1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
3 * New feature: Experimental support for bivalent streams: streams
4 opened with :element-type :default now allow character and binary
6 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
7 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
9 * bug fix: no more highly sporadic "couldn't check whether ~S is
10 readable" when reading a stream and an interrupt hits in the middle
11 of a select system call
12 * compiler better recognizes complex arrays (reported by Eduardo
14 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
15 for variables on SBCL built with linkage-tables. (reported by Luis
18 ** added x86-64 support
19 ** incompatible change: the threading api now works with thread
20 objects instead of thread ids
21 ** bug fix: threads are protected from signals and interruption when
22 starting up or going down
23 ** bug fix: a race where an exiting thread could lose its stack to gc
24 ** fixed numerous gc deadlocks introduced in the pthread merge
26 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
27 * numerous signal handling fixes to increase stability
28 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
29 * minor incompatible change: we now correctly canonize default
30 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
31 as per the MOP, rather than the historical (INITARG INITFUNCTION
32 INITFORM). (reported by Bruno Haible)
33 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
34 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
36 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
37 pathname is a directory pathname.
38 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
39 * dynamic space size on PPC has been increased to 768Mb. (thanks to
41 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
42 :METACLASS argument in addition to a class name. (reported by
43 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
44 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
45 (reported by Bruno Haible; patch from Gerd Moellmann)
46 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
48 * bug fix: more cleanups to the floating point exception handling on
49 x86-64 (thanks to James Knight)
50 * bug fix: the compiler does not try to constant fold calls of
51 COERCE and BIT-* functions when they return freshly constructed
52 objects. (reported by Nikodemus Siivola and Paul Dietz)
53 * optimization: improved the MIPS versions of generic (in the
54 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
55 * optimization: direct conversion of (unsigned-byte 32) to floats on
57 * optimization: structure instances with raw slots now use less
58 memory, and probably show better memory locality. (thanks to
60 * optimization: DYNAMIC-EXTENT declarations for lists and closures
61 are treated as requests for stack allocation on the x86-64,
62 Alpha-32, PPC and SPARC platforms.
63 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
64 the wrong file descriptor; implementation of SOCKET-OPEN-P.
65 (thanks to Tony Martinez)
67 ** gcing a dead thread can no longer lead to lockups
68 ** threads block signals until they are set up properly
69 ** errno is no longer shared by threads
70 ** interrupt-thread restores the eflags register on x86
71 ** fixed some lockups due to gc/thread interaction
72 * fixed some bugs revealed by Paul Dietz' test suite:
73 ** invalid dotted lists no longer raise a read error when
75 ** signal an error if a symbol that names a declaration is used
76 as the name of a type, or vice versa
77 ** allow using the (declare (typespec var*)) abbreviation for
78 (declare (type typespec var*)) with all type specifiers
79 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
80 ** always return NIL from PPRINT-POP when OBJECT is NIL
81 ** don't signal errors when pretty-printing malformed LABELS,
82 FLET or MACROLET forms
83 ** declarations in a DOLIST body are also in scope for the
85 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
87 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
90 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
91 * fixed cross-compiler leakages that prevented building a 32-bit
92 target with a 64-bit host compiler.
93 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
94 opened with :IF-EXISTS :OVERWRITE.
95 * fixed bug 281: error for an invalid qualifier in a short-form method
96 combination method is not signalled until the faulty method is called.
97 * bug fix: iteration variable type inferrer failed to deal with open
98 intervals. (reported by Alan Shields)
99 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
100 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
101 or more runtime options were provided to the sbcl binary.
102 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
103 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
104 * optimization: there's now a fast-path for fixnum arguments in the
105 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
106 * optimization: the code generated on x86-64 is more compact thanks
107 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
108 * bug fix: floating-point exception handling now partly works on
109 x86-64. (Thanks to James Knight)
110 * improvement to the MIPS backend from Thiemo Seufer: C-style
111 64-bit long long arguments and return values to alien functions
113 * fixed some bugs revealed by Paul Dietz' test suite:
114 ** the type-error signalled from WARN has a filled-in DATUM slot.
115 ** the type-error required when a stream is not associated with
116 a file has the stream as its datum.
117 ** type-errors on single-floats on x86-64 no longer have
118 :INVALID-OBJECT as the datum
119 ** the type-errors signalled for invalid function names now have
120 a correct expected type
121 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
122 for structures defined with a :TYPE. Documentation strings for
123 typed structures are no longer immediately discarded
124 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
125 broadcast streams and synonym streams. FILE-LENGTH now also works
126 on broadcast streams.
128 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
129 * incompatible change: the --noprogrammer option, deprecated since
130 version 0.7.5, has been removed. Please use the equivalent
131 --disable-debugger option instead.
132 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
134 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
135 inoperational for a while now) has been completely removed.
136 * null lexical environments are now printed as #<NULL-LEXENV>,
137 significantly reducing the amount of clutter in typical backtraces.
138 * documentation on weak pointers, finalization, and after GC hooks
139 has been added to the manual.
140 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
141 as other specialized array types, is much faster. SUBSEQ and
142 COPY-SEQ on such arrays have also been sped up.
143 * optimization: EQL is now more efficient when at least other argument
144 is known to be of type (OR FIXNUM (NOT NUMBER)).
145 * fixed bug: the runtime is now less vulnerable to changes in the
146 size of the SBCL object on OS X, and virtual memory is reserved for
147 all spaces that need to be at a fixed address.
148 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
149 * fixed bug: finalizers and after GC hooks that cause consing are now
151 * fixed bug: compiler error messages and summaries are now printed to
152 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
153 * fixed inference of the upper bound of an iteration variable.
154 (reported by Rajat Datta).
155 * fixed bug 211e: calling local functions with duplicated constant
156 keyword argument no longer causes a bogus style warning about an
158 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
159 ftype declaration to be disregarded. (reported by Dave Roberts)
160 * fixed bug 373: caused by erronous compilation of references to alien
161 variables in the runtime on ppc/darwin.
162 * fixed bug 376: CONJUGATE type deriver.
163 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
164 (reported by Baughn on #lisp)
165 * a cross-compiler bug on non-x86 platforms has been identified and
166 fixed. (thanks to Bruno Haible)
167 * improvements to the MIPS runtime code for increased stability.
168 (thanks to Thiemo Seufer)
169 * increased the maximimum compact environment size to allow
170 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
171 * partial workaround for bug 135: don't name the function we're
172 calling for hairy cases of EVAL, so as not to accumulate one environment
173 entry per call to EVAL. (reported by Kevin Reid)
174 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
175 * optimization: type testing for non-vector arrays should be faster.
176 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
178 * bug fix: setting 31st element of a bit vector to zero did not work
180 * bug fix: redefining a class definition which failed due to a
181 previous accessor / function clash now works (but see BUGS entry
182 #380 for more problems in this area). (thanks to Zach Beane)
183 * the long form of DEFINE-METHOD-COMBINATION disables method group
184 checking when given a single method group with pattern *. (thanks
186 * on x86 compiler supports stack allocation of results of simple
187 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
188 * support for building and running on Mac OS X 10.4 "Tiger" has been added
189 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
190 * fixed some bugs related to Unicode integration:
191 ** the restarts for recovering from input and output encoding
192 errors only appear when there is in fact such an error to
194 * fixed some bugs revealed by Paul Dietz' test suite:
195 ** MISC.549 and similar: late transformation of unsafe type
196 assertions into derived types caused unexpected code
198 ** SCALE-FLOAT type deriver is less wrong.
199 ** type derivers for EXP, LOG and similar functions compute result
200 types for complex arguments better.
201 ** (MISC.563) CONJUGATE type deriver works for very restricted
203 ** out-of-line type testers for character strings are available.
204 ** EQUAL compiler transform understands specialness of objects
206 ** accessing double-floats stored on the stack now works on x86-64.
207 ** debugger internals could sometimes create invalid lispobjs,
208 resulting in GC crashes.
209 ** MISC.548: type check weakening can convert required type into
211 ** initialization forms for bindings are not in scope of free special
214 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
215 * incompatible change: thread support for non-NPTL systems has
216 been removed - locking is buggy and unreliable. A threaded
217 SBCL build will now warn at startup and refuse to create
218 new threads, unless futex support is detected in the kernel
219 * incompatible change: the top level REPL now has only an ABORT
220 restart associated with it, not TOPLEVEL and ABORT as it used to.
221 TOP and TOPLEVEL are now available as debugger commands for
222 returning to the top level.
223 * incompatible change: forms evaluated in the REPL now use the
224 global optimization policy.
225 * incompatible change: user- and system-initialization files are
226 no longer processed with LOAD, but by READ and EVAL; hence the
227 global optimization policy, startup package, readtable, etc,
229 * The .fasl file format number has been incremented because of
230 various incompatible changes.
231 * internal entry point details and argument counts no longer appear
232 in backtraces unless explicitly requested by setting
233 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
234 * built-in and standard functions no longer have names like "top
235 level local call to FOO".
236 * fixed bug 32: functions defined in non-null lexical environments
237 now have more legible printed representation
238 * fixed bug 33: functions defined in non-null lexical environemnts
239 are now more amenable to inspection by INSPECT.
240 * workaround for bug 354: XEPs no longer appear in backtraces unless
241 explicitly requested.
242 * fixed bug: receiving the signal which results from attempting to
243 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
244 SIGBUS on other *BSDs) on architectures where the C stack is also
245 the Lisp stack (x86 and x86-64) from foreign code no longer leads
246 to debugger confusion or wild instability. (reported by Cheuksan
248 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
249 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
250 (reported by Lutz Euler)
251 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
252 compiled code. (reported by Paul Dietz)
253 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
254 specialized parameter to a method to something that is not TYPEP
255 the specializer is now possible.
256 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
257 face of package deletion.
258 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
259 pathname. (reported by David Tolpin on c.l.l)
260 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
261 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
262 STRING-STREAM and FILE-STREAM, respectively.
263 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
264 than previously; in addition, its test suite is now run on build.
265 (thanks to Luke Gorrie)
266 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
267 * optimization: added a immediate representation for single-floats
269 * fixed some bugs revealed by Paul Dietz' test suite:
270 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
271 ** attempting to create a package with a colliding nickname causes
272 correctable errors to be signalled.
273 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
274 ** several x86-64 backend bugs related to sign-extension of immediate
277 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
278 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
279 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
281 * fixed loading of multiply forward-referenced layouts.
282 (thanks to Cheuksan Wang)
283 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
284 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
285 (see COMPILE-FILE documentation for details of :PRINT option)
286 * fixed bugs 19 and 317: fixed-format floating point printing is
287 more accurate. This also fixes a bug reported by Adam Warner
288 related to the ~@F format directive.
289 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
291 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
292 dispatch table if the from-char is a dispatch macro character.
293 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
294 by the wordsize no longer ignore the last word. (reported by Lutz
296 * fixed bug in type checking of dynamic-extent variables. (reported
298 * optimization: sequence traversal functions use their freedom to
299 coerce function designators to functions.
300 * optimization: code with many calls to CLOS methods specialized on
301 CLOS classes has had redundant type checks removed.
302 * optimization: added declarations to speed up operations that access
303 the internal character database (for example STRING-UPCASE)
304 * optimization: comparison operations between floats and sufficiently small
305 fixnums no longer create extra rationals
306 * fixed some bugs related to Unicode integration:
307 ** portions of multibyte characters at the end of buffers for
308 character-based file input are correctly transferred to the
309 start of the buffer at the next read.
310 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
311 passing it through to OPEN.
312 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
313 argument given to internal calls to OPEN.
314 ** the built-in comment readers (introduced by character sequences
315 ";" and "#|") are more forgiving to encoding errors; they will
316 STYLE-WARN and then attempt to resync the stream at a character
317 boundary. (thanks to Teemu Kalvas)
318 * fixed some bugs in the x86-64 port:
319 ** Negative short int return values from c-calls are sign-extended
321 ** The stack is aligned to 16-bytes for c-calls, as required by
322 the ABI. (thanks to Cheuksan Wang)
323 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
324 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
325 for enums. (thanks to Vincent Arkesteijn)
326 ** Multiple small optimizations and bugfixes for floating point
328 * fixed some bugs revealed by Paul Dietz' test suite:
329 ** Space, Tab, Linefeed, Return and Page have the invalid
330 secondary constituent character trait.
331 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
333 ** WITH-INPUT-FROM-STRING should only update the index place on
335 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
337 ** Bit-array operations (BIT-AND and similar) worked incorrectly
338 with one-dimensional arrays with fill pointers.
339 ** TYPE-OF failed on a complex with an integer realpart and a
341 ** compiler failure during type inference for the code of form
342 (IF (EQL X (THE ...)) ...) (MISC.535).
344 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
345 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
346 platform. The port supports 61-bit fixnums, large memory spaces
347 and reloading shared object files.
348 * enhancement: saving cores with foreign code loaded is now
349 supported on ppc/Darwin in addition to the previously supported
351 * enhancement: the statistical profiler now walks deeper into the
352 call stack for more meaningful call-graphs and accrued time
353 reports (x86/x86-64 only). It also now reports time spent in
355 * enhancement: it is now possible to trace most individual methods
356 of a generic function in addition to tracing the generic function
358 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
360 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
361 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
362 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
363 produces an error. (thanks to Vincent Arkesteijn)
364 * bug fix: NAMESTRING on pathnames with :WILD components in their
365 directories works correctly. (thanks to Artem V. Andreev)
366 * fixed bug 125: compiler preserves identity of closures. (reported
368 * bug fixed: functions with &REST arguments sometimes failed with
369 "Undefined function" when compiled with (DEBUG 3). (reported by
371 * bug fix: overflow during compiling of setting element of a bit
372 vector with constant index and value. (reported by Timmy Douglas)
373 * build fix: fixed the dependence on *LOAD-PATHNAME* and
374 *COMPILE-FILE-PATHNAME* being absolute pathnames.
375 * on x86 compiler partially supports stack allocation of dynamic-extent
377 * GO and RETURN-FROM do not check the extent of their exit points
378 when compiled with SAFETY 0.
379 * fixed some bugs related to Unicode integration:
380 ** encoding and decoding errors are now much more robustly
381 handled; it should now be possible to recover even from invalid
382 input or output to the terminal. (thanks to Teemu Kalvas)
383 ** provided a first cut at implementing STRING-TO-OCTETS and
384 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
385 ** altered the SB-MD5 contributed module slightly, changing the
386 interface just enough for it to be supportable for builds where
387 lisp characters are not eight bits.
388 * fixed some bugs revealed by Paul Dietz' test suite:
389 ** the FORMATTER-generated functions for ~V[ conditionals require
390 the correct number of arguments.
391 ** READ-FROM-STRING returns the mandated second value when applied
392 to displaced strings.
393 ** the #\Rubout and #\Backspace characters are treated as invalid
394 constituent characters by the tokenizer.
396 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
397 * new feature: reloading changed shared object files with
398 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
399 * new feature: references to foreign variables and functions
400 can now be compiled and loaded before the corresponding shared
401 object file is loaded, as long as the foreign definitions are
402 available at runtime.
403 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
404 supported. (thanks to Dan Debertin)
405 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
406 just keywords. (thanks to Vincent Arkesteijn)
407 * fixed bug #331: structure-class instances corresponding to
408 DEFSTRUCT forms are now created eagerly.
409 * fixed bug #345: backtraces from calls to undefined functions work
410 on x86 as well. Related bug #61 is now also partially fixed on x86
411 (backtraces from throws to unknown catch tags.)
412 * bug fix: lambda-list parsing is now stricter vrt. order and number
413 of lambda-list keywords.
414 * bug fix: as specified by AMOP, an error is signalled if a
415 class-option appears multiple times in a DEFCLASS form. (reported
417 * bug fix: RANDOM can be compiled when the compiler derives the type
418 of its numeric argument as a disjoint set of small integers.
419 (reported by Paul Dietz)
420 * bug fix: starting a core saved with shared objects loaded when
421 those objects are not available no longer causes threaded SBCL to
422 hang. (reported by Sean Ross)
423 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
425 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
426 argument, which must always be NIL. (reported by Kalle Niemitalo)
427 * bug fix: printing 1.0d+23 no longer results in an error.
428 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
429 * bug fix: structure slot setters preserve evaluation order. (thanks
431 * fixed some bugs related to Unicode integration:
432 ** RUN-PROGRAM can allow its child to take input from a Lisp
433 stream. (reported by Stefan Scholl)
434 ** ASDF-INSTALL successfully downloads .tar.gz files in all
435 locales. (reported by Ken Causey)
436 * fixed some bugs revealed by Paul Dietz' test suite:
437 ** INCF, DECF and REMF evaluate their place form as specified in
439 ** FORMATTER expands ~{ ~} iteration directives with V or #
440 parameters correctly.
441 ** FORMATTER deals with the ~@[ ~] conditional directive where the
442 consequent uses no arguments correctly.
443 ** the system has a partial understanding of the (COMPLEX RATIO)
445 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
448 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
449 * new feature: a build-time option (controlled by the :SB-UNICODE
450 keyword feature, enabled by default) for building the system with
451 support for the entire 21-bit character space defined by the
453 * new feature: the system now has rudimentary external-format
454 support; the primary user-visible change at this time is that
455 characters with the high bit set (such as lower-case-e-acute) will
456 print correctly to a terminal in a UTF-8 environment.
457 * minor incompatible change: BASE-CHAR no longer names a class;
458 however, CHARACTER continues to do so, as required by ANSI.
459 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
460 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
461 has been moved to the SB-EXT package (temporarily re-exported from
463 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
464 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
466 * fixed bug #351: better error-handlind and reporting for malformed
468 * fixed bug #350: bignum-printing is now more memory-efficient,
469 allowing printing of very large bignums, eg. (expt 2 10000000).
470 (reported by Bruno Haible)
471 * fixed bug #302: better primitive-type selection for intersection
473 * fixed bug #308: non-graphic characters now all have names, as
474 required. (reported by Bruno Haible)
475 * bug fix: redefining a class with different superclasses now correctly
476 removes it from the direct-subclasses of its previous superclasses.
477 (reported by David Morse)
478 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
479 new value now works. (reported by Bruno Haible)
480 * bug fix: correct canonicalization of multiple non-standard slot
481 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
482 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
483 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
484 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
485 now exists, an signals an error.
486 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
487 during macroexpansion for non-iteration variables as well. (reported
488 by Bruno Haible for CMUCL)
489 * bug fix: Cyclic structures and unprintable objects in compiler
490 messages no longer cause errors. (reported by Bruno Haible)
491 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
492 and READ-FROM-STRING all now return a primary value of NIL if
493 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
494 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
495 (reported by Bruno Haible for CMUCL)
496 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
497 do not designate a condition. (reported by Bruno Haible for
499 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
500 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
501 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
502 permitting resolution in favour of any of the conflicting symbols.
503 (reported by Bruno Haible for CMUCL)
504 * bug fix: EQUAL compiler optimizations is less aggressive on
505 strings which can potentially compare true despite having distinct
506 specialized array element types.
507 * bug fix: unit enumerations can be defined without dividing by
508 zero. (thanks to Vincent Arkesteijn)
509 * FORMAT compile-time argument count checking has been enhanced.
510 (report from Bruno Haible for CMUCL)
511 * a partial workaround for the bug 262: the compiler does not try to
512 inline-expand a local function doing RETURN-FROM from a deleted
513 BLOCK. (thanks to Peter Denno for the bug report and to David
514 Wragg for the simple test case)
515 * fixed some bugs revealed by Paul Dietz' test suite:
516 ** RENAME-PACKAGE allows all package designators as new package
518 ** constraint propagation and lambda variable substitution are
519 more cautious in dealing with partially deleted code.
520 ** compiler sometimes lost reoptimization passes.
521 ** CERROR, given a condition as condition designator, passes its
522 remaining arguments to the continue format control without
524 ** Case-altering FORMAT directives work correctly on non-ASCII
526 ** The REAL type specifier handles bounds outside the floating
527 point ranges without signalling FLOATING-POINT-OVERFLOW.
528 ** Functions with IR1-transformations can create intercomponent
529 references to global functions.
530 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
532 ** FORMAT ~R treats a nil value for its first parameter correctly.
534 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
535 * enhancement: saving cores with foreign code loaded is now
536 supported on x86/NetBSD and sparc/Linux in addition to the previously
538 * bug fix: on some platforms repeated installations caused multiple
539 copies of HTML documentation to be installed -- should not happen
540 any more. (reported by Stefan Scholl)
541 * bug fix: parsing self-recursive alien record types multiple times
542 no longer causes infinite recursion. (reported by Thomas F. Burdick,
543 original patch by Helmut Eller for CMUCL)
544 * bug fix: stack-exhaustion detection works now on NetBSD as well.
545 (thanks to Richard Kreuter)
546 * bug fix: defining classes whose accessors are methods on existing
547 generic functions in other (locked) packages no longer signals
548 bogus package lock violations. (reported by François-René Rideau)
549 * bug fix: special variables as DEFMETHOD parameters no longer have
550 associated bogus type declarations. (reported by David Wragg and
552 * bug fix: read-write consistency on streams of element-type
553 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
554 * bug fix: redefiniton of the only method of a generic function with
555 no DEFGENERIC no longer emits a full WARNING. In addition,
556 redefinition of generic functions with no DEFGENERIC to an
557 incompatible lambda list now signals an error. (thanks to Zach
559 * bug fix: DEFGENERIC now works even when there's a function of the
560 same name in an enclosing lexical environment. (thanks to Zach
562 * fixed compiler failure, caused by instrumenting code during
563 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
564 * optimization: added loop analysis and improved register allocation
565 * fixed some bugs revealed by Paul Dietz' test suite:
566 ** POSITION on displaced vectors with non-zero displacement
567 returns the right answer.
568 ** (SIMPLE-STRING) is a valid type specifier for sequence
570 ** *PRINT-LEVEL* handling for slotless structures is pedantically
572 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
573 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
575 ** The justification version of the FORMAT ~< directive treats
576 non-zero minpad parameter correctly.
578 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
579 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
580 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
581 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
582 the supported interface.
583 * new feature: Single-stepping of code compiled with DEBUG 2 or
584 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
585 * new feature: saving cores with foreign code loaded is now
586 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
587 Timothy Moore's work for CMUCL)
588 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
589 parameters to * instead of NIL if no initform is supplied.
590 (reported by Johan Bockgaard)
591 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
592 correctly when FUNCALL appears as the car of the form. Note:
593 despite this FUNCALL forms are not currently subject to
594 compiler-macro expansion. (port of Raymond Toy's fix for the
595 same from CMUCL, reported by Johan Bockgaard)
596 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
597 (thanks for Teemu Kalvas)
598 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
599 case instead of terminating the iteration (thanks for Julian
600 Squires, Sean Champ and Raymond Toy)
601 * bug fix: incorrect expansion of defgeneric that caused a style
602 warning. (thanks for Zach Beane)
603 * bug fix: slot accessor effective method computation works properly
604 for classes with multiple non-standard applicable methods on
605 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
606 * on x86 compiler supports stack allocation of results of LIST and
607 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
608 CMUCL implementation by Gerd Moellmann)
609 * fixed some bugs revealed by Paul Dietz' test suite:
610 ** FORMAT strings with both the ~<~:;~> form of the justification
611 directive and pretty-printing directives cause an error.
613 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
614 * incompatible change: the internal functions
615 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
616 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
617 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
618 instead of the old functions.
619 * new feature: on platforms where "dladdr" is available foreign
620 function names now appear in backtraces. (based on Helmut Eller's
622 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
623 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
625 * documentation: ASDF manual is now distributed with SBCL.
626 * bug fix: SBCL can now load its contributed modules with REQUIRE
627 even if the system-provided entries have been removed from
628 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
630 * bug fix: backtraces involving undefined functions or assembly
631 routines are more informative. (thanks to Brian Downing)
632 * bug fix: mutually referent alien structures now work correctly.
633 (reported by Rick Taube)
634 * bug fix: structures defined by WITH-ALIEN can be referred to
635 within other definitions in the same WITH-ALIEN.
636 * bug fix: division operators (MOD, TRUNCATE and the like) with
637 constant zero divisors and integer dividends no longer generate
639 * bug fix: provide default methods for INPUT-STREAM-P and
640 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
641 * bug fix: improve the detection and resolution of MOP metacycles
642 (where effective method computation depends on the generic function
643 having its effective method computed). (reported by Bruno Haible)
644 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
645 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
646 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
647 * bug fix: pathnames with relative directory components can be
648 represented relative to default pathnames.
649 * optimization: in taking the GCD of bignums, reduce the two bignums
650 to approximately the same size (using Euclid's algorithm) before
651 applying the more sophisticated binary GCD. (thanks to Juho
653 * optimization: COUNT on bitvectors now operates word-at-a-time.
654 * optimization: ASH with a positive, but not necessarily constant,
655 (leftwards) shift, when in a modular context, is compiled to a
657 * fixed some bugs revealed by Paul Dietz' test suite:
658 ** FORMAT variable parameters ("~V<char>") are defaulted properly
659 if the corresponding argument is NIL.
660 ** FORMAT directives accepting a minpad parameter treat negative
662 ** literal commas as character parameters to FORMAT directives are
664 ** literal spaces directly after ~<Newline> directives within a
665 format-logical-block (~:< ~@:>) do not induce :FILL-style
666 conditional newlines.
667 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
668 as specified: it no longer includes conditional newlines).
669 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
671 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
672 :PER-LINE-PREFIX argument does not evaluate to a string.
674 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
675 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
676 the manual for details; for now, package locks can be disabled by
677 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
678 if no major problems are found then it is likely that they will be
679 compiled in unconditionally.
680 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
681 unsupported operators on all platforms. To load a shared library
682 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
683 object file, link it into a shared library outside of SBCL and
684 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
686 * bug workaround: ROOM T and the GC were not working together
687 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
688 reporting facility aren't preserved. That reporting has been
689 disabled, so now until and unless someone figures out how to make it
690 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
691 * minor incompatible change: as threatened around sbcl-0.8.0, the
692 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
693 The symbols are also exported from SB-PCL for backwards
694 compatibility, but more so than before SB-PCL should be treated as
695 an implementation-internal package.
696 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
698 * the SB-SPROF contrib now works on (most) non-x86 architectures.
699 It is known as of this release not to work on the Alpha, however.
700 * fixed bug #167: errors signalled due to illegal syntax in method
701 bodies are now more legible.
702 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
703 designators and can hence be used with TYPEP.
704 * fixed bug #333: CHECK-TYPE now ensures that the type error
705 signalled, if any, has the right object to be accessed by
706 TYPE-ERROR-DATUM. (reported by Tony Martinez)
707 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
708 5.1.2.3 for multiple-value place subforms. (reported by Kalle
710 * fixed bug #334: programmatic addition of slots using specialized
711 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
712 and :ALLOCATION :CLASS slots.
713 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
714 float exponents. (rereported by Peter Seibel)
715 * fixed a bug: #\Space (and other whitespace characters) are no
716 longer considered to be macro characters in standard syntax by
718 * fixed bug: initialization of condition class metaobjects no longer
719 causes an instance of the condition to be created. (reported by
721 * fixed bug: it is now possible to have slots such that
722 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
723 description is neither :INSTANCE nor :CLASS.
724 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
725 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
726 SLOT-VALUE-USING-CLASS), and no longer causes errors with
727 non-standard SLOT-DEFINITION-ALLOCATION values.
728 * fixed bugs: various race conditions handling exiting threads.
729 CL-PPCRE's thread torture test now passes.
730 * fixed arguably-a-bug: GC time proportional to number of threads in
731 system even when most of them are idle
732 * optimization: improved performance of BIT and SBIT on bit-vectors.
733 * .fasl file incompatibility: The fasl file version number has
734 been incremented because of changes associated with package locks.
736 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
737 * minor incompatible change: the system no longer provides
738 optimization hints (conditions which are TYPEP
739 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
740 mode; these hints are still emitted when higher SPEED optimization
742 * new contrib module: a sampling profiler (profiling by statistical
743 sampling, rather than by instrumenting functions) is available as
744 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
745 * the behaviour of the standard function ED is now customizeable by
746 third parties through a hook variable: see ED's documentation
747 string for information on the protocol.
748 * the compiler no longer emits efficiency notes for (FUNCALL X)
749 when the type of X is uncertain under default optimization
751 * fixed bug 276: mutating a binding of a specialized parameter to a
752 method to something that is not TYPEP the specializer is now
754 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
755 exhibit the required behaviour on the x86 platform. (thanks to
756 Peter van Eynde, Eric Marsden and Bruno Haible)
757 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
758 even for difficult arguments. (reported by Peter Graves)
759 * fixed bug 141a: the backquote printer now descends quoted
761 * fixed another bug in backquote printing: no more destructive
762 modification of the form's list structure. (reported by Brian
764 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
765 move between its address being taken and the call to
766 interrupt_thread, fixing a crashing race condition.
767 * the SB-POSIX contrib implementation has been adjusted so that it
768 no longer exhibits ridiculously poor performance when constructing
769 instances corresponding to C structs.
771 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
772 * minor incompatible change: the sb-grovel contrib now treats C
773 structures as alien (in the sense of SB-ALIEN) objects rather than
774 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
775 has implications for memory management of client code
776 (sb-grovel-returned objects must now be manually managed) and for
777 type safety (alien objects now have full types).
778 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
779 used to control emission of compiler diagnostics, rather than the
780 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
781 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
782 quality should be considered deprecated.
783 * install.sh now installs the user manual as well
784 * (not quite a new documentable feature, but worth considering in
785 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
786 beginnings of a semantically meaningful condition hierarchy is
787 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
789 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
790 designator as the defaults argument.
791 * fixed bug: Displaced arrays whose displaced-to array has become
792 too small now cause ARRAY-DIMENSION to signal an error, providing
793 for safer bounds-checking. (reported by Bruno Haible)
794 * fixed bug: DEFCLASS slot definitions with identical :READER and
795 :WRITER names now signal a reasonable error. (reported by Thomas
797 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
799 * fixed bug: Invalid slot specification errors now print correctly.
800 (thanks to Zach Beane)
801 * fixed bug 320: Shared to local slot value transfers in class
802 redefinitions now happen corrently from superclasses as
803 well. (reported by Bruno Haible)
804 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
806 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
807 improper lists correctly. (reported by Bruno Haible)
808 * fixed bug 313: source-transform for <fun-name> was erroneously
809 applied to a call of a value of a variable with name <fun-name>.
810 (reported by Antonio Menezes Leitao)
811 * fixed bug 307: The obsolete instance protocol ensures that
812 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
813 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
814 remove dead unknown-values globs from the middle of the stack.
815 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
816 easier to resume long computations after using *BREAK-ON-SIGNALS*
817 to diagnose and fix failures. (thanks to Nikodemus Siivola)
818 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
819 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
820 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
821 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
822 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
824 * fixed a (fairly theoretical) bug in string printing: if
825 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
826 does not have array-element-type equal to the most general string
828 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
829 function name. (thanks to Nikodemus Siivola)
830 * fixed bug: compile-time format string checker failed on
831 non-closed ~{. (reported by Thomas F Burdick)
832 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
833 don't warn on legal loop code involving more than one aggregate
834 boolean. (thanks to Nikodemus Siivola)
835 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
836 in keyword position to MAKE-PACKAGE was not being reported as
838 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
839 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
841 * fixed bugs 280 and 312: the checking for multiple definitions in a
842 file is less likely to become confused by uses of inline
844 * fixed bug: the #S reader macro performs the keyword coercion
845 specified for slot names. (reported by Kalle Niemitalo)
846 * fixed bug: lambda lists may contain symbols whose names start with
847 & but are not lambda-list-keywords; their occurrence triggers a
849 * fixed bug 321: define-method-combination argument lambda lists do
850 not cause invalid code to be generated when &optional or &aux
851 variables have default values. (reported by Bruno Haible)
852 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
853 have CLOS classes; this appears also to have fixed reported
854 instability in stack exhaustion detection.
855 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
856 MUFFLE-WARNING when no associated restart is present is now
858 * optimization: rearranged the expansion of various defining macros
859 so that each expands into only one top-level form in a
860 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
862 * optimization: used a previously little-used slot in symbols to
863 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
865 * fixed some bugs revealed by Paul Dietz' test suite:
866 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
868 ** Fixed an optimization bug related to inheritance of initforms
869 from local to shared slots.
870 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
871 ** CAST optimizer forgot to flush argument derived type.
872 ** print/read consistency on floats is now orders of magnitude
873 more likely. (thanks also to Bruno Haible for a similar report
875 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
876 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
877 for objects of type REAL. Make it so.
878 ** SXHASH is specified (infelicitously) to respect similarity,
879 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
880 it so. (thanks to Markus Ziegler)
881 ** on the Alpha, the compiler succeeds in compiling functions
882 returning a known number of arguments greater than 63.
883 ** fixed handling of invalid NIL arguments in keyword position
885 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
887 ** provided a readably-printable representation for RANDOM-STATE
889 ** ensured that pathnames, where they have a namestring, always
890 print using #P"..." syntax.
892 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
893 * Support for the forthcoming 2.0 version of the NetBSD kernel
894 running on x86 hardware has been added. (thanks to Perry
895 E. Metzger most immediately, and others for their past work)
896 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
897 current release version; SBCL's OpenBSD support had been broken
898 since about the time of OpenBSD's switch to ELF binary format.)
899 * [placeholder for DX summary]
900 ** user code with &REST lists declared dynamic-extent, under high
901 speed or space and low safety and debug optimization policy.
902 * The manual has been converted to Texinfo format and the debugger
903 chapter from the cmucl manual has been added.
904 * A facility has been added to extract documentation strings from
905 sbcl and store them as Texinfo-formatted snippets for inclusion in
906 the manual (via Texinfo's @include directive)
907 * bug fix: compiler emitted division in optimized DEREF. (thanks for
908 the test case to Dave Roberts)
909 * bug fix: multidimensional simple arrays loaded from FASLs had fill
910 pointers. (reported by Sean Ross)
911 * bug fix: PROFILE output is printed nicely even for large numerical
912 values. (thanks to Zach Beane)
913 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
914 greater than 32 handle EOF correctly.
915 * bug fix: on X86 an immediate argument of the IMUL instruction is
916 correctly printed in disassembly. (thanks to Lutz Euler)
917 * bug fix: class slots in redefined classes preserve their old
918 values. (thanks to Bruno Haible and Nikodemus Siivola)
919 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
920 succeeds. (reported by Marco Baringer)
921 * fixed some bugs revealed by Paul Dietz' test suite:
922 ** READ-SEQUENCE now works on ECHO-STREAMs.
923 ** RATIONALIZE works more according to its specification. (thanks
926 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
927 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
928 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
929 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
930 should matter to you only if you rebind the printer control
931 variables and then find you want different bindings in the
932 debugger than in the ordinary execution of your program.)
933 * The runtime build system has been tweaked to support building
934 (on SPARC/SunOS) using a C compiler which invokes Sun's own
935 assembler and linker. (thanks to Nikodemus Siivola)
936 * Unbound, undefined, undeclared variables now trigger full
937 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
938 more often programmer error than deliberate exploitation of undefined
940 * optimization: the hash algorithm for strings has changed to one
941 that is less vulnerable to spurious collisions. (thanks to Juho
943 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
944 needless bounds checking. (thanks to Juho Snellman)
945 * optimization: implemented multiplication as a modular
946 (UNSIGNED-BYTE 32) operation on the PPC backend.
947 * fixed some bugs revealed by Paul Dietz' test suite:
948 ** ADJUST-ARRAY now copies the datum in a zero rank array if
950 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
951 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
952 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
954 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
956 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
958 ** ECHO-STREAMs no longer attempt to echo the end of file value to
959 their output stream on EOF from read.
960 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
961 have been read to end-of-file.
962 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
964 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
965 description of determination of which consecutive characters
967 ** Printing the "Space" character with escaping on now yields "#\\ ",
968 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
969 ** Reading floating-point numbers with *READ-BASE* set to a number
970 less than 10 works correctly.
971 ** Reading floating-point numbers with *READ-BASE* set to a number
972 more than 10 works correctly.
973 ** Printing with *PRINT-READABLY* targets the standard readtable, not
974 the readtable currently in effect.
976 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
977 * minor incompatible change: parsing of namestrings on a physical
978 (Unix) host has changed; numbers after the final #\. in a
979 namestring are no longer interpreted as a version field. This is
980 intented to be largely invisible to the user, except that the
981 meaning of the namestring "*.*.*" has changed: it now refers to a
982 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
983 should usually be replaced by
984 (make-pathname :name :wild :type :wild :version :wild)
985 with the added benefit that this is more likely to be portable.
986 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
987 option to OPEN now signals an error if the file being opened
988 exists; this may have an impact on existing code.
989 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
990 signals received, on the PowerPC platforms (both Linux and
991 Darwin). (thanks to Pierre Mai for pointing out the location of
993 * several fixes on OS X: The system now builds and runs cleanly on
994 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
995 (thanks to Brian Mastenbrook)
996 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
997 second-resolution: integer multiples of 1/3600 between -24 and 24.
998 (thanks to Vincent Arkesteijn)
999 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1000 type when called with 1 argument; PEEK-CHAR checked type of
1001 PEEK-TYPE only after having read first character from a
1002 stream. (reported by Peter Graves)
1003 * bug fix: the garbage collector now has much better locality
1004 behaviour, and in particular no longer treats all memory as being
1005 exhausted when many small objects point to each other in a deeply
1007 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1008 recognized as being TYPEP their class.
1009 * bug fix: the PUSHNEW documentation string has been corrected.
1010 (thanks to Vincent Arkesteijn)
1011 * bug fix: defaulting of the value for the last of an atypically
1012 large number of multiple values being bound was not being
1013 performed correctly on the Alpha or PPC platforms
1014 * optimization: implemented multiplication as a modular
1015 (UNSIGNED-BYTE 32) operation on the x86 backend.
1016 * optimization: SEARCH on simple-base-strings can now be open-coded.
1017 (see also contrib/compiler-extras.lisp for inspiration for
1018 teaching the compiler about the Boyer-Moore algorithm).
1019 * value, returned by MAX (and MIN) called with several EQUALP, but
1020 not EQL, arguments now does not depend on compiler settings.
1021 * fixed some bugs revealed by Paul Dietz' test suite:
1022 ** in stack analysis liveness information is propagated from
1023 non-local entry points.
1024 ** pathwise CAST removing failed when the CAST node did not start
1026 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1027 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1029 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1030 host is already defined.
1031 ** RENAME-FILE works on streams instead of signalling an internal
1033 ** PEEK-CHAR uses the current readtable when determining whether
1034 or not a character is whitespace.
1035 ** MERGE-PATHNAMES handles the case when the pathname does not
1036 specify a name while the default-pathname specifies a version.
1037 ** Pathnames now stand a better chance of respecting print/read
1039 ** Attempting to use standardized file system operators with a
1040 pathname with invalid :DIRECTORY components signals a
1042 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1044 ** DIRECTORY on logical pathnames is more correct.
1045 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1046 signal a TYPE-ERROR if their argument is not a stream.
1047 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1048 designator argument does not designate a stream.
1049 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1050 examining the synonym.
1051 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1053 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1054 element-type larger than ([UN]SIGNED-BYTE 32).
1056 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1057 * When built with the :SB-FUTEX feature, threaded builds now take
1058 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1059 for faster/more reliable mutex and condition variable support.
1060 * Incompatible change (but one you probably shouldn't have been using
1061 anyway): the interface and code for arbitrating between multiple
1062 threads in the same user session has been redesigned.
1063 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1064 error when called without an explicit environment argument.
1065 (thanks to Nikodemus Siivola)
1066 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1067 confused about their position. (thanks to Adam Warner and Gerd
1069 * bug fix: Pretty printing backquoted forms with unquotations in the
1070 argument list position of various code constructs such as LAMBDA
1071 now works correctly. (reported by Paul Dietz)
1072 * bug fix: Pretty printing unquotations no longer loses all
1073 stream position information.
1074 * optimization: performance of string output streams is now less
1075 poor for multiple small sequence writes.
1076 * optimization: performance of CSUBTYPEP in the presence of complex
1077 expressions involving CONS and NOT many times has been improved.
1078 (reported by Paul Dietz)
1079 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1081 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1082 streams for the REPL, for the debugger, and so on. (thanks to
1084 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1085 * fixed some bugs revealed by Paul Dietz' test suite:
1086 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1087 is ignored if it is too small, rather than propagating through
1088 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1089 ** extremely complex negations of CONS types were not being
1090 sufficiently canonicalized, leading to inconsistencies in
1092 ** VALUES tranformer lost derived type.
1094 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1095 * fixed a bootstrapping bug: the build process no longer assumes
1096 that the various BOOLE-related constants have the same value in
1097 host and target lisps. (noted by Paul Dietz' test suite on an
1098 SBCL binary built from CLISP)
1099 * The system can now be dynamically linked on the MIPS platform,
1100 which enables dynamic loading of foreign code from Lisp. (thanks
1101 to Ralf Baechle for discussions on the MIPS ABI)
1102 * The system now records debugging information for its own source
1103 files in a filesystem-position-independent manner, relative to
1104 the "SYS" logical host.
1105 * fixed a compiler bug: MV-LET convertion did not check references
1106 to the "max args" entry point. (reported by Brian Downing)
1107 * tweaked disassembly notes to be less confident about proclaiming
1108 some instruction as an LRA. (thanks to Brian Downing)
1109 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1110 now each have their own history, command character, and other
1111 characteristics. (thanks to David Lichteblau)
1112 * fixed some bugs revealed by Paul Dietz' test suite:
1113 ** compiler failure in compiling LOGAND expressions including a
1115 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1116 shift greater than 32.
1117 ** FLUSH-DEST did not mark blocks for type check regeneration.
1118 ** HANDLER-CASE failed to accept declarations in handler clauses
1119 in some circumstances.
1121 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1122 * New code in contrib/sb-introspect (still probably not entirely
1123 stable yet) provides some support for smart Lisp development
1124 environments like SLIME.
1125 * The conditions signalled for errors occurring when loading .fasl
1126 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1127 in a way which should help ASDF recover gracefully.
1128 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1129 stuff has been cleaned up. If you code contrib/ stuff, this might
1130 affect you, and you can look at contrib/README, contrib/STANDARDS,
1131 and/or the 0.8.4.27 diff to check.
1132 * In full calls the compiler now does not generate checks for declared
1133 argument types for all arguments.
1134 * various threading fixes
1135 ** and some experimental patches which didn't make it into
1136 the main tree for this release, but which are shipped in
1137 contrib/experimental-thread.patch as a possible fix for some
1138 failures (deadlock, spinning...) in GC-intensive multithreaded
1140 * fixed PPC build problem (source code incompatibility of different
1141 library versions): added offsetof() hackery which attempts to divine
1142 where glibc maintainers put uc_mcontext today
1143 * fixed bug 282: compiler does not trust type assertions while passing
1144 arguments to a full call.
1145 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1146 &OPTIONAL VALUES type parameter.
1147 * fix bug 214: algorithm for noting rejected templates is now more
1148 similar to that of template seletion. (also reported by rydis on
1150 * fixed bug 141b: printing backquoted information readably and prettily
1151 inserts a space where necessary.
1152 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1153 (CHARACTER 10) are now reported as errors, rather than propagated
1154 as unknown types. (reported by piso on #lisp)
1155 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1156 with values NIL and :ERROR. (thanks to Milan Zamazal)
1157 * fixed bug 191c: CLOS now does proper keyword argument checking as
1158 described in CLHS 7.6.5 and 7.6.5.1.
1159 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1160 longer raise an error; further, using a list as a for-as-arithmetic
1161 counter now raises a meaningful error.
1162 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1163 now understood by sequence creation functions such as MAKE-SEQUENCE
1165 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1166 a STRING-INPUT-STREAM.
1167 * compiler enhancement: SIGNUM is now better able to derive the type
1169 * type declarations inside WITH-SLOTS are checked. (reported by
1171 * fixed some bugs revealed by Paul Dietz' test suite:
1172 ** incorrect optimization of TRUNCATE for a positive first
1173 argument and negative second.
1174 ** compiler failure in let-convertion during flushing dead code.
1175 ** compiler failure while deriving type of TRUNCATE on an
1176 interval, containing 0.
1177 ** ASH of a negative bignum by a negative bignum count now returns
1179 ** intersection of CONS types now canonicalizes properly, fixing
1180 inconsistencies in SUBTYPEP.
1182 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1183 * incompatible change: The --disable-debugger command line
1184 option now clobbers the debugger at a more fundamental
1185 level, by redefining #'INVOKE-DEBUGGER instead of by
1186 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1187 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1188 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1189 enter the debugger REPL and then suffer recursive errors
1190 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1191 while under the new system, BREAK in --disable-debugger mode
1192 terminates the system just as an unhandled error would.
1193 * fixed compiler performance when processing loops with a step >1;
1194 * bug fix: DOCUMENTATION now retrieves generic function
1195 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1196 support has been systematized, and now supports the methods
1197 specified by ANSI, along with a default method and a method for
1198 slot documentation. (reported by Nathan Froyd)
1199 * bug fix: effective methods associated with a generic function are
1200 no longer cached over a change of that generic function's method
1201 combination. (reported by Andreas Fuchs)
1202 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1203 names a non-existent directory. (thanks to Andreas Fuchs)
1204 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1205 the PPC platform, lead to stack corruption; this has been fixed.
1206 (reported by Rainer Joswig)
1207 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1208 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1209 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1210 arithmetic in general, is now much more reliable on the PPC
1212 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1213 platform now returns the right answer.
1214 * optimization: restored some effective method precomputation in
1215 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1216 precomputation is now tunable.
1217 * optimization: compiler-internal data structure use has been
1218 reviewed, and changes have been made that should improve the
1219 performance of the compiler by about 20%.
1220 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1221 simple-base-strings and simple-bit-vectors is improved.
1222 * optimization: the optimization of 32-bit logical and arithmetic
1223 functions introduced in version 0.8.3 on the x86 has been
1224 implemented on the mips, ppc and sparc platforms; an
1225 implementation of the same facility, but for 64-bit arithmetic,
1226 has been added for the alpha.
1227 * microoptimization: the compiler is better able to make use of the
1228 x86 LEA instruction for multiplication by constants.
1229 * bug fix: in some situations compiler did not report usage of
1230 generic arithmetic in (SPEED 3) policy.
1231 * bug 145b fix: compiler used wrong type specifier while converting
1232 MEMBER-types to numeric.
1233 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1235 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1236 index argument now works properly on the Alpha platform.
1237 * bug fix: floating point exception treatment on the Alpha platform
1239 * bug fix: FILE-POSITION works much better on string input and
1240 output streams. (thanks to Nikodemus Siivola)
1241 * bug fix: many threading/garbage collection symptoms sorted.
1242 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1243 might be pseudo-atomic.
1244 * internal change: Stopping for GC is now done with signals not
1245 ptrace. GC is now done in whichever thread wanted it, instead of
1247 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1249 * bug fix: result form in DO is not contained in the implicit
1251 * incompatible change: ICR structure is changed; the value part of
1252 CONTINUATION is now called LVAR; corresponding functions are
1253 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1254 * added type deriver for ISQRT (thanks to Robert E. Brown).
1255 * bug fix: better support for loading from the command line when an
1256 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1258 * fixed some bugs revealed by Paul Dietz' test suite:
1259 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1260 ** ROUND and FROUND now give the right answer when given very
1261 small float arguments.
1262 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1264 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1265 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1266 ** LCM should return a non-negative integer.
1267 ** PARSE-INTEGER returned the index of a terminator instead of the
1268 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1269 ** PARSE-INTEGER returned an incorrect index being applied to a
1271 ** LCM with two arguments of 0 returns 0 rather than signalling
1273 ** unsigned addition of a 32-bit constant with the high bit set no
1274 longer causes an internal compiler error on the x86.
1275 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1276 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1277 with negative last argument.
1278 ** byte specifiers with zero size and position no longer cause
1279 an error during type derivation.
1280 ** bignum multiplication on the Alpha platform now returns the
1282 * porting: The system now builds on SuSE AMD64, although it still
1283 generates a 32-bit binary.
1284 * .fasl file incompatibility: The fasl file version number has
1285 been incremented (because of the changes to internal compiler
1286 data structures referred to above).
1288 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1289 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1290 more accurately, on the Darwin kernel running on PowerPC hardware.
1291 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1292 * Compiler code deletion notes now signal a condition of type
1293 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1294 an associated MUFFLE-WARNING restart.
1295 * The compiler now performs limited argument count validation of
1296 constant format strings in FORMAT, and where appropriate in ERROR,
1297 CERROR and WARN. (thanks to Gerd Moellmann)
1298 * New ASDF-INSTALL contrib can be used for automatic download and
1299 installation of third-party Lisp code from CCLAN or other sites
1301 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1302 which forces another thread to execute a function supplied by the
1304 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1305 now accept and act upon their :ELEMENT-TYPE keyword argument.
1306 (reported by Martin Atzmueller, Edi Weitz)
1307 * bug fix: FILE-POSITION now accepts position designators up to
1308 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1309 is the greater. (thanks to Patrik Nordebo)
1310 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1311 Kalle Olavi Niemitalo)
1312 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1313 printable. (reported by Eric Marsden)
1314 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1315 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1316 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1317 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1318 (thanks to Stig Sandoe)
1319 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1320 UNREAD-CHAR now correctly includes the unread character in the
1321 target sequence. (thanks to Gerd Moellmann)
1322 * bug fix in threaded builds: the system can now be suspended and
1323 resumed by shell job control with minimal disruption.
1324 * bug fixes in times and timezones >2038AD
1325 * better handling of "where is GNU make?" problem in build scripts
1326 (thanks to Nikodemus Siivola)
1327 * new optimization: inside a named function any reference to a
1328 function with the same name is considered to be a self-reference;
1329 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1330 optimization quality.
1331 * new optimization on x86: logical functions and + now have
1332 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1333 used when the result is truncated to 32 bits.
1334 * VALUES declaration is partially enabled.
1335 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1336 * fixed some bugs revealed by Paul Dietz' test suite:
1337 ** The system now obeys the constraint imposed by
1338 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1339 types form a lattice under type intersection.
1340 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1341 ** ASSOC now ignores NIL elements in an alist.
1342 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1343 and (1+ MOST-POSITIVE-FIXNUM) answers.
1344 ** The addition of a method with invalid qualifiers to a generic
1345 function does not cause an error to be signalled immediately;
1346 a warning is signalled, and the error is generated only on
1347 calling the generic function.
1348 * changed .fasl file version number, in part to add type codes for
1349 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1350 obscure ANSI requirements
1352 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1353 * fixed bug 148: failure to inline-expand a local function left
1354 garbage, confusing the compiler.
1355 * fixed bugs 3cd: structure slot readers perform type check if the
1356 slot can have an invalid value (i.e. it is either not initialized
1357 or can be written with a less specific slot writer).
1358 * bug fix: the compiler now traps array references to elements off
1359 the end of an array; previously, the bounds checking in some
1360 circumstances could go off-by-one.
1361 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1363 * type declarations for array element types now obey the description
1364 on the CLHS page "Declaration TYPE", as per discussions on
1365 sbcl-help around 2003-05-08. This means that a declaration
1366 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1367 declaration, all references to BAR will be asserted or assumed
1368 (with THE, so dependent on compiler policy) to involve objects of
1369 type FOO. Note that no such declaration is implied in
1370 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1371 * declared types of functions from the "Conditions"
1372 chapter. (reported by Paul Dietz)
1373 * bug fix: CERROR accepts a function as its first argument.
1374 * bug fix: NTH an NTHCDR accept a bignum as index
1375 arguments. (reported by Adam Warner)
1376 * optimization: character compare routines now optimize comparing
1377 against a constant character. (reported by Gilbert Baumann)
1378 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1379 argument now works properly on the MIPS platform.
1380 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1381 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1382 declared to have a functional type, does not cause an error
1384 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1385 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1386 * fixed bug: if last continuation of a deleted block has a
1387 destination, this destination should be deleted too. (reported by
1389 * fixed a bug in the bootstrap process: the host compiler's values
1390 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1391 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1392 test case from Patrik Nordebo)
1393 * improved the ability of the disassembler on the PPC platform to
1394 provide helpful disassembly notes.
1395 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1396 the class in more cases than previously.
1397 * bug fix: FILE-POSITION now understands :START and :END for
1398 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1399 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1400 8) data. (thanks to David Lichteblau)
1401 * bug fix: it is possible to add a method to a generic function
1402 without lambda list.
1403 * bug fix: reader failed to signal END-OF-FILE inside an
1404 object representation. (reported by Nikodemus Siivola)
1405 * fixed some bugs revealed by Paul Dietz' test suite:
1406 ** LAST and [N]BUTLAST should accept a bignum.
1407 ** condition slot accessors are methods.
1408 ** (VECTOR NIL) is a subtype of STRING.
1410 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1411 * minor incompatible change: some nonsensical specialized lambda
1412 lists (used in DEFMETHOD) which were previously ignored now signal
1414 * minor incompatible change: the system is now aware of the types of
1415 variables in the COMMON-LISP package, and will signal errors for
1416 most violations of these type constraints (where previously they
1417 were silently accepted).
1418 * minor incompatible change: COMPILE-FILE now uses the freedom
1419 afforded (ANSI 3.2.2.3) to use derived function types for
1420 functions defined in the same file. This also permits the system
1421 to warn on static type mismatches and function
1422 redefinition. (Currently it does not work with high DEBUG level.)
1423 * minor incompatible change: VALUES declaration is disabled.
1424 * When issuing notes, the compiler now signals a condition of type
1425 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1426 restart for use in user handlers. It is expected that the
1427 COMPILER-NOTE condition will eventually become a condition
1428 supertype to a hierarchy of note types, which will then be
1429 handleable in a similar fashion. However, at the moment, no such
1430 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1431 but it's an implementation detail, not a classification for the
1433 * Changes in type checking closed the following bugs:
1434 ** type checking of unused values (192b, 194d, 203);
1435 ** template selection based on unsafe type assertions (192c, 236);
1436 ** type checking in branches (194bc).
1437 * A short form of VALUES type specifier has ANSI meaning (it has
1438 increased the number of situations when SBCL cannot perform type
1440 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1441 keywords or constants is permissible.
1442 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1443 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1444 classes. (thanks to Antonio Martinez)
1445 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1446 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1447 argument to a pretty-print stream. Instead, it leaves any such
1448 operations to DESCRIBE-OBJECT methods.
1449 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1451 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1452 lambda lists. (thanks to David Lichteblau)
1453 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1454 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1455 (thanks to Gerd Moellmann)
1456 * fixed bug 246: increased compilation speed of long
1457 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1459 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1461 * DEFCONSTANT now throws a condition of type
1462 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1463 constant to a non-EQL value; CONTINUE and ABORT restarts
1464 respectively change and preserve the value.
1465 * fixed bug 63: The code walker, part of the implementation of CLOS,
1466 is now better at handling symbol macros.
1467 * bug fix: There is no longer an internal implementation type named
1468 CL:LENGTH. (reported by Raymond Toy)
1469 * bug fix: In macro-like defining macros/special operators the
1470 implicit block does not enclose the lambda list.
1471 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1472 enclosed in parentheses) are not suitable as type specifiers, and
1473 their use properly signals an error now.
1474 * bug fix: An argument count mismatch for a type specifier in code
1475 being compiled no longer causes an unhandled error at compile
1476 time, but signals a compile-time warning.
1477 * fixed simple vector readable printing
1478 * bug fix: DESCRIBE takes more care over whether the class
1479 precedence list slot of a class is bound before accessing it.
1480 (reported by Markus Krummenacker)
1481 * bug fix: FORMATTER can successfully compile pretty-printer format
1482 strings which use variants of the ~* directive inside.
1483 * bug fix: SEARCH now applies its TEST predicate to the elements of
1484 the arguments in the correct order. (thanks to Wolfhard Buss)
1485 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1486 in type inference. (thanks to Robert E. Brown)
1487 * bug fix: Reading in symbols with an explicit package name of ""
1488 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1489 * fixed some bugs revealed by Paul Dietz' test suite:
1490 ** NIL is now allowed as a structure slot name.
1491 ** Arbitrary numbers, not just REALs, are allowed in certain
1492 circumstances in LOOP for-as-arithmetic clauses.
1493 ** Multiple class redefinitions before slot access no longer
1494 causes a type error.
1495 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1496 association between the name and a class.
1497 ** Generic functions with non-standard method-combination and over
1498 five methods all of which return constants no longer return NIL
1499 after the first few invocations. (thanks to Gerd Moellmann)
1500 ** CALL-NEXT-METHOD with no arguments now passes the original
1501 values of the arguments, even in the presence of assignment.
1502 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1504 ** DISASSEMBLE works with closures and funcallable instances.
1505 ** ADD-METHOD now returns the generic function, not the new method.
1506 ** FIND-METHOD signals an error if the lengths of the specializers
1507 is incompatible with the generic function, even if the ERRORP
1509 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1510 which its argument is a member.
1511 ** DEFCLASS only redefines the class named by its class-name
1512 argument if that name is the proper name of the class;
1513 otherwise, it creates a new class.
1514 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1515 of the UNBOUND-SLOT condition to the name of the slot.
1516 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1517 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1518 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1519 SLOT-MAKUNBOUND in the specified fashion.
1521 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1522 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1523 cross-compilation host. As a consequence, we can now bootstrap our
1524 way up to SBCL starting with a bare gcc toolchain and human-readable
1525 source code (first the source to CLISP, then the source to SBCL).
1526 * A contributed module containing a partial implementation of the
1527 simple-streams interface has been included. (thanks to Rudi
1529 * A contributed module implementing the RFC1321 Message Digest
1530 Algorithm, known as MD5, has been included.
1531 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1532 no longer has any effect, as the code controlled by this feature
1533 has been deleted. (As far as we know, no-one has ever built using
1534 this feature, and its semantics were confused in any case).
1535 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1536 work on conditions (as required by the ANSI specification),
1537 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1538 expected behaviour on conditions. Users should note, however,
1539 that such behaviour is not required by the ANSI specification,
1540 and so use of this behaviour may render their code unportable.
1541 * fixed some bugs revealed by Paul Dietz' test suite:
1542 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1544 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1545 ** SXHASH on condition objects no longer returns NIL.
1546 ** :ALLOCATION :CLASS slots are better treated; their values are
1547 updated on class redefinition, and initforms inherited from
1548 superclasses are applied.
1549 ** REMOVE-METHOD returns its generic function argument even when
1550 no method was removed.
1551 ** SHARED-INITIALIZE now initializes the values of the requested
1552 slots, including those with :ALLOCATION :CLASS.
1553 ** ALLOCATE-INSTANCE now works on structure classes defined via
1554 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1556 ** SLOT-EXISTS-P now works on conditions, as well as structures
1558 ** MAKE-LOAD-FORM now has the required methods on
1559 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1560 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1561 keyword, and now implements the SLOT-NAMES argument.
1562 ** methods with &OPTIONAL arguments no longer allow too many
1563 arguments to be passed in the call without error.
1564 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1565 option is consistent with the required arguments of the generic
1566 function lambda list.
1567 * bug fix: REQUIRE accepts a string designator. (Thanks to
1569 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1570 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1571 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1572 * bug fix: The compiler now checks for duplicated variables in macro
1574 * bug fix: SETQ on globals returns the correct value.
1575 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1576 not a COMPILER-ERROR (followed by some other strange error on
1577 choosing the CONTINUE restart).
1578 * bug fix: make.sh and friends are now more consistent in the way that
1579 they look for GNU "make".
1581 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1582 * experimental native threads support (on x86 Linux >=2.4 only).
1583 This is not compiled in by default: you need to add :SB-THREAD to
1584 the target features. See the "Beyond ANSI" chapter of the manual
1586 * fix for longstanding nonANSIism: The old distinction between
1587 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1588 The return value from CL:FIND-CLASS is now a CLOS class, and
1589 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1590 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1591 * An interface to the MetaObject Protocol, as described in Kiczales,
1592 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1593 MIT Press, 1991, is available from the SB-MOP package.
1594 * incompatible change: the SB-PCL package should now be considered
1595 a private implementation detail, and no longer a semi-private MOP
1597 * minor incompatible change: due to rearrangement for threads, the
1598 control stack and binding stack are now allocated at arbitrary
1599 addresses instead of being hardcoded per-port. Users affected by
1600 this probably have to be doing advanced things with shared
1601 libraries, and will know who they are.
1602 * minor incompatible change: Previously, all --eval forms used were
1603 processed with READ before any of them were processed with EVAL.
1604 Now each --eval form is processed with both READ and EVAL before
1605 the next --eval form is processed. (Thus package operations like
1606 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1607 work as the user might reasonably expect.)
1608 * minor incompatible change: *STANDARD-INPUT* is now only an
1609 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1611 * minor incompatible change: Y-OR-N-P is now character-oriented,
1612 not line oriented. Also, YES-OR-NO-P now works without errors.
1613 (thanks to Antonio Martinez)
1614 * sb-aclrepl module improvements: an integrated inspector, added
1615 repl features, and a bug fix to :trace command.
1616 * Known functions, which cannot be open coded by the backend, are
1617 considered to be able to check types of their arguments. (fixing
1618 a bug report by Nathan J. Froyd)
1619 * fixed a bug in computing method discriminating functions: It is
1620 now possible to define methods specialized on classes which have
1621 forward-referenced superclasses. (thanks to Gerd Moellmann)
1622 * fixed evaluation order in optional entries (reported by Gilbert
1624 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1625 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1626 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1627 required-by-AMOP NAME argument, as well as CLASS and
1628 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1629 * fixed bug 20: DEFMETHOD can define methods using names that are
1630 not the proper names of classes to designate class specializers.
1631 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1632 Unix file descriptors, instead of blowing up. (thanks to Antonio
1634 * Garbage collection refactoring: user-visible change is that a
1635 call to the GC function during WITHOUT-GCING will not do garbage
1636 collection until the end of the WITHOUT-GCING. If you were doing
1637 this you were probably losing anyway.
1638 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1639 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1640 TYPEP the latter but not the former.
1641 * The compiler issues a full WARNING for calls to undefined functions
1642 with names from the CL package.
1643 * MAP-INTO for a vector destination is open coded. (reported by
1644 Brian Downing on c.l.l)
1645 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1646 documentation string.
1647 * fixed some bugs revealed by Paul Dietz' test suite:
1648 ** COPY-ALIST now signals an error if its argument is a dotted
1650 ** Condition slots are now accessed more correctly in the presence
1651 of multiple initargs for a given slot.
1652 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1653 exclude restarts of the same name associated with a different
1655 ** DEFCLASS of forward-referenced classes with another
1656 forward-referenced class in the superclasses list no longer
1658 ** Condition slots are now initialized once each, not multiple
1659 times. (thanks to Gerd Moellmann)
1660 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1662 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1663 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1664 arguments contain duplicated elements.
1665 ** RESTART-CASE understands local macros.
1666 ** RESTART-CASE associates exactly its own restarts with a condition.
1667 ** ENDP in safe mode checks its argument to be of type LIST.
1668 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1669 in question is unbound.
1670 ** Optimized MAKE-INSTANCE functions no longer cause internal
1671 assertion failures in the presence of duplicate initargs.
1672 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1673 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1674 propagated even in the ctor optimized implementation of
1676 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1678 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1679 * a better implementation of SXHASH on (simple) bit vectors,
1680 measured both in execution speed and in distribution of results
1681 over the positive fixnums, has been installed. Likewise, a better
1682 implementation of EQUAL for simple bit vectors is now available.
1683 * fixed CEILING optimization for a divisor of form 2^k.
1684 * fixed bug 240 (emitting extra style warnings "using the lexical
1685 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1686 by Antonio Martinez)
1687 * fixed SXHASH, giving different results for NIL depending on type
1688 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1689 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1690 variable. (found by Rolf Wester)
1691 * SBCL does not ignore type declarations for special
1692 variables. (reported by rif on c.l.l 2003-03-05)
1693 * some bug fixes in contrib/sb-aclrepl/
1694 * fixed some bugs revealed by Paul Dietz' test suite:
1695 ** a bug in the CONS type specifier, whereby the CAR and CDR
1696 types got intertwined, has been fixed;
1697 ** the type system is now able to reason about the interaction
1698 between INTEGER and RATIO types more completely;
1699 ** APPEND, [N]REVERSE and NRECONC check that those their
1700 arguments, which must be proper lists, are really so;
1701 ** An array specialized to be unable to hold elements has been
1702 implemented, as required -- yes, really -- by ANSI;
1703 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1704 on malformed property lists;
1706 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1707 * incompatible packaging change: in line with Unix convention,
1708 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1709 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1710 * REQUIRE and PROVIDE are now optionally capable of doing something
1711 useful. See the documentation string for REQUIRE.
1712 * infrastructure for a managed SBCL contrib system: contributed
1713 modules in this release include:
1714 ** the ASDF system definition facility;
1715 ** an interface to the BSD Sockets API;
1716 ** an ACL-like convenience interface to the repl;
1717 (thanks to Kevin Rosenberg)
1718 ** an implementation of ROTATE-BYTE, with efficient implementation
1720 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1721 gives a closer approximation to the right answer than previously.
1722 (thanks to Raymond Toy)
1723 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1724 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1725 optional environment arguments, as required by ANSI.
1726 * fixed bugs in other functions taking environment objects, allowing
1727 calls with an explicit NIL environment argument to be compiled
1729 * fixed bug 228: primary return values from
1730 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1731 COMPILE or FUNCTION.
1732 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1733 :TYPE will no longer signal errors on innocuous objects.
1734 * fixed bug 231b: SETQ is better at respecting type declarations in
1735 the lexical environment.
1736 * fixed a bug in DEFCLASS: classes named by symbols with no or
1737 unprintable packages can now be defined.
1738 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1739 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1740 * fixed compiler failure related to checking types of functions.
1741 (reported by Robert E. Brown)
1742 * the compiler is now much more consistent in its error-checking
1743 treatment of bounding index arguments to sequence functions: in
1744 (SAFETY 3) code, errors will be signalled in almost all cases if
1745 invalid sequence bounding indices are passed to functions defined
1746 by ANSI to operate on sequences.
1747 * fixed a bug in the build procedure: documentation of SBCL-specific
1748 packages is now preserved and available in the final Lisp image.
1749 * lifted FDEFINITION lookup out of loops in the implementation of
1750 many list operations. (thanks to Robert E. Brown)
1751 * fixed a bug in the reader: the #n# reader macro now works for
1752 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1753 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1754 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1755 E. Brown and Tony Martinez respectively)
1756 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1757 SLOT-BOUNDP: in particular, it is now able to optimize them much
1758 better, and is now not vulnerable to having packages renamed.
1759 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1760 always called when a slot is not present in an instance. (thanks
1762 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1763 longer signal an error on unreading a character following EOF on
1764 the previous constituent stream. (thanks to Tony Martinez)
1765 * fixed some bugs revealed by Paul Dietz' test suite:
1766 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1767 not just nonnegative fixnums;
1768 ** the logical bit-array operators such as BIT-AND now accept an
1769 explicit NIL for their "opt-arg" argument (to indicate a
1770 freshly-consed result bit-array);
1771 ** ELT now signals an error on an invalid sequence index in safe
1773 ** the type system is now cleverer about negations of numeric
1774 types, and consequently understands the BIGNUM and RATIO types
1776 ** the type system is now cleverer about the interaction between
1777 INTEGER and RATIO types: while bugs still remain, many more
1778 cases are accurately computed;
1779 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1780 if it is in the last clause;
1781 ** CONSTANTLY now correctly returns a side-effect-free function in
1783 ** DECLARE is no longer treated as a special-operator; in
1784 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1785 * incremented fasl file version number due to the change in the
1786 DEFSTRUCT-SLOT-DESCRIPTION structure.
1788 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1789 * minor incompatible change: code processed by the "interpreter" or
1790 EVAL now has a compilation optimization policy of (DEBUG 2)
1791 (changed from (DEBUG 1)) to improve debuggability of interactive
1792 development, and to allow the use of the debug RETURN command in
1794 * an experimental implementation of the RETURN command for the
1795 debugger has been included. (thanks to Frederik Kuivinen)
1796 * fixed bug 62: constraints were not propagated into a loop.
1797 * fixed bug in embedded calls of SORT (reported and investigated by
1799 * fixed some bugs revealed by Paul Dietz' test suite:
1800 ** printing and reading of arrays with some dimensions having
1801 length 0 (thanks to Gerd Moellmann);
1802 ** BOA constructor with &AUX argument without a default value does
1803 not cause a type error;
1804 ** CONSTANTP now returns true for all self-evaluating objects.
1806 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1807 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1808 accessors that are related by inheritance, as specified in the
1809 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1810 Valtteri Vuorikoski)
1811 * The compiler is now able to inline functions that were defined in
1812 a complex lexical environment (e.g. inside a MACROLET).
1813 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1815 * Support for the upcoming FreeBSD-5.0 release has been included.
1816 (thanks to Dag-Erling Smorgrav)
1817 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1818 effect when it is not in a toplevel context.
1819 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1820 stand a better chance of being correct. (thanks to Gerd
1822 * fixed bug in COERCE, which now signals an error on coercing a
1823 rational to a bounded real type which excludes the expected
1825 * The compiler is now able to derive types more accurately from the
1826 COERCE and COMPILE functions.
1827 * fixed bug 223: functional binding is considered to be constant
1828 only for symbols in the CL package.
1829 * fixed bug 231: SETQ did not check the type of a variable being set
1830 (reported by Robert E. Brown)
1831 * A new optimization for MAKE-INSTANCE has been included, fixing
1832 various bugs (including relating to :ALLOCATION :CLASS slots and
1833 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1835 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1836 ** As required by ANSI, LOOP now disallows anonymous collection
1837 clauses such as COLLECT I in conjunction with aggregate boolean
1838 clauses such as THEREIS (= I 1);
1839 ** LOOP now signals an error when any variable is reused in the
1840 same loop (including the potentially useful construct analogous
1841 to WITH A = 1 WITH A = (1+ A);
1842 ** IT is only a special loop symbol within the first clause of a
1843 conditional loop clause;
1844 ** LOOP with a typed iteration variable over a hashtable now
1845 signals a type error iff it should.
1846 * fixed some other bugs revealed by Paul Dietz' test suite:
1847 ** FILE-STREAM now names the class previously known as FD-STREAM;
1848 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1849 argument) no longer signals an error;
1850 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1851 of the slot symbol, rather than using the current package
1852 ((:CONC-NAME "") continues to intern the slot's name in the
1854 * incremented fasl file version number, because of the incompatible
1855 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1856 of the new implementation of DEFINE-COMPILER-MACRO.
1858 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1859 * Support for building SBCL for MIPS platforms running in
1860 little-endian mode has now been checked in, and basic
1861 functionality on said platforms verified.
1862 * minor incompatible change: PCL now records the pathname of a file
1863 in which methods and the like are defined, rather than its
1865 * minor incompatible change: TRUENAME now considers the truename of
1866 a file naming a directory to be the pathname with :DIRECTORY
1867 component indicating that directory.
1868 * minor incompatible change: a NAMED clause in the extended form of
1869 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1870 reason for the previous behaviour is unclear.
1871 * more systematization and improvement of CLOS and MOP conformance
1872 in PCL (thanks to Gerd Moellman and Pierre Mai):
1873 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1875 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1876 primary methods with no specializers;
1877 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1879 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1880 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1881 FORWARD-REFERENCED-CLASSes; error reporting on
1882 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1884 ** SXHASH on CLOS instances now uses a slot internal to the
1885 instance to return different numbers on distinct instances,
1886 while preserving the same return value through invocations of
1888 ** DEFMETHOD signals errors when methods with longer incongruent
1889 lambda lists are added to generic functions;
1890 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1891 CLASS, as specified in AMOP;
1892 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1893 on the order returned by the primary method for classes of
1894 class STANDARD-CLASS;
1895 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1896 * fixed some bugs shown by Paul Dietz' test suite:
1897 ** DOLIST puts its body in TAGBODY;
1898 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1900 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1901 value producing form;
1902 ** if more variables are given to PROGV than values, extra
1903 variables are bound and made to have no value;
1904 ** NSUBSTITUTE on list arguments gets the right answer with
1906 ** ELT signals an error of type TYPE-ERROR when the index argument
1907 is not a valid sequence index;
1908 ** LOOP signals (at macroexpansion time) an error of type
1909 PROGRAM-ERROR when duplicate variable names are found;
1910 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1911 ** FUNCALL of special-operators now cause an error of type
1913 ** PSETQ now works as required in the presence of side-effecting
1914 symbol-macro places;
1915 ** NCONC accepts any object as its last argument;
1916 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1918 ** loop-for-as-package does not require a package to be explicitely
1920 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1922 * fixed bug 166: compiler preserves "there is a way to go"
1923 invariant when deleting code.
1924 * fixed bug 172: macro lambda lists with required arguments after
1925 &REST arguments now cause an error to be signalled. (thanks to
1927 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1928 bug. (thanks to Matthew Danish)
1929 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1931 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1932 arguments are lexically rebound. (thanks to Gerd Moellmann and
1934 * fixed bug 194: error messages are now more informative when there
1935 is no primary method applicable in a call to a generic
1936 function. (thanks to Gerd Moellmann)
1937 * fixed bug in command line argument checking (thanks to Julian
1939 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1940 (thanks to Matthew Danish)
1941 * incremented fasl file version number, because of the
1942 SXHASH-related changes in the layout of CLOS data structures
1944 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1945 * minor incompatible change: The runtime (the Unix executable named
1946 "sbcl") is now much pickier about the .core files it will load.
1947 Essentially it now requires .core files to descend from the same
1948 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1949 as the runtime does. (The intent is to prevent the crashes which
1950 can occur, and which can even be reported as mysterious failures,
1951 when people patch the sources or change the build parameters
1952 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1953 sbcl and .core files.)
1954 * fixed bug: VALUES-LIST is no longer optimized away.
1955 * fixed bug 142: The FFI conversion of C string values to Lisp
1956 string values no longer conses excessively. (thanks to Nathan
1957 Froyd porting Raymond Toy's fix to CMU CL)
1958 * began to systematize and improve MOP conformance in PCL (thanks to
1959 Nathan Froyd, Gerd Moellman and Pierre Mai):
1960 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1962 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1963 ** FINALIZE-INHERITANCE is now called on class finalization;
1964 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1965 argument precedence order.
1966 * fixed bug 202: The compiler no longer fails on functions whose
1967 derived types contradict their declared type.
1968 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1969 so it can be non-toplevel.
1970 * The fasl file version number has changed (because of the new
1971 implementation of DEFMACRO).
1972 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1973 safe code, that any length requirement by their type-specifier
1974 argument is valid. The exceptions to this are described in bug
1976 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1977 functions now check the types of their inputs as required by ANSI.
1978 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1979 introduced symbol is DECLAREd to be SPECIAL.
1980 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1981 * fixed bug: :COUNT argument to sequence functions may be negative
1982 * fixed bug: body of DO-SYMBOLS may contain declarations
1983 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1984 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1985 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1986 symbol macro only once
1987 * fixed printing of call frame when argument list is unavailable
1988 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1989 * compiler no longer signals WARNING on unknown keyword
1992 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1993 * A beta-quality port to the mips architecture running Linux,
1994 based on the old CMUCL backend, has been made. It has been tested
1995 on a big-endian kernel, and works sufficiently well to be able to
1996 rebuild itself; it has not been tested in little-endian mode.
1997 * fixed an inconsistency between gencgc.c and purify.c which made
1998 dumping/loading .core files unreliable
1999 * fixed bug 120a: The compiler now deals correctly with IFs where
2000 the consequent is the same as the alternative, instead of
2001 misderiving the return type. (thanks to Alexey Dejneka)
2002 * fixed bug 113: Logical pathnames are now dumpable (the logical
2003 host is resolved at load-time, throwing an error if it is not
2005 * fixed bug 174: FORMAT's error message is slightly clearer when a
2006 non-printing character is used in a format directive.
2007 * fixed several bugs in compiler checking of type declarations, i.e.
2008 violations of the Python "declarations are assertions" principle
2009 (thanks to Alexey Dejneka)
2010 * fixed several bugs in PCL's error checking (thanks to Gerd
2012 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2014 * fixed bug in compilation of functions as first class values
2015 (thanks to Antonio Martinez-Shotton)
2016 * The compiler's handling TYPE-ERRORs which it can prove will
2017 inevitably happen at runtime has been cleaned up and corrected
2018 in several ways. (thanks to Alexey Dejneka)
2019 * improved argument type checking for various basic arithmetic
2020 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2021 lavished on them in the past that they can be compiled in many
2022 ways in different special cases
2023 * fixed bug 181: compiler checks validity of user supplied type
2025 * cleaned up code flushing in optimization: Function calls which
2026 should signal errors for safety purposes (e.g. which ANSI says
2027 should signal errors when their arguments are of incorrect type)
2028 are no longer optimized away.
2029 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2030 * incremented fasl file version number, because changes in the
2031 implementation of sequence functions like COERCE caused
2032 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2033 in old inline expansions) to become undefined. (Actually these
2034 changes were later undone, so we might very well be binary
2035 compatible with 0.7.7 after all, but leaving the version number
2036 incremented seemed like the simplest and most conservative
2039 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2040 * An alpha-quality port to the parisc architecture running Linux,
2041 based on the old CMUCL backend, has been made. This, even more so
2042 than the other backends, should be considered still a work in
2043 progress; known problems include that the Linux kernel in 64-bit
2044 mode does not propagate the correct sigcontext structure to
2045 userspace, and consequently SBCL on a parisc64 kernel will not
2047 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2048 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2049 suppressed.) Also now that the compiler is looking at declarations
2050 in the environment, it checks optimization declarations as well,
2051 and suppresses inlining when (> DEBUG SPEED).
2052 * More fixes have been made to treatment of floating point exception
2053 treatment and other Unix signals. In particular, floating point
2054 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2055 * The detection and handling of control stack exhaustion (infinite
2056 or very deeply nested recursion) has changed. Stack exhaustion
2057 detection is now done by write-protecting pages at the OS level
2058 and applies at all optimization settings; when found, a
2059 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2060 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2061 be caught using IGNORE-ERRORS.
2062 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2063 correctly when transforming recursive LABELS functions to LETs.
2064 (thanks to Alexey Dejneka)
2065 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2066 that are names of constants or global variables.
2067 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2068 alien routines with docstrings.
2069 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2070 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2072 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2073 (thanks to Eric Marsden)
2074 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2075 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2076 * Bug fix: The SPARC backend can now compile functions involving
2077 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2078 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2079 to a functions expecting a PATHNAME-DESIGNATOR.
2080 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2081 lambda lists. (thanks to Alexey Dejneka)
2082 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2083 functions so that EXPORTing the name of the class doesn't cause
2084 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2085 to Antonio Martinez for reporting this)
2086 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2087 constant-folded without failing an assertion. (thanks to Einar
2088 Floystad Dorum for reporting this)
2089 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2090 types are now dealt with more correctly by the compiler.
2091 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2092 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2093 behaviour analogous to RENAME-FILE. This puts its behaviour more
2094 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2096 * The fasl file version number has changed again. (because of the
2097 bug fix involving the names of PCL MAKE-INSTANCE functions)
2099 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2100 * bug fix: Floating point exceptions are treated much more
2101 consistently on the x86/Linux and PPC/Linux platforms.
2102 * Array initialization with :INITIAL-ELEMENT is now much faster for
2103 cases when the compiler cannot open code the array creation, but
2104 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2105 array accesses have also seen a speed increase.
2106 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2107 specified. (This was at the root of some bad interactions between
2108 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2110 * bug fix: Internal error arguments for undefined functions are now
2111 computed correctly on the PPC/Linux platform.
2112 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2113 Raymond Toy's patch for CMU CL)
2114 * Support for the Solaris 9 operating environment has been included
2115 (thanks to Daniel Merritt)
2116 * A very ugly but hopefully complete draft of the missing FFI chapter
2117 of the manual has been created by reformatting the corresponding
2118 CMU CL manual chapter into (currently very ugly and incoherent)
2119 DocBook and bringing it up to date for SBCL behavior. Thus, the
2120 manual is now essentially complete, at least by my extreme
2121 once-and-only-once standards, whereby it's acceptable to refer to
2122 the doc strings of SB-EXT functions as the primary documentation.
2123 * The fasl file version number has changed again, due to cleanup of
2124 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2125 is no longer a static symbol.)
2127 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2128 * SBCL now builds with OpenMCL (version 0.12) as the
2129 cross-compilation host; also, more progress has been made toward
2130 bootstrapping under CLISP.
2131 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2133 * bug 158 fixed: The compiler can now deal with integer loop
2134 increments different from 1; fixing this turned out also to fix
2136 * bug 169 fixed: no more bogus warnings about using lexical bindings
2137 despite the presence of perfectly good SPECIAL declarations (thanks
2138 to David Lichteblau)
2139 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2140 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2141 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2143 * bug 180 fixed: Method combination specifications no longer ignore
2144 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2145 * bug fix: Structure type predicate functions now check their argument
2146 count as they should.
2147 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2148 correctly. (thanks to Pierre Mai)
2149 * minor incompatible change: The --noprogrammer option is deprecated
2150 in favor of the new --disable-debugger option, which is very similar.
2151 (The major difference is that it takes effect at a slightly different
2152 time at startup, causing handling of errors in --sysinit and
2153 --userinit files will be affected differently.) The
2154 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2155 been added to allow this functionality to be controlled from ordinary
2156 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2157 maintainers, who might want to run non-interactive scripts to
2158 build SBCL cores which will later be used interactively.)
2159 * minor incompatible change: The LOAD function no longer, when given
2160 a wild pathname to load, loads all files matching that pathname.
2161 Instead, an error of type FILE-ERROR is signalled.
2163 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2164 * bug 147 fixed: The compiler preserves its block link/count
2165 invariants more correctly now so that it doesn't crash. (thanks
2167 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2169 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2170 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2171 no-value-for-this-character as NIL (as specified by ANSI).
2172 * HOST-NAMESTRING on physical pathnames now returns a string that is
2173 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2174 (thanks to Christophe Rhodes)
2175 * The Alpha port handles icache flushing more correctly. (thanks to
2177 * More progress has been made toward bootstrapping under CLISP. (thanks
2178 to Christophe Rhodes)
2179 * The fasl file format has changed again, because dynamic loading
2180 on OpenBSD (which has non-ELF object files) motivated some cleanups
2181 in the way that foreign symbols are transformed and passed around.
2182 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2183 is no longer treated as whitespace by the reader, but instead as
2184 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
2185 |A
\x7fB|, instead of A as it used to.
2187 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2188 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2189 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2190 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2191 this as of 0.7.1.45, but was left out of the previous news section
2192 (thanks to Dan Barlow)
2193 * SBCL now runs on the Solaris operating system on SPARC architectures
2194 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2195 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2196 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2197 * SPARC backend cleanups, allowing builds of cores optimized for V8
2198 and V9 SPARCS, and also emission of code targeted to a particular
2199 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2201 * SBCL is closer to bootstrapping under CLISP, thanks to various
2202 fixes by Christophe Rhodes.
2203 * The fasl file format has changed again, to allow the compiler's
2204 INFO database to support symbol macros.
2205 * The user manual (in doc/) is formatted into HTML more nicely.
2206 (thanks to coreythomas)
2207 * The system is smarter about SUBTYPEP relationships, especially
2208 those involving NOT types (including types such as ATOM which are
2209 represented internally using NOT types). Thus SUBTYPEP is less
2210 likely to return (VALUES NIL NIL) in general, and in particular
2211 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2213 * The fasl file format has changed again, because the internal
2214 representation of types now includes a new slot to support the new
2215 SUBTYPEP-of-NOT-types logic.
2216 * (not a change in the main branch of SBCL, but a related prototype
2217 which can hopefully be merged into the main branch of SBCL in the
2218 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2219 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2220 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2221 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2222 dispatch macro character combinations. (thanks to Alexey Dejneka)
2223 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2224 that superficially look like logical namestrings correctly.
2225 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2227 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2228 default supplied-p) for &optional and &key arguments. (thanks to
2231 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2232 * incompatible change: The compiler is now less aggressive about
2233 tail call optimization, doing it only when (> SPACE DEBUG) or
2234 (> SPEED DEBUG). (This is an incompatible change because there are
2235 programs which relied on the old CMU-CL-style behavior to optimize
2236 away their unbounded recursion which will now die of stack overflow.)
2237 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2238 for non-GENCGC systems has been increased to 20M (since that
2239 seems much closer to the likely performance optimum for modern
2240 systems than the old 4M value was)
2241 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2242 and *DEBUG-PRINT-LEVEL*
2243 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2244 of CMU CL's support for SPARC, and various endianness and other
2245 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2246 * new syntactic sugar for the Unix command line: --load foo.bar is now
2247 an alternate notation for --eval '(load "foo.bar")'.
2249 ** The system now detects stack overflow and handles it gracefully,
2250 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2251 optimization settings. (This is a good thing in general, and
2252 its introduction in this version should be particularly timely
2253 for anyone whose code fails because of suppression of tail
2255 ** The system now hunts for the C variable "environ" in a more
2256 devious way, to avoid segfaults when the C library version differs
2257 between compile time and run time. (thanks to Christophe Rhodes)
2258 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2259 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2260 ** The compiler no longer issues bogus style warnings for undefined
2261 classes in the same source file as the DEFCLASSes which defined
2262 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2264 ** fixes in CONDITION class precedence list for undefined function
2265 errors (thanks to Alexei Dejneka)
2266 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2267 correctly. (thanks to Dan Barlow)
2268 ** portability fixes aiming at bootstrapping under CLISP (thanks
2269 to Dave McDonald and Christophe Rhodes)
2270 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2271 ** fixes in type translation and and type inference (thanks to
2273 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2274 ** various fixes in the new ports (thanks to Dan Barlow)
2275 * several changes related to debugging:
2276 ** suppression of tail recursion, as noted above
2277 ** stack overflow detection, as noted above
2278 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2279 is now the default. (For some time encapsulation has been more
2280 reliable than the breakpoint-based :ENCAPSULATE NIL
2281 implementation, at least on X86 systems; and I just noticed that
2282 encapsulation also seems closer to the spirit of the ANSI
2285 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2287 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2288 up properly again. (There was a packaging bug in 0.7.0 which
2289 left their definitions in SB-SYS::LOAD-FOREIGN and
2290 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2291 vital for most things which interface to C-level interfaces,
2292 like extensions working with sockets or databases or
2293 Perl-compatible regexes or whatever, and the need to fix
2294 this bug is the main reason that 0.7.1 was released so
2296 ** DEFGENERIC is now choosier about the methods it redefines, so that
2297 reLOADing a previously-LOADed file containing DEFGENERICs does
2298 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2299 cycle now works as it should. (thanks to Alexey Dejneka)
2300 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2301 fixed by Christophe Rhodes. (He actually submitted this patch
2302 months ago, and I delayed until after 0.7.0.)
2303 ** Bug 111 (internal compiler confusion about runtime checks on
2304 FUNCTION types) was fixed by Alexey Dejneka.
2305 * Some internal cleanups (getting rid of variables which aren't
2306 needed now that the byte interpreter is gone) caused the fasl
2307 file format number to change again.
2309 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2310 * major incompatible change: The default fasl file extension, i.e. the
2311 default extension for files produced by COMPILE-FILE, has changed
2312 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2314 ** There are many changes in the implementation of the compiler.
2315 SBCL is now essentially a compiler-only implementation of ANSI
2316 Common Lisp. EVAL still "interprets" a few special cases, but
2317 almost all the interesting cases are handled by creating
2318 a LAMBDA expression, calling COMPILE on it, then calling
2319 FUNCALL on the result.
2320 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2321 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2322 Since the code is newer, there might still be some new bugs
2323 (though not as many as before Martin Atzmueller's fixes:-). But
2324 the new code is substantially simpler and clearer, and hopefully
2325 any remaining bugs will be simpler, less fundamental, and more
2326 fixable then the bugs in the old code.
2327 ** The revised compiler is still a little unsteady on its feet.
2329 *** The debugging information it produces (particularly the names
2330 of FUNCTION objects) is sometimes much less useful than what
2331 the old compiler produced.
2332 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2333 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2334 within a MACROLET) has been temporarily weakened.
2335 ** There are new compiler optimizations for various functions:
2336 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2337 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2338 *** the math functions TRUNCATE, FLOOR, and CEILING
2339 *** the function-of-all-trades COERCE
2340 Mostly these should be transparent, but there's one
2341 potentially-annoying problem (bug 117): when the compiler
2342 inline-expands a function and does type analysis on the result,
2343 it can create control paths which have type mismatches, and
2344 when it can't prove that those control paths aren't taken,
2345 it will issue WARNINGs about the type mismatches. This is
2346 a particular problem in practice for the new sequence functions.
2347 It's not clear how this should be fixed, and for now, a
2348 workaround is given in the entry for 117 in the BUGS file.
2349 ** (Because of the interaction between the two previous items --
2350 occasional inlining problems and new inline expansions -- some
2351 of the new sequence function optimizations won't really kick in
2352 completely until debugging information, and then inlining, are
2353 straightened out in some future version.)
2354 * minor incompatible changes:
2355 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2356 DIRECTORY no longer implicitly promotes NIL slots of its
2357 pathname argument to :WILD. In particular, when you ask for the
2358 contents of a directory (which you used to be able to do without
2359 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2360 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2361 ** changes in behavior that ANSI explicitly defines to be
2362 implementation dependent:
2363 *** The new compiler-only implementation still conforms with ANSI,
2364 but acts a little different than before. Besides the obvious
2365 changes in performance tradeoffs (that the cost per form passed
2366 to EVAL has gone up, and the cost per form executed by EVAL
2367 has gone down), the behavior of the system changes a little
2368 because there are no longer any interpreted function objects.
2369 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2370 e.g. doing COMPILE on the output of interactive DEFUN is
2372 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2374 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2375 has changed from (:CL) to NIL.
2376 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2377 e.g. #\Newline and #\Tab, don't have names specified in the
2378 ANSI Common Lisp standard, is now based on their ASCII symbolic
2379 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2380 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2381 are no longer used for output.
2382 ** changes in internal implementation constants:
2383 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2384 4 million. (If your application spends a lot of time GCing and
2385 you have a lot of RAM, you might want to experiment with
2386 increasing it even more.)
2387 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2388 However, almost all old code should still continue to work without
2389 immediate update, as SB-C-CALL is now a (deprecated) nickname
2391 ** Old operator names in the style DEF-FOO are now deprecated in
2392 favor of new corresponding names DEFINE-FOO, for consistency with
2393 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2394 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2395 internal symbols, but a few supported extensions like
2396 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2397 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2398 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2399 etc. as you get deeper into recursive calls to the debugger
2400 command loop, instead of the old "5]", "5]]", "5]]]"
2401 sequence. (I was motivated to do this when squabbles between
2402 ILISP and SBCL left me very deeply nested in the debugger. In the
2403 short term, this change will probably provoke more ILISP/SBCL
2404 squabbles, but hopefully it will be an improvement in the long run.)
2405 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2406 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2407 function, since declaiming return types involving aliens is
2408 (1) annoyingly messy to do by hand and (2) vital to efficient
2409 compilation of code which calls such functions.
2410 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2411 longer reexported by the SB-EXT package. They're solely useful
2412 for alien code, so it seems more logical that you should get
2413 them from the SB-ALIEN package, not in SB-EXT.
2414 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2415 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2416 features. Instead, the code that they used to control is always
2417 built into the system.
2418 * many other bug fixes
2419 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2420 advantage of the new EVAL-WHEN stuff and to clean them up in
2421 general, and they are now more ANSI-compliant in a number of
2422 ways. Martin Atzmueller is responsible for a lot of this.
2423 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2425 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2426 *** correct ERROR type for various file operations
2427 *** some fixes for Lisp streams
2428 *** DEFMETHOD syntax checking
2429 *** changing old weird representation of debug information as
2430 strings (which, among their other deficiencies, don't transform
2431 correctly when you rename packages, and don't change their
2432 print representation when you change things like *PACKAGE*
2433 and *PRINT-LENGTH*) to symbols and lists of symbols
2434 He also made several improvements and fixed several bugs in DESCRIBE.
2435 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2437 *** misbehavior of WRITE-STRING/WRITE-LINE
2438 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2439 and several other LOOP problems as well
2440 *** DIRECTORY when similar filenames are present
2441 *** DEFGENERIC with :METHOD options
2442 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2443 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2444 *** argument ordering in FIND with :TEST option
2445 *** mishandled package designator argument in APROPOS-LIST
2446 *** various problems in the backquote readmacro
2447 *** a bug in APROPOS
2448 *** probably some others that I'm not describing very well here,
2449 since the CVS log documents them by reference to sbcl-devel
2450 messages, and the SourceForge archives aren't working well.:-(
2451 ** Dan Barlow improved the Alpha port (and is making progress on the
2452 PPC port, for those of you who think different).
2453 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2454 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2455 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2456 continues to work on the SPARC port (for those of you in a position
2457 to look down upon our little PC-compatible boxes from a great height).
2458 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2459 and uses that copy, rather than the current dynamic value, when
2460 it's trying to decide whether to truncate output. Thus e.g.
2461 (let ((*print-lines* 50))
2462 (pprint-logical-block (stream nil)
2464 (let ((*print-lines* 8))
2465 (print (aref possiblybigthings i) stream)))))
2466 should now truncate the logical block only at 50 lines, instead of
2467 often truncating it at 8 lines, as it did before.
2468 * The doc/cmucl/ directory, containing old CMU CL documentation
2469 from the time of the fork, is no longer part of the base system.
2470 SourceForge has shut down its anonymous FTP service, and with it
2471 my original plan for distributing the old CMU CL documentation
2472 there. For now, if you need these files you can download an old
2473 SBCL source release and extract them from it.
2474 * The fasl file version number changed again, for dozens of reasons,
2475 some of which are apparent above.
2477 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2478 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2479 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2480 patch, so that DISASSEMBLE output is much nicer.
2481 * The code in the SB-PROFILE package now seems reasonably stable.
2482 I still haven't decided what the final interface should look like
2483 (I'd like PROFILE to interact cleanly with TRACE, since both
2484 facilities use function encapsulation) but if you have a need
2485 for profiling now, you can probably use it successfully with
2486 the current CMU-CL-style interface.
2487 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2488 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2489 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2490 link now returns the dangling link itself, and for similar
2491 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2492 link itself. (In these cases the old code signalled an error and
2493 looped endlessly, respectively.) Thus, DIRECTORY now works even
2494 in the presence of dangling and cyclic symbolic links.
2495 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2496 is now a supported extension again, since the consensus on
2497 sbcl-devel was that it can be useful for ordinary development
2498 work, not just for debugging SBCL itself.
2499 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2500 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2501 that currently-defined functions might be redefined later with
2502 different return types.
2503 * Hash tables can be printed readably, as inspired by CMU CL code
2504 of Eric Marsden and SBCL code of Martin Atzmueller.
2505 * better error handling in CLOS method combination, thanks to
2506 Martin Atzmueller porting Pierre Mai's CMU CL patches
2507 * more overflow fixes for >16Mbyte I/O buffers
2508 * A bug in READ has been fixed, so that now a single Ctrl-D
2509 character suffices to cause end-of-file on character streams.
2510 In particular, now you only need one Ctrl-D at the command
2511 line (not two) to exit SBCL.
2512 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2514 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2515 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2516 does the right thing.
2517 * The implementation of some type tests, especially for CONDITION
2518 types, is now tidier and maybe faster, due to CMU CL code
2519 originally by Douglas Crosher, ported by Martin Atzmueller.
2520 * Some math functions have been fixed, and there are new
2521 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2522 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2523 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2524 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2525 likely to be built into sbcl-0.7.0.)
2526 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2527 CL, ported to SBCL by Martin Atzmueller)
2528 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2529 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2530 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2531 * There's a new slam.sh hack to shorten the edit/compile/debug
2532 cycle for low-level changes to SBCL itself, and a new
2533 :SB-AFTER-XC-CORE target feature to control the generation of
2534 the after-xc.core file needed by slam.sh.
2535 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2536 COMPILE-FILE is no longer supported, so that now every function
2537 gets an entry point, so that block compilation looks a little
2538 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2539 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2540 returns the number of bytes consed since the system started,
2541 rather than the number consed since the first time the function
2542 was called. (The new definition parallels ANSI functions like
2543 CL:GET-INTERNAL-RUN-TIME.)
2544 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2545 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2546 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2547 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2548 * Fasl file version numbers are now independent of the target CPU,
2549 since historically most system changes which required version
2550 number changes have affected all CPUs equally. Similarly,
2551 the byte fasl file version is now equal to the ordinary
2554 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2555 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2556 is no longer recognized. I apologize for this, because it was
2557 listed in SB-EXT as a supported extension, but I found that
2558 its existing behavior was poorly specified, as well as incorrectly
2559 specified, and it looked like too much of a mess to straighten it
2560 out. I have enough on my hands trying to get ANSI stuff to work..
2561 * many patches ported from CMU CL by Martin Atzmueller, with
2562 half a dozen bug fixes in pretty-printing and the debugger, and
2563 half a dozen others elsewhere
2564 * fixed bug 13: Floating point infinities are now supported again.
2565 They might still be a little bit flaky, but thanks to bug reports
2566 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2567 as flaky as they were.
2568 * The --noprogrammer command line option is now supported. (Its
2569 behavior is slightly different in detail from what the old man
2570 page claimed it would do, but it's still appropriate under the
2571 same circumstances that the man page talks about.)
2572 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2573 are now supported, and enabled by default. Thus, the compiler can
2574 handle many floating point and complex operations much less
2575 inefficiently. (Thus e.g. you can implement a complex FFT
2577 * The compiler now detects type mismatches between DECLAIM FTYPE
2578 and DEFUN better, and implements CHECK-TYPE more correctly, and
2579 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2580 thanks to patches from Martin Atzmueller.
2581 * various fixes to make the cross-compiler more portable to
2582 ANSI-conforming-but-different cross-compilation hosts (notably
2583 Lispworks for Windows, following bug reports from Arthur Lemmens)
2584 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2585 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2586 fixed thanks to Pierre Mai's CMU CL patches.
2587 * fixes to keep the system from overflowing internal counters when
2588 it tries to use i/o buffers larger than 16M bytes
2589 * fixed bug 45a: Various internal functions required to support
2590 complex special functions have been merged from CMU CL sources.
2591 (When I was first setting up SBCL, I misunderstood a compile-time
2592 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2593 * improved support for type intersection and union, fixing bug 12
2594 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2595 more obscure bugs as well
2596 * some steps toward byte-compiling non-performance-critical
2597 parts of the system, courtesy of patches from Martin Atzmueller
2598 * Christophe Rhodes has made some debian packages of sbcl at
2599 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2600 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2601 stable, but are nonetheless usable. When he's ready, I'd be happy
2602 to add them to the SourceForge "File Releases" section. (And if
2603 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2604 * new fasl file format version number (because of changes in
2605 internal representation of (OR ..) types to accommodate the new
2606 support for (AND ..) types, among other things)
2608 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2609 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2611 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2613 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2614 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2615 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2616 * DESCRIBE now works on structure objects again.
2617 * Most function call argument type mismatches are now handled as
2618 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2619 know whether the function will be redefined before the call is
2620 executed. (The compiler could flag local calls with full WARNINGs,
2621 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2622 it doesn't keep track of enough information to know whether calls
2623 are local in this sense.)
2624 * Compiler output is now more verbose, with messages truncated
2625 later than before. (There should be some supported way for users
2626 to override the default verbosity, but I haven't decided how to
2627 provide it yet, so this behavior is still controlled by the internal
2628 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2629 src/compiler/ir1util.lisp.)
2630 * Fasl file format version numbers have increased again, because
2631 support for the Gray streams extension changes the layout of the
2632 system's STREAM objects.
2633 * The Gray subclassable streams extension now works, thanks to a
2634 patch from Martin Atzmueller.
2635 * The full LOAD-FOREIGN extension (not just the primitive
2636 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2637 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2638 but like most other programs, it defaults to copying the Unix
2639 environment from the original process instead of starting the
2640 new process in an empty environment.
2641 * Extensions which manipulate the Unix environment now support
2642 an :ENVIRONMENT keyword option which doesn't smash case or
2643 do other bad things. The CMU-CL-style :ENV option is retained
2644 for porting convenience.
2645 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2646 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2648 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2650 * A patch from Martin Atzmueller seems to have solved the SIGINT
2651 problem, and as far as we know, signal-handling now works cleanly.
2652 (If you find any new bugs, please report them!)
2653 * The system no longer defaults Lisp source file names to types
2654 ".l", ".cl", or ".lsp", but only to ".lisp".
2655 * The compiler no longer uses special default file extensions for
2656 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2657 seems to expect a single default extension for all compiled code,
2658 and there's no compelling reason to try to stretch the standard
2659 to allow two different extensions.) Instead, byte-compiled files
2660 default to the same extension as native-compiled files.
2661 * Fasl file format version numbers have increased again, because
2662 a rearrangement of internal implementation packages made some
2663 dumped symbols in old fasl files unreadable in new cores.
2664 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2665 many fewer weird special cases.
2666 * Bug #17 (differing COMPILE-FILE behavior between logical and
2667 physical pathnames) has been fixed, and some related misbehavior too,
2668 thanks to a patch from Martin Atzmueller.
2669 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2670 by Tim Moore, ported to SBCL by Martin Atzmueller.
2671 * Martin Atzmueller fixed several filesystem-related problems,
2672 including bug #36, in part by porting CMU CL patches, which were
2673 written in part by Paul Werkowski.
2674 * More compiler warnings in src/runtime/ are gone, thanks to
2675 more patches from Martin Atzmueller.
2676 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2679 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2681 * DESCRIBE now works on CONDITION objects.
2682 * The debugger now handles errors which arise when trying to print
2683 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2685 * The build system now uses an additional file, customize-target-features.lisp,
2686 to allow local modifications to the target *FEATURES* list. (The point of
2687 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2688 debugging features enabled, without having to worry about propagating it
2689 into everyone's system when I do a "cvs update".) When no
2690 customize-target-features.lisp file exists, the target *FEATURES* list
2691 should be constructed the same way as before.
2692 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2693 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2694 warn about behavior which is undefined under the ANSI spec. Note: This
2695 is specified by ANSI, but it's not very popular with programmers.
2696 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2697 macro in the SBCL sources for an example of a workaround which you
2698 might use to make portable ANSI-standard code which does what you want.
2699 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2700 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2701 when it's used as a non-toplevel form. (This is required in order
2702 to implement the DEFCONSTANT-EQX macro.)
2703 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2704 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2705 funny things, due to bugs in the implementation of EVAL-WHEN
2706 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2707 non-toplevel DEFCONSTANTs are uncommon.)
2708 * The core file version number and fasl file version number have been
2709 incremented, because the old noncompliant DEFCONSTANT behavior involved
2710 calling functions which no longer exist, and because I also took the
2711 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2712 * fixed bug 1 (error handling before read-eval-print loop starts), and
2713 redid debugger restarts and related debugger commands somewhat while
2715 ** The QUIT debugger command is gone, since it did something
2716 rather different than the SB-EXT:QUIT command, and since it never
2717 worked properly outside the main toplevel read/eval/print loop.
2718 Invoking the new TOPLEVEL restart provides the same functionality.
2719 ** The GO debugger command is also gone, since you can just invoke
2720 the CONTINUE restart directly instead.
2721 ** The TOP debugger command is also gone, since it's redundant with the
2722 FRAME 0 command, and since it interfered with abbreviations for the
2724 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2726 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2727 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2728 some porting work by Martin Atzmueller)
2729 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2730 with a boatload of patches which he ported to SBCL. Now that those
2732 ** The system tries to make sure that its low-priority messages
2733 are prefixed by semicolons, to help people who like to use
2734 syntax highlighting in their ILISP buffer. (This patch
2735 was originally due to Raymond Toy.)
2736 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2737 patches originally written by Raymond Toy.
2738 ** The compiler understands coercion between single-value and
2739 multiple-VALUES type expressions better, getting rid of some very
2740 weird behavior, thanks to patches originally by Robert MacLachlan
2741 and Douglas Crosher.
2742 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2743 lambda lists, thanks to a patch originally by Pierre Mai.
2744 ** The system no longer bogusly warns about "abbreviated type
2746 ** The compiler gets less confused by inlining and RETURN-FROM,
2747 thanks to some patches originally by Tim Moore.
2748 ** The system no longer hangs when dumping circular lists to fasl
2749 files, thanks to a patch originally from Douglas Crosher.
2750 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2751 undefined function error.
2752 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2753 in structure slot initforms) for now, documented workaround instead:-|
2754 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2755 * fixed bug 5: added stubs for various Gray stream functions called
2756 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2757 installed, at least appropriate type errors are generated
2758 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2759 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2760 consistently in DEFMETHOD forms.
2761 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2762 it doesn't seem to affect SBCL after all
2763 * The C runtime system now builds with better optimization and many
2764 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2766 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2768 * The system is now under CVS at SourceForge (instead of the
2769 CVS repository on my home machine).
2770 * The new signal handling code has been tweaked to treat register
2771 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2772 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2773 downstream, e.g. in debug-int.lisp, has implicit dependencies
2774 on the unsignedness of integer representation of machine words,
2775 and that caused the system to bomb out with infinite regress
2776 when trying to recover from type errors involving signed values,
2777 e.g. (BUTLAST '(1 2 3) -1).)
2778 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2779 (This was one of the bugs Peter Van Eynde reported back in July.)
2780 * The system now uses code inspired by Colin Walters' O(N)
2781 implementation of MAP (from the cmucl-imp@cons.org mailing
2782 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2783 inline the MAP operation, and there is more than one
2784 sequence argument to the MAP call (so that it can't just
2785 do ETYPECASE once and for all based on the type of the
2786 single sequence argument). (The old non-inline implementation
2787 of the general M-argument sequence-of-length-N case required
2788 O(M*N*N) time when any of the sequence arguments were LISTs.)
2789 * The QUIT :UNIX-CODE keyword argument has been renamed to
2790 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2791 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2793 * Raymond Wiker's patches to port dynamic loading from Linux to
2794 FreeBSD have been added.
2795 * The BUGS file is now more nearly up to date, thanks in large part
2796 to Martin Atzmueller's review of it.
2797 * The debugger now flushes standard output streams before it begins
2798 its output ("debugger invoked" and so forth).
2799 * The core version number and fasl file version number have both
2800 been incremented, because of incompatible changes in the layout
2802 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2803 used to not be called for a saved Lisp image.)
2804 * Martin Atzmueller's version of a patch to fix a compiler crash,
2805 as posted on sbcl-devel 13 September 2000, has been installed.
2806 * Instead of installing Martin Atzmueller's patch for the
2807 compiler transform for SUBSEQ, I deleted the compiler transform,
2808 and transforms for some similar consing operations.
2809 * A bug in signal handling which kept TRACE from working on OpenBSD
2811 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2812 away by constant folding
2813 * The system now defines its address space constants in one place
2814 (in the Lisp sources), and propagates them automatically elsewhere
2815 (through GENESIS and the sbcl.h file). Therefore, patching the
2816 address map is less unnecessarily tedious and error-prone. The
2817 Lisp names of address space constants have also been systematized.
2818 * CVS tags like dollar-Header-dollar have been removed from
2819 the sources, because they have never saved me trouble and
2820 they've been source of trouble working with patches and other
2821 diff-related operations.
2822 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2823 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2825 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2827 * The system has been ported to OpenBSD.
2828 * The system now compiles with a simple "sh make.sh" on the systems
2829 that it's supported on. I.e., now you no longer need to tweak
2830 text in the target-features.lisp-expr and symlinks in src/runtime/
2831 by hand, the make.sh takes care of it for you.
2832 * The system is no longer so grossly inefficient when compiling code
2833 involving vectors implemented as general (not simple) vectors (VECTOR T),
2834 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2835 which dares to use the various sequence functions on non-simple
2836 vectors, takes less of a performance hit.
2837 * There is now a primitive type predicate VECTOR-T-P
2838 to test for the (VECTOR T) type, so that e.g.
2839 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
2840 can now be compiled with some semblance of efficiency. (The old code
2841 turned the type declaration into a full call to %TYPEP at runtime!)
2842 * AREF on (VECTOR T) is still not fast, since it's still compiled
2843 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
2844 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
2845 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
2846 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
2847 without making full calls to SUBTYPEP at runtime.
2848 (Some analogous efficiency issues for non-simple vectors specialized to
2849 element types other than T, or for non-simple multidimensional arrays,
2850 have not been addressed. They could almost certainly be handled the
2851 same way if anyone is motivated to do so.)
2852 * The changes in array handling break binary compatibility, so
2853 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
2854 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
2855 instead of (VALUES T T).
2856 * By following the instructions that Dan Barlow posted to sbcl-devel
2857 on 2 July 2000, I was able to enable primitive dynamic object
2858 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
2859 functionality is not implemented (since it calls ld to resolve
2860 library references automatically, requiring RUN-PROGRAM for its
2861 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
2862 try to resolve library references) is now supported.
2863 * The system now flushes the standard output streams when it terminates,
2864 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
2865 them at several other probably-convenient times, e.g. in each pass of
2866 the toplevel read-eval-print loop, and after evaluating a form given
2867 as an "--eval" command-line option. (These changes were motivated by a
2868 discussion of stream flushing issues on cmucl-imp in August 2000.)
2869 * The source transform for TYPEP of array types no longer assumes
2870 that an array whose element type is a not-yet-defined type
2871 is implemented as an array of T, but instead punts, so that the
2872 type will be interpreted at runtime.
2873 * There is now some support for cross-compiling in make.sh: each of
2874 the phases of make.sh has its own script. (This should be transparent
2875 to people doing ordinary, non-cross-compile builds.)
2876 * Since my laptop doesn't have hundreds of megabytes of memory like
2877 my desktop machine, I became more motivated to do some items on
2878 my to-do list in order to reduce the size of the system a little:
2879 ** Arrange for various needed-only-at-cold-init things to be
2880 uninterned after cold init. To support this, those things have
2881 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
2882 symbols with such names are now uninterned after cold init).
2883 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
2884 fasl files for cold load.
2885 ** Remove the old compiler structure pooling code (which used to
2886 be conditional on the target feature :SB-ALLOC) completely.
2887 ** Redo the representation of some data in cold init to be more compact.
2888 (I also looked into supporting byte compiled code at bootstrap time,
2889 which would probably reduce the size of the system a lot, but that
2890 looked too complicated, so I punted for now.)
2891 * The maximum signal nesting depth in the src/runtime/ support code has
2892 been reduced from 4096 to 256. (I don't know any reason for the very
2893 large old value. If the new smaller value turns out to break something,
2894 I'll probably just bump it back up.)
2895 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
2897 * Many, many bugs reported by Peter Van Eynde have been added to
2898 the BUGS list; some have even been fixed.
2899 * While enabling dynamic object file loading, I tried to make the
2900 code easier to understand, renaming various functions and variables
2901 with less ambiguous names, and changing some function calling
2902 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
2903 * While trying to figure out how to do the OpenBSD port, I tried to
2904 clean up some of the code in src/runtime/. In particular, I dropped
2905 support for non-POSIX signal handling, added various comments,
2906 tweaked the code to reduce the number of compilation warnings, and
2907 renamed some files to increase consistency.
2908 * To support the new automatic configuration functionality in make.sh,
2909 the source file target-features.lisp-expr has been replaced with the
2910 source file base-target-features.lisp-expr and the machine-generated
2911 file local-target-features.lisp-expr.
2912 * fixed a stupid quoting error in make.sh so that using CMU CL
2913 "lisp -batch" as cross-compilation host works again
2915 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
2917 * DESCRIBE no longer tries to call itself recursively to describe
2918 bound/fbound values, so that it no longer fails on symbols which are
2919 bound to themselves (like keywords, T, and NIL).
2920 * DESCRIBE now works on generic functions.
2921 * The printer now prints less-screwed-up representations of closures
2922 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
2923 * A private symbol is used instead of the :EMPTY keyword previously
2924 used to mark empty slots in hash tables. Thus
2925 (DEFVAR *HT* (MAKE-HASH-TABLE))
2926 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
2927 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
2928 now does what ANSI says that it should. (You can still get
2929 similar noncompliant behavior if bang on the hash table
2930 implementation with all the symbols you get back from
2931 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
2932 This breaks binary compatibility, since tests for equality to
2933 :EMPTY are wired into things like the macroexpansion of
2934 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
2936 * There's now a minimal placeholder implementation for CL:STEP,
2937 as required by ANSI.
2938 * An obscure bug in the interaction of the normal compiler, the byte
2939 compiler, inlining, and structure predicates has been patched
2940 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
2941 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
2943 * Missing ordinary arguments in a macro call are now detected even
2944 when the macro lambda list contains &KEY or &REST.
2945 * The debugger no longer complains about encountering the top of the
2946 stack when you type "FRAME 0" to explicitly instruct it to go to
2947 the top of the stack. And it now prints the frame you request even
2948 if it's the current frame (instead of saying "You are here.").
2949 * As specified by ANSI, the system now always prints keywords
2950 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
2952 * The default initial SIZE of HASH-TABLEs is now smaller.
2953 * Type information from CLOS class dispatch is now propagated
2954 into DEFMETHOD bodies, so that e.g.
2955 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2957 is now basically equivalent to
2958 (DEFMETHOD FOO ((X SINGLE-FLOAT))
2959 (DECLARE (TYPE SINGLE-FLOAT X))
2961 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
2962 operation, without having to do run-time type dispatch.
2963 * The macroexpansion of DEFMETHOD has been tweaked so that it has
2964 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
2965 * Since I don't seem to be making big file reorganizations very often
2966 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
2967 is overflowing my ability to conveniently back them up), I've finally
2968 checked the system into CVS. (The CVS repository is on my home system,
2969 not at SourceForge -- putting it on SourceForge might come later.)
2970 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
2971 high-level GC-NOTIFY-FOO functions send their output. (There's
2972 still very little control of where low-level verbose GC functions
2973 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
2974 less than it used to -- the GC-NOTIFY-FOO functions are now under
2975 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
2976 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
2977 in only one place in the source code, and propagates it automatically
2978 everywhere that it's needed. Thus e.g. when I bump the version from
2979 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
2980 * The C source files now include boilerplate legalese and documentation
2981 at the head of each file (just as the Lisp source files already did).
2982 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
2983 to his page will be replaced with a link to his new CLiki service.
2985 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
2987 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
2988 * The build process now looks for GNU make under the default name "gmake",
2989 instead of "make" as it used to. If GNU make is not available as "gmake"
2990 on your system, you can change this default behavior by setting the
2991 GNUMAKE environment variable.
2992 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
2993 can build without error under CMU CL.
2995 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
2997 * There is now a partial SBCL user manual (with some new text and some
2998 text cribbed from the CMU CL manual).
2999 * The beginnings of a profiler have been added (starting with the
3000 CMU CL profiler and simplifying and cleaning up). Eventually the
3001 main interface should be through the TRACE macro, but for now,
3002 it's still accessed through vaguely CMU-CL-style functions and macros
3003 exported from the package SB-PROFILE.
3004 * Some problems left over from porting CMU CL to the new
3005 cross-compilation bootstrap process have been cleaned up:
3006 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3007 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3008 part when getting rid of the compiler *BACKEND* stuff.)
3009 ** The value of *NULL-TYPE* was screwed up, because it was
3010 being initialized before the type system knew the final
3011 definition of the 'NULL type. This screwed up several key
3012 optimizations in the compiler, causing inefficiency in all sorts
3013 of places. (I found it because I wanted to understand why
3014 GET-INTERNAL-RUN-TIME was consing.)
3015 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3016 PROCLAIM FTYPE information. Unfortunately this broke binary
3017 compatibility again, since now the forms output by DEFGENERIC
3018 to refer to functions which didn't exist in 0.6.3.
3019 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3020 can use the new (as of 0.6.3) transform for SOME into MAP into
3022 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3023 values to (INTEGER 0 1000000), so that the time code will no longer
3024 occasionally get blown up by Linux returning 1000000 microseconds
3025 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3026 its output conform to the ANSI spec. (Alas, this makes its output
3027 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3028 * A full call to MAP NIL with a single sequence argument no longer conses.
3029 * fixes to problems pointed out by Martin Atzmueller:
3030 * The manual page no longer talks about multiprocessing as though
3031 it were currently supported.
3032 * The ILISP support patches have been removed from the distribution,
3033 because as of version 5.10.1, ILISP now supports SBCL without us
3034 having to maintain patches.
3035 * added a modified version of Raymond Toy's recent CMU CL patch for
3036 EQUALP comparison of HASH-TABLE
3038 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3040 * The system still can't cross-compile itself with
3041 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3042 entails), but at least it can compile more of itself that way
3043 than it used to be able to, and various buglets which were uncovered
3044 by trying to cross-compile itself that way have now been fixed.
3045 * This release breaks binary compatibility again. This time
3046 at least I've incremented the FASL file format version to 2, so that the
3047 problem can be detected reliably instead of just causing weird errors.
3048 * various new style warnings:
3049 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3050 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3051 ** using the lexical binding of a variable named in the *FOO* style
3052 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3053 as specified by ANSI.
3054 * *RANDOM-STATE* is no longer automatically initialized from
3055 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3056 default behavior of the system is to repeat its behavior every time
3057 it's run. If you'd like to change this behavior, you can always
3058 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3059 old convention there was no comparably easy way to get the system to
3060 repeat its behavior every time it was run.
3061 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3062 FTYPE declarations has been removed, in favor of their ANSI
3063 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3064 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3065 the types of their sequence arguments can be determined at compile time.
3066 This is done through a new open code expansion for MAP which eliminates
3067 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3068 argument types can be determined at compile time.
3069 * The optimizer now transforms COERCE into an identity operation when it
3070 can prove that the coerced object is already of the correct type. (This
3071 can be a win for machine generated code, including the output of other
3072 optimization transforms, such as the MAP transform above.)
3073 * Credit information has been moved from source file headers into CREDITS.
3074 * Source file headers have been made more standard.
3075 * The CASE macro now compiles without complaining even when it has
3078 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3080 * (Note that the way that the PCL macroexpansions were rewritten
3081 to accommodate the change in DEFGENERIC below breaks binary
3082 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3083 not run under sbcl-0.6.2. Once we get out of alpha releases,
3084 i.e. hit release 1.0.0, we'll probably try to maintain binary
3085 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3086 and sbcl-1.4.4. Until then, however, it might be fairly common
3087 for maintenance releases to break binary compatibility.)
3088 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3090 * The reporting of unhandled signals has been changed to print some
3091 explanatory text as well as the report form. (Previously only
3092 the report form was printed.)
3093 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3094 it defines, so that the compiler no longer issues undefined function
3095 warnings for compiled-but-not-yet-loaded generic functions.
3096 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3097 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3098 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3099 section of the ANSI spec are supported.
3100 * The "" nickname for the "KEYWORD" package has been removed.
3101 The reader still handles symbol tokens which begin with a package marker
3102 as keywords, but it doesn't expose its mechanism for doing so in the
3103 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3104 * The system now issues STYLE-WARNINGs for contradictory TYPE
3105 proclamations. (Warnings for contradictory FTYPE proclamations would
3106 be nice too, but those can't be done usefully unless the type system
3107 is made smarter about FUNCTION types.)
3108 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3109 systematized, so that "*target-*.lisp is supposed to exist only on the
3110 target and imply that there's a related file which exists on the
3111 host, and *host-*.lisp is supposed to exist only on the host and imply
3112 that there's a related file which exists on the target. This involves a
3113 lot of renaming. Hopefully the acute confusion caused by the renaming
3114 will be justified by the reduction in chronic confusion..
3115 ** runtime-type.lisp -> early-target-type.lisp
3116 ** target-type.lisp -> late-target-type.lisp
3117 ** early-host-format.lisp -> early-format.lisp
3118 ** late-host-format.lisp -> late-format.lisp
3119 ** host-error.lisp -> misc-error.lisp
3120 ** early-error.lisp -> early-target-error.lisp
3121 ** late-error.lisp -> late-target-error.lisp
3122 ** host-defboot.lisp -> early-defboot.lisp
3123 ** code/misc.lisp -> code/target-misc.lisp
3124 ** code/host-misc.lisp -> code/misc.lisp
3125 ** code/numbers.lisp -> code/target-numbers.lisp
3126 ** code/early-numbers.lisp -> numbers.lisp
3127 ** early-host-type.lisp -> early-type.lisp
3128 ** late-host-type.lisp -> late-type.lisp
3129 ** host-typep.lisp -> typep.lisp
3130 ** load.lisp -> target-load.lisp
3131 ** host-load.lisp -> load.lisp
3132 ** host-disassem.lisp -> disassem.lisp
3133 ** host-insts.lisp -> insts.lisp
3134 ** byte-comp.lisp -> target-byte-comp.lisp
3135 ** host-byte-comp.lisp -> byte-comp.lisp
3136 ** host-signal.lisp -> signal.lisp
3137 ** host-defstruct.lisp -> defstruct.lisp
3138 ** late-target-type.lisp -> deftypes-for-target.lisp
3139 Furthermore, several other previously target-only files foo.lisp (e.g.
3140 hash-table.lisp and random.lisp) have been split into a target-and-host
3141 foo.lisp file and a target-only target-foo.lisp file, with their key type
3142 definitions in the target-and-host part, so that the cross-compiler will
3143 know more about target types.
3144 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3145 gone away. In their place are various *BACKEND-FOO* variables
3146 corresponding to the slots of the old structure.
3147 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3148 into the target SBCL has been fixed.
3149 * The chill.lisp system for loading cold code into a running SBCL
3151 * Support for the CMU CL "scavenger hook" extension has been removed.
3152 (It was undocumented and unused in the CMU CL sources that SBCL was
3153 derived from, and stale in sbcl-0.6.1.)
3154 * Various errors in the cross-compiler type system were detected
3155 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3156 (enabling various consistency checks). Many of them were fixed,
3157 but some hard problems remain, so the compiler is back to
3158 running without *TYPE-SYSTEM-INITIALIZED* for now.
3159 * As part of the cross-compiler type system cleanup, I implemented
3160 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3161 * I have started adding UNCROSS calls throughout the type system
3162 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3163 on cross-compiler input files will be able to go away, and various
3165 * CONSTANTP now returns true for quoted forms (as explicitly required
3168 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3170 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3171 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3172 (MAKE-LIST -1). In the longer term, it ought to become true
3173 that declarations are assertions even at SAFETY 1. For now, it's not
3174 quite true even at SAFETY 3, but it's at least more nearly true..
3175 (Note that this change seems to increases the size of the system by
3176 O(5%) and to decrease the speed of the compiler by 20% or more.)
3177 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3178 for the problem of printing dozens of lines of distracting information
3179 about low-level system machinery as part of the top stack frame
3180 on entry to the debugger when an undefined function was called.
3181 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3182 is not reset to COMMON-LISP-USER.
3183 * Compilation of stuff related to dyncount.lisp has been made conditional
3184 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3185 smaller. The various dyncount-related symbols have been moved into
3186 a new "SB-DYNCOUNT" package.
3187 * tty-inspect.lisp has been renamed to inspect.lisp.
3188 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3189 feature has gone away. (When we eventually port to other flavors of
3190 libc and/or Unix, we'll try to make the differences between flavors
3191 invisible at the user level.)
3192 * Various other *FEATURES* tags, and/or their associated conditionals,
3193 have been removed if obsolescent, or given better documentation, or
3194 sometimes given more-mnemonic names.
3196 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3198 * tidied up "make.sh" script
3199 * tidied up system directory structure
3200 * better "clean.sh" behavior
3201 * added doc/FOR-CMUCL-DEVELOPERS
3202 * many many small tweaks to output format, e.g. removing possibly-confusing
3203 trailing #\. character in DESCRIBE-INSTANCE
3204 * (EQUALP #\A 'A) no longer signals an error.
3205 * new hashing code, including EQUALP hashing
3206 * tidied up Lisp initialization and toplevel
3207 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3208 * command line argument processing
3209 * added POSIX-GETENV function to deal with Unix-ish environment variables
3210 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3211 terminating SBCL on EOF
3212 * non-verbose GC by default
3213 * There is no more "sbcl" shell script; the sbcl file is now the C
3214 runtime executable (just like CMU CL).
3215 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3217 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3218 symbol and package information as Lisp native symbol and package objects
3219 instead of strings naming symbols and strings naming packages. This way,
3220 whenever packages are renamed (as in warm init), debug information is
3221 transformed along with everything else.
3222 * tweaked the optimization policy declarations which control the building
3223 of SBCL itself. Now, among other things, the system no longer saves
3224 source location debugging information. (This helps two problems at once
3225 by reducing SBCL size and by keeping SBCL from trying to look for its
3226 sources -- which may not exist -- when reporting errors.)
3227 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3228 debugging and testing purposes
3229 * cleaned up printing, making the printer call PRINT-OBJECT for
3230 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3231 methods, giving nearly-ANSI behavior
3232 * converted almost all special variables to use *FOO* naming convention
3233 * deleted PARSE-TIME functionality, since it can be done portably
3234 * moved some files out of cold init into warm init
3235 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3237 * regularized formatting of source files
3238 * added an install.sh script
3239 * fixed ridiculous memory usage of cross-compiler by making
3240 compiler/alloc.lisp not try to do pooling unless it can hook
3241 itself into the GC of the cross-compilation host. Now the system
3242 builds nicely on my old laptop.
3243 * added :SB-ALLOC in target-features.lisp-expr
3244 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3245 was not implemented)
3246 * re-did condition handling and note reporting in the compiler. Notes
3247 are no longer handled by signalling conditions. Style warnings
3248 and warnings are handled more correctly and reported in such a way
3249 that it's easy to find one or the other in your output (so that you
3250 can e.g. figure out which of many problems caused COMPILE-FILE to
3252 * changed the severity of several compiler warnings from full WARNING
3253 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3254 compiler note reporting so that it doesn't use the condition system
3255 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3256 COMPILE-FILE command)
3257 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3258 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3259 no longer supported, since their implementation was deeply intertwingled
3260 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3261 * removed bogus "support" for compiler macros named (SETF FOO), and
3262 removed the compiler macro for SETF INFO (but only after making a fool
3263 of myself on the cmucl-imp mailing list by posting a bogus patch for
3264 DEFINE-COMPILER-MACRO..)
3265 * Compiled files containing forms which have side effects on the Lisp
3266 reader (such as DEFPACKAGE forms) are now handled more correctly.
3267 (Compiler queuing of top level lambdas has been suppressed by setting
3268 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3269 * deleted various currently-unused source files, e.g. gengc.lisp. They
3270 may be added back at some point e.g. when porting to other architectures,
3271 but until they are it's distracting to distribute them and to try to
3273 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3274 there were so many of them they're just distractions, and UNCROSS is
3275 known to be able to handle the current sources
3276 * moved PROFILE functionality into TRACE, so that it will be clear
3277 how the wrapping and unwrapping of functions when you profile them
3278 interacts with the wrapping and unwrapping of functions when you
3279 trace them. (Actually, the functionality isn't there yet, but at least
3280 the interface specification is there. Hopefully, the functionality will
3281 arrive with some future maintenance release.)
3282 * removed host-oops.lisp
3283 * changed signature of QUIT function to allow UNIX-CODE argument
3284 * fixed READ-SEQUENCE bug
3285 * tweaked verbose GC output so that it looks more like the progress
3286 output that ANSI specifies for functions like LOAD
3287 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3288 * added <http://sbcl.sourceforge.com> to the banner information printed by