1 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2 * fixed cross-compiler leakages that prevented building a 32-bit
3 target with a 64-bit host compiler.
4 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
5 opened with :IF-EXISTS :OVERWRITE.
6 * fixed bug 281: error for an invalid qualifier in a short-form method
7 combination method is not signalled until the faulty method is called.
8 * bug fix: iteration variable type inferrer failed to deal with open
9 intervals. (reported by Alan Shields)
10 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
11 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
12 or more runtime options were provided to the sbcl binary.
13 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
14 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
15 * optimization: There's now a fast-path for fixnum arguments in the
16 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
17 * fixed some bugs revealed by Paul Dietz' test suite:
18 ** the type-error signalled from WARN has a filled-in DATUM slot.
19 ** the type-error required when a stream is not associated with
20 a file has the stream as its datum.
21 ** type-errors on single-floats on x86-64 no longer have
22 :INVALID-OBJECT as the datum
24 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
25 * incompatible change: the --noprogrammer option, deprecated since
26 version 0.7.5, has been removed. Please use the equivalent
27 --disable-debugger option instead.
28 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
30 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
31 inoperational for a while now) has been completely removed.
32 * null lexical environments are now printed as #<NULL-LEXENV>,
33 significantly reducing the amount of clutter in typical backtraces.
34 * documentation on weak pointers, finalization, and after GC hooks
35 has been added to the manual.
36 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
37 as other specialized array types, is much faster. SUBSEQ and
38 COPY-SEQ on such arrays have also been sped up.
39 * optimization: EQL is now more efficient when at least other argument
40 is known to be of type (OR FIXNUM (NOT NUMBER)).
41 * fixed bug: the runtime is now less vulnerable to changes in the
42 size of the SBCL object on OS X, and virtual memory is reserved for
43 all spaces that need to be at a fixed address.
44 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
45 * fixed bug: finalizers and after GC hooks that cause consing are now
47 * fixed bug: compiler error messages and summaries are now printed to
48 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
49 * fixed inference of the upper bound of an iteration variable.
50 (reported by Rajat Datta).
51 * fixed bug 211e: calling local functions with duplicated constant
52 keyword argument no longer causes a bogus style warning about an
54 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
55 ftype declaration to be disregarded. (reported by Dave Roberts)
56 * fixed bug 373: caused by erronous compilation of references to alien
57 variables in the runtime on ppc/darwin.
58 * fixed bug 376: CONJUGATE type deriver.
59 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
60 (reported by Baughn on #lisp)
61 * a cross-compiler bug on non-x86 platforms has been identified and
62 fixed. (thanks to Bruno Haible)
63 * improvements to the MIPS runtime code for increased stability.
64 (thanks to Thiemo Seufer)
65 * increased the maximimum compact environment size to allow
66 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
67 * partial workaround for bug 135: don't name the function we're
68 calling for hairy cases of EVAL, so as not to accumulate one environment
69 entry per call to EVAL. (reported by Kevin Reid)
70 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
71 * optimization: type testing for non-vector arrays should be faster.
72 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
74 * bug fix: setting 31st element of a bit vector to zero did not work
76 * bug fix: redefining a class definition which failed due to a
77 previous accessor / function clash now works (but see BUGS entry
78 #380 for more problems in this area). (thanks to Zach Beane)
79 * the long form of DEFINE-METHOD-COMBINATION disables method group
80 checking when given a single method group with pattern *. (thanks
82 * on x86 compiler supports stack allocation of results of simple
83 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
84 * support for building and running on Mac OS X 10.4 "Tiger" has been added
85 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
86 * fixed some bugs related to Unicode integration:
87 ** the restarts for recovering from input and output encoding
88 errors only appear when there is in fact such an error to
90 * fixed some bugs revealed by Paul Dietz' test suite:
91 ** MISC.549 and similar: late transformation of unsafe type
92 assertions into derived types caused unexpected code
94 ** SCALE-FLOAT type deriver is less wrong.
95 ** type derivers for EXP, LOG and similar functions compute result
96 types for complex arguments better.
97 ** (MISC.563) CONJUGATE type deriver works for very restricted
99 ** out-of-line type testers for character strings are available.
100 ** EQUAL compiler transform understands specialness of objects
102 ** accessing double-floats stored on the stack now works on x86-64.
103 ** debugger internals could sometimes create invalid lispobjs,
104 resulting in GC crashes.
105 ** MISC.548: type check weakening can convert required type into
107 ** initialization forms for bindings are not in scope of free special
110 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
111 * incompatible change: thread support for non-NPTL systems has
112 been removed - locking is buggy and unreliable. A threaded
113 SBCL build will now warn at startup and refuse to create
114 new threads, unless futex support is detected in the kernel
115 * incompatible change: the top level REPL now has only an ABORT
116 restart associated with it, not TOPLEVEL and ABORT as it used to.
117 TOP and TOPLEVEL are now available as debugger commands for
118 returning to the top level.
119 * incompatible change: forms evaluated in the REPL now use the
120 global optimization policy.
121 * incompatible change: user- and system-initialization files are
122 no longer processed with LOAD, but by READ and EVAL; hence the
123 global optimization policy, startup package, readtable, etc,
125 * The .fasl file format number has been incremented because of
126 various incompatible changes.
127 * internal entry point details and argument counts no longer appear
128 in backtraces unless explicitly requested by setting
129 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
130 * built-in and standard functions no longer have names like "top
131 level local call to FOO".
132 * fixed bug 32: functions defined in non-null lexical environments
133 now have more legible printed representation
134 * fixed bug 33: functions defined in non-null lexical environemnts
135 are now more amenable to inspection by INSPECT.
136 * workaround for bug 354: XEPs no longer appear in backtraces unless
137 explicitly requested.
138 * fixed bug: receiving the signal which results from attempting to
139 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
140 SIGBUS on other *BSDs) on architectures where the C stack is also
141 the Lisp stack (x86 and x86-64) from foreign code no longer leads
142 to debugger confusion or wild instability. (reported by Cheuksan
144 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
145 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
146 (reported by Lutz Euler)
147 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
148 compiled code. (reported by Paul Dietz)
149 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
150 specialized parameter to a method to something that is not TYPEP
151 the specializer is now possible.
152 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
153 face of package deletion.
154 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
155 pathname. (reported by David Tolpin on c.l.l)
156 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
157 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
158 STRING-STREAM and FILE-STREAM, respectively.
159 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
160 than previously; in addition, its test suite is now run on build.
161 (thanks to Luke Gorrie)
162 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
163 * optimization: added a immediate representation for single-floats
165 * fixed some bugs revealed by Paul Dietz' test suite:
166 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
167 ** attempting to create a package with a colliding nickname causes
168 correctable errors to be signalled.
169 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
170 ** several x86-64 backend bugs related to sign-extension of immediate
173 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
174 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
175 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
177 * fixed loading of multiply forward-referenced layouts.
178 (thanks to Cheuksan Wang)
179 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
180 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
181 (see COMPILE-FILE documentation for details of :PRINT option)
182 * fixed bugs 19 and 317: fixed-format floating point printing is
183 more accurate. This also fixes a bug reported by Adam Warner
184 related to the ~@F format directive.
185 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
187 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
188 dispatch table if the from-char is a dispatch macro character.
189 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
190 by the wordsize no longer ignore the last word. (reported by Lutz
192 * fixed bug in type checking of dynamic-extent variables. (reported
194 * optimization: sequence traversal functions use their freedom to
195 coerce function designators to functions.
196 * optimization: code with many calls to CLOS methods specialized on
197 CLOS classes has had redundant type checks removed.
198 * optimization: added declarations to speed up operations that access
199 the internal character database (for example STRING-UPCASE)
200 * optimization: comparison operations between floats and sufficiently small
201 fixnums no longer create extra rationals
202 * fixed some bugs related to Unicode integration:
203 ** portions of multibyte characters at the end of buffers for
204 character-based file input are correctly transferred to the
205 start of the buffer at the next read.
206 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
207 passing it through to OPEN.
208 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
209 argument given to internal calls to OPEN.
210 ** the built-in comment readers (introduced by character sequences
211 ";" and "#|") are more forgiving to encoding errors; they will
212 STYLE-WARN and then attempt to resync the stream at a character
213 boundary. (thanks to Teemu Kalvas)
214 * fixed some bugs in the x86-64 port:
215 ** Negative short int return values from c-calls are sign-extended
217 ** The stack is aligned to 16-bytes for c-calls, as required by
218 the ABI. (thanks to Cheuksan Wang)
219 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
220 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
221 for enums. (thanks to Vincent Arkesteijn)
222 ** Multiple small optimizations and bugfixes for floating point
224 * fixed some bugs revealed by Paul Dietz' test suite:
225 ** Space, Tab, Linefeed, Return and Page have the invalid
226 secondary constituent character trait.
227 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
229 ** WITH-INPUT-FROM-STRING should only update the index place on
231 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
233 ** Bit-array operations (BIT-AND and similar) worked incorrectly
234 with one-dimensional arrays with fill pointers.
235 ** TYPE-OF failed on a complex with an integer realpart and a
237 ** compiler failure during type inference for the code of form
238 (IF (EQL X (THE ...)) ...) (MISC.535).
240 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
241 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
242 platform. The port supports 61-bit fixnums, large memory spaces
243 and reloading shared object files.
244 * enhancement: saving cores with foreign code loaded is now
245 supported on ppc/Darwin in addition to the previously supported
247 * enhancement: the statistical profiler now walks deeper into the
248 call stack for more meaningful call-graphs and accrued time
249 reports (x86/x86-64 only). It also now reports time spent in
251 * enhancement: it is now possible to trace most individual methods
252 of a generic function in addition to tracing the generic function
254 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
256 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
257 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
258 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
259 produces an error. (thanks to Vincent Arkesteijn)
260 * bug fix: NAMESTRING on pathnames with :WILD components in their
261 directories works correctly. (thanks to Artem V. Andreev)
262 * fixed bug 125: compiler preserves identity of closures. (reported
264 * bug fixed: functions with &REST arguments sometimes failed with
265 "Undefined function" when compiled with (DEBUG 3). (reported by
267 * bug fix: overflow during compiling of setting element of a bit
268 vector with constant index and value. (reported by Timmy Douglas)
269 * build fix: fixed the dependence on *LOAD-PATHNAME* and
270 *COMPILE-FILE-PATHNAME* being absolute pathnames.
271 * on x86 compiler partially supports stack allocation of dynamic-extent
273 * GO and RETURN-FROM do not check the extent of their exit points
274 when compiled with SAFETY 0.
275 * fixed some bugs related to Unicode integration:
276 ** encoding and decoding errors are now much more robustly
277 handled; it should now be possible to recover even from invalid
278 input or output to the terminal. (thanks to Teemu Kalvas)
279 ** provided a first cut at implementing STRING-TO-OCTETS and
280 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
281 ** altered the SB-MD5 contributed module slightly, changing the
282 interface just enough for it to be supportable for builds where
283 lisp characters are not eight bits.
284 * fixed some bugs revealed by Paul Dietz' test suite:
285 ** the FORMATTER-generated functions for ~V[ conditionals require
286 the correct number of arguments.
287 ** READ-FROM-STRING returns the mandated second value when applied
288 to displaced strings.
289 ** the #\Rubout and #\Backspace characters are treated as invalid
290 constituent characters by the tokenizer.
292 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
293 * new feature: reloading changed shared object files with
294 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
295 * new feature: references to foreign variables and functions
296 can now be compiled and loaded before the corresponding shared
297 object file is loaded, as long as the foreign definitions are
298 available at runtime.
299 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
300 supported. (thanks to Dan Debertin)
301 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
302 just keywords. (thanks to Vincent Arkesteijn)
303 * fixed bug #331: structure-class instances corresponding to
304 DEFSTRUCT forms are now created eagerly.
305 * fixed bug #345: backtraces from calls to undefined functions work
306 on x86 as well. Related bug #61 is now also partially fixed on x86
307 (backtraces from throws to unknown catch tags.)
308 * bug fix: lambda-list parsing is now stricter vrt. order and number
309 of lambda-list keywords.
310 * bug fix: as specified by AMOP, an error is signalled if a
311 class-option appears multiple times in a DEFCLASS form. (reported
313 * bug fix: RANDOM can be compiled when the compiler derives the type
314 of its numeric argument as a disjoint set of small integers.
315 (reported by Paul Dietz)
316 * bug fix: starting a core saved with shared objects loaded when
317 those objects are not available no longer causes threaded SBCL to
318 hang. (reported by Sean Ross)
319 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
321 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
322 argument, which must always be NIL. (reported by Kalle Niemitalo)
323 * bug fix: printing 1.0d+23 no longer results in an error.
324 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
325 * bug fix: structure slot setters preserve evaluation order. (thanks
327 * fixed some bugs related to Unicode integration:
328 ** RUN-PROGRAM can allow its child to take input from a Lisp
329 stream. (reported by Stefan Scholl)
330 ** ASDF-INSTALL successfully downloads .tar.gz files in all
331 locales. (reported by Ken Causey)
332 * fixed some bugs revealed by Paul Dietz' test suite:
333 ** INCF, DECF and REMF evaluate their place form as specified in
335 ** FORMATTER expands ~{ ~} iteration directives with V or #
336 parameters correctly.
337 ** FORMATTER deals with the ~@[ ~] conditional directive where the
338 consequent uses no arguments correctly.
339 ** the system has a partial understanding of the (COMPLEX RATIO)
341 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
344 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
345 * new feature: a build-time option (controlled by the :SB-UNICODE
346 keyword feature, enabled by default) for building the system with
347 support for the entire 21-bit character space defined by the
349 * new feature: the system now has rudimentary external-format
350 support; the primary user-visible change at this time is that
351 characters with the high bit set (such as lower-case-e-acute) will
352 print correctly to a terminal in a UTF-8 environment.
353 * minor incompatible change: BASE-CHAR no longer names a class;
354 however, CHARACTER continues to do so, as required by ANSI.
355 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
356 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
357 has been moved to the SB-EXT package (temporarily re-exported from
359 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
360 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
362 * fixed bug #351: better error-handlind and reporting for malformed
364 * fixed bug #350: bignum-printing is now more memory-efficient,
365 allowing printing of very large bignums, eg. (expt 2 10000000).
366 (reported by Bruno Haible)
367 * fixed bug #302: better primitive-type selection for intersection
369 * fixed bug #308: non-graphic characters now all have names, as
370 required. (reported by Bruno Haible)
371 * bug fix: redefining a class with different superclasses now correctly
372 removes it from the direct-subclasses of its previous superclasses.
373 (reported by David Morse)
374 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
375 new value now works. (reported by Bruno Haible)
376 * bug fix: correct canonicalization of multiple non-standard slot
377 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
378 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
379 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
380 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
381 now exists, an signals an error.
382 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
383 during macroexpansion for non-iteration variables as well. (reported
384 by Bruno Haible for CMUCL)
385 * bug fix: Cyclic structures and unprintable objects in compiler
386 messages no longer cause errors. (reported by Bruno Haible)
387 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
388 and READ-FROM-STRING all now return a primary value of NIL if
389 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
390 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
391 (reported by Bruno Haible for CMUCL)
392 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
393 do not designate a condition. (reported by Bruno Haible for
395 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
396 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
397 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
398 permitting resolution in favour of any of the conflicting symbols.
399 (reported by Bruno Haible for CMUCL)
400 * bug fix: EQUAL compiler optimizations is less aggressive on
401 strings which can potentially compare true despite having distinct
402 specialized array element types.
403 * bug fix: unit enumerations can be defined without dividing by
404 zero. (thanks to Vincent Arkesteijn)
405 * FORMAT compile-time argument count checking has been enhanced.
406 (report from Bruno Haible for CMUCL)
407 * a partial workaround for the bug 262: the compiler does not try to
408 inline-expand a local function doing RETURN-FROM from a deleted
409 BLOCK. (thanks to Peter Denno for the bug report and to David
410 Wragg for the simple test case)
411 * fixed some bugs revealed by Paul Dietz' test suite:
412 ** RENAME-PACKAGE allows all package designators as new package
414 ** constraint propagation and lambda variable substitution are
415 more cautious in dealing with partially deleted code.
416 ** compiler sometimes lost reoptimization passes.
417 ** CERROR, given a condition as condition designator, passes its
418 remaining arguments to the continue format control without
420 ** Case-altering FORMAT directives work correctly on non-ASCII
422 ** The REAL type specifier handles bounds outside the floating
423 point ranges without signalling FLOATING-POINT-OVERFLOW.
424 ** Functions with IR1-transformations can create intercomponent
425 references to global functions.
426 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
428 ** FORMAT ~R treats a nil value for its first parameter correctly.
430 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
431 * enhancement: saving cores with foreign code loaded is now
432 supported on x86/NetBSD and sparc/Linux in addition to the previously
434 * bug fix: on some platforms repeated installations caused multiple
435 copies of HTML documentation to be installed -- should not happen
436 any more. (reported by Stefan Scholl)
437 * bug fix: parsing self-recursive alien record types multiple times
438 no longer causes infinite recursion. (reported by Thomas F. Burdick,
439 original patch by Helmut Eller for CMUCL)
440 * bug fix: stack-exhaustion detection works now on NetBSD as well.
441 (thanks to Richard Kreuter)
442 * bug fix: defining classes whose accessors are methods on existing
443 generic functions in other (locked) packages no longer signals
444 bogus package lock violations. (reported by François-René Rideau)
445 * bug fix: special variables as DEFMETHOD parameters no longer have
446 associated bogus type declarations. (reported by David Wragg and
448 * bug fix: read-write consistency on streams of element-type
449 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
450 * bug fix: redefiniton of the only method of a generic function with
451 no DEFGENERIC no longer emits a full WARNING. In addition,
452 redefinition of generic functions with no DEFGENERIC to an
453 incompatible lambda list now signals an error. (thanks to Zach
455 * bug fix: DEFGENERIC now works even when there's a function of the
456 same name in an enclosing lexical environment. (thanks to Zach
458 * fixed compiler failure, caused by instrumenting code during
459 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
460 * optimization: added loop analysis and improved register allocation
461 * fixed some bugs revealed by Paul Dietz' test suite:
462 ** POSITION on displaced vectors with non-zero displacement
463 returns the right answer.
464 ** (SIMPLE-STRING) is a valid type specifier for sequence
466 ** *PRINT-LEVEL* handling for slotless structures is pedantically
468 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
469 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
471 ** The justification version of the FORMAT ~< directive treats
472 non-zero minpad parameter correctly.
474 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
475 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
476 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
477 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
478 the supported interface.
479 * new feature: Single-stepping of code compiled with DEBUG 2 or
480 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
481 * new feature: saving cores with foreign code loaded is now
482 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
483 Timothy Moore's work for CMUCL)
484 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
485 parameters to * instead of NIL if no initform is supplied.
486 (reported by Johan Bockgaard)
487 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
488 correctly when FUNCALL appears as the car of the form. Note:
489 despite this FUNCALL forms are not currently subject to
490 compiler-macro expansion. (port of Raymond Toy's fix for the
491 same from CMUCL, reported by Johan Bockgaard)
492 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
493 (thanks for Teemu Kalvas)
494 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
495 case instead of terminating the iteration (thanks for Julian
496 Squires, Sean Champ and Raymond Toy)
497 * bug fix: incorrect expansion of defgeneric that caused a style
498 warning. (thanks for Zach Beane)
499 * bug fix: slot accessor effective method computation works properly
500 for classes with multiple non-standard applicable methods on
501 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
502 * on x86 compiler supports stack allocation of results of LIST and
503 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
504 CMUCL implementation by Gerd Moellmann)
505 * fixed some bugs revealed by Paul Dietz' test suite:
506 ** FORMAT strings with both the ~<~:;~> form of the justification
507 directive and pretty-printing directives cause an error.
509 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
510 * incompatible change: the internal functions
511 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
512 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
513 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
514 instead of the old functions.
515 * new feature: on platforms where "dladdr" is available foreign
516 function names now appear in backtraces. (based on Helmut Eller's
518 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
519 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
521 * documentation: ASDF manual is now distributed with SBCL.
522 * bug fix: SBCL can now load its contributed modules with REQUIRE
523 even if the system-provided entries have been removed from
524 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
526 * bug fix: backtraces involving undefined functions or assembly
527 routines are more informative. (thanks to Brian Downing)
528 * bug fix: mutually referent alien structures now work correctly.
529 (reported by Rick Taube)
530 * bug fix: structures defined by WITH-ALIEN can be referred to
531 within other definitions in the same WITH-ALIEN.
532 * bug fix: division operators (MOD, TRUNCATE and the like) with
533 constant zero divisors and integer dividends no longer generate
535 * bug fix: provide default methods for INPUT-STREAM-P and
536 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
537 * bug fix: improve the detection and resolution of MOP metacycles
538 (where effective method computation depends on the generic function
539 having its effective method computed). (reported by Bruno Haible)
540 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
541 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
542 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
543 * bug fix: pathnames with relative directory components can be
544 represented relative to default pathnames.
545 * optimization: in taking the GCD of bignums, reduce the two bignums
546 to approximately the same size (using Euclid's algorithm) before
547 applying the more sophisticated binary GCD. (thanks to Juho
549 * optimization: COUNT on bitvectors now operates word-at-a-time.
550 * optimization: ASH with a positive, but not necessarily constant,
551 (leftwards) shift, when in a modular context, is compiled to a
553 * fixed some bugs revealed by Paul Dietz' test suite:
554 ** FORMAT variable parameters ("~V<char>") are defaulted properly
555 if the corresponding argument is NIL.
556 ** FORMAT directives accepting a minpad parameter treat negative
558 ** literal commas as character parameters to FORMAT directives are
560 ** literal spaces directly after ~<Newline> directives within a
561 format-logical-block (~:< ~@:>) do not induce :FILL-style
562 conditional newlines.
563 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
564 as specified: it no longer includes conditional newlines).
565 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
567 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
568 :PER-LINE-PREFIX argument does not evaluate to a string.
570 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
571 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
572 the manual for details; for now, package locks can be disabled by
573 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
574 if no major problems are found then it is likely that they will be
575 compiled in unconditionally.
576 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
577 unsupported operators on all platforms. To load a shared library
578 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
579 object file, link it into a shared library outside of SBCL and
580 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
582 * bug workaround: ROOM T and the GC were not working together
583 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
584 reporting facility aren't preserved. That reporting has been
585 disabled, so now until and unless someone figures out how to make it
586 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
587 * minor incompatible change: as threatened around sbcl-0.8.0, the
588 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
589 The symbols are also exported from SB-PCL for backwards
590 compatibility, but more so than before SB-PCL should be treated as
591 an implementation-internal package.
592 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
594 * the SB-SPROF contrib now works on (most) non-x86 architectures.
595 It is known as of this release not to work on the Alpha, however.
596 * fixed bug #167: errors signalled due to illegal syntax in method
597 bodies are now more legible.
598 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
599 designators and can hence be used with TYPEP.
600 * fixed bug #333: CHECK-TYPE now ensures that the type error
601 signalled, if any, has the right object to be accessed by
602 TYPE-ERROR-DATUM. (reported by Tony Martinez)
603 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
604 5.1.2.3 for multiple-value place subforms. (reported by Kalle
606 * fixed bug #334: programmatic addition of slots using specialized
607 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
608 and :ALLOCATION :CLASS slots.
609 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
610 float exponents. (rereported by Peter Seibel)
611 * fixed a bug: #\Space (and other whitespace characters) are no
612 longer considered to be macro characters in standard syntax by
614 * fixed bug: initialization of condition class metaobjects no longer
615 causes an instance of the condition to be created. (reported by
617 * fixed bug: it is now possible to have slots such that
618 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
619 description is neither :INSTANCE nor :CLASS.
620 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
621 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
622 SLOT-VALUE-USING-CLASS), and no longer causes errors with
623 non-standard SLOT-DEFINITION-ALLOCATION values.
624 * fixed bugs: various race conditions handling exiting threads.
625 CL-PPCRE's thread torture test now passes.
626 * fixed arguably-a-bug: GC time proportional to number of threads in
627 system even when most of them are idle
628 * optimization: improved performance of BIT and SBIT on bit-vectors.
629 * .fasl file incompatibility: The fasl file version number has
630 been incremented because of changes associated with package locks.
632 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
633 * minor incompatible change: the system no longer provides
634 optimization hints (conditions which are TYPEP
635 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
636 mode; these hints are still emitted when higher SPEED optimization
638 * new contrib module: a sampling profiler (profiling by statistical
639 sampling, rather than by instrumenting functions) is available as
640 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
641 * the behaviour of the standard function ED is now customizeable by
642 third parties through a hook variable: see ED's documentation
643 string for information on the protocol.
644 * the compiler no longer emits efficiency notes for (FUNCALL X)
645 when the type of X is uncertain under default optimization
647 * fixed bug 276: mutating a binding of a specialized parameter to a
648 method to something that is not TYPEP the specializer is now
650 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
651 exhibit the required behaviour on the x86 platform. (thanks to
652 Peter van Eynde, Eric Marsden and Bruno Haible)
653 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
654 even for difficult arguments. (reported by Peter Graves)
655 * fixed bug 141a: the backquote printer now descends quoted
657 * fixed another bug in backquote printing: no more destructive
658 modification of the form's list structure. (reported by Brian
660 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
661 move between its address being taken and the call to
662 interrupt_thread, fixing a crashing race condition.
663 * the SB-POSIX contrib implementation has been adjusted so that it
664 no longer exhibits ridiculously poor performance when constructing
665 instances corresponding to C structs.
667 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
668 * minor incompatible change: the sb-grovel contrib now treats C
669 structures as alien (in the sense of SB-ALIEN) objects rather than
670 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
671 has implications for memory management of client code
672 (sb-grovel-returned objects must now be manually managed) and for
673 type safety (alien objects now have full types).
674 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
675 used to control emission of compiler diagnostics, rather than the
676 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
677 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
678 quality should be considered deprecated.
679 * install.sh now installs the user manual as well
680 * (not quite a new documentable feature, but worth considering in
681 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
682 beginnings of a semantically meaningful condition hierarchy is
683 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
685 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
686 designator as the defaults argument.
687 * fixed bug: Displaced arrays whose displaced-to array has become
688 too small now cause ARRAY-DIMENSION to signal an error, providing
689 for safer bounds-checking. (reported by Bruno Haible)
690 * fixed bug: DEFCLASS slot definitions with identical :READER and
691 :WRITER names now signal a reasonable error. (reported by Thomas
693 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
695 * fixed bug: Invalid slot specification errors now print correctly.
696 (thanks to Zach Beane)
697 * fixed bug 320: Shared to local slot value transfers in class
698 redefinitions now happen corrently from superclasses as
699 well. (reported by Bruno Haible)
700 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
702 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
703 improper lists correctly. (reported by Bruno Haible)
704 * fixed bug 313: source-transform for <fun-name> was erroneously
705 applied to a call of a value of a variable with name <fun-name>.
706 (reported by Antonio Menezes Leitao)
707 * fixed bug 307: The obsolete instance protocol ensures that
708 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
709 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
710 remove dead unknown-values globs from the middle of the stack.
711 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
712 easier to resume long computations after using *BREAK-ON-SIGNALS*
713 to diagnose and fix failures. (thanks to Nikodemus Siivola)
714 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
715 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
716 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
717 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
718 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
720 * fixed a (fairly theoretical) bug in string printing: if
721 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
722 does not have array-element-type equal to the most general string
724 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
725 function name. (thanks to Nikodemus Siivola)
726 * fixed bug: compile-time format string checker failed on
727 non-closed ~{. (reported by Thomas F Burdick)
728 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
729 don't warn on legal loop code involving more than one aggregate
730 boolean. (thanks to Nikodemus Siivola)
731 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
732 in keyword position to MAKE-PACKAGE was not being reported as
734 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
735 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
737 * fixed bugs 280 and 312: the checking for multiple definitions in a
738 file is less likely to become confused by uses of inline
740 * fixed bug: the #S reader macro performs the keyword coercion
741 specified for slot names. (reported by Kalle Niemitalo)
742 * fixed bug: lambda lists may contain symbols whose names start with
743 & but are not lambda-list-keywords; their occurrence triggers a
745 * fixed bug 321: define-method-combination argument lambda lists do
746 not cause invalid code to be generated when &optional or &aux
747 variables have default values. (reported by Bruno Haible)
748 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
749 have CLOS classes; this appears also to have fixed reported
750 instability in stack exhaustion detection.
751 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
752 MUFFLE-WARNING when no associated restart is present is now
754 * optimization: rearranged the expansion of various defining macros
755 so that each expands into only one top-level form in a
756 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
758 * optimization: used a previously little-used slot in symbols to
759 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
761 * fixed some bugs revealed by Paul Dietz' test suite:
762 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
764 ** Fixed an optimization bug related to inheritance of initforms
765 from local to shared slots.
766 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
767 ** CAST optimizer forgot to flush argument derived type.
768 ** print/read consistency on floats is now orders of magnitude
769 more likely. (thanks also to Bruno Haible for a similar report
771 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
772 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
773 for objects of type REAL. Make it so.
774 ** SXHASH is specified (infelicitously) to respect similarity,
775 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
776 it so. (thanks to Markus Ziegler)
777 ** on the Alpha, the compiler succeeds in compiling functions
778 returning a known number of arguments greater than 63.
779 ** fixed handling of invalid NIL arguments in keyword position
781 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
783 ** provided a readably-printable representation for RANDOM-STATE
785 ** ensured that pathnames, where they have a namestring, always
786 print using #P"..." syntax.
788 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
789 * Support for the forthcoming 2.0 version of the NetBSD kernel
790 running on x86 hardware has been added. (thanks to Perry
791 E. Metzger most immediately, and others for their past work)
792 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
793 current release version; SBCL's OpenBSD support had been broken
794 since about the time of OpenBSD's switch to ELF binary format.)
795 * [placeholder for DX summary]
796 ** user code with &REST lists declared dynamic-extent, under high
797 speed or space and low safety and debug optimization policy.
798 * The manual has been converted to Texinfo format and the debugger
799 chapter from the cmucl manual has been added.
800 * A facility has been added to extract documentation strings from
801 sbcl and store them as Texinfo-formatted snippets for inclusion in
802 the manual (via Texinfo's @include directive)
803 * bug fix: compiler emitted division in optimized DEREF. (thanks for
804 the test case to Dave Roberts)
805 * bug fix: multidimensional simple arrays loaded from FASLs had fill
806 pointers. (reported by Sean Ross)
807 * bug fix: PROFILE output is printed nicely even for large numerical
808 values. (thanks to Zach Beane)
809 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
810 greater than 32 handle EOF correctly.
811 * bug fix: on X86 an immediate argument of the IMUL instruction is
812 correctly printed in disassembly. (thanks to Lutz Euler)
813 * bug fix: class slots in redefined classes preserve their old
814 values. (thanks to Bruno Haible and Nikodemus Siivola)
815 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
816 succeeds. (reported by Marco Baringer)
817 * fixed some bugs revealed by Paul Dietz' test suite:
818 ** READ-SEQUENCE now works on ECHO-STREAMs.
819 ** RATIONALIZE works more according to its specification. (thanks
822 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
823 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
824 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
825 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
826 should matter to you only if you rebind the printer control
827 variables and then find you want different bindings in the
828 debugger than in the ordinary execution of your program.)
829 * The runtime build system has been tweaked to support building
830 (on SPARC/SunOS) using a C compiler which invokes Sun's own
831 assembler and linker. (thanks to Nikodemus Siivola)
832 * Unbound, undefined, undeclared variables now trigger full
833 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
834 more often programmer error than deliberate exploitation of undefined
836 * optimization: the hash algorithm for strings has changed to one
837 that is less vulnerable to spurious collisions. (thanks to Juho
839 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
840 needless bounds checking. (thanks to Juho Snellman)
841 * optimization: implemented multiplication as a modular
842 (UNSIGNED-BYTE 32) operation on the PPC backend.
843 * fixed some bugs revealed by Paul Dietz' test suite:
844 ** ADJUST-ARRAY now copies the datum in a zero rank array if
846 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
847 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
848 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
850 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
852 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
854 ** ECHO-STREAMs no longer attempt to echo the end of file value to
855 their output stream on EOF from read.
856 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
857 have been read to end-of-file.
858 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
860 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
861 description of determination of which consecutive characters
863 ** Printing the "Space" character with escaping on now yields "#\\ ",
864 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
865 ** Reading floating-point numbers with *READ-BASE* set to a number
866 less than 10 works correctly.
867 ** Reading floating-point numbers with *READ-BASE* set to a number
868 more than 10 works correctly.
869 ** Printing with *PRINT-READABLY* targets the standard readtable, not
870 the readtable currently in effect.
872 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
873 * minor incompatible change: parsing of namestrings on a physical
874 (Unix) host has changed; numbers after the final #\. in a
875 namestring are no longer interpreted as a version field. This is
876 intented to be largely invisible to the user, except that the
877 meaning of the namestring "*.*.*" has changed: it now refers to a
878 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
879 should usually be replaced by
880 (make-pathname :name :wild :type :wild :version :wild)
881 with the added benefit that this is more likely to be portable.
882 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
883 option to OPEN now signals an error if the file being opened
884 exists; this may have an impact on existing code.
885 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
886 signals received, on the PowerPC platforms (both Linux and
887 Darwin). (thanks to Pierre Mai for pointing out the location of
889 * several fixes on OS X: The system now builds and runs cleanly on
890 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
891 (thanks to Brian Mastenbrook)
892 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
893 second-resolution: integer multiples of 1/3600 between -24 and 24.
894 (thanks to Vincent Arkesteijn)
895 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
896 type when called with 1 argument; PEEK-CHAR checked type of
897 PEEK-TYPE only after having read first character from a
898 stream. (reported by Peter Graves)
899 * bug fix: the garbage collector now has much better locality
900 behaviour, and in particular no longer treats all memory as being
901 exhausted when many small objects point to each other in a deeply
903 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
904 recognized as being TYPEP their class.
905 * bug fix: the PUSHNEW documentation string has been corrected.
906 (thanks to Vincent Arkesteijn)
907 * bug fix: defaulting of the value for the last of an atypically
908 large number of multiple values being bound was not being
909 performed correctly on the Alpha or PPC platforms
910 * optimization: implemented multiplication as a modular
911 (UNSIGNED-BYTE 32) operation on the x86 backend.
912 * optimization: SEARCH on simple-base-strings can now be open-coded.
913 (see also contrib/compiler-extras.lisp for inspiration for
914 teaching the compiler about the Boyer-Moore algorithm).
915 * value, returned by MAX (and MIN) called with several EQUALP, but
916 not EQL, arguments now does not depend on compiler settings.
917 * fixed some bugs revealed by Paul Dietz' test suite:
918 ** in stack analysis liveness information is propagated from
919 non-local entry points.
920 ** pathwise CAST removing failed when the CAST node did not start
922 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
923 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
925 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
926 host is already defined.
927 ** RENAME-FILE works on streams instead of signalling an internal
929 ** PEEK-CHAR uses the current readtable when determining whether
930 or not a character is whitespace.
931 ** MERGE-PATHNAMES handles the case when the pathname does not
932 specify a name while the default-pathname specifies a version.
933 ** Pathnames now stand a better chance of respecting print/read
935 ** Attempting to use standardized file system operators with a
936 pathname with invalid :DIRECTORY components signals a
938 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
940 ** DIRECTORY on logical pathnames is more correct.
941 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
942 signal a TYPE-ERROR if their argument is not a stream.
943 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
944 designator argument does not designate a stream.
945 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
946 examining the synonym.
947 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
949 ** OPEN and WITH-OPEN-STREAM allow opening streams with
950 element-type larger than ([UN]SIGNED-BYTE 32).
952 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
953 * When built with the :SB-FUTEX feature, threaded builds now take
954 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
955 for faster/more reliable mutex and condition variable support.
956 * Incompatible change (but one you probably shouldn't have been using
957 anyway): the interface and code for arbitrating between multiple
958 threads in the same user session has been redesigned.
959 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
960 error when called without an explicit environment argument.
961 (thanks to Nikodemus Siivola)
962 * bug fix: buffered :DIRECTION :IO streams are less likely to become
963 confused about their position. (thanks to Adam Warner and Gerd
965 * bug fix: Pretty printing backquoted forms with unquotations in the
966 argument list position of various code constructs such as LAMBDA
967 now works correctly. (reported by Paul Dietz)
968 * bug fix: Pretty printing unquotations no longer loses all
969 stream position information.
970 * optimization: performance of string output streams is now less
971 poor for multiple small sequence writes.
972 * optimization: performance of CSUBTYPEP in the presence of complex
973 expressions involving CONS and NOT many times has been improved.
974 (reported by Paul Dietz)
975 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
977 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
978 streams for the REPL, for the debugger, and so on. (thanks to
980 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
981 * fixed some bugs revealed by Paul Dietz' test suite:
982 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
983 is ignored if it is too small, rather than propagating through
984 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
985 ** extremely complex negations of CONS types were not being
986 sufficiently canonicalized, leading to inconsistencies in
988 ** VALUES tranformer lost derived type.
990 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
991 * fixed a bootstrapping bug: the build process no longer assumes
992 that the various BOOLE-related constants have the same value in
993 host and target lisps. (noted by Paul Dietz' test suite on an
994 SBCL binary built from CLISP)
995 * The system can now be dynamically linked on the MIPS platform,
996 which enables dynamic loading of foreign code from Lisp. (thanks
997 to Ralf Baechle for discussions on the MIPS ABI)
998 * The system now records debugging information for its own source
999 files in a filesystem-position-independent manner, relative to
1000 the "SYS" logical host.
1001 * fixed a compiler bug: MV-LET convertion did not check references
1002 to the "max args" entry point. (reported by Brian Downing)
1003 * tweaked disassembly notes to be less confident about proclaiming
1004 some instruction as an LRA. (thanks to Brian Downing)
1005 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1006 now each have their own history, command character, and other
1007 characteristics. (thanks to David Lichteblau)
1008 * fixed some bugs revealed by Paul Dietz' test suite:
1009 ** compiler failure in compiling LOGAND expressions including a
1011 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1012 shift greater than 32.
1013 ** FLUSH-DEST did not mark blocks for type check regeneration.
1014 ** HANDLER-CASE failed to accept declarations in handler clauses
1015 in some circumstances.
1017 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1018 * New code in contrib/sb-introspect (still probably not entirely
1019 stable yet) provides some support for smart Lisp development
1020 environments like SLIME.
1021 * The conditions signalled for errors occurring when loading .fasl
1022 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1023 in a way which should help ASDF recover gracefully.
1024 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1025 stuff has been cleaned up. If you code contrib/ stuff, this might
1026 affect you, and you can look at contrib/README, contrib/STANDARDS,
1027 and/or the 0.8.4.27 diff to check.
1028 * In full calls the compiler now does not generate checks for declared
1029 argument types for all arguments.
1030 * various threading fixes
1031 ** and some experimental patches which didn't make it into
1032 the main tree for this release, but which are shipped in
1033 contrib/experimental-thread.patch as a possible fix for some
1034 failures (deadlock, spinning...) in GC-intensive multithreaded
1036 * fixed PPC build problem (source code incompatibility of different
1037 library versions): added offsetof() hackery which attempts to divine
1038 where glibc maintainers put uc_mcontext today
1039 * fixed bug 282: compiler does not trust type assertions while passing
1040 arguments to a full call.
1041 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1042 &OPTIONAL VALUES type parameter.
1043 * fix bug 214: algorithm for noting rejected templates is now more
1044 similar to that of template seletion. (also reported by rydis on
1046 * fixed bug 141b: printing backquoted information readably and prettily
1047 inserts a space where necessary.
1048 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1049 (CHARACTER 10) are now reported as errors, rather than propagated
1050 as unknown types. (reported by piso on #lisp)
1051 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1052 with values NIL and :ERROR. (thanks to Milan Zamazal)
1053 * fixed bug 191c: CLOS now does proper keyword argument checking as
1054 described in CLHS 7.6.5 and 7.6.5.1.
1055 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1056 longer raise an error; further, using a list as a for-as-arithmetic
1057 counter now raises a meaningful error.
1058 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1059 now understood by sequence creation functions such as MAKE-SEQUENCE
1061 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1062 a STRING-INPUT-STREAM.
1063 * compiler enhancement: SIGNUM is now better able to derive the type
1065 * type declarations inside WITH-SLOTS are checked. (reported by
1067 * fixed some bugs revealed by Paul Dietz' test suite:
1068 ** incorrect optimization of TRUNCATE for a positive first
1069 argument and negative second.
1070 ** compiler failure in let-convertion during flushing dead code.
1071 ** compiler failure while deriving type of TRUNCATE on an
1072 interval, containing 0.
1073 ** ASH of a negative bignum by a negative bignum count now returns
1075 ** intersection of CONS types now canonicalizes properly, fixing
1076 inconsistencies in SUBTYPEP.
1078 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1079 * incompatible change: The --disable-debugger command line
1080 option now clobbers the debugger at a more fundamental
1081 level, by redefining #'INVOKE-DEBUGGER instead of by
1082 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1083 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1084 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1085 enter the debugger REPL and then suffer recursive errors
1086 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1087 while under the new system, BREAK in --disable-debugger mode
1088 terminates the system just as an unhandled error would.
1089 * fixed compiler performance when processing loops with a step >1;
1090 * bug fix: DOCUMENTATION now retrieves generic function
1091 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1092 support has been systematized, and now supports the methods
1093 specified by ANSI, along with a default method and a method for
1094 slot documentation. (reported by Nathan Froyd)
1095 * bug fix: effective methods associated with a generic function are
1096 no longer cached over a change of that generic function's method
1097 combination. (reported by Andreas Fuchs)
1098 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1099 names a non-existent directory. (thanks to Andreas Fuchs)
1100 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1101 the PPC platform, lead to stack corruption; this has been fixed.
1102 (reported by Rainer Joswig)
1103 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1104 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1105 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1106 arithmetic in general, is now much more reliable on the PPC
1108 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1109 platform now returns the right answer.
1110 * optimization: restored some effective method precomputation in
1111 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1112 precomputation is now tunable.
1113 * optimization: compiler-internal data structure use has been
1114 reviewed, and changes have been made that should improve the
1115 performance of the compiler by about 20%.
1116 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1117 simple-base-strings and simple-bit-vectors is improved.
1118 * optimization: the optimization of 32-bit logical and arithmetic
1119 functions introduced in version 0.8.3 on the x86 has been
1120 implemented on the mips, ppc and sparc platforms; an
1121 implementation of the same facility, but for 64-bit arithmetic,
1122 has been added for the alpha.
1123 * microoptimization: the compiler is better able to make use of the
1124 x86 LEA instruction for multiplication by constants.
1125 * bug fix: in some situations compiler did not report usage of
1126 generic arithmetic in (SPEED 3) policy.
1127 * bug 145b fix: compiler used wrong type specifier while converting
1128 MEMBER-types to numeric.
1129 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1131 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1132 index argument now works properly on the Alpha platform.
1133 * bug fix: floating point exception treatment on the Alpha platform
1135 * bug fix: FILE-POSITION works much better on string input and
1136 output streams. (thanks to Nikodemus Siivola)
1137 * bug fix: many threading/garbage collection symptoms sorted.
1138 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1139 might be pseudo-atomic.
1140 * internal change: Stopping for GC is now done with signals not
1141 ptrace. GC is now done in whichever thread wanted it, instead of
1143 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1145 * bug fix: result form in DO is not contained in the implicit
1147 * incompatible change: ICR structure is changed; the value part of
1148 CONTINUATION is now called LVAR; corresponding functions are
1149 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1150 * added type deriver for ISQRT (thanks to Robert E. Brown).
1151 * bug fix: better support for loading from the command line when an
1152 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1154 * fixed some bugs revealed by Paul Dietz' test suite:
1155 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1156 ** ROUND and FROUND now give the right answer when given very
1157 small float arguments.
1158 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1160 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1161 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1162 ** LCM should return a non-negative integer.
1163 ** PARSE-INTEGER returned the index of a terminator instead of the
1164 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1165 ** PARSE-INTEGER returned an incorrect index being applied to a
1167 ** LCM with two arguments of 0 returns 0 rather than signalling
1169 ** unsigned addition of a 32-bit constant with the high bit set no
1170 longer causes an internal compiler error on the x86.
1171 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1172 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1173 with negative last argument.
1174 ** byte specifiers with zero size and position no longer cause
1175 an error during type derivation.
1176 ** bignum multiplication on the Alpha platform now returns the
1178 * porting: The system now builds on SuSE AMD64, although it still
1179 generates a 32-bit binary.
1180 * .fasl file incompatibility: The fasl file version number has
1181 been incremented (because of the changes to internal compiler
1182 data structures referred to above).
1184 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1185 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1186 more accurately, on the Darwin kernel running on PowerPC hardware.
1187 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1188 * Compiler code deletion notes now signal a condition of type
1189 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1190 an associated MUFFLE-WARNING restart.
1191 * The compiler now performs limited argument count validation of
1192 constant format strings in FORMAT, and where appropriate in ERROR,
1193 CERROR and WARN. (thanks to Gerd Moellmann)
1194 * New ASDF-INSTALL contrib can be used for automatic download and
1195 installation of third-party Lisp code from CCLAN or other sites
1197 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1198 which forces another thread to execute a function supplied by the
1200 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1201 now accept and act upon their :ELEMENT-TYPE keyword argument.
1202 (reported by Martin Atzmueller, Edi Weitz)
1203 * bug fix: FILE-POSITION now accepts position designators up to
1204 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1205 is the greater. (thanks to Patrik Nordebo)
1206 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1207 Kalle Olavi Niemitalo)
1208 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1209 printable. (reported by Eric Marsden)
1210 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1211 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1212 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1213 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1214 (thanks to Stig Sandoe)
1215 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1216 UNREAD-CHAR now correctly includes the unread character in the
1217 target sequence. (thanks to Gerd Moellmann)
1218 * bug fix in threaded builds: the system can now be suspended and
1219 resumed by shell job control with minimal disruption.
1220 * bug fixes in times and timezones >2038AD
1221 * better handling of "where is GNU make?" problem in build scripts
1222 (thanks to Nikodemus Siivola)
1223 * new optimization: inside a named function any reference to a
1224 function with the same name is considered to be a self-reference;
1225 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1226 optimization quality.
1227 * new optimization on x86: logical functions and + now have
1228 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1229 used when the result is truncated to 32 bits.
1230 * VALUES declaration is partially enabled.
1231 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1232 * fixed some bugs revealed by Paul Dietz' test suite:
1233 ** The system now obeys the constraint imposed by
1234 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1235 types form a lattice under type intersection.
1236 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1237 ** ASSOC now ignores NIL elements in an alist.
1238 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1239 and (1+ MOST-POSITIVE-FIXNUM) answers.
1240 ** The addition of a method with invalid qualifiers to a generic
1241 function does not cause an error to be signalled immediately;
1242 a warning is signalled, and the error is generated only on
1243 calling the generic function.
1244 * changed .fasl file version number, in part to add type codes for
1245 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1246 obscure ANSI requirements
1248 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1249 * fixed bug 148: failure to inline-expand a local function left
1250 garbage, confusing the compiler.
1251 * fixed bugs 3cd: structure slot readers perform type check if the
1252 slot can have an invalid value (i.e. it is either not initialized
1253 or can be written with a less specific slot writer).
1254 * bug fix: the compiler now traps array references to elements off
1255 the end of an array; previously, the bounds checking in some
1256 circumstances could go off-by-one.
1257 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1259 * type declarations for array element types now obey the description
1260 on the CLHS page "Declaration TYPE", as per discussions on
1261 sbcl-help around 2003-05-08. This means that a declaration
1262 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1263 declaration, all references to BAR will be asserted or assumed
1264 (with THE, so dependent on compiler policy) to involve objects of
1265 type FOO. Note that no such declaration is implied in
1266 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1267 * declared types of functions from the "Conditions"
1268 chapter. (reported by Paul Dietz)
1269 * bug fix: CERROR accepts a function as its first argument.
1270 * bug fix: NTH an NTHCDR accept a bignum as index
1271 arguments. (reported by Adam Warner)
1272 * optimization: character compare routines now optimize comparing
1273 against a constant character. (reported by Gilbert Baumann)
1274 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1275 argument now works properly on the MIPS platform.
1276 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1277 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1278 declared to have a functional type, does not cause an error
1280 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1281 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1282 * fixed bug: if last continuation of a deleted block has a
1283 destination, this destination should be deleted too. (reported by
1285 * fixed a bug in the bootstrap process: the host compiler's values
1286 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1287 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1288 test case from Patrik Nordebo)
1289 * improved the ability of the disassembler on the PPC platform to
1290 provide helpful disassembly notes.
1291 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1292 the class in more cases than previously.
1293 * bug fix: FILE-POSITION now understands :START and :END for
1294 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1295 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1296 8) data. (thanks to David Lichteblau)
1297 * bug fix: it is possible to add a method to a generic function
1298 without lambda list.
1299 * bug fix: reader failed to signal END-OF-FILE inside an
1300 object representation. (reported by Nikodemus Siivola)
1301 * fixed some bugs revealed by Paul Dietz' test suite:
1302 ** LAST and [N]BUTLAST should accept a bignum.
1303 ** condition slot accessors are methods.
1304 ** (VECTOR NIL) is a subtype of STRING.
1306 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1307 * minor incompatible change: some nonsensical specialized lambda
1308 lists (used in DEFMETHOD) which were previously ignored now signal
1310 * minor incompatible change: the system is now aware of the types of
1311 variables in the COMMON-LISP package, and will signal errors for
1312 most violations of these type constraints (where previously they
1313 were silently accepted).
1314 * minor incompatible change: COMPILE-FILE now uses the freedom
1315 afforded (ANSI 3.2.2.3) to use derived function types for
1316 functions defined in the same file. This also permits the system
1317 to warn on static type mismatches and function
1318 redefinition. (Currently it does not work with high DEBUG level.)
1319 * minor incompatible change: VALUES declaration is disabled.
1320 * When issuing notes, the compiler now signals a condition of type
1321 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1322 restart for use in user handlers. It is expected that the
1323 COMPILER-NOTE condition will eventually become a condition
1324 supertype to a hierarchy of note types, which will then be
1325 handleable in a similar fashion. However, at the moment, no such
1326 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1327 but it's an implementation detail, not a classification for the
1329 * Changes in type checking closed the following bugs:
1330 ** type checking of unused values (192b, 194d, 203);
1331 ** template selection based on unsafe type assertions (192c, 236);
1332 ** type checking in branches (194bc).
1333 * A short form of VALUES type specifier has ANSI meaning (it has
1334 increased the number of situations when SBCL cannot perform type
1336 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1337 keywords or constants is permissible.
1338 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1339 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1340 classes. (thanks to Antonio Martinez)
1341 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1342 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1343 argument to a pretty-print stream. Instead, it leaves any such
1344 operations to DESCRIBE-OBJECT methods.
1345 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1347 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1348 lambda lists. (thanks to David Lichteblau)
1349 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1350 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1351 (thanks to Gerd Moellmann)
1352 * fixed bug 246: increased compilation speed of long
1353 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1355 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1357 * DEFCONSTANT now throws a condition of type
1358 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1359 constant to a non-EQL value; CONTINUE and ABORT restarts
1360 respectively change and preserve the value.
1361 * fixed bug 63: The code walker, part of the implementation of CLOS,
1362 is now better at handling symbol macros.
1363 * bug fix: There is no longer an internal implementation type named
1364 CL:LENGTH. (reported by Raymond Toy)
1365 * bug fix: In macro-like defining macros/special operators the
1366 implicit block does not enclose the lambda list.
1367 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1368 enclosed in parentheses) are not suitable as type specifiers, and
1369 their use properly signals an error now.
1370 * bug fix: An argument count mismatch for a type specifier in code
1371 being compiled no longer causes an unhandled error at compile
1372 time, but signals a compile-time warning.
1373 * fixed simple vector readable printing
1374 * bug fix: DESCRIBE takes more care over whether the class
1375 precedence list slot of a class is bound before accessing it.
1376 (reported by Markus Krummenacker)
1377 * bug fix: FORMATTER can successfully compile pretty-printer format
1378 strings which use variants of the ~* directive inside.
1379 * bug fix: SEARCH now applies its TEST predicate to the elements of
1380 the arguments in the correct order. (thanks to Wolfhard Buss)
1381 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1382 in type inference. (thanks to Robert E. Brown)
1383 * bug fix: Reading in symbols with an explicit package name of ""
1384 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1385 * fixed some bugs revealed by Paul Dietz' test suite:
1386 ** NIL is now allowed as a structure slot name.
1387 ** Arbitrary numbers, not just REALs, are allowed in certain
1388 circumstances in LOOP for-as-arithmetic clauses.
1389 ** Multiple class redefinitions before slot access no longer
1390 causes a type error.
1391 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1392 association between the name and a class.
1393 ** Generic functions with non-standard method-combination and over
1394 five methods all of which return constants no longer return NIL
1395 after the first few invocations. (thanks to Gerd Moellmann)
1396 ** CALL-NEXT-METHOD with no arguments now passes the original
1397 values of the arguments, even in the presence of assignment.
1398 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1400 ** DISASSEMBLE works with closures and funcallable instances.
1401 ** ADD-METHOD now returns the generic function, not the new method.
1402 ** FIND-METHOD signals an error if the lengths of the specializers
1403 is incompatible with the generic function, even if the ERRORP
1405 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1406 which its argument is a member.
1407 ** DEFCLASS only redefines the class named by its class-name
1408 argument if that name is the proper name of the class;
1409 otherwise, it creates a new class.
1410 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1411 of the UNBOUND-SLOT condition to the name of the slot.
1412 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1413 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1414 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1415 SLOT-MAKUNBOUND in the specified fashion.
1417 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1418 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1419 cross-compilation host. As a consequence, we can now bootstrap our
1420 way up to SBCL starting with a bare gcc toolchain and human-readable
1421 source code (first the source to CLISP, then the source to SBCL).
1422 * A contributed module containing a partial implementation of the
1423 simple-streams interface has been included. (thanks to Rudi
1425 * A contributed module implementing the RFC1321 Message Digest
1426 Algorithm, known as MD5, has been included.
1427 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1428 no longer has any effect, as the code controlled by this feature
1429 has been deleted. (As far as we know, no-one has ever built using
1430 this feature, and its semantics were confused in any case).
1431 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1432 work on conditions (as required by the ANSI specification),
1433 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1434 expected behaviour on conditions. Users should note, however,
1435 that such behaviour is not required by the ANSI specification,
1436 and so use of this behaviour may render their code unportable.
1437 * fixed some bugs revealed by Paul Dietz' test suite:
1438 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1440 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1441 ** SXHASH on condition objects no longer returns NIL.
1442 ** :ALLOCATION :CLASS slots are better treated; their values are
1443 updated on class redefinition, and initforms inherited from
1444 superclasses are applied.
1445 ** REMOVE-METHOD returns its generic function argument even when
1446 no method was removed.
1447 ** SHARED-INITIALIZE now initializes the values of the requested
1448 slots, including those with :ALLOCATION :CLASS.
1449 ** ALLOCATE-INSTANCE now works on structure classes defined via
1450 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1452 ** SLOT-EXISTS-P now works on conditions, as well as structures
1454 ** MAKE-LOAD-FORM now has the required methods on
1455 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1456 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1457 keyword, and now implements the SLOT-NAMES argument.
1458 ** methods with &OPTIONAL arguments no longer allow too many
1459 arguments to be passed in the call without error.
1460 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1461 option is consistent with the required arguments of the generic
1462 function lambda list.
1463 * bug fix: REQUIRE accepts a string designator. (Thanks to
1465 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1466 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1467 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1468 * bug fix: The compiler now checks for duplicated variables in macro
1470 * bug fix: SETQ on globals returns the correct value.
1471 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1472 not a COMPILER-ERROR (followed by some other strange error on
1473 choosing the CONTINUE restart).
1474 * bug fix: make.sh and friends are now more consistent in the way that
1475 they look for GNU "make".
1477 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1478 * experimental native threads support (on x86 Linux >=2.4 only).
1479 This is not compiled in by default: you need to add :SB-THREAD to
1480 the target features. See the "Beyond ANSI" chapter of the manual
1482 * fix for longstanding nonANSIism: The old distinction between
1483 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1484 The return value from CL:FIND-CLASS is now a CLOS class, and
1485 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1486 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1487 * An interface to the MetaObject Protocol, as described in Kiczales,
1488 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1489 MIT Press, 1991, is available from the SB-MOP package.
1490 * incompatible change: the SB-PCL package should now be considered
1491 a private implementation detail, and no longer a semi-private MOP
1493 * minor incompatible change: due to rearrangement for threads, the
1494 control stack and binding stack are now allocated at arbitrary
1495 addresses instead of being hardcoded per-port. Users affected by
1496 this probably have to be doing advanced things with shared
1497 libraries, and will know who they are.
1498 * minor incompatible change: Previously, all --eval forms used were
1499 processed with READ before any of them were processed with EVAL.
1500 Now each --eval form is processed with both READ and EVAL before
1501 the next --eval form is processed. (Thus package operations like
1502 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1503 work as the user might reasonably expect.)
1504 * minor incompatible change: *STANDARD-INPUT* is now only an
1505 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1507 * minor incompatible change: Y-OR-N-P is now character-oriented,
1508 not line oriented. Also, YES-OR-NO-P now works without errors.
1509 (thanks to Antonio Martinez)
1510 * sb-aclrepl module improvements: an integrated inspector, added
1511 repl features, and a bug fix to :trace command.
1512 * Known functions, which cannot be open coded by the backend, are
1513 considered to be able to check types of their arguments. (fixing
1514 a bug report by Nathan J. Froyd)
1515 * fixed a bug in computing method discriminating functions: It is
1516 now possible to define methods specialized on classes which have
1517 forward-referenced superclasses. (thanks to Gerd Moellmann)
1518 * fixed evaluation order in optional entries (reported by Gilbert
1520 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1521 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1522 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1523 required-by-AMOP NAME argument, as well as CLASS and
1524 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1525 * fixed bug 20: DEFMETHOD can define methods using names that are
1526 not the proper names of classes to designate class specializers.
1527 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1528 Unix file descriptors, instead of blowing up. (thanks to Antonio
1530 * Garbage collection refactoring: user-visible change is that a
1531 call to the GC function during WITHOUT-GCING will not do garbage
1532 collection until the end of the WITHOUT-GCING. If you were doing
1533 this you were probably losing anyway.
1534 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1535 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1536 TYPEP the latter but not the former.
1537 * The compiler issues a full WARNING for calls to undefined functions
1538 with names from the CL package.
1539 * MAP-INTO for a vector destination is open coded. (reported by
1540 Brian Downing on c.l.l)
1541 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1542 documentation string.
1543 * fixed some bugs revealed by Paul Dietz' test suite:
1544 ** COPY-ALIST now signals an error if its argument is a dotted
1546 ** Condition slots are now accessed more correctly in the presence
1547 of multiple initargs for a given slot.
1548 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1549 exclude restarts of the same name associated with a different
1551 ** DEFCLASS of forward-referenced classes with another
1552 forward-referenced class in the superclasses list no longer
1554 ** Condition slots are now initialized once each, not multiple
1555 times. (thanks to Gerd Moellmann)
1556 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1558 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1559 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1560 arguments contain duplicated elements.
1561 ** RESTART-CASE understands local macros.
1562 ** RESTART-CASE associates exactly its own restarts with a condition.
1563 ** ENDP in safe mode checks its argument to be of type LIST.
1564 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1565 in question is unbound.
1566 ** Optimized MAKE-INSTANCE functions no longer cause internal
1567 assertion failures in the presence of duplicate initargs.
1568 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1569 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1570 propagated even in the ctor optimized implementation of
1572 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1574 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1575 * a better implementation of SXHASH on (simple) bit vectors,
1576 measured both in execution speed and in distribution of results
1577 over the positive fixnums, has been installed. Likewise, a better
1578 implementation of EQUAL for simple bit vectors is now available.
1579 * fixed CEILING optimization for a divisor of form 2^k.
1580 * fixed bug 240 (emitting extra style warnings "using the lexical
1581 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1582 by Antonio Martinez)
1583 * fixed SXHASH, giving different results for NIL depending on type
1584 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1585 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1586 variable. (found by Rolf Wester)
1587 * SBCL does not ignore type declarations for special
1588 variables. (reported by rif on c.l.l 2003-03-05)
1589 * some bug fixes in contrib/sb-aclrepl/
1590 * fixed some bugs revealed by Paul Dietz' test suite:
1591 ** a bug in the CONS type specifier, whereby the CAR and CDR
1592 types got intertwined, has been fixed;
1593 ** the type system is now able to reason about the interaction
1594 between INTEGER and RATIO types more completely;
1595 ** APPEND, [N]REVERSE and NRECONC check that those their
1596 arguments, which must be proper lists, are really so;
1597 ** An array specialized to be unable to hold elements has been
1598 implemented, as required -- yes, really -- by ANSI;
1599 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1600 on malformed property lists;
1602 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1603 * incompatible packaging change: in line with Unix convention,
1604 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1605 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1606 * REQUIRE and PROVIDE are now optionally capable of doing something
1607 useful. See the documentation string for REQUIRE.
1608 * infrastructure for a managed SBCL contrib system: contributed
1609 modules in this release include:
1610 ** the ASDF system definition facility;
1611 ** an interface to the BSD Sockets API;
1612 ** an ACL-like convenience interface to the repl;
1613 (thanks to Kevin Rosenberg)
1614 ** an implementation of ROTATE-BYTE, with efficient implementation
1616 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1617 gives a closer approximation to the right answer than previously.
1618 (thanks to Raymond Toy)
1619 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1620 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1621 optional environment arguments, as required by ANSI.
1622 * fixed bugs in other functions taking environment objects, allowing
1623 calls with an explicit NIL environment argument to be compiled
1625 * fixed bug 228: primary return values from
1626 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1627 COMPILE or FUNCTION.
1628 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1629 :TYPE will no longer signal errors on innocuous objects.
1630 * fixed bug 231b: SETQ is better at respecting type declarations in
1631 the lexical environment.
1632 * fixed a bug in DEFCLASS: classes named by symbols with no or
1633 unprintable packages can now be defined.
1634 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1635 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1636 * fixed compiler failure related to checking types of functions.
1637 (reported by Robert E. Brown)
1638 * the compiler is now much more consistent in its error-checking
1639 treatment of bounding index arguments to sequence functions: in
1640 (SAFETY 3) code, errors will be signalled in almost all cases if
1641 invalid sequence bounding indices are passed to functions defined
1642 by ANSI to operate on sequences.
1643 * fixed a bug in the build procedure: documentation of SBCL-specific
1644 packages is now preserved and available in the final Lisp image.
1645 * lifted FDEFINITION lookup out of loops in the implementation of
1646 many list operations. (thanks to Robert E. Brown)
1647 * fixed a bug in the reader: the #n# reader macro now works for
1648 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1649 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1650 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1651 E. Brown and Tony Martinez respectively)
1652 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1653 SLOT-BOUNDP: in particular, it is now able to optimize them much
1654 better, and is now not vulnerable to having packages renamed.
1655 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1656 always called when a slot is not present in an instance. (thanks
1658 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1659 longer signal an error on unreading a character following EOF on
1660 the previous constituent stream. (thanks to Tony Martinez)
1661 * fixed some bugs revealed by Paul Dietz' test suite:
1662 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1663 not just nonnegative fixnums;
1664 ** the logical bit-array operators such as BIT-AND now accept an
1665 explicit NIL for their "opt-arg" argument (to indicate a
1666 freshly-consed result bit-array);
1667 ** ELT now signals an error on an invalid sequence index in safe
1669 ** the type system is now cleverer about negations of numeric
1670 types, and consequently understands the BIGNUM and RATIO types
1672 ** the type system is now cleverer about the interaction between
1673 INTEGER and RATIO types: while bugs still remain, many more
1674 cases are accurately computed;
1675 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1676 if it is in the last clause;
1677 ** CONSTANTLY now correctly returns a side-effect-free function in
1679 ** DECLARE is no longer treated as a special-operator; in
1680 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1681 * incremented fasl file version number due to the change in the
1682 DEFSTRUCT-SLOT-DESCRIPTION structure.
1684 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1685 * minor incompatible change: code processed by the "interpreter" or
1686 EVAL now has a compilation optimization policy of (DEBUG 2)
1687 (changed from (DEBUG 1)) to improve debuggability of interactive
1688 development, and to allow the use of the debug RETURN command in
1690 * an experimental implementation of the RETURN command for the
1691 debugger has been included. (thanks to Frederik Kuivinen)
1692 * fixed bug 62: constraints were not propagated into a loop.
1693 * fixed bug in embedded calls of SORT (reported and investigated by
1695 * fixed some bugs revealed by Paul Dietz' test suite:
1696 ** printing and reading of arrays with some dimensions having
1697 length 0 (thanks to Gerd Moellmann);
1698 ** BOA constructor with &AUX argument without a default value does
1699 not cause a type error;
1700 ** CONSTANTP now returns true for all self-evaluating objects.
1702 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1703 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1704 accessors that are related by inheritance, as specified in the
1705 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1706 Valtteri Vuorikoski)
1707 * The compiler is now able to inline functions that were defined in
1708 a complex lexical environment (e.g. inside a MACROLET).
1709 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1711 * Support for the upcoming FreeBSD-5.0 release has been included.
1712 (thanks to Dag-Erling Smorgrav)
1713 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1714 effect when it is not in a toplevel context.
1715 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1716 stand a better chance of being correct. (thanks to Gerd
1718 * fixed bug in COERCE, which now signals an error on coercing a
1719 rational to a bounded real type which excludes the expected
1721 * The compiler is now able to derive types more accurately from the
1722 COERCE and COMPILE functions.
1723 * fixed bug 223: functional binding is considered to be constant
1724 only for symbols in the CL package.
1725 * fixed bug 231: SETQ did not check the type of a variable being set
1726 (reported by Robert E. Brown)
1727 * A new optimization for MAKE-INSTANCE has been included, fixing
1728 various bugs (including relating to :ALLOCATION :CLASS slots and
1729 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1731 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1732 ** As required by ANSI, LOOP now disallows anonymous collection
1733 clauses such as COLLECT I in conjunction with aggregate boolean
1734 clauses such as THEREIS (= I 1);
1735 ** LOOP now signals an error when any variable is reused in the
1736 same loop (including the potentially useful construct analogous
1737 to WITH A = 1 WITH A = (1+ A);
1738 ** IT is only a special loop symbol within the first clause of a
1739 conditional loop clause;
1740 ** LOOP with a typed iteration variable over a hashtable now
1741 signals a type error iff it should.
1742 * fixed some other bugs revealed by Paul Dietz' test suite:
1743 ** FILE-STREAM now names the class previously known as FD-STREAM;
1744 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1745 argument) no longer signals an error;
1746 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1747 of the slot symbol, rather than using the current package
1748 ((:CONC-NAME "") continues to intern the slot's name in the
1750 * incremented fasl file version number, because of the incompatible
1751 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1752 of the new implementation of DEFINE-COMPILER-MACRO.
1754 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1755 * Support for building SBCL for MIPS platforms running in
1756 little-endian mode has now been checked in, and basic
1757 functionality on said platforms verified.
1758 * minor incompatible change: PCL now records the pathname of a file
1759 in which methods and the like are defined, rather than its
1761 * minor incompatible change: TRUENAME now considers the truename of
1762 a file naming a directory to be the pathname with :DIRECTORY
1763 component indicating that directory.
1764 * minor incompatible change: a NAMED clause in the extended form of
1765 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1766 reason for the previous behaviour is unclear.
1767 * more systematization and improvement of CLOS and MOP conformance
1768 in PCL (thanks to Gerd Moellman and Pierre Mai):
1769 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1771 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1772 primary methods with no specializers;
1773 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1775 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1776 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1777 FORWARD-REFERENCED-CLASSes; error reporting on
1778 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1780 ** SXHASH on CLOS instances now uses a slot internal to the
1781 instance to return different numbers on distinct instances,
1782 while preserving the same return value through invocations of
1784 ** DEFMETHOD signals errors when methods with longer incongruent
1785 lambda lists are added to generic functions;
1786 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1787 CLASS, as specified in AMOP;
1788 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1789 on the order returned by the primary method for classes of
1790 class STANDARD-CLASS;
1791 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1792 * fixed some bugs shown by Paul Dietz' test suite:
1793 ** DOLIST puts its body in TAGBODY;
1794 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1796 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1797 value producing form;
1798 ** if more variables are given to PROGV than values, extra
1799 variables are bound and made to have no value;
1800 ** NSUBSTITUTE on list arguments gets the right answer with
1802 ** ELT signals an error of type TYPE-ERROR when the index argument
1803 is not a valid sequence index;
1804 ** LOOP signals (at macroexpansion time) an error of type
1805 PROGRAM-ERROR when duplicate variable names are found;
1806 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1807 ** FUNCALL of special-operators now cause an error of type
1809 ** PSETQ now works as required in the presence of side-effecting
1810 symbol-macro places;
1811 ** NCONC accepts any object as its last argument;
1812 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1814 ** loop-for-as-package does not require a package to be explicitely
1816 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1818 * fixed bug 166: compiler preserves "there is a way to go"
1819 invariant when deleting code.
1820 * fixed bug 172: macro lambda lists with required arguments after
1821 &REST arguments now cause an error to be signalled. (thanks to
1823 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1824 bug. (thanks to Matthew Danish)
1825 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1827 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1828 arguments are lexically rebound. (thanks to Gerd Moellmann and
1830 * fixed bug 194: error messages are now more informative when there
1831 is no primary method applicable in a call to a generic
1832 function. (thanks to Gerd Moellmann)
1833 * fixed bug in command line argument checking (thanks to Julian
1835 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1836 (thanks to Matthew Danish)
1837 * incremented fasl file version number, because of the
1838 SXHASH-related changes in the layout of CLOS data structures
1840 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1841 * minor incompatible change: The runtime (the Unix executable named
1842 "sbcl") is now much pickier about the .core files it will load.
1843 Essentially it now requires .core files to descend from the same
1844 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1845 as the runtime does. (The intent is to prevent the crashes which
1846 can occur, and which can even be reported as mysterious failures,
1847 when people patch the sources or change the build parameters
1848 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1849 sbcl and .core files.)
1850 * fixed bug: VALUES-LIST is no longer optimized away.
1851 * fixed bug 142: The FFI conversion of C string values to Lisp
1852 string values no longer conses excessively. (thanks to Nathan
1853 Froyd porting Raymond Toy's fix to CMU CL)
1854 * began to systematize and improve MOP conformance in PCL (thanks to
1855 Nathan Froyd, Gerd Moellman and Pierre Mai):
1856 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1858 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1859 ** FINALIZE-INHERITANCE is now called on class finalization;
1860 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1861 argument precedence order.
1862 * fixed bug 202: The compiler no longer fails on functions whose
1863 derived types contradict their declared type.
1864 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1865 so it can be non-toplevel.
1866 * The fasl file version number has changed (because of the new
1867 implementation of DEFMACRO).
1868 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1869 safe code, that any length requirement by their type-specifier
1870 argument is valid. The exceptions to this are described in bug
1872 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1873 functions now check the types of their inputs as required by ANSI.
1874 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1875 introduced symbol is DECLAREd to be SPECIAL.
1876 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1877 * fixed bug: :COUNT argument to sequence functions may be negative
1878 * fixed bug: body of DO-SYMBOLS may contain declarations
1879 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1880 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1881 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1882 symbol macro only once
1883 * fixed printing of call frame when argument list is unavailable
1884 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1885 * compiler no longer signals WARNING on unknown keyword
1888 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1889 * A beta-quality port to the mips architecture running Linux,
1890 based on the old CMUCL backend, has been made. It has been tested
1891 on a big-endian kernel, and works sufficiently well to be able to
1892 rebuild itself; it has not been tested in little-endian mode.
1893 * fixed an inconsistency between gencgc.c and purify.c which made
1894 dumping/loading .core files unreliable
1895 * fixed bug 120a: The compiler now deals correctly with IFs where
1896 the consequent is the same as the alternative, instead of
1897 misderiving the return type. (thanks to Alexey Dejneka)
1898 * fixed bug 113: Logical pathnames are now dumpable (the logical
1899 host is resolved at load-time, throwing an error if it is not
1901 * fixed bug 174: FORMAT's error message is slightly clearer when a
1902 non-printing character is used in a format directive.
1903 * fixed several bugs in compiler checking of type declarations, i.e.
1904 violations of the Python "declarations are assertions" principle
1905 (thanks to Alexey Dejneka)
1906 * fixed several bugs in PCL's error checking (thanks to Gerd
1908 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1910 * fixed bug in compilation of functions as first class values
1911 (thanks to Antonio Martinez-Shotton)
1912 * The compiler's handling TYPE-ERRORs which it can prove will
1913 inevitably happen at runtime has been cleaned up and corrected
1914 in several ways. (thanks to Alexey Dejneka)
1915 * improved argument type checking for various basic arithmetic
1916 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1917 lavished on them in the past that they can be compiled in many
1918 ways in different special cases
1919 * fixed bug 181: compiler checks validity of user supplied type
1921 * cleaned up code flushing in optimization: Function calls which
1922 should signal errors for safety purposes (e.g. which ANSI says
1923 should signal errors when their arguments are of incorrect type)
1924 are no longer optimized away.
1925 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1926 * incremented fasl file version number, because changes in the
1927 implementation of sequence functions like COERCE caused
1928 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1929 in old inline expansions) to become undefined. (Actually these
1930 changes were later undone, so we might very well be binary
1931 compatible with 0.7.7 after all, but leaving the version number
1932 incremented seemed like the simplest and most conservative
1935 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1936 * An alpha-quality port to the parisc architecture running Linux,
1937 based on the old CMUCL backend, has been made. This, even more so
1938 than the other backends, should be considered still a work in
1939 progress; known problems include that the Linux kernel in 64-bit
1940 mode does not propagate the correct sigcontext structure to
1941 userspace, and consequently SBCL on a parisc64 kernel will not
1943 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1944 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1945 suppressed.) Also now that the compiler is looking at declarations
1946 in the environment, it checks optimization declarations as well,
1947 and suppresses inlining when (> DEBUG SPEED).
1948 * More fixes have been made to treatment of floating point exception
1949 treatment and other Unix signals. In particular, floating point
1950 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1951 * The detection and handling of control stack exhaustion (infinite
1952 or very deeply nested recursion) has changed. Stack exhaustion
1953 detection is now done by write-protecting pages at the OS level
1954 and applies at all optimization settings; when found, a
1955 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1956 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1957 be caught using IGNORE-ERRORS.
1958 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1959 correctly when transforming recursive LABELS functions to LETs.
1960 (thanks to Alexey Dejneka)
1961 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1962 that are names of constants or global variables.
1963 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1964 alien routines with docstrings.
1965 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1966 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1968 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1969 (thanks to Eric Marsden)
1970 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1971 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1972 * Bug fix: The SPARC backend can now compile functions involving
1973 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1974 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1975 to a functions expecting a PATHNAME-DESIGNATOR.
1976 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1977 lambda lists. (thanks to Alexey Dejneka)
1978 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1979 functions so that EXPORTing the name of the class doesn't cause
1980 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1981 to Antonio Martinez for reporting this)
1982 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1983 constant-folded without failing an assertion. (thanks to Einar
1984 Floystad Dorum for reporting this)
1985 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1986 types are now dealt with more correctly by the compiler.
1987 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1988 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1989 behaviour analogous to RENAME-FILE. This puts its behaviour more
1990 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1992 * The fasl file version number has changed again. (because of the
1993 bug fix involving the names of PCL MAKE-INSTANCE functions)
1995 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1996 * bug fix: Floating point exceptions are treated much more
1997 consistently on the x86/Linux and PPC/Linux platforms.
1998 * Array initialization with :INITIAL-ELEMENT is now much faster for
1999 cases when the compiler cannot open code the array creation, but
2000 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2001 array accesses have also seen a speed increase.
2002 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2003 specified. (This was at the root of some bad interactions between
2004 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2006 * bug fix: Internal error arguments for undefined functions are now
2007 computed correctly on the PPC/Linux platform.
2008 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2009 Raymond Toy's patch for CMU CL)
2010 * Support for the Solaris 9 operating environment has been included
2011 (thanks to Daniel Merritt)
2012 * A very ugly but hopefully complete draft of the missing FFI chapter
2013 of the manual has been created by reformatting the corresponding
2014 CMU CL manual chapter into (currently very ugly and incoherent)
2015 DocBook and bringing it up to date for SBCL behavior. Thus, the
2016 manual is now essentially complete, at least by my extreme
2017 once-and-only-once standards, whereby it's acceptable to refer to
2018 the doc strings of SB-EXT functions as the primary documentation.
2019 * The fasl file version number has changed again, due to cleanup of
2020 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2021 is no longer a static symbol.)
2023 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2024 * SBCL now builds with OpenMCL (version 0.12) as the
2025 cross-compilation host; also, more progress has been made toward
2026 bootstrapping under CLISP.
2027 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2029 * bug 158 fixed: The compiler can now deal with integer loop
2030 increments different from 1; fixing this turned out also to fix
2032 * bug 169 fixed: no more bogus warnings about using lexical bindings
2033 despite the presence of perfectly good SPECIAL declarations (thanks
2034 to David Lichteblau)
2035 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2036 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2037 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2039 * bug 180 fixed: Method combination specifications no longer ignore
2040 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2041 * bug fix: Structure type predicate functions now check their argument
2042 count as they should.
2043 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2044 correctly. (thanks to Pierre Mai)
2045 * minor incompatible change: The --noprogrammer option is deprecated
2046 in favor of the new --disable-debugger option, which is very similar.
2047 (The major difference is that it takes effect at a slightly different
2048 time at startup, causing handling of errors in --sysinit and
2049 --userinit files will be affected differently.) The
2050 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2051 been added to allow this functionality to be controlled from ordinary
2052 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2053 maintainers, who might want to run non-interactive scripts to
2054 build SBCL cores which will later be used interactively.)
2055 * minor incompatible change: The LOAD function no longer, when given
2056 a wild pathname to load, loads all files matching that pathname.
2057 Instead, an error of type FILE-ERROR is signalled.
2059 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2060 * bug 147 fixed: The compiler preserves its block link/count
2061 invariants more correctly now so that it doesn't crash. (thanks
2063 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2065 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2066 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2067 no-value-for-this-character as NIL (as specified by ANSI).
2068 * HOST-NAMESTRING on physical pathnames now returns a string that is
2069 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2070 (thanks to Christophe Rhodes)
2071 * The Alpha port handles icache flushing more correctly. (thanks to
2073 * More progress has been made toward bootstrapping under CLISP. (thanks
2074 to Christophe Rhodes)
2075 * The fasl file format has changed again, because dynamic loading
2076 on OpenBSD (which has non-ELF object files) motivated some cleanups
2077 in the way that foreign symbols are transformed and passed around.
2078 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2079 is no longer treated as whitespace by the reader, but instead as
2080 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2081 |A
\7fB|, instead of A as it used to.
2083 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2084 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2085 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2086 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2087 this as of 0.7.1.45, but was left out of the previous news section
2088 (thanks to Dan Barlow)
2089 * SBCL now runs on the Solaris operating system on SPARC architectures
2090 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2091 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2092 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2093 * SPARC backend cleanups, allowing builds of cores optimized for V8
2094 and V9 SPARCS, and also emission of code targeted to a particular
2095 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2097 * SBCL is closer to bootstrapping under CLISP, thanks to various
2098 fixes by Christophe Rhodes.
2099 * The fasl file format has changed again, to allow the compiler's
2100 INFO database to support symbol macros.
2101 * The user manual (in doc/) is formatted into HTML more nicely.
2102 (thanks to coreythomas)
2103 * The system is smarter about SUBTYPEP relationships, especially
2104 those involving NOT types (including types such as ATOM which are
2105 represented internally using NOT types). Thus SUBTYPEP is less
2106 likely to return (VALUES NIL NIL) in general, and in particular
2107 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2109 * The fasl file format has changed again, because the internal
2110 representation of types now includes a new slot to support the new
2111 SUBTYPEP-of-NOT-types logic.
2112 * (not a change in the main branch of SBCL, but a related prototype
2113 which can hopefully be merged into the main branch of SBCL in the
2114 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2115 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2116 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2117 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2118 dispatch macro character combinations. (thanks to Alexey Dejneka)
2119 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2120 that superficially look like logical namestrings correctly.
2121 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2123 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2124 default supplied-p) for &optional and &key arguments. (thanks to
2127 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2128 * incompatible change: The compiler is now less aggressive about
2129 tail call optimization, doing it only when (> SPACE DEBUG) or
2130 (> SPEED DEBUG). (This is an incompatible change because there are
2131 programs which relied on the old CMU-CL-style behavior to optimize
2132 away their unbounded recursion which will now die of stack overflow.)
2133 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2134 for non-GENCGC systems has been increased to 20M (since that
2135 seems much closer to the likely performance optimum for modern
2136 systems than the old 4M value was)
2137 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2138 and *DEBUG-PRINT-LEVEL*
2139 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2140 of CMU CL's support for SPARC, and various endianness and other
2141 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2142 * new syntactic sugar for the Unix command line: --load foo.bar is now
2143 an alternate notation for --eval '(load "foo.bar")'.
2145 ** The system now detects stack overflow and handles it gracefully,
2146 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2147 optimization settings. (This is a good thing in general, and
2148 its introduction in this version should be particularly timely
2149 for anyone whose code fails because of suppression of tail
2151 ** The system now hunts for the C variable "environ" in a more
2152 devious way, to avoid segfaults when the C library version differs
2153 between compile time and run time. (thanks to Christophe Rhodes)
2154 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2155 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2156 ** The compiler no longer issues bogus style warnings for undefined
2157 classes in the same source file as the DEFCLASSes which defined
2158 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2160 ** fixes in CONDITION class precedence list for undefined function
2161 errors (thanks to Alexei Dejneka)
2162 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2163 correctly. (thanks to Dan Barlow)
2164 ** portability fixes aiming at bootstrapping under CLISP (thanks
2165 to Dave McDonald and Christophe Rhodes)
2166 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2167 ** fixes in type translation and and type inference (thanks to
2169 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2170 ** various fixes in the new ports (thanks to Dan Barlow)
2171 * several changes related to debugging:
2172 ** suppression of tail recursion, as noted above
2173 ** stack overflow detection, as noted above
2174 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2175 is now the default. (For some time encapsulation has been more
2176 reliable than the breakpoint-based :ENCAPSULATE NIL
2177 implementation, at least on X86 systems; and I just noticed that
2178 encapsulation also seems closer to the spirit of the ANSI
2181 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2183 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2184 up properly again. (There was a packaging bug in 0.7.0 which
2185 left their definitions in SB-SYS::LOAD-FOREIGN and
2186 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2187 vital for most things which interface to C-level interfaces,
2188 like extensions working with sockets or databases or
2189 Perl-compatible regexes or whatever, and the need to fix
2190 this bug is the main reason that 0.7.1 was released so
2192 ** DEFGENERIC is now choosier about the methods it redefines, so that
2193 reLOADing a previously-LOADed file containing DEFGENERICs does
2194 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2195 cycle now works as it should. (thanks to Alexey Dejneka)
2196 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2197 fixed by Christophe Rhodes. (He actually submitted this patch
2198 months ago, and I delayed until after 0.7.0.)
2199 ** Bug 111 (internal compiler confusion about runtime checks on
2200 FUNCTION types) was fixed by Alexey Dejneka.
2201 * Some internal cleanups (getting rid of variables which aren't
2202 needed now that the byte interpreter is gone) caused the fasl
2203 file format number to change again.
2205 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2206 * major incompatible change: The default fasl file extension, i.e. the
2207 default extension for files produced by COMPILE-FILE, has changed
2208 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2210 ** There are many changes in the implementation of the compiler.
2211 SBCL is now essentially a compiler-only implementation of ANSI
2212 Common Lisp. EVAL still "interprets" a few special cases, but
2213 almost all the interesting cases are handled by creating
2214 a LAMBDA expression, calling COMPILE on it, then calling
2215 FUNCALL on the result.
2216 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2217 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2218 Since the code is newer, there might still be some new bugs
2219 (though not as many as before Martin Atzmueller's fixes:-). But
2220 the new code is substantially simpler and clearer, and hopefully
2221 any remaining bugs will be simpler, less fundamental, and more
2222 fixable then the bugs in the old code.
2223 ** The revised compiler is still a little unsteady on its feet.
2225 *** The debugging information it produces (particularly the names
2226 of FUNCTION objects) is sometimes much less useful than what
2227 the old compiler produced.
2228 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2229 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2230 within a MACROLET) has been temporarily weakened.
2231 ** There are new compiler optimizations for various functions:
2232 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2233 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2234 *** the math functions TRUNCATE, FLOOR, and CEILING
2235 *** the function-of-all-trades COERCE
2236 Mostly these should be transparent, but there's one
2237 potentially-annoying problem (bug 117): when the compiler
2238 inline-expands a function and does type analysis on the result,
2239 it can create control paths which have type mismatches, and
2240 when it can't prove that those control paths aren't taken,
2241 it will issue WARNINGs about the type mismatches. This is
2242 a particular problem in practice for the new sequence functions.
2243 It's not clear how this should be fixed, and for now, a
2244 workaround is given in the entry for 117 in the BUGS file.
2245 ** (Because of the interaction between the two previous items --
2246 occasional inlining problems and new inline expansions -- some
2247 of the new sequence function optimizations won't really kick in
2248 completely until debugging information, and then inlining, are
2249 straightened out in some future version.)
2250 * minor incompatible changes:
2251 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2252 DIRECTORY no longer implicitly promotes NIL slots of its
2253 pathname argument to :WILD. In particular, when you ask for the
2254 contents of a directory (which you used to be able to do without
2255 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2256 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2257 ** changes in behavior that ANSI explicitly defines to be
2258 implementation dependent:
2259 *** The new compiler-only implementation still conforms with ANSI,
2260 but acts a little different than before. Besides the obvious
2261 changes in performance tradeoffs (that the cost per form passed
2262 to EVAL has gone up, and the cost per form executed by EVAL
2263 has gone down), the behavior of the system changes a little
2264 because there are no longer any interpreted function objects.
2265 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2266 e.g. doing COMPILE on the output of interactive DEFUN is
2268 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2270 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2271 has changed from (:CL) to NIL.
2272 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2273 e.g. #\Newline and #\Tab, don't have names specified in the
2274 ANSI Common Lisp standard, is now based on their ASCII symbolic
2275 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2276 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2277 are no longer used for output.
2278 ** changes in internal implementation constants:
2279 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2280 4 million. (If your application spends a lot of time GCing and
2281 you have a lot of RAM, you might want to experiment with
2282 increasing it even more.)
2283 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2284 However, almost all old code should still continue to work without
2285 immediate update, as SB-C-CALL is now a (deprecated) nickname
2287 ** Old operator names in the style DEF-FOO are now deprecated in
2288 favor of new corresponding names DEFINE-FOO, for consistency with
2289 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2290 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2291 internal symbols, but a few supported extensions like
2292 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2293 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2294 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2295 etc. as you get deeper into recursive calls to the debugger
2296 command loop, instead of the old "5]", "5]]", "5]]]"
2297 sequence. (I was motivated to do this when squabbles between
2298 ILISP and SBCL left me very deeply nested in the debugger. In the
2299 short term, this change will probably provoke more ILISP/SBCL
2300 squabbles, but hopefully it will be an improvement in the long run.)
2301 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2302 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2303 function, since declaiming return types involving aliens is
2304 (1) annoyingly messy to do by hand and (2) vital to efficient
2305 compilation of code which calls such functions.
2306 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2307 longer reexported by the SB-EXT package. They're solely useful
2308 for alien code, so it seems more logical that you should get
2309 them from the SB-ALIEN package, not in SB-EXT.
2310 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2311 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2312 features. Instead, the code that they used to control is always
2313 built into the system.
2314 * many other bug fixes
2315 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2316 advantage of the new EVAL-WHEN stuff and to clean them up in
2317 general, and they are now more ANSI-compliant in a number of
2318 ways. Martin Atzmueller is responsible for a lot of this.
2319 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2321 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2322 *** correct ERROR type for various file operations
2323 *** some fixes for Lisp streams
2324 *** DEFMETHOD syntax checking
2325 *** changing old weird representation of debug information as
2326 strings (which, among their other deficiencies, don't transform
2327 correctly when you rename packages, and don't change their
2328 print representation when you change things like *PACKAGE*
2329 and *PRINT-LENGTH*) to symbols and lists of symbols
2330 He also made several improvements and fixed several bugs in DESCRIBE.
2331 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2333 *** misbehavior of WRITE-STRING/WRITE-LINE
2334 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2335 and several other LOOP problems as well
2336 *** DIRECTORY when similar filenames are present
2337 *** DEFGENERIC with :METHOD options
2338 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2339 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2340 *** argument ordering in FIND with :TEST option
2341 *** mishandled package designator argument in APROPOS-LIST
2342 *** various problems in the backquote readmacro
2343 *** a bug in APROPOS
2344 *** probably some others that I'm not describing very well here,
2345 since the CVS log documents them by reference to sbcl-devel
2346 messages, and the SourceForge archives aren't working well.:-(
2347 ** Dan Barlow improved the Alpha port (and is making progress on the
2348 PPC port, for those of you who think different).
2349 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2350 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2351 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2352 continues to work on the SPARC port (for those of you in a position
2353 to look down upon our little PC-compatible boxes from a great height).
2354 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2355 and uses that copy, rather than the current dynamic value, when
2356 it's trying to decide whether to truncate output. Thus e.g.
2357 (let ((*print-lines* 50))
2358 (pprint-logical-block (stream nil)
2360 (let ((*print-lines* 8))
2361 (print (aref possiblybigthings i) stream)))))
2362 should now truncate the logical block only at 50 lines, instead of
2363 often truncating it at 8 lines, as it did before.
2364 * The doc/cmucl/ directory, containing old CMU CL documentation
2365 from the time of the fork, is no longer part of the base system.
2366 SourceForge has shut down its anonymous FTP service, and with it
2367 my original plan for distributing the old CMU CL documentation
2368 there. For now, if you need these files you can download an old
2369 SBCL source release and extract them from it.
2370 * The fasl file version number changed again, for dozens of reasons,
2371 some of which are apparent above.
2373 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2374 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2375 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2376 patch, so that DISASSEMBLE output is much nicer.
2377 * The code in the SB-PROFILE package now seems reasonably stable.
2378 I still haven't decided what the final interface should look like
2379 (I'd like PROFILE to interact cleanly with TRACE, since both
2380 facilities use function encapsulation) but if you have a need
2381 for profiling now, you can probably use it successfully with
2382 the current CMU-CL-style interface.
2383 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2384 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2385 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2386 link now returns the dangling link itself, and for similar
2387 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2388 link itself. (In these cases the old code signalled an error and
2389 looped endlessly, respectively.) Thus, DIRECTORY now works even
2390 in the presence of dangling and cyclic symbolic links.
2391 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2392 is now a supported extension again, since the consensus on
2393 sbcl-devel was that it can be useful for ordinary development
2394 work, not just for debugging SBCL itself.
2395 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2396 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2397 that currently-defined functions might be redefined later with
2398 different return types.
2399 * Hash tables can be printed readably, as inspired by CMU CL code
2400 of Eric Marsden and SBCL code of Martin Atzmueller.
2401 * better error handling in CLOS method combination, thanks to
2402 Martin Atzmueller porting Pierre Mai's CMU CL patches
2403 * more overflow fixes for >16Mbyte I/O buffers
2404 * A bug in READ has been fixed, so that now a single Ctrl-D
2405 character suffices to cause end-of-file on character streams.
2406 In particular, now you only need one Ctrl-D at the command
2407 line (not two) to exit SBCL.
2408 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2410 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2411 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2412 does the right thing.
2413 * The implementation of some type tests, especially for CONDITION
2414 types, is now tidier and maybe faster, due to CMU CL code
2415 originally by Douglas Crosher, ported by Martin Atzmueller.
2416 * Some math functions have been fixed, and there are new
2417 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2418 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2419 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2420 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2421 likely to be built into sbcl-0.7.0.)
2422 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2423 CL, ported to SBCL by Martin Atzmueller)
2424 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2425 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2426 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2427 * There's a new slam.sh hack to shorten the edit/compile/debug
2428 cycle for low-level changes to SBCL itself, and a new
2429 :SB-AFTER-XC-CORE target feature to control the generation of
2430 the after-xc.core file needed by slam.sh.
2431 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2432 COMPILE-FILE is no longer supported, so that now every function
2433 gets an entry point, so that block compilation looks a little
2434 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2435 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2436 returns the number of bytes consed since the system started,
2437 rather than the number consed since the first time the function
2438 was called. (The new definition parallels ANSI functions like
2439 CL:GET-INTERNAL-RUN-TIME.)
2440 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2441 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2442 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2443 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2444 * Fasl file version numbers are now independent of the target CPU,
2445 since historically most system changes which required version
2446 number changes have affected all CPUs equally. Similarly,
2447 the byte fasl file version is now equal to the ordinary
2450 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2451 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2452 is no longer recognized. I apologize for this, because it was
2453 listed in SB-EXT as a supported extension, but I found that
2454 its existing behavior was poorly specified, as well as incorrectly
2455 specified, and it looked like too much of a mess to straighten it
2456 out. I have enough on my hands trying to get ANSI stuff to work..
2457 * many patches ported from CMU CL by Martin Atzmueller, with
2458 half a dozen bug fixes in pretty-printing and the debugger, and
2459 half a dozen others elsewhere
2460 * fixed bug 13: Floating point infinities are now supported again.
2461 They might still be a little bit flaky, but thanks to bug reports
2462 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2463 as flaky as they were.
2464 * The --noprogrammer command line option is now supported. (Its
2465 behavior is slightly different in detail from what the old man
2466 page claimed it would do, but it's still appropriate under the
2467 same circumstances that the man page talks about.)
2468 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2469 are now supported, and enabled by default. Thus, the compiler can
2470 handle many floating point and complex operations much less
2471 inefficiently. (Thus e.g. you can implement a complex FFT
2473 * The compiler now detects type mismatches between DECLAIM FTYPE
2474 and DEFUN better, and implements CHECK-TYPE more correctly, and
2475 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2476 thanks to patches from Martin Atzmueller.
2477 * various fixes to make the cross-compiler more portable to
2478 ANSI-conforming-but-different cross-compilation hosts (notably
2479 Lispworks for Windows, following bug reports from Arthur Lemmens)
2480 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2481 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2482 fixed thanks to Pierre Mai's CMU CL patches.
2483 * fixes to keep the system from overflowing internal counters when
2484 it tries to use i/o buffers larger than 16M bytes
2485 * fixed bug 45a: Various internal functions required to support
2486 complex special functions have been merged from CMU CL sources.
2487 (When I was first setting up SBCL, I misunderstood a compile-time
2488 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2489 * improved support for type intersection and union, fixing bug 12
2490 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2491 more obscure bugs as well
2492 * some steps toward byte-compiling non-performance-critical
2493 parts of the system, courtesy of patches from Martin Atzmueller
2494 * Christophe Rhodes has made some debian packages of sbcl at
2495 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2496 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2497 stable, but are nonetheless usable. When he's ready, I'd be happy
2498 to add them to the SourceForge "File Releases" section. (And if
2499 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2500 * new fasl file format version number (because of changes in
2501 internal representation of (OR ..) types to accommodate the new
2502 support for (AND ..) types, among other things)
2504 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2505 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2507 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2509 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2510 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2511 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2512 * DESCRIBE now works on structure objects again.
2513 * Most function call argument type mismatches are now handled as
2514 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2515 know whether the function will be redefined before the call is
2516 executed. (The compiler could flag local calls with full WARNINGs,
2517 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2518 it doesn't keep track of enough information to know whether calls
2519 are local in this sense.)
2520 * Compiler output is now more verbose, with messages truncated
2521 later than before. (There should be some supported way for users
2522 to override the default verbosity, but I haven't decided how to
2523 provide it yet, so this behavior is still controlled by the internal
2524 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2525 src/compiler/ir1util.lisp.)
2526 * Fasl file format version numbers have increased again, because
2527 support for the Gray streams extension changes the layout of the
2528 system's STREAM objects.
2529 * The Gray subclassable streams extension now works, thanks to a
2530 patch from Martin Atzmueller.
2531 * The full LOAD-FOREIGN extension (not just the primitive
2532 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2533 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2534 but like most other programs, it defaults to copying the Unix
2535 environment from the original process instead of starting the
2536 new process in an empty environment.
2537 * Extensions which manipulate the Unix environment now support
2538 an :ENVIRONMENT keyword option which doesn't smash case or
2539 do other bad things. The CMU-CL-style :ENV option is retained
2540 for porting convenience.
2541 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2542 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2544 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2546 * A patch from Martin Atzmueller seems to have solved the SIGINT
2547 problem, and as far as we know, signal-handling now works cleanly.
2548 (If you find any new bugs, please report them!)
2549 * The system no longer defaults Lisp source file names to types
2550 ".l", ".cl", or ".lsp", but only to ".lisp".
2551 * The compiler no longer uses special default file extensions for
2552 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2553 seems to expect a single default extension for all compiled code,
2554 and there's no compelling reason to try to stretch the standard
2555 to allow two different extensions.) Instead, byte-compiled files
2556 default to the same extension as native-compiled files.
2557 * Fasl file format version numbers have increased again, because
2558 a rearrangement of internal implementation packages made some
2559 dumped symbols in old fasl files unreadable in new cores.
2560 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2561 many fewer weird special cases.
2562 * Bug #17 (differing COMPILE-FILE behavior between logical and
2563 physical pathnames) has been fixed, and some related misbehavior too,
2564 thanks to a patch from Martin Atzmueller.
2565 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2566 by Tim Moore, ported to SBCL by Martin Atzmueller.
2567 * Martin Atzmueller fixed several filesystem-related problems,
2568 including bug #36, in part by porting CMU CL patches, which were
2569 written in part by Paul Werkowski.
2570 * More compiler warnings in src/runtime/ are gone, thanks to
2571 more patches from Martin Atzmueller.
2572 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2575 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2577 * DESCRIBE now works on CONDITION objects.
2578 * The debugger now handles errors which arise when trying to print
2579 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2581 * The build system now uses an additional file, customize-target-features.lisp,
2582 to allow local modifications to the target *FEATURES* list. (The point of
2583 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2584 debugging features enabled, without having to worry about propagating it
2585 into everyone's system when I do a "cvs update".) When no
2586 customize-target-features.lisp file exists, the target *FEATURES* list
2587 should be constructed the same way as before.
2588 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2589 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2590 warn about behavior which is undefined under the ANSI spec. Note: This
2591 is specified by ANSI, but it's not very popular with programmers.
2592 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2593 macro in the SBCL sources for an example of a workaround which you
2594 might use to make portable ANSI-standard code which does what you want.
2595 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2596 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2597 when it's used as a non-toplevel form. (This is required in order
2598 to implement the DEFCONSTANT-EQX macro.)
2599 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2600 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2601 funny things, due to bugs in the implementation of EVAL-WHEN
2602 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2603 non-toplevel DEFCONSTANTs are uncommon.)
2604 * The core file version number and fasl file version number have been
2605 incremented, because the old noncompliant DEFCONSTANT behavior involved
2606 calling functions which no longer exist, and because I also took the
2607 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2608 * fixed bug 1 (error handling before read-eval-print loop starts), and
2609 redid debugger restarts and related debugger commands somewhat while
2611 ** The QUIT debugger command is gone, since it did something
2612 rather different than the SB-EXT:QUIT command, and since it never
2613 worked properly outside the main toplevel read/eval/print loop.
2614 Invoking the new TOPLEVEL restart provides the same functionality.
2615 ** The GO debugger command is also gone, since you can just invoke
2616 the CONTINUE restart directly instead.
2617 ** The TOP debugger command is also gone, since it's redundant with the
2618 FRAME 0 command, and since it interfered with abbreviations for the
2620 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2622 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2623 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2624 some porting work by Martin Atzmueller)
2625 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2626 with a boatload of patches which he ported to SBCL. Now that those
2628 ** The system tries to make sure that its low-priority messages
2629 are prefixed by semicolons, to help people who like to use
2630 syntax highlighting in their ILISP buffer. (This patch
2631 was originally due to Raymond Toy.)
2632 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2633 patches originally written by Raymond Toy.
2634 ** The compiler understands coercion between single-value and
2635 multiple-VALUES type expressions better, getting rid of some very
2636 weird behavior, thanks to patches originally by Robert MacLachlan
2637 and Douglas Crosher.
2638 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2639 lambda lists, thanks to a patch originally by Pierre Mai.
2640 ** The system no longer bogusly warns about "abbreviated type
2642 ** The compiler gets less confused by inlining and RETURN-FROM,
2643 thanks to some patches originally by Tim Moore.
2644 ** The system no longer hangs when dumping circular lists to fasl
2645 files, thanks to a patch originally from Douglas Crosher.
2646 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2647 undefined function error.
2648 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2649 in structure slot initforms) for now, documented workaround instead:-|
2650 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2651 * fixed bug 5: added stubs for various Gray stream functions called
2652 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2653 installed, at least appropriate type errors are generated
2654 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2655 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2656 consistently in DEFMETHOD forms.
2657 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2658 it doesn't seem to affect SBCL after all
2659 * The C runtime system now builds with better optimization and many
2660 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2662 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2664 * The system is now under CVS at SourceForge (instead of the
2665 CVS repository on my home machine).
2666 * The new signal handling code has been tweaked to treat register
2667 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2668 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2669 downstream, e.g. in debug-int.lisp, has implicit dependencies
2670 on the unsignedness of integer representation of machine words,
2671 and that caused the system to bomb out with infinite regress
2672 when trying to recover from type errors involving signed values,
2673 e.g. (BUTLAST '(1 2 3) -1).)
2674 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2675 (This was one of the bugs Peter Van Eynde reported back in July.)
2676 * The system now uses code inspired by Colin Walters' O(N)
2677 implementation of MAP (from the cmucl-imp@cons.org mailing
2678 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2679 inline the MAP operation, and there is more than one
2680 sequence argument to the MAP call (so that it can't just
2681 do ETYPECASE once and for all based on the type of the
2682 single sequence argument). (The old non-inline implementation
2683 of the general M-argument sequence-of-length-N case required
2684 O(M*N*N) time when any of the sequence arguments were LISTs.)
2685 * The QUIT :UNIX-CODE keyword argument has been renamed to
2686 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2687 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2689 * Raymond Wiker's patches to port dynamic loading from Linux to
2690 FreeBSD have been added.
2691 * The BUGS file is now more nearly up to date, thanks in large part
2692 to Martin Atzmueller's review of it.
2693 * The debugger now flushes standard output streams before it begins
2694 its output ("debugger invoked" and so forth).
2695 * The core version number and fasl file version number have both
2696 been incremented, because of incompatible changes in the layout
2698 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2699 used to not be called for a saved Lisp image.)
2700 * Martin Atzmueller's version of a patch to fix a compiler crash,
2701 as posted on sbcl-devel 13 September 2000, has been installed.
2702 * Instead of installing Martin Atzmueller's patch for the
2703 compiler transform for SUBSEQ, I deleted the compiler transform,
2704 and transforms for some similar consing operations.
2705 * A bug in signal handling which kept TRACE from working on OpenBSD
2707 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2708 away by constant folding
2709 * The system now defines its address space constants in one place
2710 (in the Lisp sources), and propagates them automatically elsewhere
2711 (through GENESIS and the sbcl.h file). Therefore, patching the
2712 address map is less unnecessarily tedious and error-prone. The
2713 Lisp names of address space constants have also been systematized.
2714 * CVS tags like dollar-Header-dollar have been removed from
2715 the sources, because they have never saved me trouble and
2716 they've been source of trouble working with patches and other
2717 diff-related operations.
2718 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2719 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2721 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2723 * The system has been ported to OpenBSD.
2724 * The system now compiles with a simple "sh make.sh" on the systems
2725 that it's supported on. I.e., now you no longer need to tweak
2726 text in the target-features.lisp-expr and symlinks in src/runtime/
2727 by hand, the make.sh takes care of it for you.
2728 * The system is no longer so grossly inefficient when compiling code
2729 involving vectors implemented as general (not simple) vectors (VECTOR T),
2730 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2731 which dares to use the various sequence functions on non-simple
2732 vectors, takes less of a performance hit.
2733 * There is now a primitive type predicate VECTOR-T-P
2734 to test for the (VECTOR T) type, so that e.g.
2735 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2736 can now be compiled with some semblance of efficiency. (The old code
2737 turned the type declaration into a full call to %TYPEP at runtime!)
2738 * AREF on (VECTOR T) is still not fast, since it's still compiled
2739 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2740 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2741 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2742 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2743 without making full calls to SUBTYPEP at runtime.
2744 (Some analogous efficiency issues for non-simple vectors specialized to
2745 element types other than T, or for non-simple multidimensional arrays,
2746 have not been addressed. They could almost certainly be handled the
2747 same way if anyone is motivated to do so.)
2748 * The changes in array handling break binary compatibility, so
2749 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2750 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2751 instead of (VALUES T T).
2752 * By following the instructions that Dan Barlow posted to sbcl-devel
2753 on 2 July 2000, I was able to enable primitive dynamic object
2754 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2755 functionality is not implemented (since it calls ld to resolve
2756 library references automatically, requiring RUN-PROGRAM for its
2757 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2758 try to resolve library references) is now supported.
2759 * The system now flushes the standard output streams when it terminates,
2760 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2761 them at several other probably-convenient times, e.g. in each pass of
2762 the toplevel read-eval-print loop, and after evaluating a form given
2763 as an "--eval" command-line option. (These changes were motivated by a
2764 discussion of stream flushing issues on cmucl-imp in August 2000.)
2765 * The source transform for TYPEP of array types no longer assumes
2766 that an array whose element type is a not-yet-defined type
2767 is implemented as an array of T, but instead punts, so that the
2768 type will be interpreted at runtime.
2769 * There is now some support for cross-compiling in make.sh: each of
2770 the phases of make.sh has its own script. (This should be transparent
2771 to people doing ordinary, non-cross-compile builds.)
2772 * Since my laptop doesn't have hundreds of megabytes of memory like
2773 my desktop machine, I became more motivated to do some items on
2774 my to-do list in order to reduce the size of the system a little:
2775 ** Arrange for various needed-only-at-cold-init things to be
2776 uninterned after cold init. To support this, those things have
2777 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2778 symbols with such names are now uninterned after cold init).
2779 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2780 fasl files for cold load.
2781 ** Remove the old compiler structure pooling code (which used to
2782 be conditional on the target feature :SB-ALLOC) completely.
2783 ** Redo the representation of some data in cold init to be more compact.
2784 (I also looked into supporting byte compiled code at bootstrap time,
2785 which would probably reduce the size of the system a lot, but that
2786 looked too complicated, so I punted for now.)
2787 * The maximum signal nesting depth in the src/runtime/ support code has
2788 been reduced from 4096 to 256. (I don't know any reason for the very
2789 large old value. If the new smaller value turns out to break something,
2790 I'll probably just bump it back up.)
2791 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2793 * Many, many bugs reported by Peter Van Eynde have been added to
2794 the BUGS list; some have even been fixed.
2795 * While enabling dynamic object file loading, I tried to make the
2796 code easier to understand, renaming various functions and variables
2797 with less ambiguous names, and changing some function calling
2798 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2799 * While trying to figure out how to do the OpenBSD port, I tried to
2800 clean up some of the code in src/runtime/. In particular, I dropped
2801 support for non-POSIX signal handling, added various comments,
2802 tweaked the code to reduce the number of compilation warnings, and
2803 renamed some files to increase consistency.
2804 * To support the new automatic configuration functionality in make.sh,
2805 the source file target-features.lisp-expr has been replaced with the
2806 source file base-target-features.lisp-expr and the machine-generated
2807 file local-target-features.lisp-expr.
2808 * fixed a stupid quoting error in make.sh so that using CMU CL
2809 "lisp -batch" as cross-compilation host works again
2811 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2813 * DESCRIBE no longer tries to call itself recursively to describe
2814 bound/fbound values, so that it no longer fails on symbols which are
2815 bound to themselves (like keywords, T, and NIL).
2816 * DESCRIBE now works on generic functions.
2817 * The printer now prints less-screwed-up representations of closures
2818 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2819 * A private symbol is used instead of the :EMPTY keyword previously
2820 used to mark empty slots in hash tables. Thus
2821 (DEFVAR *HT* (MAKE-HASH-TABLE))
2822 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2823 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2824 now does what ANSI says that it should. (You can still get
2825 similar noncompliant behavior if bang on the hash table
2826 implementation with all the symbols you get back from
2827 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2828 This breaks binary compatibility, since tests for equality to
2829 :EMPTY are wired into things like the macroexpansion of
2830 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2832 * There's now a minimal placeholder implementation for CL:STEP,
2833 as required by ANSI.
2834 * An obscure bug in the interaction of the normal compiler, the byte
2835 compiler, inlining, and structure predicates has been patched
2836 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2837 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2839 * Missing ordinary arguments in a macro call are now detected even
2840 when the macro lambda list contains &KEY or &REST.
2841 * The debugger no longer complains about encountering the top of the
2842 stack when you type "FRAME 0" to explicitly instruct it to go to
2843 the top of the stack. And it now prints the frame you request even
2844 if it's the current frame (instead of saying "You are here.").
2845 * As specified by ANSI, the system now always prints keywords
2846 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2848 * The default initial SIZE of HASH-TABLEs is now smaller.
2849 * Type information from CLOS class dispatch is now propagated
2850 into DEFMETHOD bodies, so that e.g.
2851 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2853 is now basically equivalent to
2854 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2855 (DECLARE (TYPE SINGLE-FLOAT X))
2857 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2858 operation, without having to do run-time type dispatch.
2859 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2860 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2861 * Since I don't seem to be making big file reorganizations very often
2862 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2863 is overflowing my ability to conveniently back them up), I've finally
2864 checked the system into CVS. (The CVS repository is on my home system,
2865 not at SourceForge -- putting it on SourceForge might come later.)
2866 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2867 high-level GC-NOTIFY-FOO functions send their output. (There's
2868 still very little control of where low-level verbose GC functions
2869 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2870 less than it used to -- the GC-NOTIFY-FOO functions are now under
2871 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2872 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2873 in only one place in the source code, and propagates it automatically
2874 everywhere that it's needed. Thus e.g. when I bump the version from
2875 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2876 * The C source files now include boilerplate legalese and documentation
2877 at the head of each file (just as the Lisp source files already did).
2878 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2879 to his page will be replaced with a link to his new CLiki service.
2881 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2883 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2884 * The build process now looks for GNU make under the default name "gmake",
2885 instead of "make" as it used to. If GNU make is not available as "gmake"
2886 on your system, you can change this default behavior by setting the
2887 GNUMAKE environment variable.
2888 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2889 can build without error under CMU CL.
2891 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2893 * There is now a partial SBCL user manual (with some new text and some
2894 text cribbed from the CMU CL manual).
2895 * The beginnings of a profiler have been added (starting with the
2896 CMU CL profiler and simplifying and cleaning up). Eventually the
2897 main interface should be through the TRACE macro, but for now,
2898 it's still accessed through vaguely CMU-CL-style functions and macros
2899 exported from the package SB-PROFILE.
2900 * Some problems left over from porting CMU CL to the new
2901 cross-compilation bootstrap process have been cleaned up:
2902 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
2903 instead of SB!XC:DEFMACRO, compounded by an oversight on my
2904 part when getting rid of the compiler *BACKEND* stuff.)
2905 ** The value of *NULL-TYPE* was screwed up, because it was
2906 being initialized before the type system knew the final
2907 definition of the 'NULL type. This screwed up several key
2908 optimizations in the compiler, causing inefficiency in all sorts
2909 of places. (I found it because I wanted to understand why
2910 GET-INTERNAL-RUN-TIME was consing.)
2911 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
2912 PROCLAIM FTYPE information. Unfortunately this broke binary
2913 compatibility again, since now the forms output by DEFGENERIC
2914 to refer to functions which didn't exist in 0.6.3.
2915 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
2916 can use the new (as of 0.6.3) transform for SOME into MAP into
2918 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
2919 values to (INTEGER 0 1000000), so that the time code will no longer
2920 occasionally get blown up by Linux returning 1000000 microseconds
2921 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
2922 its output conform to the ANSI spec. (Alas, this makes its output
2923 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
2924 * A full call to MAP NIL with a single sequence argument no longer conses.
2925 * fixes to problems pointed out by Martin Atzmueller:
2926 * The manual page no longer talks about multiprocessing as though
2927 it were currently supported.
2928 * The ILISP support patches have been removed from the distribution,
2929 because as of version 5.10.1, ILISP now supports SBCL without us
2930 having to maintain patches.
2931 * added a modified version of Raymond Toy's recent CMU CL patch for
2932 EQUALP comparison of HASH-TABLE
2934 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
2936 * The system still can't cross-compile itself with
2937 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
2938 entails), but at least it can compile more of itself that way
2939 than it used to be able to, and various buglets which were uncovered
2940 by trying to cross-compile itself that way have now been fixed.
2941 * This release breaks binary compatibility again. This time
2942 at least I've incremented the FASL file format version to 2, so that the
2943 problem can be detected reliably instead of just causing weird errors.
2944 * various new style warnings:
2945 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
2946 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
2947 ** using the lexical binding of a variable named in the *FOO* style
2948 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
2949 as specified by ANSI.
2950 * *RANDOM-STATE* is no longer automatically initialized from
2951 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
2952 default behavior of the system is to repeat its behavior every time
2953 it's run. If you'd like to change this behavior, you can always
2954 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
2955 old convention there was no comparably easy way to get the system to
2956 repeat its behavior every time it was run.
2957 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
2958 FTYPE declarations has been removed, in favor of their ANSI
2959 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
2960 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
2961 the types of their sequence arguments can be determined at compile time.
2962 This is done through a new open code expansion for MAP which eliminates
2963 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
2964 argument types can be determined at compile time.
2965 * The optimizer now transforms COERCE into an identity operation when it
2966 can prove that the coerced object is already of the correct type. (This
2967 can be a win for machine generated code, including the output of other
2968 optimization transforms, such as the MAP transform above.)
2969 * Credit information has been moved from source file headers into CREDITS.
2970 * Source file headers have been made more standard.
2971 * The CASE macro now compiles without complaining even when it has
2974 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
2976 * (Note that the way that the PCL macroexpansions were rewritten
2977 to accommodate the change in DEFGENERIC below breaks binary
2978 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
2979 not run under sbcl-0.6.2. Once we get out of alpha releases,
2980 i.e. hit release 1.0.0, we'll probably try to maintain binary
2981 compatibility between maintenance releases, e.g. between sbcl-1.4.3
2982 and sbcl-1.4.4. Until then, however, it might be fairly common
2983 for maintenance releases to break binary compatibility.)
2984 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
2986 * The reporting of unhandled signals has been changed to print some
2987 explanatory text as well as the report form. (Previously only
2988 the report form was printed.)
2989 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
2990 it defines, so that the compiler no longer issues undefined function
2991 warnings for compiled-but-not-yet-loaded generic functions.
2992 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
2993 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
2994 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
2995 section of the ANSI spec are supported.
2996 * The "" nickname for the "KEYWORD" package has been removed.
2997 The reader still handles symbol tokens which begin with a package marker
2998 as keywords, but it doesn't expose its mechanism for doing so in the
2999 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3000 * The system now issues STYLE-WARNINGs for contradictory TYPE
3001 proclamations. (Warnings for contradictory FTYPE proclamations would
3002 be nice too, but those can't be done usefully unless the type system
3003 is made smarter about FUNCTION types.)
3004 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3005 systematized, so that "*target-*.lisp is supposed to exist only on the
3006 target and imply that there's a related file which exists on the
3007 host, and *host-*.lisp is supposed to exist only on the host and imply
3008 that there's a related file which exists on the target. This involves a
3009 lot of renaming. Hopefully the acute confusion caused by the renaming
3010 will be justified by the reduction in chronic confusion..
3011 ** runtime-type.lisp -> early-target-type.lisp
3012 ** target-type.lisp -> late-target-type.lisp
3013 ** early-host-format.lisp -> early-format.lisp
3014 ** late-host-format.lisp -> late-format.lisp
3015 ** host-error.lisp -> misc-error.lisp
3016 ** early-error.lisp -> early-target-error.lisp
3017 ** late-error.lisp -> late-target-error.lisp
3018 ** host-defboot.lisp -> early-defboot.lisp
3019 ** code/misc.lisp -> code/target-misc.lisp
3020 ** code/host-misc.lisp -> code/misc.lisp
3021 ** code/numbers.lisp -> code/target-numbers.lisp
3022 ** code/early-numbers.lisp -> numbers.lisp
3023 ** early-host-type.lisp -> early-type.lisp
3024 ** late-host-type.lisp -> late-type.lisp
3025 ** host-typep.lisp -> typep.lisp
3026 ** load.lisp -> target-load.lisp
3027 ** host-load.lisp -> load.lisp
3028 ** host-disassem.lisp -> disassem.lisp
3029 ** host-insts.lisp -> insts.lisp
3030 ** byte-comp.lisp -> target-byte-comp.lisp
3031 ** host-byte-comp.lisp -> byte-comp.lisp
3032 ** host-signal.lisp -> signal.lisp
3033 ** host-defstruct.lisp -> defstruct.lisp
3034 ** late-target-type.lisp -> deftypes-for-target.lisp
3035 Furthermore, several other previously target-only files foo.lisp (e.g.
3036 hash-table.lisp and random.lisp) have been split into a target-and-host
3037 foo.lisp file and a target-only target-foo.lisp file, with their key type
3038 definitions in the target-and-host part, so that the cross-compiler will
3039 know more about target types.
3040 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3041 gone away. In their place are various *BACKEND-FOO* variables
3042 corresponding to the slots of the old structure.
3043 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3044 into the target SBCL has been fixed.
3045 * The chill.lisp system for loading cold code into a running SBCL
3047 * Support for the CMU CL "scavenger hook" extension has been removed.
3048 (It was undocumented and unused in the CMU CL sources that SBCL was
3049 derived from, and stale in sbcl-0.6.1.)
3050 * Various errors in the cross-compiler type system were detected
3051 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3052 (enabling various consistency checks). Many of them were fixed,
3053 but some hard problems remain, so the compiler is back to
3054 running without *TYPE-SYSTEM-INITIALIZED* for now.
3055 * As part of the cross-compiler type system cleanup, I implemented
3056 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3057 * I have started adding UNCROSS calls throughout the type system
3058 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3059 on cross-compiler input files will be able to go away, and various
3061 * CONSTANTP now returns true for quoted forms (as explicitly required
3064 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3066 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3067 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3068 (MAKE-LIST -1). In the longer term, it ought to become true
3069 that declarations are assertions even at SAFETY 1. For now, it's not
3070 quite true even at SAFETY 3, but it's at least more nearly true..
3071 (Note that this change seems to increases the size of the system by
3072 O(5%) and to decrease the speed of the compiler by 20% or more.)
3073 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3074 for the problem of printing dozens of lines of distracting information
3075 about low-level system machinery as part of the top stack frame
3076 on entry to the debugger when an undefined function was called.
3077 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3078 is not reset to COMMON-LISP-USER.
3079 * Compilation of stuff related to dyncount.lisp has been made conditional
3080 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3081 smaller. The various dyncount-related symbols have been moved into
3082 a new "SB-DYNCOUNT" package.
3083 * tty-inspect.lisp has been renamed to inspect.lisp.
3084 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3085 feature has gone away. (When we eventually port to other flavors of
3086 libc and/or Unix, we'll try to make the differences between flavors
3087 invisible at the user level.)
3088 * Various other *FEATURES* tags, and/or their associated conditionals,
3089 have been removed if obsolescent, or given better documentation, or
3090 sometimes given more-mnemonic names.
3092 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3094 * tidied up "make.sh" script
3095 * tidied up system directory structure
3096 * better "clean.sh" behavior
3097 * added doc/FOR-CMUCL-DEVELOPERS
3098 * many many small tweaks to output format, e.g. removing possibly-confusing
3099 trailing #\. character in DESCRIBE-INSTANCE
3100 * (EQUALP #\A 'A) no longer signals an error.
3101 * new hashing code, including EQUALP hashing
3102 * tidied up Lisp initialization and toplevel
3103 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3104 * command line argument processing
3105 * added POSIX-GETENV function to deal with Unix-ish environment variables
3106 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3107 terminating SBCL on EOF
3108 * non-verbose GC by default
3109 * There is no more "sbcl" shell script; the sbcl file is now the C
3110 runtime executable (just like CMU CL).
3111 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3113 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3114 symbol and package information as Lisp native symbol and package objects
3115 instead of strings naming symbols and strings naming packages. This way,
3116 whenever packages are renamed (as in warm init), debug information is
3117 transformed along with everything else.
3118 * tweaked the optimization policy declarations which control the building
3119 of SBCL itself. Now, among other things, the system no longer saves
3120 source location debugging information. (This helps two problems at once
3121 by reducing SBCL size and by keeping SBCL from trying to look for its
3122 sources -- which may not exist -- when reporting errors.)
3123 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3124 debugging and testing purposes
3125 * cleaned up printing, making the printer call PRINT-OBJECT for
3126 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3127 methods, giving nearly-ANSI behavior
3128 * converted almost all special variables to use *FOO* naming convention
3129 * deleted PARSE-TIME functionality, since it can be done portably
3130 * moved some files out of cold init into warm init
3131 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3133 * regularized formatting of source files
3134 * added an install.sh script
3135 * fixed ridiculous memory usage of cross-compiler by making
3136 compiler/alloc.lisp not try to do pooling unless it can hook
3137 itself into the GC of the cross-compilation host. Now the system
3138 builds nicely on my old laptop.
3139 * added :SB-ALLOC in target-features.lisp-expr
3140 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3141 was not implemented)
3142 * re-did condition handling and note reporting in the compiler. Notes
3143 are no longer handled by signalling conditions. Style warnings
3144 and warnings are handled more correctly and reported in such a way
3145 that it's easy to find one or the other in your output (so that you
3146 can e.g. figure out which of many problems caused COMPILE-FILE to
3148 * changed the severity of several compiler warnings from full WARNING
3149 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3150 compiler note reporting so that it doesn't use the condition system
3151 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3152 COMPILE-FILE command)
3153 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3154 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3155 no longer supported, since their implementation was deeply intertwingled
3156 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3157 * removed bogus "support" for compiler macros named (SETF FOO), and
3158 removed the compiler macro for SETF INFO (but only after making a fool
3159 of myself on the cmucl-imp mailing list by posting a bogus patch for
3160 DEFINE-COMPILER-MACRO..)
3161 * Compiled files containing forms which have side effects on the Lisp
3162 reader (such as DEFPACKAGE forms) are now handled more correctly.
3163 (Compiler queuing of top level lambdas has been suppressed by setting
3164 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3165 * deleted various currently-unused source files, e.g. gengc.lisp. They
3166 may be added back at some point e.g. when porting to other architectures,
3167 but until they are it's distracting to distribute them and to try to
3169 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3170 there were so many of them they're just distractions, and UNCROSS is
3171 known to be able to handle the current sources
3172 * moved PROFILE functionality into TRACE, so that it will be clear
3173 how the wrapping and unwrapping of functions when you profile them
3174 interacts with the wrapping and unwrapping of functions when you
3175 trace them. (Actually, the functionality isn't there yet, but at least
3176 the interface specification is there. Hopefully, the functionality will
3177 arrive with some future maintenance release.)
3178 * removed host-oops.lisp
3179 * changed signature of QUIT function to allow UNIX-CODE argument
3180 * fixed READ-SEQUENCE bug
3181 * tweaked verbose GC output so that it looks more like the progress
3182 output that ANSI specifies for functions like LOAD
3183 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3184 * added <http://sbcl.sourceforge.com> to the banner information printed by