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
584 * the SB-SPROF contrib now works on (most) non-x86 architectures.
585 It is known as of this release not to work on the Alpha, however.
586 * fixed bug #167: errors signalled due to illegal syntax in method
587 bodies are now more legible.
588 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
589 designators and can hence be used with TYPEP.
590 * fixed bug #333: CHECK-TYPE now ensures that the type error
591 signalled, if any, has the right object to be accessed by
592 TYPE-ERROR-DATUM. (reported by Tony Martinez)
593 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
594 5.1.2.3 for multiple-value place subforms. (reported by Kalle
596 * fixed bug #334: programmatic addition of slots using specialized
597 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
598 and :ALLOCATION :CLASS slots.
599 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
600 float exponents. (rereported by Peter Seibel)
601 * fixed a bug: #\Space (and other whitespace characters) are no
602 longer considered to be macro characters in standard syntax by
604 * fixed bug: initialization of condition class metaobjects no longer
605 causes an instance of the condition to be created. (reported by
607 * fixed bug: it is now possible to have slots such that
608 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
609 description is neither :INSTANCE nor :CLASS.
610 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
611 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
612 SLOT-VALUE-USING-CLASS), and no longer causes errors with
613 non-standard SLOT-DEFINITION-ALLOCATION values.
614 * fixed bugs: various race conditions handling exiting threads.
615 CL-PPCRE's thread torture test now passes.
616 * fixed arguably-a-bug: GC time proportional to number of threads in
617 system even when most of them are idle
618 * optimization: improved performance of BIT and SBIT on bit-vectors.
619 * .fasl file incompatibility: The fasl file version number has
620 been incremented because of changes associated with package locks.
622 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
623 * minor incompatible change: the system no longer provides
624 optimization hints (conditions which are TYPEP
625 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
626 mode; these hints are still emitted when higher SPEED optimization
628 * new contrib module: a sampling profiler (profiling by statistical
629 sampling, rather than by instrumenting functions) is available as
630 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
631 * the behaviour of the standard function ED is now customizeable by
632 third parties through a hook variable: see ED's documentation
633 string for information on the protocol.
634 * the compiler no longer emits efficiency notes for (FUNCALL X)
635 when the type of X is uncertain under default optimization
637 * fixed bug 276: mutating a binding of a specialized parameter to a
638 method to something that is not TYPEP the specializer is now
640 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
641 exhibit the required behaviour on the x86 platform. (thanks to
642 Peter van Eynde, Eric Marsden and Bruno Haible)
643 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
644 even for difficult arguments. (reported by Peter Graves)
645 * fixed bug 141a: the backquote printer now descends quoted
647 * fixed another bug in backquote printing: no more destructive
648 modification of the form's list structure. (reported by Brian
650 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
651 move between its address being taken and the call to
652 interrupt_thread, fixing a crashing race condition.
653 * the SB-POSIX contrib implementation has been adjusted so that it
654 no longer exhibits ridiculously poor performance when constructing
655 instances corresponding to C structs.
657 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
658 * minor incompatible change: the sb-grovel contrib now treats C
659 structures as alien (in the sense of SB-ALIEN) objects rather than
660 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
661 has implications for memory management of client code
662 (sb-grovel-returned objects must now be manually managed) and for
663 type safety (alien objects now have full types).
664 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
665 used to control emission of compiler diagnostics, rather than the
666 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
667 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
668 quality should be considered deprecated.
669 * install.sh now installs the user manual as well
670 * (not quite a new documentable feature, but worth considering in
671 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
672 beginnings of a semantically meaningful condition hierarchy is
673 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
675 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
676 designator as the defaults argument.
677 * fixed bug: Displaced arrays whose displaced-to array has become
678 too small now cause ARRAY-DIMENSION to signal an error, providing
679 for safer bounds-checking. (reported by Bruno Haible)
680 * fixed bug: DEFCLASS slot definitions with identical :READER and
681 :WRITER names now signal a reasonable error. (reported by Thomas
683 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
685 * fixed bug: Invalid slot specification errors now print correctly.
686 (thanks to Zach Beane)
687 * fixed bug 320: Shared to local slot value transfers in class
688 redefinitions now happen corrently from superclasses as
689 well. (reported by Bruno Haible)
690 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
692 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
693 improper lists correctly. (reported by Bruno Haible)
694 * fixed bug 313: source-transform for <fun-name> was erroneously
695 applied to a call of a value of a variable with name <fun-name>.
696 (reported by Antonio Menezes Leitao)
697 * fixed bug 307: The obsolete instance protocol ensures that
698 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
699 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
700 remove dead unknown-values globs from the middle of the stack.
701 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
702 easier to resume long computations after using *BREAK-ON-SIGNALS*
703 to diagnose and fix failures. (thanks to Nikodemus Siivola)
704 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
705 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
706 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
707 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
708 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
710 * fixed a (fairly theoretical) bug in string printing: if
711 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
712 does not have array-element-type equal to the most general string
714 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
715 function name. (thanks to Nikodemus Siivola)
716 * fixed bug: compile-time format string checker failed on
717 non-closed ~{. (reported by Thomas F Burdick)
718 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
719 don't warn on legal loop code involving more than one aggregate
720 boolean. (thanks to Nikodemus Siivola)
721 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
722 in keyword position to MAKE-PACKAGE was not being reported as
724 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
725 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
727 * fixed bugs 280 and 312: the checking for multiple definitions in a
728 file is less likely to become confused by uses of inline
730 * fixed bug: the #S reader macro performs the keyword coercion
731 specified for slot names. (reported by Kalle Niemitalo)
732 * fixed bug: lambda lists may contain symbols whose names start with
733 & but are not lambda-list-keywords; their occurrence triggers a
735 * fixed bug 321: define-method-combination argument lambda lists do
736 not cause invalid code to be generated when &optional or &aux
737 variables have default values. (reported by Bruno Haible)
738 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
739 have CLOS classes; this appears also to have fixed reported
740 instability in stack exhaustion detection.
741 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
742 MUFFLE-WARNING when no associated restart is present is now
744 * optimization: rearranged the expansion of various defining macros
745 so that each expands into only one top-level form in a
746 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
748 * optimization: used a previously little-used slot in symbols to
749 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
751 * fixed some bugs revealed by Paul Dietz' test suite:
752 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
754 ** Fixed an optimization bug related to inheritance of initforms
755 from local to shared slots.
756 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
757 ** CAST optimizer forgot to flush argument derived type.
758 ** print/read consistency on floats is now orders of magnitude
759 more likely. (thanks also to Bruno Haible for a similar report
761 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
762 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
763 for objects of type REAL. Make it so.
764 ** SXHASH is specified (infelicitously) to respect similarity,
765 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
766 it so. (thanks to Markus Ziegler)
767 ** on the Alpha, the compiler succeeds in compiling functions
768 returning a known number of arguments greater than 63.
769 ** fixed handling of invalid NIL arguments in keyword position
771 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
773 ** provided a readably-printable representation for RANDOM-STATE
775 ** ensured that pathnames, where they have a namestring, always
776 print using #P"..." syntax.
778 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
779 * Support for the forthcoming 2.0 version of the NetBSD kernel
780 running on x86 hardware has been added. (thanks to Perry
781 E. Metzger most immediately, and others for their past work)
782 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
783 current release version; SBCL's OpenBSD support had been broken
784 since about the time of OpenBSD's switch to ELF binary format.)
785 * [placeholder for DX summary]
786 ** user code with &REST lists declared dynamic-extent, under high
787 speed or space and low safety and debug optimization policy.
788 * The manual has been converted to Texinfo format and the debugger
789 chapter from the cmucl manual has been added.
790 * A facility has been added to extract documentation strings from
791 sbcl and store them as Texinfo-formatted snippets for inclusion in
792 the manual (via Texinfo's @include directive)
793 * bug fix: compiler emitted division in optimized DEREF. (thanks for
794 the test case to Dave Roberts)
795 * bug fix: multidimensional simple arrays loaded from FASLs had fill
796 pointers. (reported by Sean Ross)
797 * bug fix: PROFILE output is printed nicely even for large numerical
798 values. (thanks to Zach Beane)
799 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
800 greater than 32 handle EOF correctly.
801 * bug fix: on X86 an immediate argument of the IMUL instruction is
802 correctly printed in disassembly. (thanks to Lutz Euler)
803 * bug fix: class slots in redefined classes preserve their old
804 values. (thanks to Bruno Haible and Nikodemus Siivola)
805 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
806 succeeds. (reported by Marco Baringer)
807 * fixed some bugs revealed by Paul Dietz' test suite:
808 ** READ-SEQUENCE now works on ECHO-STREAMs.
809 ** RATIONALIZE works more according to its specification. (thanks
812 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
813 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
814 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
815 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
816 should matter to you only if you rebind the printer control
817 variables and then find you want different bindings in the
818 debugger than in the ordinary execution of your program.)
819 * The runtime build system has been tweaked to support building
820 (on SPARC/SunOS) using a C compiler which invokes Sun's own
821 assembler and linker. (thanks to Nikodemus Siivola)
822 * Unbound, undefined, undeclared variables now trigger full
823 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
824 more often programmer error than deliberate exploitation of undefined
826 * optimization: the hash algorithm for strings has changed to one
827 that is less vulnerable to spurious collisions. (thanks to Juho
829 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
830 needless bounds checking. (thanks to Juho Snellman)
831 * optimization: implemented multiplication as a modular
832 (UNSIGNED-BYTE 32) operation on the PPC backend.
833 * fixed some bugs revealed by Paul Dietz' test suite:
834 ** ADJUST-ARRAY now copies the datum in a zero rank array if
836 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
837 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
838 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
840 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
842 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
844 ** ECHO-STREAMs no longer attempt to echo the end of file value to
845 their output stream on EOF from read.
846 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
847 have been read to end-of-file.
848 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
850 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
851 description of determination of which consecutive characters
853 ** Printing the "Space" character with escaping on now yields "#\\ ",
854 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
855 ** Reading floating-point numbers with *READ-BASE* set to a number
856 less than 10 works correctly.
857 ** Reading floating-point numbers with *READ-BASE* set to a number
858 more than 10 works correctly.
859 ** Printing with *PRINT-READABLY* targets the standard readtable, not
860 the readtable currently in effect.
862 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
863 * minor incompatible change: parsing of namestrings on a physical
864 (Unix) host has changed; numbers after the final #\. in a
865 namestring are no longer interpreted as a version field. This is
866 intented to be largely invisible to the user, except that the
867 meaning of the namestring "*.*.*" has changed: it now refers to a
868 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
869 should usually be replaced by
870 (make-pathname :name :wild :type :wild :version :wild)
871 with the added benefit that this is more likely to be portable.
872 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
873 option to OPEN now signals an error if the file being opened
874 exists; this may have an impact on existing code.
875 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
876 signals received, on the PowerPC platforms (both Linux and
877 Darwin). (thanks to Pierre Mai for pointing out the location of
879 * several fixes on OS X: The system now builds and runs cleanly on
880 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
881 (thanks to Brian Mastenbrook)
882 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
883 second-resolution: integer multiples of 1/3600 between -24 and 24.
884 (thanks to Vincent Arkesteijn)
885 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
886 type when called with 1 argument; PEEK-CHAR checked type of
887 PEEK-TYPE only after having read first character from a
888 stream. (reported by Peter Graves)
889 * bug fix: the garbage collector now has much better locality
890 behaviour, and in particular no longer treats all memory as being
891 exhausted when many small objects point to each other in a deeply
893 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
894 recognized as being TYPEP their class.
895 * bug fix: the PUSHNEW documentation string has been corrected.
896 (thanks to Vincent Arkesteijn)
897 * bug fix: defaulting of the value for the last of an atypically
898 large number of multiple values being bound was not being
899 performed correctly on the Alpha or PPC platforms
900 * optimization: implemented multiplication as a modular
901 (UNSIGNED-BYTE 32) operation on the x86 backend.
902 * optimization: SEARCH on simple-base-strings can now be open-coded.
903 (see also contrib/compiler-extras.lisp for inspiration for
904 teaching the compiler about the Boyer-Moore algorithm).
905 * value, returned by MAX (and MIN) called with several EQUALP, but
906 not EQL, arguments now does not depend on compiler settings.
907 * fixed some bugs revealed by Paul Dietz' test suite:
908 ** in stack analysis liveness information is propagated from
909 non-local entry points.
910 ** pathwise CAST removing failed when the CAST node did not start
912 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
913 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
915 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
916 host is already defined.
917 ** RENAME-FILE works on streams instead of signalling an internal
919 ** PEEK-CHAR uses the current readtable when determining whether
920 or not a character is whitespace.
921 ** MERGE-PATHNAMES handles the case when the pathname does not
922 specify a name while the default-pathname specifies a version.
923 ** Pathnames now stand a better chance of respecting print/read
925 ** Attempting to use standardized file system operators with a
926 pathname with invalid :DIRECTORY components signals a
928 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
930 ** DIRECTORY on logical pathnames is more correct.
931 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
932 signal a TYPE-ERROR if their argument is not a stream.
933 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
934 designator argument does not designate a stream.
935 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
936 examining the synonym.
937 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
939 ** OPEN and WITH-OPEN-STREAM allow opening streams with
940 element-type larger than ([UN]SIGNED-BYTE 32).
942 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
943 * When built with the :SB-FUTEX feature, threaded builds now take
944 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
945 for faster/more reliable mutex and condition variable support.
946 * Incompatible change (but one you probably shouldn't have been using
947 anyway): the interface and code for arbitrating between multiple
948 threads in the same user session has been redesigned.
949 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
950 error when called without an explicit environment argument.
951 (thanks to Nikodemus Siivola)
952 * bug fix: buffered :DIRECTION :IO streams are less likely to become
953 confused about their position. (thanks to Adam Warner and Gerd
955 * bug fix: Pretty printing backquoted forms with unquotations in the
956 argument list position of various code constructs such as LAMBDA
957 now works correctly. (reported by Paul Dietz)
958 * bug fix: Pretty printing unquotations no longer loses all
959 stream position information.
960 * optimization: performance of string output streams is now less
961 poor for multiple small sequence writes.
962 * optimization: performance of CSUBTYPEP in the presence of complex
963 expressions involving CONS and NOT many times has been improved.
964 (reported by Paul Dietz)
965 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
967 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
968 streams for the REPL, for the debugger, and so on. (thanks to
970 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
971 * fixed some bugs revealed by Paul Dietz' test suite:
972 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
973 is ignored if it is too small, rather than propagating through
974 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
975 ** extremely complex negations of CONS types were not being
976 sufficiently canonicalized, leading to inconsistencies in
978 ** VALUES tranformer lost derived type.
980 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
981 * fixed a bootstrapping bug: the build process no longer assumes
982 that the various BOOLE-related constants have the same value in
983 host and target lisps. (noted by Paul Dietz' test suite on an
984 SBCL binary built from CLISP)
985 * The system can now be dynamically linked on the MIPS platform,
986 which enables dynamic loading of foreign code from Lisp. (thanks
987 to Ralf Baechle for discussions on the MIPS ABI)
988 * The system now records debugging information for its own source
989 files in a filesystem-position-independent manner, relative to
990 the "SYS" logical host.
991 * fixed a compiler bug: MV-LET convertion did not check references
992 to the "max args" entry point. (reported by Brian Downing)
993 * tweaked disassembly notes to be less confident about proclaiming
994 some instruction as an LRA. (thanks to Brian Downing)
995 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
996 now each have their own history, command character, and other
997 characteristics. (thanks to David Lichteblau)
998 * fixed some bugs revealed by Paul Dietz' test suite:
999 ** compiler failure in compiling LOGAND expressions including a
1001 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1002 shift greater than 32.
1003 ** FLUSH-DEST did not mark blocks for type check regeneration.
1004 ** HANDLER-CASE failed to accept declarations in handler clauses
1005 in some circumstances.
1007 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1008 * New code in contrib/sb-introspect (still probably not entirely
1009 stable yet) provides some support for smart Lisp development
1010 environments like SLIME.
1011 * The conditions signalled for errors occurring when loading .fasl
1012 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1013 in a way which should help ASDF recover gracefully.
1014 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1015 stuff has been cleaned up. If you code contrib/ stuff, this might
1016 affect you, and you can look at contrib/README, contrib/STANDARDS,
1017 and/or the 0.8.4.27 diff to check.
1018 * In full calls the compiler now does not generate checks for declared
1019 argument types for all arguments.
1020 * various threading fixes
1021 ** and some experimental patches which didn't make it into
1022 the main tree for this release, but which are shipped in
1023 contrib/experimental-thread.patch as a possible fix for some
1024 failures (deadlock, spinning...) in GC-intensive multithreaded
1026 * fixed PPC build problem (source code incompatibility of different
1027 library versions): added offsetof() hackery which attempts to divine
1028 where glibc maintainers put uc_mcontext today
1029 * fixed bug 282: compiler does not trust type assertions while passing
1030 arguments to a full call.
1031 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1032 &OPTIONAL VALUES type parameter.
1033 * fix bug 214: algorithm for noting rejected templates is now more
1034 similar to that of template seletion. (also reported by rydis on
1036 * fixed bug 141b: printing backquoted information readably and prettily
1037 inserts a space where necessary.
1038 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1039 (CHARACTER 10) are now reported as errors, rather than propagated
1040 as unknown types. (reported by piso on #lisp)
1041 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1042 with values NIL and :ERROR. (thanks to Milan Zamazal)
1043 * fixed bug 191c: CLOS now does proper keyword argument checking as
1044 described in CLHS 7.6.5 and 7.6.5.1.
1045 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1046 longer raise an error; further, using a list as a for-as-arithmetic
1047 counter now raises a meaningful error.
1048 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1049 now understood by sequence creation functions such as MAKE-SEQUENCE
1051 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1052 a STRING-INPUT-STREAM.
1053 * compiler enhancement: SIGNUM is now better able to derive the type
1055 * type declarations inside WITH-SLOTS are checked. (reported by
1057 * fixed some bugs revealed by Paul Dietz' test suite:
1058 ** incorrect optimization of TRUNCATE for a positive first
1059 argument and negative second.
1060 ** compiler failure in let-convertion during flushing dead code.
1061 ** compiler failure while deriving type of TRUNCATE on an
1062 interval, containing 0.
1063 ** ASH of a negative bignum by a negative bignum count now returns
1065 ** intersection of CONS types now canonicalizes properly, fixing
1066 inconsistencies in SUBTYPEP.
1068 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1069 * incompatible change: The --disable-debugger command line
1070 option now clobbers the debugger at a more fundamental
1071 level, by redefining #'INVOKE-DEBUGGER instead of by
1072 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1073 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1074 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1075 enter the debugger REPL and then suffer recursive errors
1076 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1077 while under the new system, BREAK in --disable-debugger mode
1078 terminates the system just as an unhandled error would.
1079 * fixed compiler performance when processing loops with a step >1;
1080 * bug fix: DOCUMENTATION now retrieves generic function
1081 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1082 support has been systematized, and now supports the methods
1083 specified by ANSI, along with a default method and a method for
1084 slot documentation. (reported by Nathan Froyd)
1085 * bug fix: effective methods associated with a generic function are
1086 no longer cached over a change of that generic function's method
1087 combination. (reported by Andreas Fuchs)
1088 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1089 names a non-existent directory. (thanks to Andreas Fuchs)
1090 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1091 the PPC platform, lead to stack corruption; this has been fixed.
1092 (reported by Rainer Joswig)
1093 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1094 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1095 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1096 arithmetic in general, is now much more reliable on the PPC
1098 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1099 platform now returns the right answer.
1100 * optimization: restored some effective method precomputation in
1101 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1102 precomputation is now tunable.
1103 * optimization: compiler-internal data structure use has been
1104 reviewed, and changes have been made that should improve the
1105 performance of the compiler by about 20%.
1106 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1107 simple-base-strings and simple-bit-vectors is improved.
1108 * optimization: the optimization of 32-bit logical and arithmetic
1109 functions introduced in version 0.8.3 on the x86 has been
1110 implemented on the mips, ppc and sparc platforms; an
1111 implementation of the same facility, but for 64-bit arithmetic,
1112 has been added for the alpha.
1113 * microoptimization: the compiler is better able to make use of the
1114 x86 LEA instruction for multiplication by constants.
1115 * bug fix: in some situations compiler did not report usage of
1116 generic arithmetic in (SPEED 3) policy.
1117 * bug 145b fix: compiler used wrong type specifier while converting
1118 MEMBER-types to numeric.
1119 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1121 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1122 index argument now works properly on the Alpha platform.
1123 * bug fix: floating point exception treatment on the Alpha platform
1125 * bug fix: FILE-POSITION works much better on string input and
1126 output streams. (thanks to Nikodemus Siivola)
1127 * bug fix: many threading/garbage collection symptoms sorted.
1128 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1129 might be pseudo-atomic.
1130 * internal change: Stopping for GC is now done with signals not
1131 ptrace. GC is now done in whichever thread wanted it, instead of
1133 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1135 * bug fix: result form in DO is not contained in the implicit
1137 * incompatible change: ICR structure is changed; the value part of
1138 CONTINUATION is now called LVAR; corresponding functions are
1139 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1140 * added type deriver for ISQRT (thanks to Robert E. Brown).
1141 * bug fix: better support for loading from the command line when an
1142 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1144 * fixed some bugs revealed by Paul Dietz' test suite:
1145 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1146 ** ROUND and FROUND now give the right answer when given very
1147 small float arguments.
1148 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1150 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1151 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1152 ** LCM should return a non-negative integer.
1153 ** PARSE-INTEGER returned the index of a terminator instead of the
1154 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1155 ** PARSE-INTEGER returned an incorrect index being applied to a
1157 ** LCM with two arguments of 0 returns 0 rather than signalling
1159 ** unsigned addition of a 32-bit constant with the high bit set no
1160 longer causes an internal compiler error on the x86.
1161 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1162 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1163 with negative last argument.
1164 ** byte specifiers with zero size and position no longer cause
1165 an error during type derivation.
1166 ** bignum multiplication on the Alpha platform now returns the
1168 * porting: The system now builds on SuSE AMD64, although it still
1169 generates a 32-bit binary.
1170 * .fasl file incompatibility: The fasl file version number has
1171 been incremented (because of the changes to internal compiler
1172 data structures referred to above).
1174 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1175 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1176 more accurately, on the Darwin kernel running on PowerPC hardware.
1177 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1178 * Compiler code deletion notes now signal a condition of type
1179 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1180 an associated MUFFLE-WARNING restart.
1181 * The compiler now performs limited argument count validation of
1182 constant format strings in FORMAT, and where appropriate in ERROR,
1183 CERROR and WARN. (thanks to Gerd Moellmann)
1184 * New ASDF-INSTALL contrib can be used for automatic download and
1185 installation of third-party Lisp code from CCLAN or other sites
1187 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1188 which forces another thread to execute a function supplied by the
1190 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1191 now accept and act upon their :ELEMENT-TYPE keyword argument.
1192 (reported by Martin Atzmueller, Edi Weitz)
1193 * bug fix: FILE-POSITION now accepts position designators up to
1194 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1195 is the greater. (thanks to Patrik Nordebo)
1196 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1197 Kalle Olavi Niemitalo)
1198 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1199 printable. (reported by Eric Marsden)
1200 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1201 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1202 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1203 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1204 (thanks to Stig Sandoe)
1205 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1206 UNREAD-CHAR now correctly includes the unread character in the
1207 target sequence. (thanks to Gerd Moellmann)
1208 * bug fix in threaded builds: the system can now be suspended and
1209 resumed by shell job control with minimal disruption.
1210 * bug fixes in times and timezones >2038AD
1211 * better handling of "where is GNU make?" problem in build scripts
1212 (thanks to Nikodemus Siivola)
1213 * new optimization: inside a named function any reference to a
1214 function with the same name is considered to be a self-reference;
1215 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1216 optimization quality.
1217 * new optimization on x86: logical functions and + now have
1218 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1219 used when the result is truncated to 32 bits.
1220 * VALUES declaration is partially enabled.
1221 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1222 * fixed some bugs revealed by Paul Dietz' test suite:
1223 ** The system now obeys the constraint imposed by
1224 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1225 types form a lattice under type intersection.
1226 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1227 ** ASSOC now ignores NIL elements in an alist.
1228 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1229 and (1+ MOST-POSITIVE-FIXNUM) answers.
1230 ** The addition of a method with invalid qualifiers to a generic
1231 function does not cause an error to be signalled immediately;
1232 a warning is signalled, and the error is generated only on
1233 calling the generic function.
1234 * changed .fasl file version number, in part to add type codes for
1235 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1236 obscure ANSI requirements
1238 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1239 * fixed bug 148: failure to inline-expand a local function left
1240 garbage, confusing the compiler.
1241 * fixed bugs 3cd: structure slot readers perform type check if the
1242 slot can have an invalid value (i.e. it is either not initialized
1243 or can be written with a less specific slot writer).
1244 * bug fix: the compiler now traps array references to elements off
1245 the end of an array; previously, the bounds checking in some
1246 circumstances could go off-by-one.
1247 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1249 * type declarations for array element types now obey the description
1250 on the CLHS page "Declaration TYPE", as per discussions on
1251 sbcl-help around 2003-05-08. This means that a declaration
1252 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1253 declaration, all references to BAR will be asserted or assumed
1254 (with THE, so dependent on compiler policy) to involve objects of
1255 type FOO. Note that no such declaration is implied in
1256 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1257 * declared types of functions from the "Conditions"
1258 chapter. (reported by Paul Dietz)
1259 * bug fix: CERROR accepts a function as its first argument.
1260 * bug fix: NTH an NTHCDR accept a bignum as index
1261 arguments. (reported by Adam Warner)
1262 * optimization: character compare routines now optimize comparing
1263 against a constant character. (reported by Gilbert Baumann)
1264 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1265 argument now works properly on the MIPS platform.
1266 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1267 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1268 declared to have a functional type, does not cause an error
1270 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1271 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1272 * fixed bug: if last continuation of a deleted block has a
1273 destination, this destination should be deleted too. (reported by
1275 * fixed a bug in the bootstrap process: the host compiler's values
1276 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1277 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1278 test case from Patrik Nordebo)
1279 * improved the ability of the disassembler on the PPC platform to
1280 provide helpful disassembly notes.
1281 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1282 the class in more cases than previously.
1283 * bug fix: FILE-POSITION now understands :START and :END for
1284 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1285 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1286 8) data. (thanks to David Lichteblau)
1287 * bug fix: it is possible to add a method to a generic function
1288 without lambda list.
1289 * bug fix: reader failed to signal END-OF-FILE inside an
1290 object representation. (reported by Nikodemus Siivola)
1291 * fixed some bugs revealed by Paul Dietz' test suite:
1292 ** LAST and [N]BUTLAST should accept a bignum.
1293 ** condition slot accessors are methods.
1294 ** (VECTOR NIL) is a subtype of STRING.
1296 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1297 * minor incompatible change: some nonsensical specialized lambda
1298 lists (used in DEFMETHOD) which were previously ignored now signal
1300 * minor incompatible change: the system is now aware of the types of
1301 variables in the COMMON-LISP package, and will signal errors for
1302 most violations of these type constraints (where previously they
1303 were silently accepted).
1304 * minor incompatible change: COMPILE-FILE now uses the freedom
1305 afforded (ANSI 3.2.2.3) to use derived function types for
1306 functions defined in the same file. This also permits the system
1307 to warn on static type mismatches and function
1308 redefinition. (Currently it does not work with high DEBUG level.)
1309 * minor incompatible change: VALUES declaration is disabled.
1310 * When issuing notes, the compiler now signals a condition of type
1311 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1312 restart for use in user handlers. It is expected that the
1313 COMPILER-NOTE condition will eventually become a condition
1314 supertype to a hierarchy of note types, which will then be
1315 handleable in a similar fashion. However, at the moment, no such
1316 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1317 but it's an implementation detail, not a classification for the
1319 * Changes in type checking closed the following bugs:
1320 ** type checking of unused values (192b, 194d, 203);
1321 ** template selection based on unsafe type assertions (192c, 236);
1322 ** type checking in branches (194bc).
1323 * A short form of VALUES type specifier has ANSI meaning (it has
1324 increased the number of situations when SBCL cannot perform type
1326 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1327 keywords or constants is permissible.
1328 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1329 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1330 classes. (thanks to Antonio Martinez)
1331 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1332 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1333 argument to a pretty-print stream. Instead, it leaves any such
1334 operations to DESCRIBE-OBJECT methods.
1335 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1337 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1338 lambda lists. (thanks to David Lichteblau)
1339 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1340 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1341 (thanks to Gerd Moellmann)
1342 * fixed bug 246: increased compilation speed of long
1343 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1345 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1347 * DEFCONSTANT now throws a condition of type
1348 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1349 constant to a non-EQL value; CONTINUE and ABORT restarts
1350 respectively change and preserve the value.
1351 * fixed bug 63: The code walker, part of the implementation of CLOS,
1352 is now better at handling symbol macros.
1353 * bug fix: There is no longer an internal implementation type named
1354 CL:LENGTH. (reported by Raymond Toy)
1355 * bug fix: In macro-like defining macros/special operators the
1356 implicit block does not enclose the lambda list.
1357 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1358 enclosed in parentheses) are not suitable as type specifiers, and
1359 their use properly signals an error now.
1360 * bug fix: An argument count mismatch for a type specifier in code
1361 being compiled no longer causes an unhandled error at compile
1362 time, but signals a compile-time warning.
1363 * fixed simple vector readable printing
1364 * bug fix: DESCRIBE takes more care over whether the class
1365 precedence list slot of a class is bound before accessing it.
1366 (reported by Markus Krummenacker)
1367 * bug fix: FORMATTER can successfully compile pretty-printer format
1368 strings which use variants of the ~* directive inside.
1369 * bug fix: SEARCH now applies its TEST predicate to the elements of
1370 the arguments in the correct order. (thanks to Wolfhard Buss)
1371 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1372 in type inference. (thanks to Robert E. Brown)
1373 * bug fix: Reading in symbols with an explicit package name of ""
1374 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1375 * fixed some bugs revealed by Paul Dietz' test suite:
1376 ** NIL is now allowed as a structure slot name.
1377 ** Arbitrary numbers, not just REALs, are allowed in certain
1378 circumstances in LOOP for-as-arithmetic clauses.
1379 ** Multiple class redefinitions before slot access no longer
1380 causes a type error.
1381 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1382 association between the name and a class.
1383 ** Generic functions with non-standard method-combination and over
1384 five methods all of which return constants no longer return NIL
1385 after the first few invocations. (thanks to Gerd Moellmann)
1386 ** CALL-NEXT-METHOD with no arguments now passes the original
1387 values of the arguments, even in the presence of assignment.
1388 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1390 ** DISASSEMBLE works with closures and funcallable instances.
1391 ** ADD-METHOD now returns the generic function, not the new method.
1392 ** FIND-METHOD signals an error if the lengths of the specializers
1393 is incompatible with the generic function, even if the ERRORP
1395 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1396 which its argument is a member.
1397 ** DEFCLASS only redefines the class named by its class-name
1398 argument if that name is the proper name of the class;
1399 otherwise, it creates a new class.
1400 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1401 of the UNBOUND-SLOT condition to the name of the slot.
1402 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1403 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1404 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1405 SLOT-MAKUNBOUND in the specified fashion.
1407 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1408 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1409 cross-compilation host. As a consequence, we can now bootstrap our
1410 way up to SBCL starting with a bare gcc toolchain and human-readable
1411 source code (first the source to CLISP, then the source to SBCL).
1412 * A contributed module containing a partial implementation of the
1413 simple-streams interface has been included. (thanks to Rudi
1415 * A contributed module implementing the RFC1321 Message Digest
1416 Algorithm, known as MD5, has been included.
1417 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1418 no longer has any effect, as the code controlled by this feature
1419 has been deleted. (As far as we know, no-one has ever built using
1420 this feature, and its semantics were confused in any case).
1421 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1422 work on conditions (as required by the ANSI specification),
1423 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1424 expected behaviour on conditions. Users should note, however,
1425 that such behaviour is not required by the ANSI specification,
1426 and so use of this behaviour may render their code unportable.
1427 * fixed some bugs revealed by Paul Dietz' test suite:
1428 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1430 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1431 ** SXHASH on condition objects no longer returns NIL.
1432 ** :ALLOCATION :CLASS slots are better treated; their values are
1433 updated on class redefinition, and initforms inherited from
1434 superclasses are applied.
1435 ** REMOVE-METHOD returns its generic function argument even when
1436 no method was removed.
1437 ** SHARED-INITIALIZE now initializes the values of the requested
1438 slots, including those with :ALLOCATION :CLASS.
1439 ** ALLOCATE-INSTANCE now works on structure classes defined via
1440 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1442 ** SLOT-EXISTS-P now works on conditions, as well as structures
1444 ** MAKE-LOAD-FORM now has the required methods on
1445 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1446 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1447 keyword, and now implements the SLOT-NAMES argument.
1448 ** methods with &OPTIONAL arguments no longer allow too many
1449 arguments to be passed in the call without error.
1450 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1451 option is consistent with the required arguments of the generic
1452 function lambda list.
1453 * bug fix: REQUIRE accepts a string designator. (Thanks to
1455 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1456 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1457 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1458 * bug fix: The compiler now checks for duplicated variables in macro
1460 * bug fix: SETQ on globals returns the correct value.
1461 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1462 not a COMPILER-ERROR (followed by some other strange error on
1463 choosing the CONTINUE restart).
1464 * bug fix: make.sh and friends are now more consistent in the way that
1465 they look for GNU "make".
1467 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1468 * experimental native threads support (on x86 Linux >=2.4 only).
1469 This is not compiled in by default: you need to add :SB-THREAD to
1470 the target features. See the "Beyond ANSI" chapter of the manual
1472 * fix for longstanding nonANSIism: The old distinction between
1473 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1474 The return value from CL:FIND-CLASS is now a CLOS class, and
1475 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1476 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1477 * An interface to the MetaObject Protocol, as described in Kiczales,
1478 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1479 MIT Press, 1991, is available from the SB-MOP package.
1480 * incompatible change: the SB-PCL package should now be considered
1481 a private implementation detail, and no longer a semi-private MOP
1483 * minor incompatible change: due to rearrangement for threads, the
1484 control stack and binding stack are now allocated at arbitrary
1485 addresses instead of being hardcoded per-port. Users affected by
1486 this probably have to be doing advanced things with shared
1487 libraries, and will know who they are.
1488 * minor incompatible change: Previously, all --eval forms used were
1489 processed with READ before any of them were processed with EVAL.
1490 Now each --eval form is processed with both READ and EVAL before
1491 the next --eval form is processed. (Thus package operations like
1492 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1493 work as the user might reasonably expect.)
1494 * minor incompatible change: *STANDARD-INPUT* is now only an
1495 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1497 * minor incompatible change: Y-OR-N-P is now character-oriented,
1498 not line oriented. Also, YES-OR-NO-P now works without errors.
1499 (thanks to Antonio Martinez)
1500 * sb-aclrepl module improvements: an integrated inspector, added
1501 repl features, and a bug fix to :trace command.
1502 * Known functions, which cannot be open coded by the backend, are
1503 considered to be able to check types of their arguments. (fixing
1504 a bug report by Nathan J. Froyd)
1505 * fixed a bug in computing method discriminating functions: It is
1506 now possible to define methods specialized on classes which have
1507 forward-referenced superclasses. (thanks to Gerd Moellmann)
1508 * fixed evaluation order in optional entries (reported by Gilbert
1510 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1511 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1512 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1513 required-by-AMOP NAME argument, as well as CLASS and
1514 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1515 * fixed bug 20: DEFMETHOD can define methods using names that are
1516 not the proper names of classes to designate class specializers.
1517 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1518 Unix file descriptors, instead of blowing up. (thanks to Antonio
1520 * Garbage collection refactoring: user-visible change is that a
1521 call to the GC function during WITHOUT-GCING will not do garbage
1522 collection until the end of the WITHOUT-GCING. If you were doing
1523 this you were probably losing anyway.
1524 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1525 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1526 TYPEP the latter but not the former.
1527 * The compiler issues a full WARNING for calls to undefined functions
1528 with names from the CL package.
1529 * MAP-INTO for a vector destination is open coded. (reported by
1530 Brian Downing on c.l.l)
1531 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1532 documentation string.
1533 * fixed some bugs revealed by Paul Dietz' test suite:
1534 ** COPY-ALIST now signals an error if its argument is a dotted
1536 ** Condition slots are now accessed more correctly in the presence
1537 of multiple initargs for a given slot.
1538 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1539 exclude restarts of the same name associated with a different
1541 ** DEFCLASS of forward-referenced classes with another
1542 forward-referenced class in the superclasses list no longer
1544 ** Condition slots are now initialized once each, not multiple
1545 times. (thanks to Gerd Moellmann)
1546 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1548 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1549 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1550 arguments contain duplicated elements.
1551 ** RESTART-CASE understands local macros.
1552 ** RESTART-CASE associates exactly its own restarts with a condition.
1553 ** ENDP in safe mode checks its argument to be of type LIST.
1554 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1555 in question is unbound.
1556 ** Optimized MAKE-INSTANCE functions no longer cause internal
1557 assertion failures in the presence of duplicate initargs.
1558 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1559 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1560 propagated even in the ctor optimized implementation of
1562 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1564 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1565 * a better implementation of SXHASH on (simple) bit vectors,
1566 measured both in execution speed and in distribution of results
1567 over the positive fixnums, has been installed. Likewise, a better
1568 implementation of EQUAL for simple bit vectors is now available.
1569 * fixed CEILING optimization for a divisor of form 2^k.
1570 * fixed bug 240 (emitting extra style warnings "using the lexical
1571 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1572 by Antonio Martinez)
1573 * fixed SXHASH, giving different results for NIL depending on type
1574 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1575 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1576 variable. (found by Rolf Wester)
1577 * SBCL does not ignore type declarations for special
1578 variables. (reported by rif on c.l.l 2003-03-05)
1579 * some bug fixes in contrib/sb-aclrepl/
1580 * fixed some bugs revealed by Paul Dietz' test suite:
1581 ** a bug in the CONS type specifier, whereby the CAR and CDR
1582 types got intertwined, has been fixed;
1583 ** the type system is now able to reason about the interaction
1584 between INTEGER and RATIO types more completely;
1585 ** APPEND, [N]REVERSE and NRECONC check that those their
1586 arguments, which must be proper lists, are really so;
1587 ** An array specialized to be unable to hold elements has been
1588 implemented, as required -- yes, really -- by ANSI;
1589 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1590 on malformed property lists;
1592 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1593 * incompatible packaging change: in line with Unix convention,
1594 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1595 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1596 * REQUIRE and PROVIDE are now optionally capable of doing something
1597 useful. See the documentation string for REQUIRE.
1598 * infrastructure for a managed SBCL contrib system: contributed
1599 modules in this release include:
1600 ** the ASDF system definition facility;
1601 ** an interface to the BSD Sockets API;
1602 ** an ACL-like convenience interface to the repl;
1603 (thanks to Kevin Rosenberg)
1604 ** an implementation of ROTATE-BYTE, with efficient implementation
1606 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1607 gives a closer approximation to the right answer than previously.
1608 (thanks to Raymond Toy)
1609 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1610 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1611 optional environment arguments, as required by ANSI.
1612 * fixed bugs in other functions taking environment objects, allowing
1613 calls with an explicit NIL environment argument to be compiled
1615 * fixed bug 228: primary return values from
1616 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1617 COMPILE or FUNCTION.
1618 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1619 :TYPE will no longer signal errors on innocuous objects.
1620 * fixed bug 231b: SETQ is better at respecting type declarations in
1621 the lexical environment.
1622 * fixed a bug in DEFCLASS: classes named by symbols with no or
1623 unprintable packages can now be defined.
1624 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1625 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1626 * fixed compiler failure related to checking types of functions.
1627 (reported by Robert E. Brown)
1628 * the compiler is now much more consistent in its error-checking
1629 treatment of bounding index arguments to sequence functions: in
1630 (SAFETY 3) code, errors will be signalled in almost all cases if
1631 invalid sequence bounding indices are passed to functions defined
1632 by ANSI to operate on sequences.
1633 * fixed a bug in the build procedure: documentation of SBCL-specific
1634 packages is now preserved and available in the final Lisp image.
1635 * lifted FDEFINITION lookup out of loops in the implementation of
1636 many list operations. (thanks to Robert E. Brown)
1637 * fixed a bug in the reader: the #n# reader macro now works for
1638 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1639 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1640 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1641 E. Brown and Tony Martinez respectively)
1642 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1643 SLOT-BOUNDP: in particular, it is now able to optimize them much
1644 better, and is now not vulnerable to having packages renamed.
1645 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1646 always called when a slot is not present in an instance. (thanks
1648 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1649 longer signal an error on unreading a character following EOF on
1650 the previous constituent stream. (thanks to Tony Martinez)
1651 * fixed some bugs revealed by Paul Dietz' test suite:
1652 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1653 not just nonnegative fixnums;
1654 ** the logical bit-array operators such as BIT-AND now accept an
1655 explicit NIL for their "opt-arg" argument (to indicate a
1656 freshly-consed result bit-array);
1657 ** ELT now signals an error on an invalid sequence index in safe
1659 ** the type system is now cleverer about negations of numeric
1660 types, and consequently understands the BIGNUM and RATIO types
1662 ** the type system is now cleverer about the interaction between
1663 INTEGER and RATIO types: while bugs still remain, many more
1664 cases are accurately computed;
1665 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1666 if it is in the last clause;
1667 ** CONSTANTLY now correctly returns a side-effect-free function in
1669 ** DECLARE is no longer treated as a special-operator; in
1670 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1671 * incremented fasl file version number due to the change in the
1672 DEFSTRUCT-SLOT-DESCRIPTION structure.
1674 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1675 * minor incompatible change: code processed by the "interpreter" or
1676 EVAL now has a compilation optimization policy of (DEBUG 2)
1677 (changed from (DEBUG 1)) to improve debuggability of interactive
1678 development, and to allow the use of the debug RETURN command in
1680 * an experimental implementation of the RETURN command for the
1681 debugger has been included. (thanks to Frederik Kuivinen)
1682 * fixed bug 62: constraints were not propagated into a loop.
1683 * fixed bug in embedded calls of SORT (reported and investigated by
1685 * fixed some bugs revealed by Paul Dietz' test suite:
1686 ** printing and reading of arrays with some dimensions having
1687 length 0 (thanks to Gerd Moellmann);
1688 ** BOA constructor with &AUX argument without a default value does
1689 not cause a type error;
1690 ** CONSTANTP now returns true for all self-evaluating objects.
1692 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1693 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1694 accessors that are related by inheritance, as specified in the
1695 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1696 Valtteri Vuorikoski)
1697 * The compiler is now able to inline functions that were defined in
1698 a complex lexical environment (e.g. inside a MACROLET).
1699 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1701 * Support for the upcoming FreeBSD-5.0 release has been included.
1702 (thanks to Dag-Erling Smorgrav)
1703 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1704 effect when it is not in a toplevel context.
1705 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1706 stand a better chance of being correct. (thanks to Gerd
1708 * fixed bug in COERCE, which now signals an error on coercing a
1709 rational to a bounded real type which excludes the expected
1711 * The compiler is now able to derive types more accurately from the
1712 COERCE and COMPILE functions.
1713 * fixed bug 223: functional binding is considered to be constant
1714 only for symbols in the CL package.
1715 * fixed bug 231: SETQ did not check the type of a variable being set
1716 (reported by Robert E. Brown)
1717 * A new optimization for MAKE-INSTANCE has been included, fixing
1718 various bugs (including relating to :ALLOCATION :CLASS slots and
1719 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1721 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1722 ** As required by ANSI, LOOP now disallows anonymous collection
1723 clauses such as COLLECT I in conjunction with aggregate boolean
1724 clauses such as THEREIS (= I 1);
1725 ** LOOP now signals an error when any variable is reused in the
1726 same loop (including the potentially useful construct analogous
1727 to WITH A = 1 WITH A = (1+ A);
1728 ** IT is only a special loop symbol within the first clause of a
1729 conditional loop clause;
1730 ** LOOP with a typed iteration variable over a hashtable now
1731 signals a type error iff it should.
1732 * fixed some other bugs revealed by Paul Dietz' test suite:
1733 ** FILE-STREAM now names the class previously known as FD-STREAM;
1734 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1735 argument) no longer signals an error;
1736 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1737 of the slot symbol, rather than using the current package
1738 ((:CONC-NAME "") continues to intern the slot's name in the
1740 * incremented fasl file version number, because of the incompatible
1741 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1742 of the new implementation of DEFINE-COMPILER-MACRO.
1744 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1745 * Support for building SBCL for MIPS platforms running in
1746 little-endian mode has now been checked in, and basic
1747 functionality on said platforms verified.
1748 * minor incompatible change: PCL now records the pathname of a file
1749 in which methods and the like are defined, rather than its
1751 * minor incompatible change: TRUENAME now considers the truename of
1752 a file naming a directory to be the pathname with :DIRECTORY
1753 component indicating that directory.
1754 * minor incompatible change: a NAMED clause in the extended form of
1755 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1756 reason for the previous behaviour is unclear.
1757 * more systematization and improvement of CLOS and MOP conformance
1758 in PCL (thanks to Gerd Moellman and Pierre Mai):
1759 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1761 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1762 primary methods with no specializers;
1763 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1765 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1766 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1767 FORWARD-REFERENCED-CLASSes; error reporting on
1768 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1770 ** SXHASH on CLOS instances now uses a slot internal to the
1771 instance to return different numbers on distinct instances,
1772 while preserving the same return value through invocations of
1774 ** DEFMETHOD signals errors when methods with longer incongruent
1775 lambda lists are added to generic functions;
1776 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1777 CLASS, as specified in AMOP;
1778 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1779 on the order returned by the primary method for classes of
1780 class STANDARD-CLASS;
1781 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1782 * fixed some bugs shown by Paul Dietz' test suite:
1783 ** DOLIST puts its body in TAGBODY;
1784 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1786 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1787 value producing form;
1788 ** if more variables are given to PROGV than values, extra
1789 variables are bound and made to have no value;
1790 ** NSUBSTITUTE on list arguments gets the right answer with
1792 ** ELT signals an error of type TYPE-ERROR when the index argument
1793 is not a valid sequence index;
1794 ** LOOP signals (at macroexpansion time) an error of type
1795 PROGRAM-ERROR when duplicate variable names are found;
1796 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1797 ** FUNCALL of special-operators now cause an error of type
1799 ** PSETQ now works as required in the presence of side-effecting
1800 symbol-macro places;
1801 ** NCONC accepts any object as its last argument;
1802 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1804 ** loop-for-as-package does not require a package to be explicitely
1806 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1808 * fixed bug 166: compiler preserves "there is a way to go"
1809 invariant when deleting code.
1810 * fixed bug 172: macro lambda lists with required arguments after
1811 &REST arguments now cause an error to be signalled. (thanks to
1813 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1814 bug. (thanks to Matthew Danish)
1815 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1817 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1818 arguments are lexically rebound. (thanks to Gerd Moellmann and
1820 * fixed bug 194: error messages are now more informative when there
1821 is no primary method applicable in a call to a generic
1822 function. (thanks to Gerd Moellmann)
1823 * fixed bug in command line argument checking (thanks to Julian
1825 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1826 (thanks to Matthew Danish)
1827 * incremented fasl file version number, because of the
1828 SXHASH-related changes in the layout of CLOS data structures
1830 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1831 * minor incompatible change: The runtime (the Unix executable named
1832 "sbcl") is now much pickier about the .core files it will load.
1833 Essentially it now requires .core files to descend from the same
1834 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1835 as the runtime does. (The intent is to prevent the crashes which
1836 can occur, and which can even be reported as mysterious failures,
1837 when people patch the sources or change the build parameters
1838 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1839 sbcl and .core files.)
1840 * fixed bug: VALUES-LIST is no longer optimized away.
1841 * fixed bug 142: The FFI conversion of C string values to Lisp
1842 string values no longer conses excessively. (thanks to Nathan
1843 Froyd porting Raymond Toy's fix to CMU CL)
1844 * began to systematize and improve MOP conformance in PCL (thanks to
1845 Nathan Froyd, Gerd Moellman and Pierre Mai):
1846 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1848 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1849 ** FINALIZE-INHERITANCE is now called on class finalization;
1850 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1851 argument precedence order.
1852 * fixed bug 202: The compiler no longer fails on functions whose
1853 derived types contradict their declared type.
1854 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1855 so it can be non-toplevel.
1856 * The fasl file version number has changed (because of the new
1857 implementation of DEFMACRO).
1858 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1859 safe code, that any length requirement by their type-specifier
1860 argument is valid. The exceptions to this are described in bug
1862 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1863 functions now check the types of their inputs as required by ANSI.
1864 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1865 introduced symbol is DECLAREd to be SPECIAL.
1866 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1867 * fixed bug: :COUNT argument to sequence functions may be negative
1868 * fixed bug: body of DO-SYMBOLS may contain declarations
1869 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1870 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1871 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1872 symbol macro only once
1873 * fixed printing of call frame when argument list is unavailable
1874 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1875 * compiler no longer signals WARNING on unknown keyword
1878 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1879 * A beta-quality port to the mips architecture running Linux,
1880 based on the old CMUCL backend, has been made. It has been tested
1881 on a big-endian kernel, and works sufficiently well to be able to
1882 rebuild itself; it has not been tested in little-endian mode.
1883 * fixed an inconsistency between gencgc.c and purify.c which made
1884 dumping/loading .core files unreliable
1885 * fixed bug 120a: The compiler now deals correctly with IFs where
1886 the consequent is the same as the alternative, instead of
1887 misderiving the return type. (thanks to Alexey Dejneka)
1888 * fixed bug 113: Logical pathnames are now dumpable (the logical
1889 host is resolved at load-time, throwing an error if it is not
1891 * fixed bug 174: FORMAT's error message is slightly clearer when a
1892 non-printing character is used in a format directive.
1893 * fixed several bugs in compiler checking of type declarations, i.e.
1894 violations of the Python "declarations are assertions" principle
1895 (thanks to Alexey Dejneka)
1896 * fixed several bugs in PCL's error checking (thanks to Gerd
1898 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1900 * fixed bug in compilation of functions as first class values
1901 (thanks to Antonio Martinez-Shotton)
1902 * The compiler's handling TYPE-ERRORs which it can prove will
1903 inevitably happen at runtime has been cleaned up and corrected
1904 in several ways. (thanks to Alexey Dejneka)
1905 * improved argument type checking for various basic arithmetic
1906 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1907 lavished on them in the past that they can be compiled in many
1908 ways in different special cases
1909 * fixed bug 181: compiler checks validity of user supplied type
1911 * cleaned up code flushing in optimization: Function calls which
1912 should signal errors for safety purposes (e.g. which ANSI says
1913 should signal errors when their arguments are of incorrect type)
1914 are no longer optimized away.
1915 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1916 * incremented fasl file version number, because changes in the
1917 implementation of sequence functions like COERCE caused
1918 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1919 in old inline expansions) to become undefined. (Actually these
1920 changes were later undone, so we might very well be binary
1921 compatible with 0.7.7 after all, but leaving the version number
1922 incremented seemed like the simplest and most conservative
1925 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1926 * An alpha-quality port to the parisc architecture running Linux,
1927 based on the old CMUCL backend, has been made. This, even more so
1928 than the other backends, should be considered still a work in
1929 progress; known problems include that the Linux kernel in 64-bit
1930 mode does not propagate the correct sigcontext structure to
1931 userspace, and consequently SBCL on a parisc64 kernel will not
1933 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1934 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1935 suppressed.) Also now that the compiler is looking at declarations
1936 in the environment, it checks optimization declarations as well,
1937 and suppresses inlining when (> DEBUG SPEED).
1938 * More fixes have been made to treatment of floating point exception
1939 treatment and other Unix signals. In particular, floating point
1940 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1941 * The detection and handling of control stack exhaustion (infinite
1942 or very deeply nested recursion) has changed. Stack exhaustion
1943 detection is now done by write-protecting pages at the OS level
1944 and applies at all optimization settings; when found, a
1945 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1946 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1947 be caught using IGNORE-ERRORS.
1948 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1949 correctly when transforming recursive LABELS functions to LETs.
1950 (thanks to Alexey Dejneka)
1951 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1952 that are names of constants or global variables.
1953 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1954 alien routines with docstrings.
1955 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1956 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1958 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1959 (thanks to Eric Marsden)
1960 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1961 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1962 * Bug fix: The SPARC backend can now compile functions involving
1963 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1964 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1965 to a functions expecting a PATHNAME-DESIGNATOR.
1966 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1967 lambda lists. (thanks to Alexey Dejneka)
1968 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1969 functions so that EXPORTing the name of the class doesn't cause
1970 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1971 to Antonio Martinez for reporting this)
1972 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1973 constant-folded without failing an assertion. (thanks to Einar
1974 Floystad Dorum for reporting this)
1975 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1976 types are now dealt with more correctly by the compiler.
1977 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1978 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1979 behaviour analogous to RENAME-FILE. This puts its behaviour more
1980 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1982 * The fasl file version number has changed again. (because of the
1983 bug fix involving the names of PCL MAKE-INSTANCE functions)
1985 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1986 * bug fix: Floating point exceptions are treated much more
1987 consistently on the x86/Linux and PPC/Linux platforms.
1988 * Array initialization with :INITIAL-ELEMENT is now much faster for
1989 cases when the compiler cannot open code the array creation, but
1990 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1991 array accesses have also seen a speed increase.
1992 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1993 specified. (This was at the root of some bad interactions between
1994 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1996 * bug fix: Internal error arguments for undefined functions are now
1997 computed correctly on the PPC/Linux platform.
1998 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1999 Raymond Toy's patch for CMU CL)
2000 * Support for the Solaris 9 operating environment has been included
2001 (thanks to Daniel Merritt)
2002 * A very ugly but hopefully complete draft of the missing FFI chapter
2003 of the manual has been created by reformatting the corresponding
2004 CMU CL manual chapter into (currently very ugly and incoherent)
2005 DocBook and bringing it up to date for SBCL behavior. Thus, the
2006 manual is now essentially complete, at least by my extreme
2007 once-and-only-once standards, whereby it's acceptable to refer to
2008 the doc strings of SB-EXT functions as the primary documentation.
2009 * The fasl file version number has changed again, due to cleanup of
2010 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2011 is no longer a static symbol.)
2013 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2014 * SBCL now builds with OpenMCL (version 0.12) as the
2015 cross-compilation host; also, more progress has been made toward
2016 bootstrapping under CLISP.
2017 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2019 * bug 158 fixed: The compiler can now deal with integer loop
2020 increments different from 1; fixing this turned out also to fix
2022 * bug 169 fixed: no more bogus warnings about using lexical bindings
2023 despite the presence of perfectly good SPECIAL declarations (thanks
2024 to David Lichteblau)
2025 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2026 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2027 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2029 * bug 180 fixed: Method combination specifications no longer ignore
2030 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2031 * bug fix: Structure type predicate functions now check their argument
2032 count as they should.
2033 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2034 correctly. (thanks to Pierre Mai)
2035 * minor incompatible change: The --noprogrammer option is deprecated
2036 in favor of the new --disable-debugger option, which is very similar.
2037 (The major difference is that it takes effect at a slightly different
2038 time at startup, causing handling of errors in --sysinit and
2039 --userinit files will be affected differently.) The
2040 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2041 been added to allow this functionality to be controlled from ordinary
2042 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2043 maintainers, who might want to run non-interactive scripts to
2044 build SBCL cores which will later be used interactively.)
2045 * minor incompatible change: The LOAD function no longer, when given
2046 a wild pathname to load, loads all files matching that pathname.
2047 Instead, an error of type FILE-ERROR is signalled.
2049 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2050 * bug 147 fixed: The compiler preserves its block link/count
2051 invariants more correctly now so that it doesn't crash. (thanks
2053 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2055 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2056 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2057 no-value-for-this-character as NIL (as specified by ANSI).
2058 * HOST-NAMESTRING on physical pathnames now returns a string that is
2059 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2060 (thanks to Christophe Rhodes)
2061 * The Alpha port handles icache flushing more correctly. (thanks to
2063 * More progress has been made toward bootstrapping under CLISP. (thanks
2064 to Christophe Rhodes)
2065 * The fasl file format has changed again, because dynamic loading
2066 on OpenBSD (which has non-ELF object files) motivated some cleanups
2067 in the way that foreign symbols are transformed and passed around.
2068 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2069 is no longer treated as whitespace by the reader, but instead as
2070 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2071 |A
\7fB|, instead of A as it used to.
2073 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2074 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2075 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2076 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2077 this as of 0.7.1.45, but was left out of the previous news section
2078 (thanks to Dan Barlow)
2079 * SBCL now runs on the Solaris operating system on SPARC architectures
2080 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2081 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2082 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2083 * SPARC backend cleanups, allowing builds of cores optimized for V8
2084 and V9 SPARCS, and also emission of code targeted to a particular
2085 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2087 * SBCL is closer to bootstrapping under CLISP, thanks to various
2088 fixes by Christophe Rhodes.
2089 * The fasl file format has changed again, to allow the compiler's
2090 INFO database to support symbol macros.
2091 * The user manual (in doc/) is formatted into HTML more nicely.
2092 (thanks to coreythomas)
2093 * The system is smarter about SUBTYPEP relationships, especially
2094 those involving NOT types (including types such as ATOM which are
2095 represented internally using NOT types). Thus SUBTYPEP is less
2096 likely to return (VALUES NIL NIL) in general, and in particular
2097 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2099 * The fasl file format has changed again, because the internal
2100 representation of types now includes a new slot to support the new
2101 SUBTYPEP-of-NOT-types logic.
2102 * (not a change in the main branch of SBCL, but a related prototype
2103 which can hopefully be merged into the main branch of SBCL in the
2104 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2105 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2106 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2107 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2108 dispatch macro character combinations. (thanks to Alexey Dejneka)
2109 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2110 that superficially look like logical namestrings correctly.
2111 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2113 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2114 default supplied-p) for &optional and &key arguments. (thanks to
2117 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2118 * incompatible change: The compiler is now less aggressive about
2119 tail call optimization, doing it only when (> SPACE DEBUG) or
2120 (> SPEED DEBUG). (This is an incompatible change because there are
2121 programs which relied on the old CMU-CL-style behavior to optimize
2122 away their unbounded recursion which will now die of stack overflow.)
2123 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2124 for non-GENCGC systems has been increased to 20M (since that
2125 seems much closer to the likely performance optimum for modern
2126 systems than the old 4M value was)
2127 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2128 and *DEBUG-PRINT-LEVEL*
2129 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2130 of CMU CL's support for SPARC, and various endianness and other
2131 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2132 * new syntactic sugar for the Unix command line: --load foo.bar is now
2133 an alternate notation for --eval '(load "foo.bar")'.
2135 ** The system now detects stack overflow and handles it gracefully,
2136 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2137 optimization settings. (This is a good thing in general, and
2138 its introduction in this version should be particularly timely
2139 for anyone whose code fails because of suppression of tail
2141 ** The system now hunts for the C variable "environ" in a more
2142 devious way, to avoid segfaults when the C library version differs
2143 between compile time and run time. (thanks to Christophe Rhodes)
2144 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2145 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2146 ** The compiler no longer issues bogus style warnings for undefined
2147 classes in the same source file as the DEFCLASSes which defined
2148 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2150 ** fixes in CONDITION class precedence list for undefined function
2151 errors (thanks to Alexei Dejneka)
2152 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2153 correctly. (thanks to Dan Barlow)
2154 ** portability fixes aiming at bootstrapping under CLISP (thanks
2155 to Dave McDonald and Christophe Rhodes)
2156 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2157 ** fixes in type translation and and type inference (thanks to
2159 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2160 ** various fixes in the new ports (thanks to Dan Barlow)
2161 * several changes related to debugging:
2162 ** suppression of tail recursion, as noted above
2163 ** stack overflow detection, as noted above
2164 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2165 is now the default. (For some time encapsulation has been more
2166 reliable than the breakpoint-based :ENCAPSULATE NIL
2167 implementation, at least on X86 systems; and I just noticed that
2168 encapsulation also seems closer to the spirit of the ANSI
2171 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2173 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2174 up properly again. (There was a packaging bug in 0.7.0 which
2175 left their definitions in SB-SYS::LOAD-FOREIGN and
2176 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2177 vital for most things which interface to C-level interfaces,
2178 like extensions working with sockets or databases or
2179 Perl-compatible regexes or whatever, and the need to fix
2180 this bug is the main reason that 0.7.1 was released so
2182 ** DEFGENERIC is now choosier about the methods it redefines, so that
2183 reLOADing a previously-LOADed file containing DEFGENERICs does
2184 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2185 cycle now works as it should. (thanks to Alexey Dejneka)
2186 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2187 fixed by Christophe Rhodes. (He actually submitted this patch
2188 months ago, and I delayed until after 0.7.0.)
2189 ** Bug 111 (internal compiler confusion about runtime checks on
2190 FUNCTION types) was fixed by Alexey Dejneka.
2191 * Some internal cleanups (getting rid of variables which aren't
2192 needed now that the byte interpreter is gone) caused the fasl
2193 file format number to change again.
2195 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2196 * major incompatible change: The default fasl file extension, i.e. the
2197 default extension for files produced by COMPILE-FILE, has changed
2198 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2200 ** There are many changes in the implementation of the compiler.
2201 SBCL is now essentially a compiler-only implementation of ANSI
2202 Common Lisp. EVAL still "interprets" a few special cases, but
2203 almost all the interesting cases are handled by creating
2204 a LAMBDA expression, calling COMPILE on it, then calling
2205 FUNCALL on the result.
2206 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2207 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2208 Since the code is newer, there might still be some new bugs
2209 (though not as many as before Martin Atzmueller's fixes:-). But
2210 the new code is substantially simpler and clearer, and hopefully
2211 any remaining bugs will be simpler, less fundamental, and more
2212 fixable then the bugs in the old code.
2213 ** The revised compiler is still a little unsteady on its feet.
2215 *** The debugging information it produces (particularly the names
2216 of FUNCTION objects) is sometimes much less useful than what
2217 the old compiler produced.
2218 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2219 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2220 within a MACROLET) has been temporarily weakened.
2221 ** There are new compiler optimizations for various functions:
2222 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2223 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2224 *** the math functions TRUNCATE, FLOOR, and CEILING
2225 *** the function-of-all-trades COERCE
2226 Mostly these should be transparent, but there's one
2227 potentially-annoying problem (bug 117): when the compiler
2228 inline-expands a function and does type analysis on the result,
2229 it can create control paths which have type mismatches, and
2230 when it can't prove that those control paths aren't taken,
2231 it will issue WARNINGs about the type mismatches. This is
2232 a particular problem in practice for the new sequence functions.
2233 It's not clear how this should be fixed, and for now, a
2234 workaround is given in the entry for 117 in the BUGS file.
2235 ** (Because of the interaction between the two previous items --
2236 occasional inlining problems and new inline expansions -- some
2237 of the new sequence function optimizations won't really kick in
2238 completely until debugging information, and then inlining, are
2239 straightened out in some future version.)
2240 * minor incompatible changes:
2241 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2242 DIRECTORY no longer implicitly promotes NIL slots of its
2243 pathname argument to :WILD. In particular, when you ask for the
2244 contents of a directory (which you used to be able to do without
2245 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2246 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2247 ** changes in behavior that ANSI explicitly defines to be
2248 implementation dependent:
2249 *** The new compiler-only implementation still conforms with ANSI,
2250 but acts a little different than before. Besides the obvious
2251 changes in performance tradeoffs (that the cost per form passed
2252 to EVAL has gone up, and the cost per form executed by EVAL
2253 has gone down), the behavior of the system changes a little
2254 because there are no longer any interpreted function objects.
2255 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2256 e.g. doing COMPILE on the output of interactive DEFUN is
2258 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2260 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2261 has changed from (:CL) to NIL.
2262 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2263 e.g. #\Newline and #\Tab, don't have names specified in the
2264 ANSI Common Lisp standard, is now based on their ASCII symbolic
2265 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2266 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2267 are no longer used for output.
2268 ** changes in internal implementation constants:
2269 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2270 4 million. (If your application spends a lot of time GCing and
2271 you have a lot of RAM, you might want to experiment with
2272 increasing it even more.)
2273 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2274 However, almost all old code should still continue to work without
2275 immediate update, as SB-C-CALL is now a (deprecated) nickname
2277 ** Old operator names in the style DEF-FOO are now deprecated in
2278 favor of new corresponding names DEFINE-FOO, for consistency with
2279 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2280 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2281 internal symbols, but a few supported extensions like
2282 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2283 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2284 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2285 etc. as you get deeper into recursive calls to the debugger
2286 command loop, instead of the old "5]", "5]]", "5]]]"
2287 sequence. (I was motivated to do this when squabbles between
2288 ILISP and SBCL left me very deeply nested in the debugger. In the
2289 short term, this change will probably provoke more ILISP/SBCL
2290 squabbles, but hopefully it will be an improvement in the long run.)
2291 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2292 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2293 function, since declaiming return types involving aliens is
2294 (1) annoyingly messy to do by hand and (2) vital to efficient
2295 compilation of code which calls such functions.
2296 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2297 longer reexported by the SB-EXT package. They're solely useful
2298 for alien code, so it seems more logical that you should get
2299 them from the SB-ALIEN package, not in SB-EXT.
2300 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2301 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2302 features. Instead, the code that they used to control is always
2303 built into the system.
2304 * many other bug fixes
2305 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2306 advantage of the new EVAL-WHEN stuff and to clean them up in
2307 general, and they are now more ANSI-compliant in a number of
2308 ways. Martin Atzmueller is responsible for a lot of this.
2309 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2311 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2312 *** correct ERROR type for various file operations
2313 *** some fixes for Lisp streams
2314 *** DEFMETHOD syntax checking
2315 *** changing old weird representation of debug information as
2316 strings (which, among their other deficiencies, don't transform
2317 correctly when you rename packages, and don't change their
2318 print representation when you change things like *PACKAGE*
2319 and *PRINT-LENGTH*) to symbols and lists of symbols
2320 He also made several improvements and fixed several bugs in DESCRIBE.
2321 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2323 *** misbehavior of WRITE-STRING/WRITE-LINE
2324 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2325 and several other LOOP problems as well
2326 *** DIRECTORY when similar filenames are present
2327 *** DEFGENERIC with :METHOD options
2328 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2329 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2330 *** argument ordering in FIND with :TEST option
2331 *** mishandled package designator argument in APROPOS-LIST
2332 *** various problems in the backquote readmacro
2333 *** a bug in APROPOS
2334 *** probably some others that I'm not describing very well here,
2335 since the CVS log documents them by reference to sbcl-devel
2336 messages, and the SourceForge archives aren't working well.:-(
2337 ** Dan Barlow improved the Alpha port (and is making progress on the
2338 PPC port, for those of you who think different).
2339 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2340 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2341 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2342 continues to work on the SPARC port (for those of you in a position
2343 to look down upon our little PC-compatible boxes from a great height).
2344 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2345 and uses that copy, rather than the current dynamic value, when
2346 it's trying to decide whether to truncate output. Thus e.g.
2347 (let ((*print-lines* 50))
2348 (pprint-logical-block (stream nil)
2350 (let ((*print-lines* 8))
2351 (print (aref possiblybigthings i) stream)))))
2352 should now truncate the logical block only at 50 lines, instead of
2353 often truncating it at 8 lines, as it did before.
2354 * The doc/cmucl/ directory, containing old CMU CL documentation
2355 from the time of the fork, is no longer part of the base system.
2356 SourceForge has shut down its anonymous FTP service, and with it
2357 my original plan for distributing the old CMU CL documentation
2358 there. For now, if you need these files you can download an old
2359 SBCL source release and extract them from it.
2360 * The fasl file version number changed again, for dozens of reasons,
2361 some of which are apparent above.
2363 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2364 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2365 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2366 patch, so that DISASSEMBLE output is much nicer.
2367 * The code in the SB-PROFILE package now seems reasonably stable.
2368 I still haven't decided what the final interface should look like
2369 (I'd like PROFILE to interact cleanly with TRACE, since both
2370 facilities use function encapsulation) but if you have a need
2371 for profiling now, you can probably use it successfully with
2372 the current CMU-CL-style interface.
2373 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2374 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2375 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2376 link now returns the dangling link itself, and for similar
2377 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2378 link itself. (In these cases the old code signalled an error and
2379 looped endlessly, respectively.) Thus, DIRECTORY now works even
2380 in the presence of dangling and cyclic symbolic links.
2381 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2382 is now a supported extension again, since the consensus on
2383 sbcl-devel was that it can be useful for ordinary development
2384 work, not just for debugging SBCL itself.
2385 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2386 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2387 that currently-defined functions might be redefined later with
2388 different return types.
2389 * Hash tables can be printed readably, as inspired by CMU CL code
2390 of Eric Marsden and SBCL code of Martin Atzmueller.
2391 * better error handling in CLOS method combination, thanks to
2392 Martin Atzmueller porting Pierre Mai's CMU CL patches
2393 * more overflow fixes for >16Mbyte I/O buffers
2394 * A bug in READ has been fixed, so that now a single Ctrl-D
2395 character suffices to cause end-of-file on character streams.
2396 In particular, now you only need one Ctrl-D at the command
2397 line (not two) to exit SBCL.
2398 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2400 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2401 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2402 does the right thing.
2403 * The implementation of some type tests, especially for CONDITION
2404 types, is now tidier and maybe faster, due to CMU CL code
2405 originally by Douglas Crosher, ported by Martin Atzmueller.
2406 * Some math functions have been fixed, and there are new
2407 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2408 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2409 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2410 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2411 likely to be built into sbcl-0.7.0.)
2412 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2413 CL, ported to SBCL by Martin Atzmueller)
2414 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2415 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2416 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2417 * There's a new slam.sh hack to shorten the edit/compile/debug
2418 cycle for low-level changes to SBCL itself, and a new
2419 :SB-AFTER-XC-CORE target feature to control the generation of
2420 the after-xc.core file needed by slam.sh.
2421 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2422 COMPILE-FILE is no longer supported, so that now every function
2423 gets an entry point, so that block compilation looks a little
2424 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2425 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2426 returns the number of bytes consed since the system started,
2427 rather than the number consed since the first time the function
2428 was called. (The new definition parallels ANSI functions like
2429 CL:GET-INTERNAL-RUN-TIME.)
2430 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2431 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2432 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2433 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2434 * Fasl file version numbers are now independent of the target CPU,
2435 since historically most system changes which required version
2436 number changes have affected all CPUs equally. Similarly,
2437 the byte fasl file version is now equal to the ordinary
2440 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2441 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2442 is no longer recognized. I apologize for this, because it was
2443 listed in SB-EXT as a supported extension, but I found that
2444 its existing behavior was poorly specified, as well as incorrectly
2445 specified, and it looked like too much of a mess to straighten it
2446 out. I have enough on my hands trying to get ANSI stuff to work..
2447 * many patches ported from CMU CL by Martin Atzmueller, with
2448 half a dozen bug fixes in pretty-printing and the debugger, and
2449 half a dozen others elsewhere
2450 * fixed bug 13: Floating point infinities are now supported again.
2451 They might still be a little bit flaky, but thanks to bug reports
2452 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2453 as flaky as they were.
2454 * The --noprogrammer command line option is now supported. (Its
2455 behavior is slightly different in detail from what the old man
2456 page claimed it would do, but it's still appropriate under the
2457 same circumstances that the man page talks about.)
2458 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2459 are now supported, and enabled by default. Thus, the compiler can
2460 handle many floating point and complex operations much less
2461 inefficiently. (Thus e.g. you can implement a complex FFT
2463 * The compiler now detects type mismatches between DECLAIM FTYPE
2464 and DEFUN better, and implements CHECK-TYPE more correctly, and
2465 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2466 thanks to patches from Martin Atzmueller.
2467 * various fixes to make the cross-compiler more portable to
2468 ANSI-conforming-but-different cross-compilation hosts (notably
2469 Lispworks for Windows, following bug reports from Arthur Lemmens)
2470 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2471 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2472 fixed thanks to Pierre Mai's CMU CL patches.
2473 * fixes to keep the system from overflowing internal counters when
2474 it tries to use i/o buffers larger than 16M bytes
2475 * fixed bug 45a: Various internal functions required to support
2476 complex special functions have been merged from CMU CL sources.
2477 (When I was first setting up SBCL, I misunderstood a compile-time
2478 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2479 * improved support for type intersection and union, fixing bug 12
2480 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2481 more obscure bugs as well
2482 * some steps toward byte-compiling non-performance-critical
2483 parts of the system, courtesy of patches from Martin Atzmueller
2484 * Christophe Rhodes has made some debian packages of sbcl at
2485 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2486 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2487 stable, but are nonetheless usable. When he's ready, I'd be happy
2488 to add them to the SourceForge "File Releases" section. (And if
2489 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2490 * new fasl file format version number (because of changes in
2491 internal representation of (OR ..) types to accommodate the new
2492 support for (AND ..) types, among other things)
2494 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2495 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2497 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2499 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2500 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2501 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2502 * DESCRIBE now works on structure objects again.
2503 * Most function call argument type mismatches are now handled as
2504 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2505 know whether the function will be redefined before the call is
2506 executed. (The compiler could flag local calls with full WARNINGs,
2507 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2508 it doesn't keep track of enough information to know whether calls
2509 are local in this sense.)
2510 * Compiler output is now more verbose, with messages truncated
2511 later than before. (There should be some supported way for users
2512 to override the default verbosity, but I haven't decided how to
2513 provide it yet, so this behavior is still controlled by the internal
2514 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2515 src/compiler/ir1util.lisp.)
2516 * Fasl file format version numbers have increased again, because
2517 support for the Gray streams extension changes the layout of the
2518 system's STREAM objects.
2519 * The Gray subclassable streams extension now works, thanks to a
2520 patch from Martin Atzmueller.
2521 * The full LOAD-FOREIGN extension (not just the primitive
2522 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2523 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2524 but like most other programs, it defaults to copying the Unix
2525 environment from the original process instead of starting the
2526 new process in an empty environment.
2527 * Extensions which manipulate the Unix environment now support
2528 an :ENVIRONMENT keyword option which doesn't smash case or
2529 do other bad things. The CMU-CL-style :ENV option is retained
2530 for porting convenience.
2531 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2532 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2534 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2536 * A patch from Martin Atzmueller seems to have solved the SIGINT
2537 problem, and as far as we know, signal-handling now works cleanly.
2538 (If you find any new bugs, please report them!)
2539 * The system no longer defaults Lisp source file names to types
2540 ".l", ".cl", or ".lsp", but only to ".lisp".
2541 * The compiler no longer uses special default file extensions for
2542 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2543 seems to expect a single default extension for all compiled code,
2544 and there's no compelling reason to try to stretch the standard
2545 to allow two different extensions.) Instead, byte-compiled files
2546 default to the same extension as native-compiled files.
2547 * Fasl file format version numbers have increased again, because
2548 a rearrangement of internal implementation packages made some
2549 dumped symbols in old fasl files unreadable in new cores.
2550 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2551 many fewer weird special cases.
2552 * Bug #17 (differing COMPILE-FILE behavior between logical and
2553 physical pathnames) has been fixed, and some related misbehavior too,
2554 thanks to a patch from Martin Atzmueller.
2555 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2556 by Tim Moore, ported to SBCL by Martin Atzmueller.
2557 * Martin Atzmueller fixed several filesystem-related problems,
2558 including bug #36, in part by porting CMU CL patches, which were
2559 written in part by Paul Werkowski.
2560 * More compiler warnings in src/runtime/ are gone, thanks to
2561 more patches from Martin Atzmueller.
2562 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2565 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2567 * DESCRIBE now works on CONDITION objects.
2568 * The debugger now handles errors which arise when trying to print
2569 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2571 * The build system now uses an additional file, customize-target-features.lisp,
2572 to allow local modifications to the target *FEATURES* list. (The point of
2573 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2574 debugging features enabled, without having to worry about propagating it
2575 into everyone's system when I do a "cvs update".) When no
2576 customize-target-features.lisp file exists, the target *FEATURES* list
2577 should be constructed the same way as before.
2578 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2579 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2580 warn about behavior which is undefined under the ANSI spec. Note: This
2581 is specified by ANSI, but it's not very popular with programmers.
2582 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2583 macro in the SBCL sources for an example of a workaround which you
2584 might use to make portable ANSI-standard code which does what you want.
2585 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2586 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2587 when it's used as a non-toplevel form. (This is required in order
2588 to implement the DEFCONSTANT-EQX macro.)
2589 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2590 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2591 funny things, due to bugs in the implementation of EVAL-WHEN
2592 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2593 non-toplevel DEFCONSTANTs are uncommon.)
2594 * The core file version number and fasl file version number have been
2595 incremented, because the old noncompliant DEFCONSTANT behavior involved
2596 calling functions which no longer exist, and because I also took the
2597 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2598 * fixed bug 1 (error handling before read-eval-print loop starts), and
2599 redid debugger restarts and related debugger commands somewhat while
2601 ** The QUIT debugger command is gone, since it did something
2602 rather different than the SB-EXT:QUIT command, and since it never
2603 worked properly outside the main toplevel read/eval/print loop.
2604 Invoking the new TOPLEVEL restart provides the same functionality.
2605 ** The GO debugger command is also gone, since you can just invoke
2606 the CONTINUE restart directly instead.
2607 ** The TOP debugger command is also gone, since it's redundant with the
2608 FRAME 0 command, and since it interfered with abbreviations for the
2610 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2612 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2613 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2614 some porting work by Martin Atzmueller)
2615 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2616 with a boatload of patches which he ported to SBCL. Now that those
2618 ** The system tries to make sure that its low-priority messages
2619 are prefixed by semicolons, to help people who like to use
2620 syntax highlighting in their ILISP buffer. (This patch
2621 was originally due to Raymond Toy.)
2622 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2623 patches originally written by Raymond Toy.
2624 ** The compiler understands coercion between single-value and
2625 multiple-VALUES type expressions better, getting rid of some very
2626 weird behavior, thanks to patches originally by Robert MacLachlan
2627 and Douglas Crosher.
2628 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2629 lambda lists, thanks to a patch originally by Pierre Mai.
2630 ** The system no longer bogusly warns about "abbreviated type
2632 ** The compiler gets less confused by inlining and RETURN-FROM,
2633 thanks to some patches originally by Tim Moore.
2634 ** The system no longer hangs when dumping circular lists to fasl
2635 files, thanks to a patch originally from Douglas Crosher.
2636 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2637 undefined function error.
2638 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2639 in structure slot initforms) for now, documented workaround instead:-|
2640 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2641 * fixed bug 5: added stubs for various Gray stream functions called
2642 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2643 installed, at least appropriate type errors are generated
2644 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2645 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2646 consistently in DEFMETHOD forms.
2647 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2648 it doesn't seem to affect SBCL after all
2649 * The C runtime system now builds with better optimization and many
2650 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2652 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2654 * The system is now under CVS at SourceForge (instead of the
2655 CVS repository on my home machine).
2656 * The new signal handling code has been tweaked to treat register
2657 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2658 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2659 downstream, e.g. in debug-int.lisp, has implicit dependencies
2660 on the unsignedness of integer representation of machine words,
2661 and that caused the system to bomb out with infinite regress
2662 when trying to recover from type errors involving signed values,
2663 e.g. (BUTLAST '(1 2 3) -1).)
2664 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2665 (This was one of the bugs Peter Van Eynde reported back in July.)
2666 * The system now uses code inspired by Colin Walters' O(N)
2667 implementation of MAP (from the cmucl-imp@cons.org mailing
2668 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2669 inline the MAP operation, and there is more than one
2670 sequence argument to the MAP call (so that it can't just
2671 do ETYPECASE once and for all based on the type of the
2672 single sequence argument). (The old non-inline implementation
2673 of the general M-argument sequence-of-length-N case required
2674 O(M*N*N) time when any of the sequence arguments were LISTs.)
2675 * The QUIT :UNIX-CODE keyword argument has been renamed to
2676 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2677 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2679 * Raymond Wiker's patches to port dynamic loading from Linux to
2680 FreeBSD have been added.
2681 * The BUGS file is now more nearly up to date, thanks in large part
2682 to Martin Atzmueller's review of it.
2683 * The debugger now flushes standard output streams before it begins
2684 its output ("debugger invoked" and so forth).
2685 * The core version number and fasl file version number have both
2686 been incremented, because of incompatible changes in the layout
2688 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2689 used to not be called for a saved Lisp image.)
2690 * Martin Atzmueller's version of a patch to fix a compiler crash,
2691 as posted on sbcl-devel 13 September 2000, has been installed.
2692 * Instead of installing Martin Atzmueller's patch for the
2693 compiler transform for SUBSEQ, I deleted the compiler transform,
2694 and transforms for some similar consing operations.
2695 * A bug in signal handling which kept TRACE from working on OpenBSD
2697 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2698 away by constant folding
2699 * The system now defines its address space constants in one place
2700 (in the Lisp sources), and propagates them automatically elsewhere
2701 (through GENESIS and the sbcl.h file). Therefore, patching the
2702 address map is less unnecessarily tedious and error-prone. The
2703 Lisp names of address space constants have also been systematized.
2704 * CVS tags like dollar-Header-dollar have been removed from
2705 the sources, because they have never saved me trouble and
2706 they've been source of trouble working with patches and other
2707 diff-related operations.
2708 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2709 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2711 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2713 * The system has been ported to OpenBSD.
2714 * The system now compiles with a simple "sh make.sh" on the systems
2715 that it's supported on. I.e., now you no longer need to tweak
2716 text in the target-features.lisp-expr and symlinks in src/runtime/
2717 by hand, the make.sh takes care of it for you.
2718 * The system is no longer so grossly inefficient when compiling code
2719 involving vectors implemented as general (not simple) vectors (VECTOR T),
2720 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2721 which dares to use the various sequence functions on non-simple
2722 vectors, takes less of a performance hit.
2723 * There is now a primitive type predicate VECTOR-T-P
2724 to test for the (VECTOR T) type, so that e.g.
2725 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2726 can now be compiled with some semblance of efficiency. (The old code
2727 turned the type declaration into a full call to %TYPEP at runtime!)
2728 * AREF on (VECTOR T) is still not fast, since it's still compiled
2729 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2730 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2731 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2732 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2733 without making full calls to SUBTYPEP at runtime.
2734 (Some analogous efficiency issues for non-simple vectors specialized to
2735 element types other than T, or for non-simple multidimensional arrays,
2736 have not been addressed. They could almost certainly be handled the
2737 same way if anyone is motivated to do so.)
2738 * The changes in array handling break binary compatibility, so
2739 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2740 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2741 instead of (VALUES T T).
2742 * By following the instructions that Dan Barlow posted to sbcl-devel
2743 on 2 July 2000, I was able to enable primitive dynamic object
2744 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2745 functionality is not implemented (since it calls ld to resolve
2746 library references automatically, requiring RUN-PROGRAM for its
2747 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2748 try to resolve library references) is now supported.
2749 * The system now flushes the standard output streams when it terminates,
2750 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2751 them at several other probably-convenient times, e.g. in each pass of
2752 the toplevel read-eval-print loop, and after evaluating a form given
2753 as an "--eval" command-line option. (These changes were motivated by a
2754 discussion of stream flushing issues on cmucl-imp in August 2000.)
2755 * The source transform for TYPEP of array types no longer assumes
2756 that an array whose element type is a not-yet-defined type
2757 is implemented as an array of T, but instead punts, so that the
2758 type will be interpreted at runtime.
2759 * There is now some support for cross-compiling in make.sh: each of
2760 the phases of make.sh has its own script. (This should be transparent
2761 to people doing ordinary, non-cross-compile builds.)
2762 * Since my laptop doesn't have hundreds of megabytes of memory like
2763 my desktop machine, I became more motivated to do some items on
2764 my to-do list in order to reduce the size of the system a little:
2765 ** Arrange for various needed-only-at-cold-init things to be
2766 uninterned after cold init. To support this, those things have
2767 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2768 symbols with such names are now uninterned after cold init).
2769 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2770 fasl files for cold load.
2771 ** Remove the old compiler structure pooling code (which used to
2772 be conditional on the target feature :SB-ALLOC) completely.
2773 ** Redo the representation of some data in cold init to be more compact.
2774 (I also looked into supporting byte compiled code at bootstrap time,
2775 which would probably reduce the size of the system a lot, but that
2776 looked too complicated, so I punted for now.)
2777 * The maximum signal nesting depth in the src/runtime/ support code has
2778 been reduced from 4096 to 256. (I don't know any reason for the very
2779 large old value. If the new smaller value turns out to break something,
2780 I'll probably just bump it back up.)
2781 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2783 * Many, many bugs reported by Peter Van Eynde have been added to
2784 the BUGS list; some have even been fixed.
2785 * While enabling dynamic object file loading, I tried to make the
2786 code easier to understand, renaming various functions and variables
2787 with less ambiguous names, and changing some function calling
2788 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2789 * While trying to figure out how to do the OpenBSD port, I tried to
2790 clean up some of the code in src/runtime/. In particular, I dropped
2791 support for non-POSIX signal handling, added various comments,
2792 tweaked the code to reduce the number of compilation warnings, and
2793 renamed some files to increase consistency.
2794 * To support the new automatic configuration functionality in make.sh,
2795 the source file target-features.lisp-expr has been replaced with the
2796 source file base-target-features.lisp-expr and the machine-generated
2797 file local-target-features.lisp-expr.
2798 * fixed a stupid quoting error in make.sh so that using CMU CL
2799 "lisp -batch" as cross-compilation host works again
2801 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2803 * DESCRIBE no longer tries to call itself recursively to describe
2804 bound/fbound values, so that it no longer fails on symbols which are
2805 bound to themselves (like keywords, T, and NIL).
2806 * DESCRIBE now works on generic functions.
2807 * The printer now prints less-screwed-up representations of closures
2808 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2809 * A private symbol is used instead of the :EMPTY keyword previously
2810 used to mark empty slots in hash tables. Thus
2811 (DEFVAR *HT* (MAKE-HASH-TABLE))
2812 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2813 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2814 now does what ANSI says that it should. (You can still get
2815 similar noncompliant behavior if bang on the hash table
2816 implementation with all the symbols you get back from
2817 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2818 This breaks binary compatibility, since tests for equality to
2819 :EMPTY are wired into things like the macroexpansion of
2820 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2822 * There's now a minimal placeholder implementation for CL:STEP,
2823 as required by ANSI.
2824 * An obscure bug in the interaction of the normal compiler, the byte
2825 compiler, inlining, and structure predicates has been patched
2826 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2827 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2829 * Missing ordinary arguments in a macro call are now detected even
2830 when the macro lambda list contains &KEY or &REST.
2831 * The debugger no longer complains about encountering the top of the
2832 stack when you type "FRAME 0" to explicitly instruct it to go to
2833 the top of the stack. And it now prints the frame you request even
2834 if it's the current frame (instead of saying "You are here.").
2835 * As specified by ANSI, the system now always prints keywords
2836 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2838 * The default initial SIZE of HASH-TABLEs is now smaller.
2839 * Type information from CLOS class dispatch is now propagated
2840 into DEFMETHOD bodies, so that e.g.
2841 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2843 is now basically equivalent to
2844 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2845 (DECLARE (TYPE SINGLE-FLOAT X))
2847 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2848 operation, without having to do run-time type dispatch.
2849 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2850 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2851 * Since I don't seem to be making big file reorganizations very often
2852 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2853 is overflowing my ability to conveniently back them up), I've finally
2854 checked the system into CVS. (The CVS repository is on my home system,
2855 not at SourceForge -- putting it on SourceForge might come later.)
2856 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2857 high-level GC-NOTIFY-FOO functions send their output. (There's
2858 still very little control of where low-level verbose GC functions
2859 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2860 less than it used to -- the GC-NOTIFY-FOO functions are now under
2861 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2862 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2863 in only one place in the source code, and propagates it automatically
2864 everywhere that it's needed. Thus e.g. when I bump the version from
2865 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2866 * The C source files now include boilerplate legalese and documentation
2867 at the head of each file (just as the Lisp source files already did).
2868 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2869 to his page will be replaced with a link to his new CLiki service.
2871 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2873 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2874 * The build process now looks for GNU make under the default name "gmake",
2875 instead of "make" as it used to. If GNU make is not available as "gmake"
2876 on your system, you can change this default behavior by setting the
2877 GNUMAKE environment variable.
2878 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2879 can build without error under CMU CL.
2881 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2883 * There is now a partial SBCL user manual (with some new text and some
2884 text cribbed from the CMU CL manual).
2885 * The beginnings of a profiler have been added (starting with the
2886 CMU CL profiler and simplifying and cleaning up). Eventually the
2887 main interface should be through the TRACE macro, but for now,
2888 it's still accessed through vaguely CMU-CL-style functions and macros
2889 exported from the package SB-PROFILE.
2890 * Some problems left over from porting CMU CL to the new
2891 cross-compilation bootstrap process have been cleaned up:
2892 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2893 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2894 part when getting rid of the compiler *BACKEND* stuff.)
2895 ** The value of *NULL-TYPE* was screwed up, because it was
2896 being initialized before the type system knew the final
2897 definition of the 'NULL type. This screwed up several key
2898 optimizations in the compiler, causing inefficiency in all sorts
2899 of places. (I found it because I wanted to understand why
2900 GET-INTERNAL-RUN-TIME was consing.)
2901 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2902 PROCLAIM FTYPE information. Unfortunately this broke binary
2903 compatibility again, since now the forms output by DEFGENERIC
2904 to refer to functions which didn't exist in 0.6.3.
2905 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2906 can use the new (as of 0.6.3) transform for SOME into MAP into
2908 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2909 values to (INTEGER 0 1000000), so that the time code will no longer
2910 occasionally get blown up by Linux returning 1000000 microseconds
2911 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2912 its output conform to the ANSI spec. (Alas, this makes its output
2913 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2914 * A full call to MAP NIL with a single sequence argument no longer conses.
2915 * fixes to problems pointed out by Martin Atzmueller:
2916 * The manual page no longer talks about multiprocessing as though
2917 it were currently supported.
2918 * The ILISP support patches have been removed from the distribution,
2919 because as of version 5.10.1, ILISP now supports SBCL without us
2920 having to maintain patches.
2921 * added a modified version of Raymond Toy's recent CMU CL patch for
2922 EQUALP comparison of HASH-TABLE
2924 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2926 * The system still can't cross-compile itself with
2927 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2928 entails), but at least it can compile more of itself that way
2929 than it used to be able to, and various buglets which were uncovered
2930 by trying to cross-compile itself that way have now been fixed.
2931 * This release breaks binary compatibility again. This time
2932 at least I've incremented the FASL file format version to 2, so that the
2933 problem can be detected reliably instead of just causing weird errors.
2934 * various new style warnings:
2935 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2936 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2937 ** using the lexical binding of a variable named in the *FOO* style
2938 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2939 as specified by ANSI.
2940 * *RANDOM-STATE* is no longer automatically initialized from
2941 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2942 default behavior of the system is to repeat its behavior every time
2943 it's run. If you'd like to change this behavior, you can always
2944 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2945 old convention there was no comparably easy way to get the system to
2946 repeat its behavior every time it was run.
2947 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2948 FTYPE declarations has been removed, in favor of their ANSI
2949 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2950 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2951 the types of their sequence arguments can be determined at compile time.
2952 This is done through a new open code expansion for MAP which eliminates
2953 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2954 argument types can be determined at compile time.
2955 * The optimizer now transforms COERCE into an identity operation when it
2956 can prove that the coerced object is already of the correct type. (This
2957 can be a win for machine generated code, including the output of other
2958 optimization transforms, such as the MAP transform above.)
2959 * Credit information has been moved from source file headers into CREDITS.
2960 * Source file headers have been made more standard.
2961 * The CASE macro now compiles without complaining even when it has
2964 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2966 * (Note that the way that the PCL macroexpansions were rewritten
2967 to accommodate the change in DEFGENERIC below breaks binary
2968 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2969 not run under sbcl-0.6.2. Once we get out of alpha releases,
2970 i.e. hit release 1.0.0, we'll probably try to maintain binary
2971 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2972 and sbcl-1.4.4. Until then, however, it might be fairly common
2973 for maintenance releases to break binary compatibility.)
2974 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2976 * The reporting of unhandled signals has been changed to print some
2977 explanatory text as well as the report form. (Previously only
2978 the report form was printed.)
2979 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2980 it defines, so that the compiler no longer issues undefined function
2981 warnings for compiled-but-not-yet-loaded generic functions.
2982 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2983 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2984 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2985 section of the ANSI spec are supported.
2986 * The "" nickname for the "KEYWORD" package has been removed.
2987 The reader still handles symbol tokens which begin with a package marker
2988 as keywords, but it doesn't expose its mechanism for doing so in the
2989 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
2990 * The system now issues STYLE-WARNINGs for contradictory TYPE
2991 proclamations. (Warnings for contradictory FTYPE proclamations would
2992 be nice too, but those can't be done usefully unless the type system
2993 is made smarter about FUNCTION types.)
2994 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
2995 systematized, so that "*target-*.lisp is supposed to exist only on the
2996 target and imply that there's a related file which exists on the
2997 host, and *host-*.lisp is supposed to exist only on the host and imply
2998 that there's a related file which exists on the target. This involves a
2999 lot of renaming. Hopefully the acute confusion caused by the renaming
3000 will be justified by the reduction in chronic confusion..
3001 ** runtime-type.lisp -> early-target-type.lisp
3002 ** target-type.lisp -> late-target-type.lisp
3003 ** early-host-format.lisp -> early-format.lisp
3004 ** late-host-format.lisp -> late-format.lisp
3005 ** host-error.lisp -> misc-error.lisp
3006 ** early-error.lisp -> early-target-error.lisp
3007 ** late-error.lisp -> late-target-error.lisp
3008 ** host-defboot.lisp -> early-defboot.lisp
3009 ** code/misc.lisp -> code/target-misc.lisp
3010 ** code/host-misc.lisp -> code/misc.lisp
3011 ** code/numbers.lisp -> code/target-numbers.lisp
3012 ** code/early-numbers.lisp -> numbers.lisp
3013 ** early-host-type.lisp -> early-type.lisp
3014 ** late-host-type.lisp -> late-type.lisp
3015 ** host-typep.lisp -> typep.lisp
3016 ** load.lisp -> target-load.lisp
3017 ** host-load.lisp -> load.lisp
3018 ** host-disassem.lisp -> disassem.lisp
3019 ** host-insts.lisp -> insts.lisp
3020 ** byte-comp.lisp -> target-byte-comp.lisp
3021 ** host-byte-comp.lisp -> byte-comp.lisp
3022 ** host-signal.lisp -> signal.lisp
3023 ** host-defstruct.lisp -> defstruct.lisp
3024 ** late-target-type.lisp -> deftypes-for-target.lisp
3025 Furthermore, several other previously target-only files foo.lisp (e.g.
3026 hash-table.lisp and random.lisp) have been split into a target-and-host
3027 foo.lisp file and a target-only target-foo.lisp file, with their key type
3028 definitions in the target-and-host part, so that the cross-compiler will
3029 know more about target types.
3030 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3031 gone away. In their place are various *BACKEND-FOO* variables
3032 corresponding to the slots of the old structure.
3033 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3034 into the target SBCL has been fixed.
3035 * The chill.lisp system for loading cold code into a running SBCL
3037 * Support for the CMU CL "scavenger hook" extension has been removed.
3038 (It was undocumented and unused in the CMU CL sources that SBCL was
3039 derived from, and stale in sbcl-0.6.1.)
3040 * Various errors in the cross-compiler type system were detected
3041 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3042 (enabling various consistency checks). Many of them were fixed,
3043 but some hard problems remain, so the compiler is back to
3044 running without *TYPE-SYSTEM-INITIALIZED* for now.
3045 * As part of the cross-compiler type system cleanup, I implemented
3046 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3047 * I have started adding UNCROSS calls throughout the type system
3048 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3049 on cross-compiler input files will be able to go away, and various
3051 * CONSTANTP now returns true for quoted forms (as explicitly required
3054 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3056 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3057 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3058 (MAKE-LIST -1). In the longer term, it ought to become true
3059 that declarations are assertions even at SAFETY 1. For now, it's not
3060 quite true even at SAFETY 3, but it's at least more nearly true..
3061 (Note that this change seems to increases the size of the system by
3062 O(5%) and to decrease the speed of the compiler by 20% or more.)
3063 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3064 for the problem of printing dozens of lines of distracting information
3065 about low-level system machinery as part of the top stack frame
3066 on entry to the debugger when an undefined function was called.
3067 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3068 is not reset to COMMON-LISP-USER.
3069 * Compilation of stuff related to dyncount.lisp has been made conditional
3070 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3071 smaller. The various dyncount-related symbols have been moved into
3072 a new "SB-DYNCOUNT" package.
3073 * tty-inspect.lisp has been renamed to inspect.lisp.
3074 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3075 feature has gone away. (When we eventually port to other flavors of
3076 libc and/or Unix, we'll try to make the differences between flavors
3077 invisible at the user level.)
3078 * Various other *FEATURES* tags, and/or their associated conditionals,
3079 have been removed if obsolescent, or given better documentation, or
3080 sometimes given more-mnemonic names.
3082 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3084 * tidied up "make.sh" script
3085 * tidied up system directory structure
3086 * better "clean.sh" behavior
3087 * added doc/FOR-CMUCL-DEVELOPERS
3088 * many many small tweaks to output format, e.g. removing possibly-confusing
3089 trailing #\. character in DESCRIBE-INSTANCE
3090 * (EQUALP #\A 'A) no longer signals an error.
3091 * new hashing code, including EQUALP hashing
3092 * tidied up Lisp initialization and toplevel
3093 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3094 * command line argument processing
3095 * added POSIX-GETENV function to deal with Unix-ish environment variables
3096 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3097 terminating SBCL on EOF
3098 * non-verbose GC by default
3099 * There is no more "sbcl" shell script; the sbcl file is now the C
3100 runtime executable (just like CMU CL).
3101 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3103 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3104 symbol and package information as Lisp native symbol and package objects
3105 instead of strings naming symbols and strings naming packages. This way,
3106 whenever packages are renamed (as in warm init), debug information is
3107 transformed along with everything else.
3108 * tweaked the optimization policy declarations which control the building
3109 of SBCL itself. Now, among other things, the system no longer saves
3110 source location debugging information. (This helps two problems at once
3111 by reducing SBCL size and by keeping SBCL from trying to look for its
3112 sources -- which may not exist -- when reporting errors.)
3113 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3114 debugging and testing purposes
3115 * cleaned up printing, making the printer call PRINT-OBJECT for
3116 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3117 methods, giving nearly-ANSI behavior
3118 * converted almost all special variables to use *FOO* naming convention
3119 * deleted PARSE-TIME functionality, since it can be done portably
3120 * moved some files out of cold init into warm init
3121 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3123 * regularized formatting of source files
3124 * added an install.sh script
3125 * fixed ridiculous memory usage of cross-compiler by making
3126 compiler/alloc.lisp not try to do pooling unless it can hook
3127 itself into the GC of the cross-compilation host. Now the system
3128 builds nicely on my old laptop.
3129 * added :SB-ALLOC in target-features.lisp-expr
3130 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3131 was not implemented)
3132 * re-did condition handling and note reporting in the compiler. Notes
3133 are no longer handled by signalling conditions. Style warnings
3134 and warnings are handled more correctly and reported in such a way
3135 that it's easy to find one or the other in your output (so that you
3136 can e.g. figure out which of many problems caused COMPILE-FILE to
3138 * changed the severity of several compiler warnings from full WARNING
3139 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3140 compiler note reporting so that it doesn't use the condition system
3141 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3142 COMPILE-FILE command)
3143 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3144 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3145 no longer supported, since their implementation was deeply intertwingled
3146 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3147 * removed bogus "support" for compiler macros named (SETF FOO), and
3148 removed the compiler macro for SETF INFO (but only after making a fool
3149 of myself on the cmucl-imp mailing list by posting a bogus patch for
3150 DEFINE-COMPILER-MACRO..)
3151 * Compiled files containing forms which have side effects on the Lisp
3152 reader (such as DEFPACKAGE forms) are now handled more correctly.
3153 (Compiler queuing of top level lambdas has been suppressed by setting
3154 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3155 * deleted various currently-unused source files, e.g. gengc.lisp. They
3156 may be added back at some point e.g. when porting to other architectures,
3157 but until they are it's distracting to distribute them and to try to
3159 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3160 there were so many of them they're just distractions, and UNCROSS is
3161 known to be able to handle the current sources
3162 * moved PROFILE functionality into TRACE, so that it will be clear
3163 how the wrapping and unwrapping of functions when you profile them
3164 interacts with the wrapping and unwrapping of functions when you
3165 trace them. (Actually, the functionality isn't there yet, but at least
3166 the interface specification is there. Hopefully, the functionality will
3167 arrive with some future maintenance release.)
3168 * removed host-oops.lisp
3169 * changed signature of QUIT function to allow UNIX-CODE argument
3170 * fixed READ-SEQUENCE bug
3171 * tweaked verbose GC output so that it looks more like the progress
3172 output that ANSI specifies for functions like LOAD
3173 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3174 * added <http://sbcl.sourceforge.com> to the banner information printed by