1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
3 * new feature: source path information is generated for macro-expansion
4 errors for use in IDE's like Slime (thanks to Helmut Eller)
6 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
7 * minor incompatible change: in sbcl-0.9.11 (but not earlier
8 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
9 into subclasses of the finalized class. Now user calls to
10 FINALIZE-INHERITANCE finalize just the one class, and calls by the
11 system return before any subclasses are finalized.
12 * minor incompatible change: The reader no longer ignores errors
13 regarding non-existent packages in #+ and #- feature tests.
14 * new feature: command line options --no-sysinit, --no-userinit to
15 inhibit loading the corresponding init files
16 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
17 for sending data through UDP sockets (thanks to François-René Rideau)
18 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
19 error is signalled instead (thanks to François-René Rideau)
20 * minor incompatible change: Error signalling behaviour of lexical
21 operations violating package locks has changed slightly. Refer to
22 documentation on package locks for details.
23 * enhancement: EVAL can process IF-expressions without resorting to the
25 * bug fix: Some operations involving SIN, COS, or TAN weren't being
26 constant-folded properly due to missing out-of-line functions.
27 (reported by Mika Pihlajamäki)
28 * bug fix: LISTEN sometimes returned T even in cases where no data was
29 immediately available from the stream
30 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
31 were reversed. (reported by Levente Mészáros)
32 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
33 when given a vector displaced to an adjustable array. (reported
35 * bug fix: derivation of float boundaries from numbers outside the
36 appropriate float range (reported by John Wiseman)
37 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
39 * bug fix: precomputing cache entries for generic functions with
40 some subclasses of specializers as yet invalid does not attempt to
41 fill a cache line with a negative offset. (reported by Levente
43 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
44 allow condition class objects as arguments to DOCUMENTATION and
45 (SETF DOCUMENTATION); only find and set documentation for
46 structure names for the STRUCTURE doc-type. (suggested by Gary
48 * improvements to the Win32/x86 port:
49 ** (user-homedir-pathname) and default initialization file
50 locations now know about the user's "Documents and Settings"
51 directory (thanks to Yaroslav Kavenchuk)
52 ** run-program is implemented (thanks to Mike Thomas)
53 ** sockets support (thanks to Timothy Ritchey)
54 ** better backtrace support (thanks to Alastair Bridgewater)
55 ** sb-grovel supported
56 ** asdf-install and sb-posix work somewhat
57 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
58 * improvements to the Solaris/x86 port:
59 ** works on Solaris 11/Solaris Express
60 ** floating-point exception handling support
61 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
62 * fixed some bugs revealed by Paul Dietz' test suite:
63 ** REMOVE-METHOD returns its generic function argument even when
64 the method is not one of the generic functions' methods.
65 ** objects declared as MEMBER types can be admissible arguments to
67 ** printing characters should simply be printed by the FORMAT ~:C
69 ** compiler failure when compiling functions with hairy constant
70 defaults for optional parameters.
71 ** compiler produces wrong code when MAYBE-INLINE-expanding a
72 function, which is already optimized.
74 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
75 * new platform: experimental support for SBCL x86/Darwin, including
76 MacOS X 10.4.5 on Intel.
77 * new feature: Unicode character names are now known to the system
78 (through CHAR-NAME and NAME-CHAR).
79 * new feature: the filesystem location of SBCL's core file is
80 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
81 * minor incompatible change: the contrib modules SB-POSIX and
82 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
83 this change is to make it easier to distribute
84 location-independent binaries.
85 * bug fix: as implied by AMOP, standardized classes no longer have
86 slots named by external symbols of public packages. (reported by
88 * optimization: calling functions via constant symbols -- as in
89 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
90 via the function object as in (FUNCALL #'FOO).
91 * optimization: on x86 and x86-64, the calling convention for the
92 case of dealing with an unknown number of values has been altered
93 to be friendlier to the prediction heuristics implemented,
94 particularly with reference to CALL and RET pairing. (thanks to
96 * optimization: on x86, the code for access to thread-local symbol
97 values and binding/unbinding of thread-local symbols is smaller.
98 (thanks to Alastair Bridgewater)
99 * enhancement: CONSTANTP is now able to determine constantness of
100 more complex forms, including calls to constant-foldable standardized
101 functions and some special forms beyond QUOTE.
102 * fixed bug: occasional GC crashes on Solaris/x86
103 * optimization: x86-64 supports stack allocation of results of simple
104 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
105 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
106 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
107 (thanks to James Knight)
108 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
109 by the standard (thanks to Stephen Compall)
111 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
112 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
113 be used for bundling the runtime and the core file into one
114 executable binary. This feature is not currently supported on all SBCL
115 platforms. (thanks to James Bielman and NIIMI Satoshi)
116 * new feature: a generational or ephemeral garbage collector is now
117 the default on the PowerPC platform (both Linux and Darwin). The
118 old Cheney (stop and copy) collector is a build-time option.
119 (thanks to Cyrus Harmon, after Raymond Toy)
120 * minor incompatible change: the method by which SBCL finds its
121 contributed modules has changed; it no longer relies on symbolic
122 links from an $SBCL_HOME/systems directory, but searches directly
123 in the subdirectories of $SBCL_HOME.
124 * enhancement: the dynamic heap size on the Linux/PPC platform is
125 markedly larger, even using the older Cheney garbage collector.
126 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
127 certain complicated string types. (reported by Gary King)
128 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
129 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
130 * fixed bug: DOCUMENTATION on structures no longer signals an
131 error if no structure type of the right name exists. (reported by
133 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
134 applicable non-standard primary or :AROUND method on
135 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
136 no longer cause unbound variable errors. (reported by Kevin Reid)
137 * fixed bug: metacircle resolution in cases where methods have slots
138 added before the slots from STANDARD-METHOD. (reported by Jean
140 * fixed bug: the Power PC assembly code for calling into lisp
141 sometimes computed its return address wrongly (depending on the
142 15th bit of the address, fixed at link time). This used to
143 manifest itself in a segmentation violation while building PCL.
144 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
145 many others over the years)
146 * fixed bug: ROOM no longer reports silly numbers for stack usage if
147 the stack pointer (treated as a signed integer) is negative.
148 (thanks to Peter van Eynde)
150 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
151 * new platform: experimental support for the Windows operating
152 system has been added. (thanks to Alastair Bridgewater)
153 * fixed several bugs in and robustified the PPC FFI (including
154 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
155 * bug fix: allow non-simple string symbol names (reported by Paul
157 * bug fix: interrupt handling on NetBSD (thanks to Richard M
159 * bug fix: saving a core corrupted callbacks on x86/x86-64
160 * bug fix: closed a loophole in metacircularity detection and
161 grounding in the PCL implementation of CLOS.
162 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
163 (reported by tomppa on #lisp)
164 * bug fix: clear the direction flag on Lisp -> C transitions, as
165 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
166 (reported by Andrej Grozin and Hendrik Maryns)
167 * optimization: major improvements to GC efficiency on GENCGC platforms
168 * optimization: faster implementation of EQUAL
169 * optimization: emit more efficient opcodes for some common
170 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
171 * optimization: several other minor code-generation improvements on x86-64
172 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
174 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
175 * minor incompatible change: (SETF CLASS-NAME) and (SETF
176 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
177 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
178 by ANSI as a generic function, and for consistency (SETF
179 GENERIC-FUNCTION-NAME) is treated likewise.
180 * fixed bug #233.b: make constraint propagation notice when a variable
181 value is changed after it is referenced but before it is used
182 * fixed bug #296: no more arbitrary behaviour from filenames with
183 odd characters as --load arguments. (reported by Adam Warner)
184 * fixed bug #390: :CHARACTER-SET pathname components now work as
185 expected. (reported by Tim Daly Jr)
186 * fixed bug #391: complicated :TYPE intersections in slot
187 definitions no longer cause an error in PCL internals.
188 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
189 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
190 returning the number of octets which would be written to the
191 file-stream. (thanks to Robert J. Macomber)
192 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
193 arguments to the internal stream functions. (thanks to David
195 * bug fix: stack exhaustion is now handled on the x86/SunOS
196 platform. (thanks to Daisuke Homma)
197 * bug fix: more accurate ROOM results on GENCGC platforms
198 * optimization: improved type inference for arithmetic-for
199 index variables in LOOP
200 * optimization: faster floating-point SQRT on x86-64
201 * fixed some bugs revealed by Paul Dietz' test suite:
202 ** DOCUMENTATION returns NIL instead of "" for method combinations
203 that don't have a docstring
205 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
206 * minor incompatible change: (SETF CLASS-NAME) and (SETF
207 GENERIC-FUNCTION-NAME) are no longer generic functions, and
208 therefore are not customizeable by user code (as seems to be at
209 least permitted and maybe required by AMOP). As a consolation,
210 however, the SBCL implementation of these functions now calls
211 REINITIALIZE-INSTANCE as specified by AMOP.
212 * bug fix: slot-definition documentation is propagated as per ANSI
213 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
214 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
215 argument, as do eslotd initialization methods. (from Pascal
216 Costanza's "Closer" project)
217 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
218 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
220 * bug fix: it is now possible to have more than one subclass of
221 STANDARD-GENERIC-FUNCTION without causing stack overflow.
222 (reported by Bruno Haible, Pascal Costanza and others)
223 * bug fix: the dependent update protocol now works for generic
224 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
226 * bug fix: condition-class instances corresponding to
227 DEFINE-CONDITION forms are now created eagerly. (reported by
228 Kalle Olavi Niemitalo on comp.lang.lisp)
229 * bug fix: floating point printing is more accurate in some
230 circumstances. (thanks to Simon Alexander)
231 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
232 merged with *DEFAULT-PATHNAME-DEFAULTS*.
233 * bug fix: callbacks on OS X now preserve stack-alignment.
234 (thanks to Cyrus Harmon)
235 * enhancement: the x86-64 disassembler is much better at
236 disassembling SSE instructions. (thanks to Lutz Euler)
237 * enhancement: CHECK-TYPE on a variable now allows type-inference
238 (thanks to Vincent Arkesteijn)
239 * optimization: improved performance of EUC-JP external format.
240 (thanks to NIIMI Satoshi)
241 * optimization: performance improvements to IO on file streams of
242 :ELEMENT-TYPE CHARACTER
243 * optimization: much faster memory allocation on x86-64
244 * optimization: faster garbage collections (latency and throughput)
246 * optimization: faster bignum subtraction and division on x86 and x86-64
247 (thanks to Lutz Euler)
248 * optimization: more accurate type-inference for loops that use a
249 floating point index variable or a negative step.
251 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
252 * bug fix: add a workaround to SBCL looping infinitely at startup on
253 Linux kernels with apparently buggy implementations of personality().
254 (thanks to Svein Ove Aas)
255 * bug fix: Unicode symbols are correctly printed in LDB backtraces
256 (thanks to David Lichteblau)
257 * bug fix: local bindings shadow global symbol macros for
258 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
259 * optimization: non-open coded uses of numeric comparison operators
260 (e.g. >) no longer cons when called with more than one parameter
261 on platforms supporting dynamic-extent allocation.
262 * enhancement: saving cores with foreign code loaded is now supported
263 on MIPS/Linux in addition to the previously supported platforms.
264 * bug fix: FIND-RESTART now tests for activity, not applicability when given
265 a restart object as identifier. (reported by Helmut Eller for CMUCL)
266 * bug fix: division by zero in sb-sprof when no samples were collected
267 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
268 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
269 (thanks to Alexander Kjeldaas)
270 * new feature: ignoring the return values of destructive functions
271 like NREVERSE signals a compile-time style-warning.
272 (thanks to Kevin Reid)
273 * new feature: the alignment of alien structure fields can be
274 explicitly specified. (thanks to Cyrus Harmon)
275 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
276 and dump core on SIGQUIT
278 ** incompatible change: threads do not inherit values of specials
279 from their parents (see manual)
280 ** bug fix: threads stacks belonging to dead threads are freed by the
281 next exiting thread, no need to gc to collect thread stacks anymore
282 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
283 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
284 ** bug fix: dynamic variable and thread start related gc lossage
285 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
287 * fixed some bugs revealed by Paul Dietz' test suite:
288 ** SUBTYPEP is slightly more accurate on heinously complicated
289 CONS types where some of the members have uncertain (in the
290 NIL, NIL sense) type relationships to each other.
292 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
293 no longer purified unless :PURIFY T is explicitly specified.
294 ** Non-purified cores are significantly smaller than before
296 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
297 * new feature: timers based on Zach Beane's excellent timer package
298 * added support for the following external formats: koi8-u,
299 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
300 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
301 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
302 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
303 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
304 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
305 * incompatible change: a threaded SBCL will no longer revert to
306 non-threaded mode on non-NPTL systems, but refuse to start entirely.
307 * bug fix: interrupts are disabled until startup is complete; no
308 more sigsegvs when receiving a signal to soon
309 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
311 * bug fix: add a workaround for the memory randomization features in
312 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
313 workaround will only be in effect on systems with the proc filesystem
315 * bug fix: printing objects of type HASH-TABLE signals a
316 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
318 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
319 have been running for over 50 days. (reported by Gilbert Baumann)
320 * bug fix: the logic for getting names of functions gets less
321 confused when confronded with alternate-metaclass
322 funcallable-instances. (reported by Cyrus Harmon)
323 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
324 consistent, even on internal alternate-metaclass objects.
325 * bug fix: flush closure information collected by physical
326 environment analysis prepass before the main pass. (bug reported
328 * bug fix: compiler pack phase does not modify a hashtable, which is
329 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
330 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
331 may be expanded. (reported by Alexander <malishev> on c.l.l)
332 * bug fix: no more style-warnings when compiling code using LAST
333 under high-SPEED low-SPACE settings. (reported by David Wallin
335 * fixed some incompatibilities between SBCL's MOP and the MOP
337 ** the METAOBJECT class is now implemented;
338 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
339 STANDARD-OBJECT, as required;
340 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
341 now compatible (as required by VALIDATE-SUPERCLASS); there
342 remains a consistency requirement over the presence or absence
343 of the FUNCTION class in the superclass list of finalized
344 classes; see the manual for more details;
345 ** the :AROUND method for COMPUTE-SLOTS on
346 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
347 requested slot ordering.
349 ** bug fix: parent thread now can be gc'ed even with a live
351 ** bug fix: binding a special with PROGV to no value is not
352 the same as that symbol not having been bound (thanks to
354 ** bug fix: binding specials is thread safe (thanks to
356 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
357 and SB-SPROF do not die with 'no handler for signal XX in
358 interrupt_handle_now(..)' anymore
359 ** bug fix: WITH-TIMEOUT works with multiple running threads
360 * fixed some bugs revealed by Paul Dietz' test suite:
361 ** ENSURE-GENERIC-FUNCTION should take a method class object for
362 the :method-class keyword argument.
364 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
365 * new port: the Solaris operating system on x86 processors is now
366 mostly supported, though some rough edges in the environment
367 remain. (thanks to Daisuke Homma)
368 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
369 space for its heap. (thanks to Thiemo Seufer)
370 * optimization: SBCL on MIPS platforms now supports dynamic-extent
371 closures. (thanks to Thiemo Seufer)
372 * minor incompatible change: eof selects abort in the debugger.
373 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
374 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
376 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
377 deprecated, and will go away in a future revision of SBCL.
378 * minor incompatible change: GC-ON and GC-OFF are no longer
379 implemented with a counter, it does not matter how many times gc
380 is switched on or off
381 * bug fix: discriminating functions for generic function classes
382 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
383 make invalid assumptions about method precedence order. (reported
385 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
386 platforms. (thanks to Thiemo Seufer)
387 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
388 * bug fix: correct pathname printing with printer escaping is on.
389 (thanks to Kevin Reid)
390 * bug fix: complex VOP definitions in "user-space" no longer trigger
391 package locks. (reported by Zach Beane)
392 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
393 for funcallable-instance functions, and is no different from
395 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
396 characters in the current readtable. (reported by Nicholas Neuss)
397 * bug fix: SUBTYPEP on various CONS types returns more a more
398 accurate acknowledgment of its certainty.
399 * optimizations: REMOVE-DUPLICATES now runs in linear time on
400 lists in some cases. This partially fixes bug 384.
401 * flush all standard streams before prompting in the REPL and the
403 * bug fix: signal handling and triggering gc do not conflict
404 directly anymore, in particular a high frequency sb-sprof does
405 not prevent gc from running
406 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
407 approximation for timezone and DST information between the
408 universal time epoch and the smallest negative 32-bit time_t.
409 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
410 year 1899 so long as the encoded time is non-negative.
411 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
412 the correct answer for non-constant rotate amounts when performing
413 an inline 32-bit rotation.
415 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
416 there is only one thread in the session
417 ** bug fix: memory leak for streams created in one thread and
418 written to in another
419 ** bug fix: lockup when compiled with gcc4
420 ** bug fix: race that allows the gc to be triggered when gc is
422 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
423 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
425 ** bug fix: finished threads are gc'ed properly
426 * fixed some bugs revealed by Paul Dietz' test suite:
427 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
428 has arguments with a different set of applicable methods from
429 the orignal arguments.
430 ** The type error thrown by MAP now has a correct expected-type
432 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
433 name a compiled function.
434 ** Three MISC tests where a large bignum was improperly coerced to
435 a float that couldn't represent that bignum during type
436 derivation were fixed.
437 ** SUBTYPEP can now handle types involving the negation of a
438 list-form FUNCTION type.
439 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
440 as (COMPLEX (AND RATIO (NOT FIXNUM))).
441 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
443 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
444 * New feature: Experimental support for bivalent streams: streams
445 opened with :element-type :default now allow character and binary
446 (unsigned-byte 8) I/O
447 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
448 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
450 * bug fix: no more highly sporadic "couldn't check whether ~S is
451 readable" when reading a stream and an interrupt hits in the middle
452 of a select system call
453 * compiler better recognizes complex arrays (reported by Eduardo
455 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
456 for variables on SBCL built with linkage-tables. (reported by Luis
458 * various error reporting improvements.
459 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
460 (Thanks to Thiemo Seufer)
461 * bug fix: nested reader invokations work correctly
462 * bug fix: it is possible to have simultaneous references to foreign
463 code and foreign data with the same name.
465 ** added x86-64 support
466 ** incompatible change: the threading api now works with thread
467 objects instead of thread ids
468 ** bug fix: threads are protected from signals and interruption when
469 starting up or going down
470 ** bug fix: a race where an exiting thread could lose its stack to gc
471 ** bug fix: don't halt on infinite error in threads if possible
472 ** fixed numerous gc deadlocks introduced in the pthread merge
473 ** bug fix: fixed thread safety issues in read and print
474 ** bug fix: debugger doesn't hang on session lock if interrupted at
475 an inappropriate moment
476 ** bug fix: run-program is now thread safe(r)
477 ** bug fix: inner with-recursive-lock no longer releases the mutex
478 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
479 * fixed some bugs revealed by Paul Dietz' test suite:
480 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
481 EXPECTED-TYPE slots filled.
482 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
484 ** MAKE-CONDITION accepts classes as type-designators.
485 ** COMPILE may never return NIL.
486 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
487 range before calling Unix time functions
489 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
490 * numerous signal handling fixes to increase stability
491 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
492 * minor incompatible change: we now correctly canonize default
493 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
494 as per the MOP, rather than the historical (INITARG INITFUNCTION
495 INITFORM). (reported by Bruno Haible)
496 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
497 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
498 for more information.
499 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
500 pathname is a directory pathname.
501 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
502 * dynamic space size on PPC has been increased to 768Mb. (thanks to
504 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
505 :METACLASS argument in addition to a class name. (reported by
506 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
507 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
508 (reported by Bruno Haible; patch from Gerd Moellmann)
509 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
511 * bug fix: more cleanups to the floating point exception handling on
512 x86-64 (thanks to James Knight)
513 * bug fix: the compiler does not try to constant fold calls of
514 COERCE and BIT-* functions when they return freshly constructed
515 objects. (reported by Nikodemus Siivola and Paul Dietz)
516 * optimization: improved the MIPS versions of generic (in the
517 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
518 * optimization: direct conversion of (unsigned-byte 32) to floats on
519 the PowerPC platform.
520 * optimization: structure instances with raw slots now use less
521 memory, and probably show better memory locality. (thanks to
523 * optimization: DYNAMIC-EXTENT declarations for lists and closures
524 are treated as requests for stack allocation on the x86-64,
525 Alpha-32, PPC and SPARC platforms.
526 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
527 the wrong file descriptor; implementation of SOCKET-OPEN-P.
528 (thanks to Tony Martinez)
530 ** gcing a dead thread can no longer lead to lockups
531 ** threads block signals until they are set up properly
532 ** errno is no longer shared by threads
533 ** interrupt-thread restores the eflags register on x86
534 ** fixed some lockups due to gc/thread interaction
535 * fixed some bugs revealed by Paul Dietz' test suite:
536 ** invalid dotted lists no longer raise a read error when
538 ** signal an error if a symbol that names a declaration is used
539 as the name of a type, or vice versa
540 ** allow using the (declare (typespec var*)) abbreviation for
541 (declare (type typespec var*)) with all type specifiers
542 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
543 ** always return NIL from PPRINT-POP when OBJECT is NIL
544 ** don't signal errors when pretty-printing malformed LABELS,
545 FLET or MACROLET forms
546 ** declarations in a DOLIST body are also in scope for the
548 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
550 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
553 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
554 * fixed cross-compiler leakages that prevented building a 32-bit
555 target with a 64-bit host compiler.
556 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
557 opened with :IF-EXISTS :OVERWRITE.
558 * fixed bug 281: error for an invalid qualifier in a short-form method
559 combination method is not signalled until the faulty method is called.
560 * bug fix: iteration variable type inferrer failed to deal with open
561 intervals. (reported by Alan Shields)
562 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
563 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
564 or more runtime options were provided to the sbcl binary.
565 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
566 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
567 * optimization: there's now a fast-path for fixnum arguments in the
568 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
569 * optimization: the code generated on x86-64 is more compact thanks
570 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
571 * bug fix: floating-point exception handling now partly works on
572 x86-64. (Thanks to James Knight)
573 * improvement to the MIPS backend from Thiemo Seufer: C-style
574 64-bit long long arguments and return values to alien functions
576 * fixed some bugs revealed by Paul Dietz' test suite:
577 ** the type-error signalled from WARN has a filled-in DATUM slot.
578 ** the type-error required when a stream is not associated with
579 a file has the stream as its datum.
580 ** type-errors on single-floats on x86-64 no longer have
581 :INVALID-OBJECT as the datum
582 ** the type-errors signalled for invalid function names now have
583 a correct expected type
584 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
585 for structures defined with a :TYPE. Documentation strings for
586 typed structures are no longer immediately discarded
587 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
588 broadcast streams and synonym streams. FILE-LENGTH now also works
589 on broadcast streams.
591 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
592 * incompatible change: the --noprogrammer option, deprecated since
593 version 0.7.5, has been removed. Please use the equivalent
594 --disable-debugger option instead.
595 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
597 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
598 inoperational for a while now) has been completely removed.
599 * null lexical environments are now printed as #<NULL-LEXENV>,
600 significantly reducing the amount of clutter in typical backtraces.
601 * documentation on weak pointers, finalization, and after GC hooks
602 has been added to the manual.
603 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
604 as other specialized array types, is much faster. SUBSEQ and
605 COPY-SEQ on such arrays have also been sped up.
606 * optimization: EQL is now more efficient when at least other argument
607 is known to be of type (OR FIXNUM (NOT NUMBER)).
608 * fixed bug: the runtime is now less vulnerable to changes in the
609 size of the SBCL object on OS X, and virtual memory is reserved for
610 all spaces that need to be at a fixed address.
611 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
612 * fixed bug: finalizers and after GC hooks that cause consing are now
614 * fixed bug: compiler error messages and summaries are now printed to
615 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
616 * fixed inference of the upper bound of an iteration variable.
617 (reported by Rajat Datta).
618 * fixed bug 211e: calling local functions with duplicated constant
619 keyword argument no longer causes a bogus style warning about an
621 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
622 ftype declaration to be disregarded. (reported by Dave Roberts)
623 * fixed bug 373: caused by erronous compilation of references to alien
624 variables in the runtime on ppc/darwin.
625 * fixed bug 376: CONJUGATE type deriver.
626 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
627 (reported by Baughn on #lisp)
628 * a cross-compiler bug on non-x86 platforms has been identified and
629 fixed. (thanks to Bruno Haible)
630 * improvements to the MIPS runtime code for increased stability.
631 (thanks to Thiemo Seufer)
632 * increased the maximimum compact environment size to allow
633 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
634 * partial workaround for bug 135: don't name the function we're
635 calling for hairy cases of EVAL, so as not to accumulate one environment
636 entry per call to EVAL. (reported by Kevin Reid)
637 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
638 * optimization: type testing for non-vector arrays should be faster.
639 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
641 * bug fix: setting 31st element of a bit vector to zero did not work
643 * bug fix: redefining a class definition which failed due to a
644 previous accessor / function clash now works (but see BUGS entry
645 #380 for more problems in this area). (thanks to Zach Beane)
646 * the long form of DEFINE-METHOD-COMBINATION disables method group
647 checking when given a single method group with pattern *. (thanks
649 * on x86 compiler supports stack allocation of results of simple
650 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
651 * support for building and running on Mac OS X 10.4 "Tiger" has been added
652 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
653 * fixed some bugs related to Unicode integration:
654 ** the restarts for recovering from input and output encoding
655 errors only appear when there is in fact such an error to
657 * fixed some bugs revealed by Paul Dietz' test suite:
658 ** MISC.549 and similar: late transformation of unsafe type
659 assertions into derived types caused unexpected code
661 ** SCALE-FLOAT type deriver is less wrong.
662 ** type derivers for EXP, LOG and similar functions compute result
663 types for complex arguments better.
664 ** (MISC.563) CONJUGATE type deriver works for very restricted
666 ** out-of-line type testers for character strings are available.
667 ** EQUAL compiler transform understands specialness of objects
669 ** accessing double-floats stored on the stack now works on x86-64.
670 ** debugger internals could sometimes create invalid lispobjs,
671 resulting in GC crashes.
672 ** MISC.548: type check weakening can convert required type into
674 ** initialization forms for bindings are not in scope of free special
677 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
678 * incompatible change: thread support for non-NPTL systems has
679 been removed - locking is buggy and unreliable. A threaded
680 SBCL build will now warn at startup and refuse to create
681 new threads, unless futex support is detected in the kernel
682 * incompatible change: the top level REPL now has only an ABORT
683 restart associated with it, not TOPLEVEL and ABORT as it used to.
684 TOP and TOPLEVEL are now available as debugger commands for
685 returning to the top level.
686 * incompatible change: forms evaluated in the REPL now use the
687 global optimization policy.
688 * incompatible change: user- and system-initialization files are
689 no longer processed with LOAD, but by READ and EVAL; hence the
690 global optimization policy, startup package, readtable, etc,
692 * The .fasl file format number has been incremented because of
693 various incompatible changes.
694 * internal entry point details and argument counts no longer appear
695 in backtraces unless explicitly requested by setting
696 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
697 * built-in and standard functions no longer have names like "top
698 level local call to FOO".
699 * fixed bug 32: functions defined in non-null lexical environments
700 now have more legible printed representation
701 * fixed bug 33: functions defined in non-null lexical environemnts
702 are now more amenable to inspection by INSPECT.
703 * workaround for bug 354: XEPs no longer appear in backtraces unless
704 explicitly requested.
705 * fixed bug: receiving the signal which results from attempting to
706 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
707 SIGBUS on other *BSDs) on architectures where the C stack is also
708 the Lisp stack (x86 and x86-64) from foreign code no longer leads
709 to debugger confusion or wild instability. (reported by Cheuksan
711 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
712 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
713 (reported by Lutz Euler)
714 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
715 compiled code. (reported by Paul Dietz)
716 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
717 specialized parameter to a method to something that is not TYPEP
718 the specializer is now possible.
719 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
720 face of package deletion.
721 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
722 pathname. (reported by David Tolpin on c.l.l)
723 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
724 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
725 STRING-STREAM and FILE-STREAM, respectively.
726 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
727 than previously; in addition, its test suite is now run on build.
728 (thanks to Luke Gorrie)
729 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
730 * optimization: added a immediate representation for single-floats
732 * fixed some bugs revealed by Paul Dietz' test suite:
733 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
734 ** attempting to create a package with a colliding nickname causes
735 correctable errors to be signalled.
736 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
737 ** several x86-64 backend bugs related to sign-extension of immediate
740 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
741 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
742 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
744 * fixed loading of multiply forward-referenced layouts.
745 (thanks to Cheuksan Wang)
746 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
747 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
748 (see COMPILE-FILE documentation for details of :PRINT option)
749 * fixed bugs 19 and 317: fixed-format floating point printing is
750 more accurate. This also fixes a bug reported by Adam Warner
751 related to the ~@F format directive.
752 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
754 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
755 dispatch table if the from-char is a dispatch macro character.
756 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
757 by the wordsize no longer ignore the last word. (reported by Lutz
759 * fixed bug in type checking of dynamic-extent variables. (reported
761 * optimization: sequence traversal functions use their freedom to
762 coerce function designators to functions.
763 * optimization: code with many calls to CLOS methods specialized on
764 CLOS classes has had redundant type checks removed.
765 * optimization: added declarations to speed up operations that access
766 the internal character database (for example STRING-UPCASE)
767 * optimization: comparison operations between floats and sufficiently small
768 fixnums no longer create extra rationals
769 * fixed some bugs related to Unicode integration:
770 ** portions of multibyte characters at the end of buffers for
771 character-based file input are correctly transferred to the
772 start of the buffer at the next read.
773 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
774 passing it through to OPEN.
775 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
776 argument given to internal calls to OPEN.
777 ** the built-in comment readers (introduced by character sequences
778 ";" and "#|") are more forgiving to encoding errors; they will
779 STYLE-WARN and then attempt to resync the stream at a character
780 boundary. (thanks to Teemu Kalvas)
781 * fixed some bugs in the x86-64 port:
782 ** Negative short int return values from c-calls are sign-extended
784 ** The stack is aligned to 16-bytes for c-calls, as required by
785 the ABI. (thanks to Cheuksan Wang)
786 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
787 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
788 for enums. (thanks to Vincent Arkesteijn)
789 ** Multiple small optimizations and bugfixes for floating point
791 * fixed some bugs revealed by Paul Dietz' test suite:
792 ** Space, Tab, Linefeed, Return and Page have the invalid
793 secondary constituent character trait.
794 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
796 ** WITH-INPUT-FROM-STRING should only update the index place on
798 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
800 ** Bit-array operations (BIT-AND and similar) worked incorrectly
801 with one-dimensional arrays with fill pointers.
802 ** TYPE-OF failed on a complex with an integer realpart and a
804 ** compiler failure during type inference for the code of form
805 (IF (EQL X (THE ...)) ...) (MISC.535).
807 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
808 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
809 platform. The port supports 61-bit fixnums, large memory spaces
810 and reloading shared object files.
811 * enhancement: saving cores with foreign code loaded is now
812 supported on ppc/Darwin in addition to the previously supported
814 * enhancement: the statistical profiler now walks deeper into the
815 call stack for more meaningful call-graphs and accrued time
816 reports (x86/x86-64 only). It also now reports time spent in
818 * enhancement: it is now possible to trace most individual methods
819 of a generic function in addition to tracing the generic function
821 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
823 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
824 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
825 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
826 produces an error. (thanks to Vincent Arkesteijn)
827 * bug fix: NAMESTRING on pathnames with :WILD components in their
828 directories works correctly. (thanks to Artem V. Andreev)
829 * fixed bug 125: compiler preserves identity of closures. (reported
831 * bug fixed: functions with &REST arguments sometimes failed with
832 "Undefined function" when compiled with (DEBUG 3). (reported by
834 * bug fix: overflow during compiling of setting element of a bit
835 vector with constant index and value. (reported by Timmy Douglas)
836 * build fix: fixed the dependence on *LOAD-PATHNAME* and
837 *COMPILE-FILE-PATHNAME* being absolute pathnames.
838 * on x86 compiler partially supports stack allocation of dynamic-extent
840 * GO and RETURN-FROM do not check the extent of their exit points
841 when compiled with SAFETY 0.
842 * fixed some bugs related to Unicode integration:
843 ** encoding and decoding errors are now much more robustly
844 handled; it should now be possible to recover even from invalid
845 input or output to the terminal. (thanks to Teemu Kalvas)
846 ** provided a first cut at implementing STRING-TO-OCTETS and
847 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
848 ** altered the SB-MD5 contributed module slightly, changing the
849 interface just enough for it to be supportable for builds where
850 lisp characters are not eight bits.
851 * fixed some bugs revealed by Paul Dietz' test suite:
852 ** the FORMATTER-generated functions for ~V[ conditionals require
853 the correct number of arguments.
854 ** READ-FROM-STRING returns the mandated second value when applied
855 to displaced strings.
856 ** the #\Rubout and #\Backspace characters are treated as invalid
857 constituent characters by the tokenizer.
859 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
860 * new feature: reloading changed shared object files with
861 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
862 * new feature: references to foreign variables and functions
863 can now be compiled and loaded before the corresponding shared
864 object file is loaded, as long as the foreign definitions are
865 available at runtime.
866 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
867 supported. (thanks to Dan Debertin)
868 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
869 just keywords. (thanks to Vincent Arkesteijn)
870 * fixed bug #331: structure-class instances corresponding to
871 DEFSTRUCT forms are now created eagerly.
872 * fixed bug #345: backtraces from calls to undefined functions work
873 on x86 as well. Related bug #61 is now also partially fixed on x86
874 (backtraces from throws to unknown catch tags.)
875 * bug fix: lambda-list parsing is now stricter vrt. order and number
876 of lambda-list keywords.
877 * bug fix: as specified by AMOP, an error is signalled if a
878 class-option appears multiple times in a DEFCLASS form. (reported
880 * bug fix: RANDOM can be compiled when the compiler derives the type
881 of its numeric argument as a disjoint set of small integers.
882 (reported by Paul Dietz)
883 * bug fix: starting a core saved with shared objects loaded when
884 those objects are not available no longer causes threaded SBCL to
885 hang. (reported by Sean Ross)
886 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
888 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
889 argument, which must always be NIL. (reported by Kalle Niemitalo)
890 * bug fix: printing 1.0d+23 no longer results in an error.
891 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
892 * bug fix: structure slot setters preserve evaluation order. (thanks
894 * fixed some bugs related to Unicode integration:
895 ** RUN-PROGRAM can allow its child to take input from a Lisp
896 stream. (reported by Stefan Scholl)
897 ** ASDF-INSTALL successfully downloads .tar.gz files in all
898 locales. (reported by Ken Causey)
899 * fixed some bugs revealed by Paul Dietz' test suite:
900 ** INCF, DECF and REMF evaluate their place form as specified in
902 ** FORMATTER expands ~{ ~} iteration directives with V or #
903 parameters correctly.
904 ** FORMATTER deals with the ~@[ ~] conditional directive where the
905 consequent uses no arguments correctly.
906 ** the system has a partial understanding of the (COMPLEX RATIO)
908 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
911 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
912 * new feature: a build-time option (controlled by the :SB-UNICODE
913 keyword feature, enabled by default) for building the system with
914 support for the entire 21-bit character space defined by the
916 * new feature: the system now has rudimentary external-format
917 support; the primary user-visible change at this time is that
918 characters with the high bit set (such as lower-case-e-acute) will
919 print correctly to a terminal in a UTF-8 environment.
920 * minor incompatible change: BASE-CHAR no longer names a class;
921 however, CHARACTER continues to do so, as required by ANSI.
922 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
923 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
924 has been moved to the SB-EXT package (temporarily re-exported from
926 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
927 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
929 * fixed bug #351: better error-handlind and reporting for malformed
931 * fixed bug #350: bignum-printing is now more memory-efficient,
932 allowing printing of very large bignums, eg. (expt 2 10000000).
933 (reported by Bruno Haible)
934 * fixed bug #302: better primitive-type selection for intersection
936 * fixed bug #308: non-graphic characters now all have names, as
937 required. (reported by Bruno Haible)
938 * bug fix: redefining a class with different superclasses now correctly
939 removes it from the direct-subclasses of its previous superclasses.
940 (reported by David Morse)
941 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
942 new value now works. (reported by Bruno Haible)
943 * bug fix: correct canonicalization of multiple non-standard slot
944 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
945 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
946 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
947 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
948 now exists, an signals an error.
949 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
950 during macroexpansion for non-iteration variables as well. (reported
951 by Bruno Haible for CMUCL)
952 * bug fix: Cyclic structures and unprintable objects in compiler
953 messages no longer cause errors. (reported by Bruno Haible)
954 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
955 and READ-FROM-STRING all now return a primary value of NIL if
956 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
957 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
958 (reported by Bruno Haible for CMUCL)
959 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
960 do not designate a condition. (reported by Bruno Haible for
962 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
963 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
964 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
965 permitting resolution in favour of any of the conflicting symbols.
966 (reported by Bruno Haible for CMUCL)
967 * bug fix: EQUAL compiler optimizations is less aggressive on
968 strings which can potentially compare true despite having distinct
969 specialized array element types.
970 * bug fix: unit enumerations can be defined without dividing by
971 zero. (thanks to Vincent Arkesteijn)
972 * FORMAT compile-time argument count checking has been enhanced.
973 (report from Bruno Haible for CMUCL)
974 * a partial workaround for the bug 262: the compiler does not try to
975 inline-expand a local function doing RETURN-FROM from a deleted
976 BLOCK. (thanks to Peter Denno for the bug report and to David
977 Wragg for the simple test case)
978 * fixed some bugs revealed by Paul Dietz' test suite:
979 ** RENAME-PACKAGE allows all package designators as new package
981 ** constraint propagation and lambda variable substitution are
982 more cautious in dealing with partially deleted code.
983 ** compiler sometimes lost reoptimization passes.
984 ** CERROR, given a condition as condition designator, passes its
985 remaining arguments to the continue format control without
987 ** Case-altering FORMAT directives work correctly on non-ASCII
989 ** The REAL type specifier handles bounds outside the floating
990 point ranges without signalling FLOATING-POINT-OVERFLOW.
991 ** Functions with IR1-transformations can create intercomponent
992 references to global functions.
993 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
995 ** FORMAT ~R treats a nil value for its first parameter correctly.
997 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
998 * enhancement: saving cores with foreign code loaded is now
999 supported on x86/NetBSD and sparc/Linux in addition to the previously
1000 supported platforms.
1001 * bug fix: on some platforms repeated installations caused multiple
1002 copies of HTML documentation to be installed -- should not happen
1003 any more. (reported by Stefan Scholl)
1004 * bug fix: parsing self-recursive alien record types multiple times
1005 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1006 original patch by Helmut Eller for CMUCL)
1007 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1008 (thanks to Richard Kreuter)
1009 * bug fix: defining classes whose accessors are methods on existing
1010 generic functions in other (locked) packages no longer signals
1011 bogus package lock violations. (reported by François-René Rideau)
1012 * bug fix: special variables as DEFMETHOD parameters no longer have
1013 associated bogus type declarations. (reported by David Wragg and
1015 * bug fix: read-write consistency on streams of element-type
1016 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1017 * bug fix: redefiniton of the only method of a generic function with
1018 no DEFGENERIC no longer emits a full WARNING. In addition,
1019 redefinition of generic functions with no DEFGENERIC to an
1020 incompatible lambda list now signals an error. (thanks to Zach
1022 * bug fix: DEFGENERIC now works even when there's a function of the
1023 same name in an enclosing lexical environment. (thanks to Zach
1025 * fixed compiler failure, caused by instrumenting code during
1026 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1027 * optimization: added loop analysis and improved register allocation
1028 * fixed some bugs revealed by Paul Dietz' test suite:
1029 ** POSITION on displaced vectors with non-zero displacement
1030 returns the right answer.
1031 ** (SIMPLE-STRING) is a valid type specifier for sequence
1033 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1035 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1036 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1038 ** The justification version of the FORMAT ~< directive treats
1039 non-zero minpad parameter correctly.
1041 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1042 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1043 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1044 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1045 the supported interface.
1046 * new feature: Single-stepping of code compiled with DEBUG 2 or
1047 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1048 * new feature: saving cores with foreign code loaded is now
1049 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1050 Timothy Moore's work for CMUCL)
1051 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1052 parameters to * instead of NIL if no initform is supplied.
1053 (reported by Johan Bockgaard)
1054 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1055 correctly when FUNCALL appears as the car of the form. Note:
1056 despite this FUNCALL forms are not currently subject to
1057 compiler-macro expansion. (port of Raymond Toy's fix for the
1058 same from CMUCL, reported by Johan Bockgaard)
1059 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1060 (thanks for Teemu Kalvas)
1061 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1062 case instead of terminating the iteration (thanks for Julian
1063 Squires, Sean Champ and Raymond Toy)
1064 * bug fix: incorrect expansion of defgeneric that caused a style
1065 warning. (thanks for Zach Beane)
1066 * bug fix: slot accessor effective method computation works properly
1067 for classes with multiple non-standard applicable methods on
1068 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1069 * on x86 compiler supports stack allocation of results of LIST and
1070 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1071 CMUCL implementation by Gerd Moellmann)
1072 * fixed some bugs revealed by Paul Dietz' test suite:
1073 ** FORMAT strings with both the ~<~:;~> form of the justification
1074 directive and pretty-printing directives cause an error.
1076 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1077 * incompatible change: the internal functions
1078 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1079 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1080 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1081 instead of the old functions.
1082 * new feature: on platforms where "dladdr" is available foreign
1083 function names now appear in backtraces. (based on Helmut Eller's
1085 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1086 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1088 * documentation: ASDF manual is now distributed with SBCL.
1089 * bug fix: SBCL can now load its contributed modules with REQUIRE
1090 even if the system-provided entries have been removed from
1091 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1093 * bug fix: backtraces involving undefined functions or assembly
1094 routines are more informative. (thanks to Brian Downing)
1095 * bug fix: mutually referent alien structures now work correctly.
1096 (reported by Rick Taube)
1097 * bug fix: structures defined by WITH-ALIEN can be referred to
1098 within other definitions in the same WITH-ALIEN.
1099 * bug fix: division operators (MOD, TRUNCATE and the like) with
1100 constant zero divisors and integer dividends no longer generate
1102 * bug fix: provide default methods for INPUT-STREAM-P and
1103 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1104 * bug fix: improve the detection and resolution of MOP metacycles
1105 (where effective method computation depends on the generic function
1106 having its effective method computed). (reported by Bruno Haible)
1107 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1108 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1109 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1110 * bug fix: pathnames with relative directory components can be
1111 represented relative to default pathnames.
1112 * optimization: in taking the GCD of bignums, reduce the two bignums
1113 to approximately the same size (using Euclid's algorithm) before
1114 applying the more sophisticated binary GCD. (thanks to Juho
1116 * optimization: COUNT on bitvectors now operates word-at-a-time.
1117 * optimization: ASH with a positive, but not necessarily constant,
1118 (leftwards) shift, when in a modular context, is compiled to a
1120 * fixed some bugs revealed by Paul Dietz' test suite:
1121 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1122 if the corresponding argument is NIL.
1123 ** FORMAT directives accepting a minpad parameter treat negative
1125 ** literal commas as character parameters to FORMAT directives are
1127 ** literal spaces directly after ~<Newline> directives within a
1128 format-logical-block (~:< ~@:>) do not induce :FILL-style
1129 conditional newlines.
1130 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1131 as specified: it no longer includes conditional newlines).
1132 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1134 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1135 :PER-LINE-PREFIX argument does not evaluate to a string.
1137 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1138 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1139 the manual for details; for now, package locks can be disabled by
1140 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1141 if no major problems are found then it is likely that they will be
1142 compiled in unconditionally.
1143 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1144 unsupported operators on all platforms. To load a shared library
1145 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1146 object file, link it into a shared library outside of SBCL and
1147 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1149 * bug workaround: ROOM T and the GC were not working together
1150 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1151 reporting facility aren't preserved. That reporting has been
1152 disabled, so now until and unless someone figures out how to make it
1153 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1154 * minor incompatible change: as threatened around sbcl-0.8.0, the
1155 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1156 The symbols are also exported from SB-PCL for backwards
1157 compatibility, but more so than before SB-PCL should be treated as
1158 an implementation-internal package.
1159 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1161 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1162 It is known as of this release not to work on the Alpha, however.
1163 * fixed bug #167: errors signalled due to illegal syntax in method
1164 bodies are now more legible.
1165 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1166 designators and can hence be used with TYPEP.
1167 * fixed bug #333: CHECK-TYPE now ensures that the type error
1168 signalled, if any, has the right object to be accessed by
1169 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1170 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1171 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1173 * fixed bug #334: programmatic addition of slots using specialized
1174 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1175 and :ALLOCATION :CLASS slots.
1176 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1177 float exponents. (rereported by Peter Seibel)
1178 * fixed a bug: #\Space (and other whitespace characters) are no
1179 longer considered to be macro characters in standard syntax by
1180 GET-MACRO-CHARACTER.
1181 * fixed bug: initialization of condition class metaobjects no longer
1182 causes an instance of the condition to be created. (reported by
1184 * fixed bug: it is now possible to have slots such that
1185 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1186 description is neither :INSTANCE nor :CLASS.
1187 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1188 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1189 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1190 non-standard SLOT-DEFINITION-ALLOCATION values.
1191 * fixed bugs: various race conditions handling exiting threads.
1192 CL-PPCRE's thread torture test now passes.
1193 * fixed arguably-a-bug: GC time proportional to number of threads in
1194 system even when most of them are idle
1195 * optimization: improved performance of BIT and SBIT on bit-vectors.
1196 * .fasl file incompatibility: The fasl file version number has
1197 been incremented because of changes associated with package locks.
1199 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1200 * minor incompatible change: the system no longer provides
1201 optimization hints (conditions which are TYPEP
1202 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1203 mode; these hints are still emitted when higher SPEED optimization
1205 * new contrib module: a sampling profiler (profiling by statistical
1206 sampling, rather than by instrumenting functions) is available as
1207 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1208 * the behaviour of the standard function ED is now customizeable by
1209 third parties through a hook variable: see ED's documentation
1210 string for information on the protocol.
1211 * the compiler no longer emits efficiency notes for (FUNCALL X)
1212 when the type of X is uncertain under default optimization
1214 * fixed bug 276: mutating a binding of a specialized parameter to a
1215 method to something that is not TYPEP the specializer is now
1217 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1218 exhibit the required behaviour on the x86 platform. (thanks to
1219 Peter van Eynde, Eric Marsden and Bruno Haible)
1220 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1221 even for difficult arguments. (reported by Peter Graves)
1222 * fixed bug 141a: the backquote printer now descends quoted
1224 * fixed another bug in backquote printing: no more destructive
1225 modification of the form's list structure. (reported by Brian
1227 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1228 move between its address being taken and the call to
1229 interrupt_thread, fixing a crashing race condition.
1230 * the SB-POSIX contrib implementation has been adjusted so that it
1231 no longer exhibits ridiculously poor performance when constructing
1232 instances corresponding to C structs.
1234 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1235 * minor incompatible change: the sb-grovel contrib now treats C
1236 structures as alien (in the sense of SB-ALIEN) objects rather than
1237 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1238 has implications for memory management of client code
1239 (sb-grovel-returned objects must now be manually managed) and for
1240 type safety (alien objects now have full types).
1241 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1242 used to control emission of compiler diagnostics, rather than the
1243 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1244 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1245 quality should be considered deprecated.
1246 * install.sh now installs the user manual as well
1247 * (not quite a new documentable feature, but worth considering in
1248 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1249 beginnings of a semantically meaningful condition hierarchy is
1250 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1252 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1253 designator as the defaults argument.
1254 * fixed bug: Displaced arrays whose displaced-to array has become
1255 too small now cause ARRAY-DIMENSION to signal an error, providing
1256 for safer bounds-checking. (reported by Bruno Haible)
1257 * fixed bug: DEFCLASS slot definitions with identical :READER and
1258 :WRITER names now signal a reasonable error. (reported by Thomas
1260 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1262 * fixed bug: Invalid slot specification errors now print correctly.
1263 (thanks to Zach Beane)
1264 * fixed bug 320: Shared to local slot value transfers in class
1265 redefinitions now happen corrently from superclasses as
1266 well. (reported by Bruno Haible)
1267 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1269 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1270 improper lists correctly. (reported by Bruno Haible)
1271 * fixed bug 313: source-transform for <fun-name> was erroneously
1272 applied to a call of a value of a variable with name <fun-name>.
1273 (reported by Antonio Menezes Leitao)
1274 * fixed bug 307: The obsolete instance protocol ensures that
1275 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1276 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1277 remove dead unknown-values globs from the middle of the stack.
1278 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1279 easier to resume long computations after using *BREAK-ON-SIGNALS*
1280 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1281 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1282 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1283 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1284 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1285 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1287 * fixed a (fairly theoretical) bug in string printing: if
1288 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1289 does not have array-element-type equal to the most general string
1291 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1292 function name. (thanks to Nikodemus Siivola)
1293 * fixed bug: compile-time format string checker failed on
1294 non-closed ~{. (reported by Thomas F Burdick)
1295 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1296 don't warn on legal loop code involving more than one aggregate
1297 boolean. (thanks to Nikodemus Siivola)
1298 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1299 in keyword position to MAKE-PACKAGE was not being reported as
1301 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1302 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1304 * fixed bugs 280 and 312: the checking for multiple definitions in a
1305 file is less likely to become confused by uses of inline
1307 * fixed bug: the #S reader macro performs the keyword coercion
1308 specified for slot names. (reported by Kalle Niemitalo)
1309 * fixed bug: lambda lists may contain symbols whose names start with
1310 & but are not lambda-list-keywords; their occurrence triggers a
1312 * fixed bug 321: define-method-combination argument lambda lists do
1313 not cause invalid code to be generated when &optional or &aux
1314 variables have default values. (reported by Bruno Haible)
1315 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1316 have CLOS classes; this appears also to have fixed reported
1317 instability in stack exhaustion detection.
1318 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1319 MUFFLE-WARNING when no associated restart is present is now
1321 * optimization: rearranged the expansion of various defining macros
1322 so that each expands into only one top-level form in a
1323 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1325 * optimization: used a previously little-used slot in symbols to
1326 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1328 * fixed some bugs revealed by Paul Dietz' test suite:
1329 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1331 ** Fixed an optimization bug related to inheritance of initforms
1332 from local to shared slots.
1333 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1334 ** CAST optimizer forgot to flush argument derived type.
1335 ** print/read consistency on floats is now orders of magnitude
1336 more likely. (thanks also to Bruno Haible for a similar report
1338 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1339 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1340 for objects of type REAL. Make it so.
1341 ** SXHASH is specified (infelicitously) to respect similarity,
1342 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1343 it so. (thanks to Markus Ziegler)
1344 ** on the Alpha, the compiler succeeds in compiling functions
1345 returning a known number of arguments greater than 63.
1346 ** fixed handling of invalid NIL arguments in keyword position
1348 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1350 ** provided a readably-printable representation for RANDOM-STATE
1352 ** ensured that pathnames, where they have a namestring, always
1353 print using #P"..." syntax.
1355 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1356 * Support for the forthcoming 2.0 version of the NetBSD kernel
1357 running on x86 hardware has been added. (thanks to Perry
1358 E. Metzger most immediately, and others for their past work)
1359 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1360 current release version; SBCL's OpenBSD support had been broken
1361 since about the time of OpenBSD's switch to ELF binary format.)
1362 * [placeholder for DX summary]
1363 ** user code with &REST lists declared dynamic-extent, under high
1364 speed or space and low safety and debug optimization policy.
1365 * The manual has been converted to Texinfo format and the debugger
1366 chapter from the cmucl manual has been added.
1367 * A facility has been added to extract documentation strings from
1368 sbcl and store them as Texinfo-formatted snippets for inclusion in
1369 the manual (via Texinfo's @include directive)
1370 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1371 the test case to Dave Roberts)
1372 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1373 pointers. (reported by Sean Ross)
1374 * bug fix: PROFILE output is printed nicely even for large numerical
1375 values. (thanks to Zach Beane)
1376 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1377 greater than 32 handle EOF correctly.
1378 * bug fix: on X86 an immediate argument of the IMUL instruction is
1379 correctly printed in disassembly. (thanks to Lutz Euler)
1380 * bug fix: class slots in redefined classes preserve their old
1381 values. (thanks to Bruno Haible and Nikodemus Siivola)
1382 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1383 succeeds. (reported by Marco Baringer)
1384 * fixed some bugs revealed by Paul Dietz' test suite:
1385 ** READ-SEQUENCE now works on ECHO-STREAMs.
1386 ** RATIONALIZE works more according to its specification. (thanks
1389 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1390 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1391 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1392 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1393 should matter to you only if you rebind the printer control
1394 variables and then find you want different bindings in the
1395 debugger than in the ordinary execution of your program.)
1396 * The runtime build system has been tweaked to support building
1397 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1398 assembler and linker. (thanks to Nikodemus Siivola)
1399 * Unbound, undefined, undeclared variables now trigger full
1400 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1401 more often programmer error than deliberate exploitation of undefined
1403 * optimization: the hash algorithm for strings has changed to one
1404 that is less vulnerable to spurious collisions. (thanks to Juho
1406 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1407 needless bounds checking. (thanks to Juho Snellman)
1408 * optimization: implemented multiplication as a modular
1409 (UNSIGNED-BYTE 32) operation on the PPC backend.
1410 * fixed some bugs revealed by Paul Dietz' test suite:
1411 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1413 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1414 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1415 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1417 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1419 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1421 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1422 their output stream on EOF from read.
1423 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1424 have been read to end-of-file.
1425 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1427 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1428 description of determination of which consecutive characters
1430 ** Printing the "Space" character with escaping on now yields "#\\ ",
1431 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1432 ** Reading floating-point numbers with *READ-BASE* set to a number
1433 less than 10 works correctly.
1434 ** Reading floating-point numbers with *READ-BASE* set to a number
1435 more than 10 works correctly.
1436 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1437 the readtable currently in effect.
1439 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1440 * minor incompatible change: parsing of namestrings on a physical
1441 (Unix) host has changed; numbers after the final #\. in a
1442 namestring are no longer interpreted as a version field. This is
1443 intented to be largely invisible to the user, except that the
1444 meaning of the namestring "*.*.*" has changed: it now refers to a
1445 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1446 should usually be replaced by
1447 (make-pathname :name :wild :type :wild :version :wild)
1448 with the added benefit that this is more likely to be portable.
1449 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1450 option to OPEN now signals an error if the file being opened
1451 exists; this may have an impact on existing code.
1452 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1453 signals received, on the PowerPC platforms (both Linux and
1454 Darwin). (thanks to Pierre Mai for pointing out the location of
1456 * several fixes on OS X: The system now builds and runs cleanly on
1457 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1458 (thanks to Brian Mastenbrook)
1459 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1460 second-resolution: integer multiples of 1/3600 between -24 and 24.
1461 (thanks to Vincent Arkesteijn)
1462 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1463 type when called with 1 argument; PEEK-CHAR checked type of
1464 PEEK-TYPE only after having read first character from a
1465 stream. (reported by Peter Graves)
1466 * bug fix: the garbage collector now has much better locality
1467 behaviour, and in particular no longer treats all memory as being
1468 exhausted when many small objects point to each other in a deeply
1470 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1471 recognized as being TYPEP their class.
1472 * bug fix: the PUSHNEW documentation string has been corrected.
1473 (thanks to Vincent Arkesteijn)
1474 * bug fix: defaulting of the value for the last of an atypically
1475 large number of multiple values being bound was not being
1476 performed correctly on the Alpha or PPC platforms
1477 * optimization: implemented multiplication as a modular
1478 (UNSIGNED-BYTE 32) operation on the x86 backend.
1479 * optimization: SEARCH on simple-base-strings can now be open-coded.
1480 (see also contrib/compiler-extras.lisp for inspiration for
1481 teaching the compiler about the Boyer-Moore algorithm).
1482 * value, returned by MAX (and MIN) called with several EQUALP, but
1483 not EQL, arguments now does not depend on compiler settings.
1484 * fixed some bugs revealed by Paul Dietz' test suite:
1485 ** in stack analysis liveness information is propagated from
1486 non-local entry points.
1487 ** pathwise CAST removing failed when the CAST node did not start
1489 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1490 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1492 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1493 host is already defined.
1494 ** RENAME-FILE works on streams instead of signalling an internal
1496 ** PEEK-CHAR uses the current readtable when determining whether
1497 or not a character is whitespace.
1498 ** MERGE-PATHNAMES handles the case when the pathname does not
1499 specify a name while the default-pathname specifies a version.
1500 ** Pathnames now stand a better chance of respecting print/read
1502 ** Attempting to use standardized file system operators with a
1503 pathname with invalid :DIRECTORY components signals a
1505 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1507 ** DIRECTORY on logical pathnames is more correct.
1508 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1509 signal a TYPE-ERROR if their argument is not a stream.
1510 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1511 designator argument does not designate a stream.
1512 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1513 examining the synonym.
1514 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1516 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1517 element-type larger than ([UN]SIGNED-BYTE 32).
1519 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1520 * When built with the :SB-FUTEX feature, threaded builds now take
1521 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1522 for faster/more reliable mutex and condition variable support.
1523 * Incompatible change (but one you probably shouldn't have been using
1524 anyway): the interface and code for arbitrating between multiple
1525 threads in the same user session has been redesigned.
1526 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1527 error when called without an explicit environment argument.
1528 (thanks to Nikodemus Siivola)
1529 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1530 confused about their position. (thanks to Adam Warner and Gerd
1532 * bug fix: Pretty printing backquoted forms with unquotations in the
1533 argument list position of various code constructs such as LAMBDA
1534 now works correctly. (reported by Paul Dietz)
1535 * bug fix: Pretty printing unquotations no longer loses all
1536 stream position information.
1537 * optimization: performance of string output streams is now less
1538 poor for multiple small sequence writes.
1539 * optimization: performance of CSUBTYPEP in the presence of complex
1540 expressions involving CONS and NOT many times has been improved.
1541 (reported by Paul Dietz)
1542 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1544 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1545 streams for the REPL, for the debugger, and so on. (thanks to
1547 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1548 * fixed some bugs revealed by Paul Dietz' test suite:
1549 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1550 is ignored if it is too small, rather than propagating through
1551 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1552 ** extremely complex negations of CONS types were not being
1553 sufficiently canonicalized, leading to inconsistencies in
1555 ** VALUES tranformer lost derived type.
1557 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1558 * fixed a bootstrapping bug: the build process no longer assumes
1559 that the various BOOLE-related constants have the same value in
1560 host and target lisps. (noted by Paul Dietz' test suite on an
1561 SBCL binary built from CLISP)
1562 * The system can now be dynamically linked on the MIPS platform,
1563 which enables dynamic loading of foreign code from Lisp. (thanks
1564 to Ralf Baechle for discussions on the MIPS ABI)
1565 * The system now records debugging information for its own source
1566 files in a filesystem-position-independent manner, relative to
1567 the "SYS" logical host.
1568 * fixed a compiler bug: MV-LET convertion did not check references
1569 to the "max args" entry point. (reported by Brian Downing)
1570 * tweaked disassembly notes to be less confident about proclaiming
1571 some instruction as an LRA. (thanks to Brian Downing)
1572 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1573 now each have their own history, command character, and other
1574 characteristics. (thanks to David Lichteblau)
1575 * fixed some bugs revealed by Paul Dietz' test suite:
1576 ** compiler failure in compiling LOGAND expressions including a
1578 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1579 shift greater than 32.
1580 ** FLUSH-DEST did not mark blocks for type check regeneration.
1581 ** HANDLER-CASE failed to accept declarations in handler clauses
1582 in some circumstances.
1584 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1585 * New code in contrib/sb-introspect (still probably not entirely
1586 stable yet) provides some support for smart Lisp development
1587 environments like SLIME.
1588 * The conditions signalled for errors occurring when loading .fasl
1589 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1590 in a way which should help ASDF recover gracefully.
1591 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1592 stuff has been cleaned up. If you code contrib/ stuff, this might
1593 affect you, and you can look at contrib/README, contrib/STANDARDS,
1594 and/or the 0.8.4.27 diff to check.
1595 * In full calls the compiler now does not generate checks for declared
1596 argument types for all arguments.
1597 * various threading fixes
1598 ** and some experimental patches which didn't make it into
1599 the main tree for this release, but which are shipped in
1600 contrib/experimental-thread.patch as a possible fix for some
1601 failures (deadlock, spinning...) in GC-intensive multithreaded
1603 * fixed PPC build problem (source code incompatibility of different
1604 library versions): added offsetof() hackery which attempts to divine
1605 where glibc maintainers put uc_mcontext today
1606 * fixed bug 282: compiler does not trust type assertions while passing
1607 arguments to a full call.
1608 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1609 &OPTIONAL VALUES type parameter.
1610 * fix bug 214: algorithm for noting rejected templates is now more
1611 similar to that of template seletion. (also reported by rydis on
1613 * fixed bug 141b: printing backquoted information readably and prettily
1614 inserts a space where necessary.
1615 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1616 (CHARACTER 10) are now reported as errors, rather than propagated
1617 as unknown types. (reported by piso on #lisp)
1618 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1619 with values NIL and :ERROR. (thanks to Milan Zamazal)
1620 * fixed bug 191c: CLOS now does proper keyword argument checking as
1621 described in CLHS 7.6.5 and 7.6.5.1.
1622 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1623 longer raise an error; further, using a list as a for-as-arithmetic
1624 counter now raises a meaningful error.
1625 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1626 now understood by sequence creation functions such as MAKE-SEQUENCE
1628 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1629 a STRING-INPUT-STREAM.
1630 * compiler enhancement: SIGNUM is now better able to derive the type
1632 * type declarations inside WITH-SLOTS are checked. (reported by
1634 * fixed some bugs revealed by Paul Dietz' test suite:
1635 ** incorrect optimization of TRUNCATE for a positive first
1636 argument and negative second.
1637 ** compiler failure in let-convertion during flushing dead code.
1638 ** compiler failure while deriving type of TRUNCATE on an
1639 interval, containing 0.
1640 ** ASH of a negative bignum by a negative bignum count now returns
1642 ** intersection of CONS types now canonicalizes properly, fixing
1643 inconsistencies in SUBTYPEP.
1645 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1646 * incompatible change: The --disable-debugger command line
1647 option now clobbers the debugger at a more fundamental
1648 level, by redefining #'INVOKE-DEBUGGER instead of by
1649 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1650 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1651 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1652 enter the debugger REPL and then suffer recursive errors
1653 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1654 while under the new system, BREAK in --disable-debugger mode
1655 terminates the system just as an unhandled error would.
1656 * fixed compiler performance when processing loops with a step >1;
1657 * bug fix: DOCUMENTATION now retrieves generic function
1658 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1659 support has been systematized, and now supports the methods
1660 specified by ANSI, along with a default method and a method for
1661 slot documentation. (reported by Nathan Froyd)
1662 * bug fix: effective methods associated with a generic function are
1663 no longer cached over a change of that generic function's method
1664 combination. (reported by Andreas Fuchs)
1665 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1666 names a non-existent directory. (thanks to Andreas Fuchs)
1667 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1668 the PPC platform, lead to stack corruption; this has been fixed.
1669 (reported by Rainer Joswig)
1670 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1671 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1672 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1673 arithmetic in general, is now much more reliable on the PPC
1675 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1676 platform now returns the right answer.
1677 * optimization: restored some effective method precomputation in
1678 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1679 precomputation is now tunable.
1680 * optimization: compiler-internal data structure use has been
1681 reviewed, and changes have been made that should improve the
1682 performance of the compiler by about 20%.
1683 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1684 simple-base-strings and simple-bit-vectors is improved.
1685 * optimization: the optimization of 32-bit logical and arithmetic
1686 functions introduced in version 0.8.3 on the x86 has been
1687 implemented on the mips, ppc and sparc platforms; an
1688 implementation of the same facility, but for 64-bit arithmetic,
1689 has been added for the alpha.
1690 * microoptimization: the compiler is better able to make use of the
1691 x86 LEA instruction for multiplication by constants.
1692 * bug fix: in some situations compiler did not report usage of
1693 generic arithmetic in (SPEED 3) policy.
1694 * bug 145b fix: compiler used wrong type specifier while converting
1695 MEMBER-types to numeric.
1696 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1698 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1699 index argument now works properly on the Alpha platform.
1700 * bug fix: floating point exception treatment on the Alpha platform
1702 * bug fix: FILE-POSITION works much better on string input and
1703 output streams. (thanks to Nikodemus Siivola)
1704 * bug fix: many threading/garbage collection symptoms sorted.
1705 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1706 might be pseudo-atomic.
1707 * internal change: Stopping for GC is now done with signals not
1708 ptrace. GC is now done in whichever thread wanted it, instead of
1710 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1712 * bug fix: result form in DO is not contained in the implicit
1714 * incompatible change: ICR structure is changed; the value part of
1715 CONTINUATION is now called LVAR; corresponding functions are
1716 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1717 * added type deriver for ISQRT (thanks to Robert E. Brown).
1718 * bug fix: better support for loading from the command line when an
1719 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1721 * fixed some bugs revealed by Paul Dietz' test suite:
1722 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1723 ** ROUND and FROUND now give the right answer when given very
1724 small float arguments.
1725 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1727 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1728 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1729 ** LCM should return a non-negative integer.
1730 ** PARSE-INTEGER returned the index of a terminator instead of the
1731 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1732 ** PARSE-INTEGER returned an incorrect index being applied to a
1734 ** LCM with two arguments of 0 returns 0 rather than signalling
1736 ** unsigned addition of a 32-bit constant with the high bit set no
1737 longer causes an internal compiler error on the x86.
1738 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1739 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1740 with negative last argument.
1741 ** byte specifiers with zero size and position no longer cause
1742 an error during type derivation.
1743 ** bignum multiplication on the Alpha platform now returns the
1745 * porting: The system now builds on SuSE AMD64, although it still
1746 generates a 32-bit binary.
1747 * .fasl file incompatibility: The fasl file version number has
1748 been incremented (because of the changes to internal compiler
1749 data structures referred to above).
1751 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1752 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1753 more accurately, on the Darwin kernel running on PowerPC hardware.
1754 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1755 * Compiler code deletion notes now signal a condition of type
1756 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1757 an associated MUFFLE-WARNING restart.
1758 * The compiler now performs limited argument count validation of
1759 constant format strings in FORMAT, and where appropriate in ERROR,
1760 CERROR and WARN. (thanks to Gerd Moellmann)
1761 * New ASDF-INSTALL contrib can be used for automatic download and
1762 installation of third-party Lisp code from CCLAN or other sites
1764 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1765 which forces another thread to execute a function supplied by the
1767 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1768 now accept and act upon their :ELEMENT-TYPE keyword argument.
1769 (reported by Martin Atzmueller, Edi Weitz)
1770 * bug fix: FILE-POSITION now accepts position designators up to
1771 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1772 is the greater. (thanks to Patrik Nordebo)
1773 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1774 Kalle Olavi Niemitalo)
1775 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1776 printable. (reported by Eric Marsden)
1777 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1778 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1779 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1780 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1781 (thanks to Stig Sandoe)
1782 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1783 UNREAD-CHAR now correctly includes the unread character in the
1784 target sequence. (thanks to Gerd Moellmann)
1785 * bug fix in threaded builds: the system can now be suspended and
1786 resumed by shell job control with minimal disruption.
1787 * bug fixes in times and timezones >2038AD
1788 * better handling of "where is GNU make?" problem in build scripts
1789 (thanks to Nikodemus Siivola)
1790 * new optimization: inside a named function any reference to a
1791 function with the same name is considered to be a self-reference;
1792 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1793 optimization quality.
1794 * new optimization on x86: logical functions and + now have
1795 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1796 used when the result is truncated to 32 bits.
1797 * VALUES declaration is partially enabled.
1798 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1799 * fixed some bugs revealed by Paul Dietz' test suite:
1800 ** The system now obeys the constraint imposed by
1801 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1802 types form a lattice under type intersection.
1803 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1804 ** ASSOC now ignores NIL elements in an alist.
1805 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1806 and (1+ MOST-POSITIVE-FIXNUM) answers.
1807 ** The addition of a method with invalid qualifiers to a generic
1808 function does not cause an error to be signalled immediately;
1809 a warning is signalled, and the error is generated only on
1810 calling the generic function.
1811 * changed .fasl file version number, in part to add type codes for
1812 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1813 obscure ANSI requirements
1815 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1816 * fixed bug 148: failure to inline-expand a local function left
1817 garbage, confusing the compiler.
1818 * fixed bugs 3cd: structure slot readers perform type check if the
1819 slot can have an invalid value (i.e. it is either not initialized
1820 or can be written with a less specific slot writer).
1821 * bug fix: the compiler now traps array references to elements off
1822 the end of an array; previously, the bounds checking in some
1823 circumstances could go off-by-one.
1824 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1826 * type declarations for array element types now obey the description
1827 on the CLHS page "Declaration TYPE", as per discussions on
1828 sbcl-help around 2003-05-08. This means that a declaration
1829 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1830 declaration, all references to BAR will be asserted or assumed
1831 (with THE, so dependent on compiler policy) to involve objects of
1832 type FOO. Note that no such declaration is implied in
1833 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1834 * declared types of functions from the "Conditions"
1835 chapter. (reported by Paul Dietz)
1836 * bug fix: CERROR accepts a function as its first argument.
1837 * bug fix: NTH an NTHCDR accept a bignum as index
1838 arguments. (reported by Adam Warner)
1839 * optimization: character compare routines now optimize comparing
1840 against a constant character. (reported by Gilbert Baumann)
1841 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1842 argument now works properly on the MIPS platform.
1843 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1844 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1845 declared to have a functional type, does not cause an error
1847 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1848 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1849 * fixed bug: if last continuation of a deleted block has a
1850 destination, this destination should be deleted too. (reported by
1852 * fixed a bug in the bootstrap process: the host compiler's values
1853 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1854 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1855 test case from Patrik Nordebo)
1856 * improved the ability of the disassembler on the PPC platform to
1857 provide helpful disassembly notes.
1858 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1859 the class in more cases than previously.
1860 * bug fix: FILE-POSITION now understands :START and :END for
1861 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1862 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1863 8) data. (thanks to David Lichteblau)
1864 * bug fix: it is possible to add a method to a generic function
1865 without lambda list.
1866 * bug fix: reader failed to signal END-OF-FILE inside an
1867 object representation. (reported by Nikodemus Siivola)
1868 * fixed some bugs revealed by Paul Dietz' test suite:
1869 ** LAST and [N]BUTLAST should accept a bignum.
1870 ** condition slot accessors are methods.
1871 ** (VECTOR NIL) is a subtype of STRING.
1873 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1874 * minor incompatible change: some nonsensical specialized lambda
1875 lists (used in DEFMETHOD) which were previously ignored now signal
1877 * minor incompatible change: the system is now aware of the types of
1878 variables in the COMMON-LISP package, and will signal errors for
1879 most violations of these type constraints (where previously they
1880 were silently accepted).
1881 * minor incompatible change: COMPILE-FILE now uses the freedom
1882 afforded (ANSI 3.2.2.3) to use derived function types for
1883 functions defined in the same file. This also permits the system
1884 to warn on static type mismatches and function
1885 redefinition. (Currently it does not work with high DEBUG level.)
1886 * minor incompatible change: VALUES declaration is disabled.
1887 * When issuing notes, the compiler now signals a condition of type
1888 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1889 restart for use in user handlers. It is expected that the
1890 COMPILER-NOTE condition will eventually become a condition
1891 supertype to a hierarchy of note types, which will then be
1892 handleable in a similar fashion. However, at the moment, no such
1893 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1894 but it's an implementation detail, not a classification for the
1896 * Changes in type checking closed the following bugs:
1897 ** type checking of unused values (192b, 194d, 203);
1898 ** template selection based on unsafe type assertions (192c, 236);
1899 ** type checking in branches (194bc).
1900 * A short form of VALUES type specifier has ANSI meaning (it has
1901 increased the number of situations when SBCL cannot perform type
1903 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1904 keywords or constants is permissible.
1905 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1906 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1907 classes. (thanks to Antonio Martinez)
1908 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1909 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1910 argument to a pretty-print stream. Instead, it leaves any such
1911 operations to DESCRIBE-OBJECT methods.
1912 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1914 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1915 lambda lists. (thanks to David Lichteblau)
1916 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1917 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1918 (thanks to Gerd Moellmann)
1919 * fixed bug 246: increased compilation speed of long
1920 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1922 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1924 * DEFCONSTANT now throws a condition of type
1925 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1926 constant to a non-EQL value; CONTINUE and ABORT restarts
1927 respectively change and preserve the value.
1928 * fixed bug 63: The code walker, part of the implementation of CLOS,
1929 is now better at handling symbol macros.
1930 * bug fix: There is no longer an internal implementation type named
1931 CL:LENGTH. (reported by Raymond Toy)
1932 * bug fix: In macro-like defining macros/special operators the
1933 implicit block does not enclose the lambda list.
1934 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1935 enclosed in parentheses) are not suitable as type specifiers, and
1936 their use properly signals an error now.
1937 * bug fix: An argument count mismatch for a type specifier in code
1938 being compiled no longer causes an unhandled error at compile
1939 time, but signals a compile-time warning.
1940 * fixed simple vector readable printing
1941 * bug fix: DESCRIBE takes more care over whether the class
1942 precedence list slot of a class is bound before accessing it.
1943 (reported by Markus Krummenacker)
1944 * bug fix: FORMATTER can successfully compile pretty-printer format
1945 strings which use variants of the ~* directive inside.
1946 * bug fix: SEARCH now applies its TEST predicate to the elements of
1947 the arguments in the correct order. (thanks to Wolfhard Buss)
1948 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1949 in type inference. (thanks to Robert E. Brown)
1950 * bug fix: Reading in symbols with an explicit package name of ""
1951 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1952 * fixed some bugs revealed by Paul Dietz' test suite:
1953 ** NIL is now allowed as a structure slot name.
1954 ** Arbitrary numbers, not just REALs, are allowed in certain
1955 circumstances in LOOP for-as-arithmetic clauses.
1956 ** Multiple class redefinitions before slot access no longer
1957 causes a type error.
1958 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1959 association between the name and a class.
1960 ** Generic functions with non-standard method-combination and over
1961 five methods all of which return constants no longer return NIL
1962 after the first few invocations. (thanks to Gerd Moellmann)
1963 ** CALL-NEXT-METHOD with no arguments now passes the original
1964 values of the arguments, even in the presence of assignment.
1965 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1967 ** DISASSEMBLE works with closures and funcallable instances.
1968 ** ADD-METHOD now returns the generic function, not the new method.
1969 ** FIND-METHOD signals an error if the lengths of the specializers
1970 is incompatible with the generic function, even if the ERRORP
1972 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1973 which its argument is a member.
1974 ** DEFCLASS only redefines the class named by its class-name
1975 argument if that name is the proper name of the class;
1976 otherwise, it creates a new class.
1977 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1978 of the UNBOUND-SLOT condition to the name of the slot.
1979 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1980 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1981 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1982 SLOT-MAKUNBOUND in the specified fashion.
1984 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1985 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1986 cross-compilation host. As a consequence, we can now bootstrap our
1987 way up to SBCL starting with a bare gcc toolchain and human-readable
1988 source code (first the source to CLISP, then the source to SBCL).
1989 * A contributed module containing a partial implementation of the
1990 simple-streams interface has been included. (thanks to Rudi
1992 * A contributed module implementing the RFC1321 Message Digest
1993 Algorithm, known as MD5, has been included.
1994 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1995 no longer has any effect, as the code controlled by this feature
1996 has been deleted. (As far as we know, no-one has ever built using
1997 this feature, and its semantics were confused in any case).
1998 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1999 work on conditions (as required by the ANSI specification),
2000 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2001 expected behaviour on conditions. Users should note, however,
2002 that such behaviour is not required by the ANSI specification,
2003 and so use of this behaviour may render their code unportable.
2004 * fixed some bugs revealed by Paul Dietz' test suite:
2005 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2007 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2008 ** SXHASH on condition objects no longer returns NIL.
2009 ** :ALLOCATION :CLASS slots are better treated; their values are
2010 updated on class redefinition, and initforms inherited from
2011 superclasses are applied.
2012 ** REMOVE-METHOD returns its generic function argument even when
2013 no method was removed.
2014 ** SHARED-INITIALIZE now initializes the values of the requested
2015 slots, including those with :ALLOCATION :CLASS.
2016 ** ALLOCATE-INSTANCE now works on structure classes defined via
2017 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2019 ** SLOT-EXISTS-P now works on conditions, as well as structures
2021 ** MAKE-LOAD-FORM now has the required methods on
2022 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2023 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2024 keyword, and now implements the SLOT-NAMES argument.
2025 ** methods with &OPTIONAL arguments no longer allow too many
2026 arguments to be passed in the call without error.
2027 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2028 option is consistent with the required arguments of the generic
2029 function lambda list.
2030 * bug fix: REQUIRE accepts a string designator. (Thanks to
2032 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2033 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2034 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2035 * bug fix: The compiler now checks for duplicated variables in macro
2037 * bug fix: SETQ on globals returns the correct value.
2038 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2039 not a COMPILER-ERROR (followed by some other strange error on
2040 choosing the CONTINUE restart).
2041 * bug fix: make.sh and friends are now more consistent in the way that
2042 they look for GNU "make".
2044 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2045 * experimental native threads support (on x86 Linux >=2.4 only).
2046 This is not compiled in by default: you need to add :SB-THREAD to
2047 the target features. See the "Beyond ANSI" chapter of the manual
2049 * fix for longstanding nonANSIism: The old distinction between
2050 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2051 The return value from CL:FIND-CLASS is now a CLOS class, and
2052 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2053 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2054 * An interface to the MetaObject Protocol, as described in Kiczales,
2055 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2056 MIT Press, 1991, is available from the SB-MOP package.
2057 * incompatible change: the SB-PCL package should now be considered
2058 a private implementation detail, and no longer a semi-private MOP
2060 * minor incompatible change: due to rearrangement for threads, the
2061 control stack and binding stack are now allocated at arbitrary
2062 addresses instead of being hardcoded per-port. Users affected by
2063 this probably have to be doing advanced things with shared
2064 libraries, and will know who they are.
2065 * minor incompatible change: Previously, all --eval forms used were
2066 processed with READ before any of them were processed with EVAL.
2067 Now each --eval form is processed with both READ and EVAL before
2068 the next --eval form is processed. (Thus package operations like
2069 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2070 work as the user might reasonably expect.)
2071 * minor incompatible change: *STANDARD-INPUT* is now only an
2072 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2074 * minor incompatible change: Y-OR-N-P is now character-oriented,
2075 not line oriented. Also, YES-OR-NO-P now works without errors.
2076 (thanks to Antonio Martinez)
2077 * sb-aclrepl module improvements: an integrated inspector, added
2078 repl features, and a bug fix to :trace command.
2079 * Known functions, which cannot be open coded by the backend, are
2080 considered to be able to check types of their arguments. (fixing
2081 a bug report by Nathan J. Froyd)
2082 * fixed a bug in computing method discriminating functions: It is
2083 now possible to define methods specialized on classes which have
2084 forward-referenced superclasses. (thanks to Gerd Moellmann)
2085 * fixed evaluation order in optional entries (reported by Gilbert
2087 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2088 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2089 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2090 required-by-AMOP NAME argument, as well as CLASS and
2091 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2092 * fixed bug 20: DEFMETHOD can define methods using names that are
2093 not the proper names of classes to designate class specializers.
2094 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2095 Unix file descriptors, instead of blowing up. (thanks to Antonio
2097 * Garbage collection refactoring: user-visible change is that a
2098 call to the GC function during WITHOUT-GCING will not do garbage
2099 collection until the end of the WITHOUT-GCING. If you were doing
2100 this you were probably losing anyway.
2101 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2102 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2103 TYPEP the latter but not the former.
2104 * The compiler issues a full WARNING for calls to undefined functions
2105 with names from the CL package.
2106 * MAP-INTO for a vector destination is open coded. (reported by
2107 Brian Downing on c.l.l)
2108 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2109 documentation string.
2110 * fixed some bugs revealed by Paul Dietz' test suite:
2111 ** COPY-ALIST now signals an error if its argument is a dotted
2113 ** Condition slots are now accessed more correctly in the presence
2114 of multiple initargs for a given slot.
2115 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2116 exclude restarts of the same name associated with a different
2118 ** DEFCLASS of forward-referenced classes with another
2119 forward-referenced class in the superclasses list no longer
2121 ** Condition slots are now initialized once each, not multiple
2122 times. (thanks to Gerd Moellmann)
2123 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2125 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2126 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2127 arguments contain duplicated elements.
2128 ** RESTART-CASE understands local macros.
2129 ** RESTART-CASE associates exactly its own restarts with a condition.
2130 ** ENDP in safe mode checks its argument to be of type LIST.
2131 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2132 in question is unbound.
2133 ** Optimized MAKE-INSTANCE functions no longer cause internal
2134 assertion failures in the presence of duplicate initargs.
2135 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2136 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2137 propagated even in the ctor optimized implementation of
2139 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2141 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2142 * a better implementation of SXHASH on (simple) bit vectors,
2143 measured both in execution speed and in distribution of results
2144 over the positive fixnums, has been installed. Likewise, a better
2145 implementation of EQUAL for simple bit vectors is now available.
2146 * fixed CEILING optimization for a divisor of form 2^k.
2147 * fixed bug 240 (emitting extra style warnings "using the lexical
2148 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2149 by Antonio Martinez)
2150 * fixed SXHASH, giving different results for NIL depending on type
2151 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2152 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2153 variable. (found by Rolf Wester)
2154 * SBCL does not ignore type declarations for special
2155 variables. (reported by rif on c.l.l 2003-03-05)
2156 * some bug fixes in contrib/sb-aclrepl/
2157 * fixed some bugs revealed by Paul Dietz' test suite:
2158 ** a bug in the CONS type specifier, whereby the CAR and CDR
2159 types got intertwined, has been fixed;
2160 ** the type system is now able to reason about the interaction
2161 between INTEGER and RATIO types more completely;
2162 ** APPEND, [N]REVERSE and NRECONC check that those their
2163 arguments, which must be proper lists, are really so;
2164 ** An array specialized to be unable to hold elements has been
2165 implemented, as required -- yes, really -- by ANSI;
2166 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2167 on malformed property lists;
2169 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2170 * incompatible packaging change: in line with Unix convention,
2171 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2172 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2173 * REQUIRE and PROVIDE are now optionally capable of doing something
2174 useful. See the documentation string for REQUIRE.
2175 * infrastructure for a managed SBCL contrib system: contributed
2176 modules in this release include:
2177 ** the ASDF system definition facility;
2178 ** an interface to the BSD Sockets API;
2179 ** an ACL-like convenience interface to the repl;
2180 (thanks to Kevin Rosenberg)
2181 ** an implementation of ROTATE-BYTE, with efficient implementation
2183 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2184 gives a closer approximation to the right answer than previously.
2185 (thanks to Raymond Toy)
2186 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2187 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2188 optional environment arguments, as required by ANSI.
2189 * fixed bugs in other functions taking environment objects, allowing
2190 calls with an explicit NIL environment argument to be compiled
2192 * fixed bug 228: primary return values from
2193 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2194 COMPILE or FUNCTION.
2195 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2196 :TYPE will no longer signal errors on innocuous objects.
2197 * fixed bug 231b: SETQ is better at respecting type declarations in
2198 the lexical environment.
2199 * fixed a bug in DEFCLASS: classes named by symbols with no or
2200 unprintable packages can now be defined.
2201 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2202 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2203 * fixed compiler failure related to checking types of functions.
2204 (reported by Robert E. Brown)
2205 * the compiler is now much more consistent in its error-checking
2206 treatment of bounding index arguments to sequence functions: in
2207 (SAFETY 3) code, errors will be signalled in almost all cases if
2208 invalid sequence bounding indices are passed to functions defined
2209 by ANSI to operate on sequences.
2210 * fixed a bug in the build procedure: documentation of SBCL-specific
2211 packages is now preserved and available in the final Lisp image.
2212 * lifted FDEFINITION lookup out of loops in the implementation of
2213 many list operations. (thanks to Robert E. Brown)
2214 * fixed a bug in the reader: the #n# reader macro now works for
2215 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2216 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2217 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2218 E. Brown and Tony Martinez respectively)
2219 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2220 SLOT-BOUNDP: in particular, it is now able to optimize them much
2221 better, and is now not vulnerable to having packages renamed.
2222 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2223 always called when a slot is not present in an instance. (thanks
2225 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2226 longer signal an error on unreading a character following EOF on
2227 the previous constituent stream. (thanks to Tony Martinez)
2228 * fixed some bugs revealed by Paul Dietz' test suite:
2229 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2230 not just nonnegative fixnums;
2231 ** the logical bit-array operators such as BIT-AND now accept an
2232 explicit NIL for their "opt-arg" argument (to indicate a
2233 freshly-consed result bit-array);
2234 ** ELT now signals an error on an invalid sequence index in safe
2236 ** the type system is now cleverer about negations of numeric
2237 types, and consequently understands the BIGNUM and RATIO types
2239 ** the type system is now cleverer about the interaction between
2240 INTEGER and RATIO types: while bugs still remain, many more
2241 cases are accurately computed;
2242 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2243 if it is in the last clause;
2244 ** CONSTANTLY now correctly returns a side-effect-free function in
2246 ** DECLARE is no longer treated as a special-operator; in
2247 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2248 * incremented fasl file version number due to the change in the
2249 DEFSTRUCT-SLOT-DESCRIPTION structure.
2251 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2252 * minor incompatible change: code processed by the "interpreter" or
2253 EVAL now has a compilation optimization policy of (DEBUG 2)
2254 (changed from (DEBUG 1)) to improve debuggability of interactive
2255 development, and to allow the use of the debug RETURN command in
2257 * an experimental implementation of the RETURN command for the
2258 debugger has been included. (thanks to Frederik Kuivinen)
2259 * fixed bug 62: constraints were not propagated into a loop.
2260 * fixed bug in embedded calls of SORT (reported and investigated by
2262 * fixed some bugs revealed by Paul Dietz' test suite:
2263 ** printing and reading of arrays with some dimensions having
2264 length 0 (thanks to Gerd Moellmann);
2265 ** BOA constructor with &AUX argument without a default value does
2266 not cause a type error;
2267 ** CONSTANTP now returns true for all self-evaluating objects.
2269 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2270 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2271 accessors that are related by inheritance, as specified in the
2272 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2273 Valtteri Vuorikoski)
2274 * The compiler is now able to inline functions that were defined in
2275 a complex lexical environment (e.g. inside a MACROLET).
2276 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2278 * Support for the upcoming FreeBSD-5.0 release has been included.
2279 (thanks to Dag-Erling Smorgrav)
2280 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2281 effect when it is not in a toplevel context.
2282 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2283 stand a better chance of being correct. (thanks to Gerd
2285 * fixed bug in COERCE, which now signals an error on coercing a
2286 rational to a bounded real type which excludes the expected
2288 * The compiler is now able to derive types more accurately from the
2289 COERCE and COMPILE functions.
2290 * fixed bug 223: functional binding is considered to be constant
2291 only for symbols in the CL package.
2292 * fixed bug 231: SETQ did not check the type of a variable being set
2293 (reported by Robert E. Brown)
2294 * A new optimization for MAKE-INSTANCE has been included, fixing
2295 various bugs (including relating to :ALLOCATION :CLASS slots and
2296 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2298 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2299 ** As required by ANSI, LOOP now disallows anonymous collection
2300 clauses such as COLLECT I in conjunction with aggregate boolean
2301 clauses such as THEREIS (= I 1);
2302 ** LOOP now signals an error when any variable is reused in the
2303 same loop (including the potentially useful construct analogous
2304 to WITH A = 1 WITH A = (1+ A);
2305 ** IT is only a special loop symbol within the first clause of a
2306 conditional loop clause;
2307 ** LOOP with a typed iteration variable over a hashtable now
2308 signals a type error iff it should.
2309 * fixed some other bugs revealed by Paul Dietz' test suite:
2310 ** FILE-STREAM now names the class previously known as FD-STREAM;
2311 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2312 argument) no longer signals an error;
2313 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2314 of the slot symbol, rather than using the current package
2315 ((:CONC-NAME "") continues to intern the slot's name in the
2317 * incremented fasl file version number, because of the incompatible
2318 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2319 of the new implementation of DEFINE-COMPILER-MACRO.
2321 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2322 * Support for building SBCL for MIPS platforms running in
2323 little-endian mode has now been checked in, and basic
2324 functionality on said platforms verified.
2325 * minor incompatible change: PCL now records the pathname of a file
2326 in which methods and the like are defined, rather than its
2328 * minor incompatible change: TRUENAME now considers the truename of
2329 a file naming a directory to be the pathname with :DIRECTORY
2330 component indicating that directory.
2331 * minor incompatible change: a NAMED clause in the extended form of
2332 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2333 reason for the previous behaviour is unclear.
2334 * more systematization and improvement of CLOS and MOP conformance
2335 in PCL (thanks to Gerd Moellman and Pierre Mai):
2336 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2338 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2339 primary methods with no specializers;
2340 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2342 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2343 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2344 FORWARD-REFERENCED-CLASSes; error reporting on
2345 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2347 ** SXHASH on CLOS instances now uses a slot internal to the
2348 instance to return different numbers on distinct instances,
2349 while preserving the same return value through invocations of
2351 ** DEFMETHOD signals errors when methods with longer incongruent
2352 lambda lists are added to generic functions;
2353 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2354 CLASS, as specified in AMOP;
2355 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2356 on the order returned by the primary method for classes of
2357 class STANDARD-CLASS;
2358 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2359 * fixed some bugs shown by Paul Dietz' test suite:
2360 ** DOLIST puts its body in TAGBODY;
2361 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2363 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2364 value producing form;
2365 ** if more variables are given to PROGV than values, extra
2366 variables are bound and made to have no value;
2367 ** NSUBSTITUTE on list arguments gets the right answer with
2369 ** ELT signals an error of type TYPE-ERROR when the index argument
2370 is not a valid sequence index;
2371 ** LOOP signals (at macroexpansion time) an error of type
2372 PROGRAM-ERROR when duplicate variable names are found;
2373 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2374 ** FUNCALL of special-operators now cause an error of type
2376 ** PSETQ now works as required in the presence of side-effecting
2377 symbol-macro places;
2378 ** NCONC accepts any object as its last argument;
2379 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2381 ** loop-for-as-package does not require a package to be explicitely
2383 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2385 * fixed bug 166: compiler preserves "there is a way to go"
2386 invariant when deleting code.
2387 * fixed bug 172: macro lambda lists with required arguments after
2388 &REST arguments now cause an error to be signalled. (thanks to
2390 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2391 bug. (thanks to Matthew Danish)
2392 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2394 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2395 arguments are lexically rebound. (thanks to Gerd Moellmann and
2397 * fixed bug 194: error messages are now more informative when there
2398 is no primary method applicable in a call to a generic
2399 function. (thanks to Gerd Moellmann)
2400 * fixed bug in command line argument checking (thanks to Julian
2402 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2403 (thanks to Matthew Danish)
2404 * incremented fasl file version number, because of the
2405 SXHASH-related changes in the layout of CLOS data structures
2407 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2408 * minor incompatible change: The runtime (the Unix executable named
2409 "sbcl") is now much pickier about the .core files it will load.
2410 Essentially it now requires .core files to descend from the same
2411 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2412 as the runtime does. (The intent is to prevent the crashes which
2413 can occur, and which can even be reported as mysterious failures,
2414 when people patch the sources or change the build parameters
2415 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2416 sbcl and .core files.)
2417 * fixed bug: VALUES-LIST is no longer optimized away.
2418 * fixed bug 142: The FFI conversion of C string values to Lisp
2419 string values no longer conses excessively. (thanks to Nathan
2420 Froyd porting Raymond Toy's fix to CMU CL)
2421 * began to systematize and improve MOP conformance in PCL (thanks to
2422 Nathan Froyd, Gerd Moellman and Pierre Mai):
2423 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2425 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2426 ** FINALIZE-INHERITANCE is now called on class finalization;
2427 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2428 argument precedence order.
2429 * fixed bug 202: The compiler no longer fails on functions whose
2430 derived types contradict their declared type.
2431 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2432 so it can be non-toplevel.
2433 * The fasl file version number has changed (because of the new
2434 implementation of DEFMACRO).
2435 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2436 safe code, that any length requirement by their type-specifier
2437 argument is valid. The exceptions to this are described in bug
2439 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2440 functions now check the types of their inputs as required by ANSI.
2441 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2442 introduced symbol is DECLAREd to be SPECIAL.
2443 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2444 * fixed bug: :COUNT argument to sequence functions may be negative
2445 * fixed bug: body of DO-SYMBOLS may contain declarations
2446 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2447 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2448 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2449 symbol macro only once
2450 * fixed printing of call frame when argument list is unavailable
2451 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2452 * compiler no longer signals WARNING on unknown keyword
2455 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2456 * A beta-quality port to the mips architecture running Linux,
2457 based on the old CMUCL backend, has been made. It has been tested
2458 on a big-endian kernel, and works sufficiently well to be able to
2459 rebuild itself; it has not been tested in little-endian mode.
2460 * fixed an inconsistency between gencgc.c and purify.c which made
2461 dumping/loading .core files unreliable
2462 * fixed bug 120a: The compiler now deals correctly with IFs where
2463 the consequent is the same as the alternative, instead of
2464 misderiving the return type. (thanks to Alexey Dejneka)
2465 * fixed bug 113: Logical pathnames are now dumpable (the logical
2466 host is resolved at load-time, throwing an error if it is not
2468 * fixed bug 174: FORMAT's error message is slightly clearer when a
2469 non-printing character is used in a format directive.
2470 * fixed several bugs in compiler checking of type declarations, i.e.
2471 violations of the Python "declarations are assertions" principle
2472 (thanks to Alexey Dejneka)
2473 * fixed several bugs in PCL's error checking (thanks to Gerd
2475 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2477 * fixed bug in compilation of functions as first class values
2478 (thanks to Antonio Martinez-Shotton)
2479 * The compiler's handling TYPE-ERRORs which it can prove will
2480 inevitably happen at runtime has been cleaned up and corrected
2481 in several ways. (thanks to Alexey Dejneka)
2482 * improved argument type checking for various basic arithmetic
2483 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2484 lavished on them in the past that they can be compiled in many
2485 ways in different special cases
2486 * fixed bug 181: compiler checks validity of user supplied type
2488 * cleaned up code flushing in optimization: Function calls which
2489 should signal errors for safety purposes (e.g. which ANSI says
2490 should signal errors when their arguments are of incorrect type)
2491 are no longer optimized away.
2492 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2493 * incremented fasl file version number, because changes in the
2494 implementation of sequence functions like COERCE caused
2495 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2496 in old inline expansions) to become undefined. (Actually these
2497 changes were later undone, so we might very well be binary
2498 compatible with 0.7.7 after all, but leaving the version number
2499 incremented seemed like the simplest and most conservative
2502 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2503 * An alpha-quality port to the parisc architecture running Linux,
2504 based on the old CMUCL backend, has been made. This, even more so
2505 than the other backends, should be considered still a work in
2506 progress; known problems include that the Linux kernel in 64-bit
2507 mode does not propagate the correct sigcontext structure to
2508 userspace, and consequently SBCL on a parisc64 kernel will not
2510 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2511 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2512 suppressed.) Also now that the compiler is looking at declarations
2513 in the environment, it checks optimization declarations as well,
2514 and suppresses inlining when (> DEBUG SPEED).
2515 * More fixes have been made to treatment of floating point exception
2516 treatment and other Unix signals. In particular, floating point
2517 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2518 * The detection and handling of control stack exhaustion (infinite
2519 or very deeply nested recursion) has changed. Stack exhaustion
2520 detection is now done by write-protecting pages at the OS level
2521 and applies at all optimization settings; when found, a
2522 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2523 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2524 be caught using IGNORE-ERRORS.
2525 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2526 correctly when transforming recursive LABELS functions to LETs.
2527 (thanks to Alexey Dejneka)
2528 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2529 that are names of constants or global variables.
2530 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2531 alien routines with docstrings.
2532 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2533 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2535 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2536 (thanks to Eric Marsden)
2537 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2538 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2539 * Bug fix: The SPARC backend can now compile functions involving
2540 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2541 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2542 to a functions expecting a PATHNAME-DESIGNATOR.
2543 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2544 lambda lists. (thanks to Alexey Dejneka)
2545 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2546 functions so that EXPORTing the name of the class doesn't cause
2547 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2548 to Antonio Martinez for reporting this)
2549 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2550 constant-folded without failing an assertion. (thanks to Einar
2551 Floystad Dorum for reporting this)
2552 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2553 types are now dealt with more correctly by the compiler.
2554 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2555 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2556 behaviour analogous to RENAME-FILE. This puts its behaviour more
2557 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2559 * The fasl file version number has changed again. (because of the
2560 bug fix involving the names of PCL MAKE-INSTANCE functions)
2562 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2563 * bug fix: Floating point exceptions are treated much more
2564 consistently on the x86/Linux and PPC/Linux platforms.
2565 * Array initialization with :INITIAL-ELEMENT is now much faster for
2566 cases when the compiler cannot open code the array creation, but
2567 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2568 array accesses have also seen a speed increase.
2569 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2570 specified. (This was at the root of some bad interactions between
2571 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2573 * bug fix: Internal error arguments for undefined functions are now
2574 computed correctly on the PPC/Linux platform.
2575 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2576 Raymond Toy's patch for CMU CL)
2577 * Support for the Solaris 9 operating environment has been included
2578 (thanks to Daniel Merritt)
2579 * A very ugly but hopefully complete draft of the missing FFI chapter
2580 of the manual has been created by reformatting the corresponding
2581 CMU CL manual chapter into (currently very ugly and incoherent)
2582 DocBook and bringing it up to date for SBCL behavior. Thus, the
2583 manual is now essentially complete, at least by my extreme
2584 once-and-only-once standards, whereby it's acceptable to refer to
2585 the doc strings of SB-EXT functions as the primary documentation.
2586 * The fasl file version number has changed again, due to cleanup of
2587 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2588 is no longer a static symbol.)
2590 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2591 * SBCL now builds with OpenMCL (version 0.12) as the
2592 cross-compilation host; also, more progress has been made toward
2593 bootstrapping under CLISP.
2594 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2596 * bug 158 fixed: The compiler can now deal with integer loop
2597 increments different from 1; fixing this turned out also to fix
2599 * bug 169 fixed: no more bogus warnings about using lexical bindings
2600 despite the presence of perfectly good SPECIAL declarations (thanks
2601 to David Lichteblau)
2602 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2603 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2604 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2606 * bug 180 fixed: Method combination specifications no longer ignore
2607 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2608 * bug fix: Structure type predicate functions now check their argument
2609 count as they should.
2610 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2611 correctly. (thanks to Pierre Mai)
2612 * minor incompatible change: The --noprogrammer option is deprecated
2613 in favor of the new --disable-debugger option, which is very similar.
2614 (The major difference is that it takes effect at a slightly different
2615 time at startup, causing handling of errors in --sysinit and
2616 --userinit files will be affected differently.) The
2617 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2618 been added to allow this functionality to be controlled from ordinary
2619 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2620 maintainers, who might want to run non-interactive scripts to
2621 build SBCL cores which will later be used interactively.)
2622 * minor incompatible change: The LOAD function no longer, when given
2623 a wild pathname to load, loads all files matching that pathname.
2624 Instead, an error of type FILE-ERROR is signalled.
2626 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2627 * bug 147 fixed: The compiler preserves its block link/count
2628 invariants more correctly now so that it doesn't crash. (thanks
2630 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2632 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2633 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2634 no-value-for-this-character as NIL (as specified by ANSI).
2635 * HOST-NAMESTRING on physical pathnames now returns a string that is
2636 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2637 (thanks to Christophe Rhodes)
2638 * The Alpha port handles icache flushing more correctly. (thanks to
2640 * More progress has been made toward bootstrapping under CLISP. (thanks
2641 to Christophe Rhodes)
2642 * The fasl file format has changed again, because dynamic loading
2643 on OpenBSD (which has non-ELF object files) motivated some cleanups
2644 in the way that foreign symbols are transformed and passed around.
2645 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2646 is no longer treated as whitespace by the reader, but instead as
2647 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2648 |A
\7fB|, instead of A as it used to.
2650 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2651 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2652 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2653 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2654 this as of 0.7.1.45, but was left out of the previous news section
2655 (thanks to Dan Barlow)
2656 * SBCL now runs on the Solaris operating system on SPARC architectures
2657 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2658 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2659 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2660 * SPARC backend cleanups, allowing builds of cores optimized for V8
2661 and V9 SPARCS, and also emission of code targeted to a particular
2662 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2664 * SBCL is closer to bootstrapping under CLISP, thanks to various
2665 fixes by Christophe Rhodes.
2666 * The fasl file format has changed again, to allow the compiler's
2667 INFO database to support symbol macros.
2668 * The user manual (in doc/) is formatted into HTML more nicely.
2669 (thanks to coreythomas)
2670 * The system is smarter about SUBTYPEP relationships, especially
2671 those involving NOT types (including types such as ATOM which are
2672 represented internally using NOT types). Thus SUBTYPEP is less
2673 likely to return (VALUES NIL NIL) in general, and in particular
2674 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2676 * The fasl file format has changed again, because the internal
2677 representation of types now includes a new slot to support the new
2678 SUBTYPEP-of-NOT-types logic.
2679 * (not a change in the main branch of SBCL, but a related prototype
2680 which can hopefully be merged into the main branch of SBCL in the
2681 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2682 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2683 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2684 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2685 dispatch macro character combinations. (thanks to Alexey Dejneka)
2686 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2687 that superficially look like logical namestrings correctly.
2688 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2690 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2691 default supplied-p) for &optional and &key arguments. (thanks to
2694 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2695 * incompatible change: The compiler is now less aggressive about
2696 tail call optimization, doing it only when (> SPACE DEBUG) or
2697 (> SPEED DEBUG). (This is an incompatible change because there are
2698 programs which relied on the old CMU-CL-style behavior to optimize
2699 away their unbounded recursion which will now die of stack overflow.)
2700 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2701 for non-GENCGC systems has been increased to 20M (since that
2702 seems much closer to the likely performance optimum for modern
2703 systems than the old 4M value was)
2704 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2705 and *DEBUG-PRINT-LEVEL*
2706 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2707 of CMU CL's support for SPARC, and various endianness and other
2708 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2709 * new syntactic sugar for the Unix command line: --load foo.bar is now
2710 an alternate notation for --eval '(load "foo.bar")'.
2712 ** The system now detects stack overflow and handles it gracefully,
2713 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2714 optimization settings. (This is a good thing in general, and
2715 its introduction in this version should be particularly timely
2716 for anyone whose code fails because of suppression of tail
2718 ** The system now hunts for the C variable "environ" in a more
2719 devious way, to avoid segfaults when the C library version differs
2720 between compile time and run time. (thanks to Christophe Rhodes)
2721 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2722 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2723 ** The compiler no longer issues bogus style warnings for undefined
2724 classes in the same source file as the DEFCLASSes which defined
2725 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2727 ** fixes in CONDITION class precedence list for undefined function
2728 errors (thanks to Alexei Dejneka)
2729 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2730 correctly. (thanks to Dan Barlow)
2731 ** portability fixes aiming at bootstrapping under CLISP (thanks
2732 to Dave McDonald and Christophe Rhodes)
2733 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2734 ** fixes in type translation and and type inference (thanks to
2736 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2737 ** various fixes in the new ports (thanks to Dan Barlow)
2738 * several changes related to debugging:
2739 ** suppression of tail recursion, as noted above
2740 ** stack overflow detection, as noted above
2741 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2742 is now the default. (For some time encapsulation has been more
2743 reliable than the breakpoint-based :ENCAPSULATE NIL
2744 implementation, at least on X86 systems; and I just noticed that
2745 encapsulation also seems closer to the spirit of the ANSI
2748 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2750 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2751 up properly again. (There was a packaging bug in 0.7.0 which
2752 left their definitions in SB-SYS::LOAD-FOREIGN and
2753 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2754 vital for most things which interface to C-level interfaces,
2755 like extensions working with sockets or databases or
2756 Perl-compatible regexes or whatever, and the need to fix
2757 this bug is the main reason that 0.7.1 was released so
2759 ** DEFGENERIC is now choosier about the methods it redefines, so that
2760 reLOADing a previously-LOADed file containing DEFGENERICs does
2761 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2762 cycle now works as it should. (thanks to Alexey Dejneka)
2763 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2764 fixed by Christophe Rhodes. (He actually submitted this patch
2765 months ago, and I delayed until after 0.7.0.)
2766 ** Bug 111 (internal compiler confusion about runtime checks on
2767 FUNCTION types) was fixed by Alexey Dejneka.
2768 * Some internal cleanups (getting rid of variables which aren't
2769 needed now that the byte interpreter is gone) caused the fasl
2770 file format number to change again.
2772 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2773 * major incompatible change: The default fasl file extension, i.e. the
2774 default extension for files produced by COMPILE-FILE, has changed
2775 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2777 ** There are many changes in the implementation of the compiler.
2778 SBCL is now essentially a compiler-only implementation of ANSI
2779 Common Lisp. EVAL still "interprets" a few special cases, but
2780 almost all the interesting cases are handled by creating
2781 a LAMBDA expression, calling COMPILE on it, then calling
2782 FUNCALL on the result.
2783 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2784 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2785 Since the code is newer, there might still be some new bugs
2786 (though not as many as before Martin Atzmueller's fixes:-). But
2787 the new code is substantially simpler and clearer, and hopefully
2788 any remaining bugs will be simpler, less fundamental, and more
2789 fixable then the bugs in the old code.
2790 ** The revised compiler is still a little unsteady on its feet.
2792 *** The debugging information it produces (particularly the names
2793 of FUNCTION objects) is sometimes much less useful than what
2794 the old compiler produced.
2795 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2796 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2797 within a MACROLET) has been temporarily weakened.
2798 ** There are new compiler optimizations for various functions:
2799 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2800 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2801 *** the math functions TRUNCATE, FLOOR, and CEILING
2802 *** the function-of-all-trades COERCE
2803 Mostly these should be transparent, but there's one
2804 potentially-annoying problem (bug 117): when the compiler
2805 inline-expands a function and does type analysis on the result,
2806 it can create control paths which have type mismatches, and
2807 when it can't prove that those control paths aren't taken,
2808 it will issue WARNINGs about the type mismatches. This is
2809 a particular problem in practice for the new sequence functions.
2810 It's not clear how this should be fixed, and for now, a
2811 workaround is given in the entry for 117 in the BUGS file.
2812 ** (Because of the interaction between the two previous items --
2813 occasional inlining problems and new inline expansions -- some
2814 of the new sequence function optimizations won't really kick in
2815 completely until debugging information, and then inlining, are
2816 straightened out in some future version.)
2817 * minor incompatible changes:
2818 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2819 DIRECTORY no longer implicitly promotes NIL slots of its
2820 pathname argument to :WILD. In particular, when you ask for the
2821 contents of a directory (which you used to be able to do without
2822 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2823 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2824 ** changes in behavior that ANSI explicitly defines to be
2825 implementation dependent:
2826 *** The new compiler-only implementation still conforms with ANSI,
2827 but acts a little different than before. Besides the obvious
2828 changes in performance tradeoffs (that the cost per form passed
2829 to EVAL has gone up, and the cost per form executed by EVAL
2830 has gone down), the behavior of the system changes a little
2831 because there are no longer any interpreted function objects.
2832 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2833 e.g. doing COMPILE on the output of interactive DEFUN is
2835 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2837 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2838 has changed from (:CL) to NIL.
2839 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2840 e.g. #\Newline and #\Tab, don't have names specified in the
2841 ANSI Common Lisp standard, is now based on their ASCII symbolic
2842 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2843 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2844 are no longer used for output.
2845 ** changes in internal implementation constants:
2846 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2847 4 million. (If your application spends a lot of time GCing and
2848 you have a lot of RAM, you might want to experiment with
2849 increasing it even more.)
2850 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2851 However, almost all old code should still continue to work without
2852 immediate update, as SB-C-CALL is now a (deprecated) nickname
2854 ** Old operator names in the style DEF-FOO are now deprecated in
2855 favor of new corresponding names DEFINE-FOO, for consistency with
2856 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2857 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2858 internal symbols, but a few supported extensions like
2859 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2860 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2861 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2862 etc. as you get deeper into recursive calls to the debugger
2863 command loop, instead of the old "5]", "5]]", "5]]]"
2864 sequence. (I was motivated to do this when squabbles between
2865 ILISP and SBCL left me very deeply nested in the debugger. In the
2866 short term, this change will probably provoke more ILISP/SBCL
2867 squabbles, but hopefully it will be an improvement in the long run.)
2868 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2869 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2870 function, since declaiming return types involving aliens is
2871 (1) annoyingly messy to do by hand and (2) vital to efficient
2872 compilation of code which calls such functions.
2873 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2874 longer reexported by the SB-EXT package. They're solely useful
2875 for alien code, so it seems more logical that you should get
2876 them from the SB-ALIEN package, not in SB-EXT.
2877 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2878 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2879 features. Instead, the code that they used to control is always
2880 built into the system.
2881 * many other bug fixes
2882 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2883 advantage of the new EVAL-WHEN stuff and to clean them up in
2884 general, and they are now more ANSI-compliant in a number of
2885 ways. Martin Atzmueller is responsible for a lot of this.
2886 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2888 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2889 *** correct ERROR type for various file operations
2890 *** some fixes for Lisp streams
2891 *** DEFMETHOD syntax checking
2892 *** changing old weird representation of debug information as
2893 strings (which, among their other deficiencies, don't transform
2894 correctly when you rename packages, and don't change their
2895 print representation when you change things like *PACKAGE*
2896 and *PRINT-LENGTH*) to symbols and lists of symbols
2897 He also made several improvements and fixed several bugs in DESCRIBE.
2898 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2900 *** misbehavior of WRITE-STRING/WRITE-LINE
2901 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2902 and several other LOOP problems as well
2903 *** DIRECTORY when similar filenames are present
2904 *** DEFGENERIC with :METHOD options
2905 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2906 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2907 *** argument ordering in FIND with :TEST option
2908 *** mishandled package designator argument in APROPOS-LIST
2909 *** various problems in the backquote readmacro
2910 *** a bug in APROPOS
2911 *** probably some others that I'm not describing very well here,
2912 since the CVS log documents them by reference to sbcl-devel
2913 messages, and the SourceForge archives aren't working well.:-(
2914 ** Dan Barlow improved the Alpha port (and is making progress on the
2915 PPC port, for those of you who think different).
2916 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2917 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2918 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2919 continues to work on the SPARC port (for those of you in a position
2920 to look down upon our little PC-compatible boxes from a great height).
2921 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2922 and uses that copy, rather than the current dynamic value, when
2923 it's trying to decide whether to truncate output. Thus e.g.
2924 (let ((*print-lines* 50))
2925 (pprint-logical-block (stream nil)
2927 (let ((*print-lines* 8))
2928 (print (aref possiblybigthings i) stream)))))
2929 should now truncate the logical block only at 50 lines, instead of
2930 often truncating it at 8 lines, as it did before.
2931 * The doc/cmucl/ directory, containing old CMU CL documentation
2932 from the time of the fork, is no longer part of the base system.
2933 SourceForge has shut down its anonymous FTP service, and with it
2934 my original plan for distributing the old CMU CL documentation
2935 there. For now, if you need these files you can download an old
2936 SBCL source release and extract them from it.
2937 * The fasl file version number changed again, for dozens of reasons,
2938 some of which are apparent above.
2940 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2941 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2942 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2943 patch, so that DISASSEMBLE output is much nicer.
2944 * The code in the SB-PROFILE package now seems reasonably stable.
2945 I still haven't decided what the final interface should look like
2946 (I'd like PROFILE to interact cleanly with TRACE, since both
2947 facilities use function encapsulation) but if you have a need
2948 for profiling now, you can probably use it successfully with
2949 the current CMU-CL-style interface.
2950 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2951 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2952 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2953 link now returns the dangling link itself, and for similar
2954 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2955 link itself. (In these cases the old code signalled an error and
2956 looped endlessly, respectively.) Thus, DIRECTORY now works even
2957 in the presence of dangling and cyclic symbolic links.
2958 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2959 is now a supported extension again, since the consensus on
2960 sbcl-devel was that it can be useful for ordinary development
2961 work, not just for debugging SBCL itself.
2962 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2963 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2964 that currently-defined functions might be redefined later with
2965 different return types.
2966 * Hash tables can be printed readably, as inspired by CMU CL code
2967 of Eric Marsden and SBCL code of Martin Atzmueller.
2968 * better error handling in CLOS method combination, thanks to
2969 Martin Atzmueller porting Pierre Mai's CMU CL patches
2970 * more overflow fixes for >16Mbyte I/O buffers
2971 * A bug in READ has been fixed, so that now a single Ctrl-D
2972 character suffices to cause end-of-file on character streams.
2973 In particular, now you only need one Ctrl-D at the command
2974 line (not two) to exit SBCL.
2975 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2977 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2978 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2979 does the right thing.
2980 * The implementation of some type tests, especially for CONDITION
2981 types, is now tidier and maybe faster, due to CMU CL code
2982 originally by Douglas Crosher, ported by Martin Atzmueller.
2983 * Some math functions have been fixed, and there are new
2984 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2985 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2986 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2987 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2988 likely to be built into sbcl-0.7.0.)
2989 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2990 CL, ported to SBCL by Martin Atzmueller)
2991 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2992 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2993 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2994 * There's a new slam.sh hack to shorten the edit/compile/debug
2995 cycle for low-level changes to SBCL itself, and a new
2996 :SB-AFTER-XC-CORE target feature to control the generation of
2997 the after-xc.core file needed by slam.sh.
2998 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2999 COMPILE-FILE is no longer supported, so that now every function
3000 gets an entry point, so that block compilation looks a little
3001 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3002 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3003 returns the number of bytes consed since the system started,
3004 rather than the number consed since the first time the function
3005 was called. (The new definition parallels ANSI functions like
3006 CL:GET-INTERNAL-RUN-TIME.)
3007 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3008 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3009 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3010 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3011 * Fasl file version numbers are now independent of the target CPU,
3012 since historically most system changes which required version
3013 number changes have affected all CPUs equally. Similarly,
3014 the byte fasl file version is now equal to the ordinary
3017 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3018 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3019 is no longer recognized. I apologize for this, because it was
3020 listed in SB-EXT as a supported extension, but I found that
3021 its existing behavior was poorly specified, as well as incorrectly
3022 specified, and it looked like too much of a mess to straighten it
3023 out. I have enough on my hands trying to get ANSI stuff to work..
3024 * many patches ported from CMU CL by Martin Atzmueller, with
3025 half a dozen bug fixes in pretty-printing and the debugger, and
3026 half a dozen others elsewhere
3027 * fixed bug 13: Floating point infinities are now supported again.
3028 They might still be a little bit flaky, but thanks to bug reports
3029 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3030 as flaky as they were.
3031 * The --noprogrammer command line option is now supported. (Its
3032 behavior is slightly different in detail from what the old man
3033 page claimed it would do, but it's still appropriate under the
3034 same circumstances that the man page talks about.)
3035 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3036 are now supported, and enabled by default. Thus, the compiler can
3037 handle many floating point and complex operations much less
3038 inefficiently. (Thus e.g. you can implement a complex FFT
3040 * The compiler now detects type mismatches between DECLAIM FTYPE
3041 and DEFUN better, and implements CHECK-TYPE more correctly, and
3042 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3043 thanks to patches from Martin Atzmueller.
3044 * various fixes to make the cross-compiler more portable to
3045 ANSI-conforming-but-different cross-compilation hosts (notably
3046 Lispworks for Windows, following bug reports from Arthur Lemmens)
3047 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3048 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3049 fixed thanks to Pierre Mai's CMU CL patches.
3050 * fixes to keep the system from overflowing internal counters when
3051 it tries to use i/o buffers larger than 16M bytes
3052 * fixed bug 45a: Various internal functions required to support
3053 complex special functions have been merged from CMU CL sources.
3054 (When I was first setting up SBCL, I misunderstood a compile-time
3055 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3056 * improved support for type intersection and union, fixing bug 12
3057 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3058 more obscure bugs as well
3059 * some steps toward byte-compiling non-performance-critical
3060 parts of the system, courtesy of patches from Martin Atzmueller
3061 * Christophe Rhodes has made some debian packages of sbcl at
3062 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3063 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3064 stable, but are nonetheless usable. When he's ready, I'd be happy
3065 to add them to the SourceForge "File Releases" section. (And if
3066 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3067 * new fasl file format version number (because of changes in
3068 internal representation of (OR ..) types to accommodate the new
3069 support for (AND ..) types, among other things)
3071 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3072 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3074 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3076 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3077 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3078 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3079 * DESCRIBE now works on structure objects again.
3080 * Most function call argument type mismatches are now handled as
3081 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3082 know whether the function will be redefined before the call is
3083 executed. (The compiler could flag local calls with full WARNINGs,
3084 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3085 it doesn't keep track of enough information to know whether calls
3086 are local in this sense.)
3087 * Compiler output is now more verbose, with messages truncated
3088 later than before. (There should be some supported way for users
3089 to override the default verbosity, but I haven't decided how to
3090 provide it yet, so this behavior is still controlled by the internal
3091 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3092 src/compiler/ir1util.lisp.)
3093 * Fasl file format version numbers have increased again, because
3094 support for the Gray streams extension changes the layout of the
3095 system's STREAM objects.
3096 * The Gray subclassable streams extension now works, thanks to a
3097 patch from Martin Atzmueller.
3098 * The full LOAD-FOREIGN extension (not just the primitive
3099 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3100 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3101 but like most other programs, it defaults to copying the Unix
3102 environment from the original process instead of starting the
3103 new process in an empty environment.
3104 * Extensions which manipulate the Unix environment now support
3105 an :ENVIRONMENT keyword option which doesn't smash case or
3106 do other bad things. The CMU-CL-style :ENV option is retained
3107 for porting convenience.
3108 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3109 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3111 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3113 * A patch from Martin Atzmueller seems to have solved the SIGINT
3114 problem, and as far as we know, signal-handling now works cleanly.
3115 (If you find any new bugs, please report them!)
3116 * The system no longer defaults Lisp source file names to types
3117 ".l", ".cl", or ".lsp", but only to ".lisp".
3118 * The compiler no longer uses special default file extensions for
3119 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3120 seems to expect a single default extension for all compiled code,
3121 and there's no compelling reason to try to stretch the standard
3122 to allow two different extensions.) Instead, byte-compiled files
3123 default to the same extension as native-compiled files.
3124 * Fasl file format version numbers have increased again, because
3125 a rearrangement of internal implementation packages made some
3126 dumped symbols in old fasl files unreadable in new cores.
3127 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3128 many fewer weird special cases.
3129 * Bug #17 (differing COMPILE-FILE behavior between logical and
3130 physical pathnames) has been fixed, and some related misbehavior too,
3131 thanks to a patch from Martin Atzmueller.
3132 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3133 by Tim Moore, ported to SBCL by Martin Atzmueller.
3134 * Martin Atzmueller fixed several filesystem-related problems,
3135 including bug #36, in part by porting CMU CL patches, which were
3136 written in part by Paul Werkowski.
3137 * More compiler warnings in src/runtime/ are gone, thanks to
3138 more patches from Martin Atzmueller.
3139 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3142 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3144 * DESCRIBE now works on CONDITION objects.
3145 * The debugger now handles errors which arise when trying to print
3146 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3148 * The build system now uses an additional file, customize-target-features.lisp,
3149 to allow local modifications to the target *FEATURES* list. (The point of
3150 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3151 debugging features enabled, without having to worry about propagating it
3152 into everyone's system when I do a "cvs update".) When no
3153 customize-target-features.lisp file exists, the target *FEATURES* list
3154 should be constructed the same way as before.
3155 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3156 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3157 warn about behavior which is undefined under the ANSI spec. Note: This
3158 is specified by ANSI, but it's not very popular with programmers.
3159 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3160 macro in the SBCL sources for an example of a workaround which you
3161 might use to make portable ANSI-standard code which does what you want.
3162 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3163 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3164 when it's used as a non-toplevel form. (This is required in order
3165 to implement the DEFCONSTANT-EQX macro.)
3166 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3167 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3168 funny things, due to bugs in the implementation of EVAL-WHEN
3169 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3170 non-toplevel DEFCONSTANTs are uncommon.)
3171 * The core file version number and fasl file version number have been
3172 incremented, because the old noncompliant DEFCONSTANT behavior involved
3173 calling functions which no longer exist, and because I also took the
3174 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3175 * fixed bug 1 (error handling before read-eval-print loop starts), and
3176 redid debugger restarts and related debugger commands somewhat while
3178 ** The QUIT debugger command is gone, since it did something
3179 rather different than the SB-EXT:QUIT command, and since it never
3180 worked properly outside the main toplevel read/eval/print loop.
3181 Invoking the new TOPLEVEL restart provides the same functionality.
3182 ** The GO debugger command is also gone, since you can just invoke
3183 the CONTINUE restart directly instead.
3184 ** The TOP debugger command is also gone, since it's redundant with the
3185 FRAME 0 command, and since it interfered with abbreviations for the
3187 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3189 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3190 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3191 some porting work by Martin Atzmueller)
3192 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3193 with a boatload of patches which he ported to SBCL. Now that those
3195 ** The system tries to make sure that its low-priority messages
3196 are prefixed by semicolons, to help people who like to use
3197 syntax highlighting in their ILISP buffer. (This patch
3198 was originally due to Raymond Toy.)
3199 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3200 patches originally written by Raymond Toy.
3201 ** The compiler understands coercion between single-value and
3202 multiple-VALUES type expressions better, getting rid of some very
3203 weird behavior, thanks to patches originally by Robert MacLachlan
3204 and Douglas Crosher.
3205 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3206 lambda lists, thanks to a patch originally by Pierre Mai.
3207 ** The system no longer bogusly warns about "abbreviated type
3209 ** The compiler gets less confused by inlining and RETURN-FROM,
3210 thanks to some patches originally by Tim Moore.
3211 ** The system no longer hangs when dumping circular lists to fasl
3212 files, thanks to a patch originally from Douglas Crosher.
3213 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3214 undefined function error.
3215 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3216 in structure slot initforms) for now, documented workaround instead:-|
3217 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3218 * fixed bug 5: added stubs for various Gray stream functions called
3219 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3220 installed, at least appropriate type errors are generated
3221 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3222 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3223 consistently in DEFMETHOD forms.
3224 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3225 it doesn't seem to affect SBCL after all
3226 * The C runtime system now builds with better optimization and many
3227 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3229 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3231 * The system is now under CVS at SourceForge (instead of the
3232 CVS repository on my home machine).
3233 * The new signal handling code has been tweaked to treat register
3234 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3235 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3236 downstream, e.g. in debug-int.lisp, has implicit dependencies
3237 on the unsignedness of integer representation of machine words,
3238 and that caused the system to bomb out with infinite regress
3239 when trying to recover from type errors involving signed values,
3240 e.g. (BUTLAST '(1 2 3) -1).)
3241 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3242 (This was one of the bugs Peter Van Eynde reported back in July.)
3243 * The system now uses code inspired by Colin Walters' O(N)
3244 implementation of MAP (from the cmucl-imp@cons.org mailing
3245 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3246 inline the MAP operation, and there is more than one
3247 sequence argument to the MAP call (so that it can't just
3248 do ETYPECASE once and for all based on the type of the
3249 single sequence argument). (The old non-inline implementation
3250 of the general M-argument sequence-of-length-N case required
3251 O(M*N*N) time when any of the sequence arguments were LISTs.)
3252 * The QUIT :UNIX-CODE keyword argument has been renamed to
3253 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3254 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3256 * Raymond Wiker's patches to port dynamic loading from Linux to
3257 FreeBSD have been added.
3258 * The BUGS file is now more nearly up to date, thanks in large part
3259 to Martin Atzmueller's review of it.
3260 * The debugger now flushes standard output streams before it begins
3261 its output ("debugger invoked" and so forth).
3262 * The core version number and fasl file version number have both
3263 been incremented, because of incompatible changes in the layout
3265 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3266 used to not be called for a saved Lisp image.)
3267 * Martin Atzmueller's version of a patch to fix a compiler crash,
3268 as posted on sbcl-devel 13 September 2000, has been installed.
3269 * Instead of installing Martin Atzmueller's patch for the
3270 compiler transform for SUBSEQ, I deleted the compiler transform,
3271 and transforms for some similar consing operations.
3272 * A bug in signal handling which kept TRACE from working on OpenBSD
3274 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3275 away by constant folding
3276 * The system now defines its address space constants in one place
3277 (in the Lisp sources), and propagates them automatically elsewhere
3278 (through GENESIS and the sbcl.h file). Therefore, patching the
3279 address map is less unnecessarily tedious and error-prone. The
3280 Lisp names of address space constants have also been systematized.
3281 * CVS tags like dollar-Header-dollar have been removed from
3282 the sources, because they have never saved me trouble and
3283 they've been source of trouble working with patches and other
3284 diff-related operations.
3285 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3286 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3288 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3290 * The system has been ported to OpenBSD.
3291 * The system now compiles with a simple "sh make.sh" on the systems
3292 that it's supported on. I.e., now you no longer need to tweak
3293 text in the target-features.lisp-expr and symlinks in src/runtime/
3294 by hand, the make.sh takes care of it for you.
3295 * The system is no longer so grossly inefficient when compiling code
3296 involving vectors implemented as general (not simple) vectors (VECTOR T),
3297 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3298 which dares to use the various sequence functions on non-simple
3299 vectors, takes less of a performance hit.
3300 * There is now a primitive type predicate VECTOR-T-P
3301 to test for the (VECTOR T) type, so that e.g.
3302 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3303 can now be compiled with some semblance of efficiency. (The old code
3304 turned the type declaration into a full call to %TYPEP at runtime!)
3305 * AREF on (VECTOR T) is still not fast, since it's still compiled
3306 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3307 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3308 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3309 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3310 without making full calls to SUBTYPEP at runtime.
3311 (Some analogous efficiency issues for non-simple vectors specialized to
3312 element types other than T, or for non-simple multidimensional arrays,
3313 have not been addressed. They could almost certainly be handled the
3314 same way if anyone is motivated to do so.)
3315 * The changes in array handling break binary compatibility, so
3316 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3317 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3318 instead of (VALUES T T).
3319 * By following the instructions that Dan Barlow posted to sbcl-devel
3320 on 2 July 2000, I was able to enable primitive dynamic object
3321 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3322 functionality is not implemented (since it calls ld to resolve
3323 library references automatically, requiring RUN-PROGRAM for its
3324 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3325 try to resolve library references) is now supported.
3326 * The system now flushes the standard output streams when it terminates,
3327 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3328 them at several other probably-convenient times, e.g. in each pass of
3329 the toplevel read-eval-print loop, and after evaluating a form given
3330 as an "--eval" command-line option. (These changes were motivated by a
3331 discussion of stream flushing issues on cmucl-imp in August 2000.)
3332 * The source transform for TYPEP of array types no longer assumes
3333 that an array whose element type is a not-yet-defined type
3334 is implemented as an array of T, but instead punts, so that the
3335 type will be interpreted at runtime.
3336 * There is now some support for cross-compiling in make.sh: each of
3337 the phases of make.sh has its own script. (This should be transparent
3338 to people doing ordinary, non-cross-compile builds.)
3339 * Since my laptop doesn't have hundreds of megabytes of memory like
3340 my desktop machine, I became more motivated to do some items on
3341 my to-do list in order to reduce the size of the system a little:
3342 ** Arrange for various needed-only-at-cold-init things to be
3343 uninterned after cold init. To support this, those things have
3344 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3345 symbols with such names are now uninterned after cold init).
3346 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3347 fasl files for cold load.
3348 ** Remove the old compiler structure pooling code (which used to
3349 be conditional on the target feature :SB-ALLOC) completely.
3350 ** Redo the representation of some data in cold init to be more compact.
3351 (I also looked into supporting byte compiled code at bootstrap time,
3352 which would probably reduce the size of the system a lot, but that
3353 looked too complicated, so I punted for now.)
3354 * The maximum signal nesting depth in the src/runtime/ support code has
3355 been reduced from 4096 to 256. (I don't know any reason for the very
3356 large old value. If the new smaller value turns out to break something,
3357 I'll probably just bump it back up.)
3358 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3360 * Many, many bugs reported by Peter Van Eynde have been added to
3361 the BUGS list; some have even been fixed.
3362 * While enabling dynamic object file loading, I tried to make the
3363 code easier to understand, renaming various functions and variables
3364 with less ambiguous names, and changing some function calling
3365 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3366 * While trying to figure out how to do the OpenBSD port, I tried to
3367 clean up some of the code in src/runtime/. In particular, I dropped
3368 support for non-POSIX signal handling, added various comments,
3369 tweaked the code to reduce the number of compilation warnings, and
3370 renamed some files to increase consistency.
3371 * To support the new automatic configuration functionality in make.sh,
3372 the source file target-features.lisp-expr has been replaced with the
3373 source file base-target-features.lisp-expr and the machine-generated
3374 file local-target-features.lisp-expr.
3375 * fixed a stupid quoting error in make.sh so that using CMU CL
3376 "lisp -batch" as cross-compilation host works again
3378 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3380 * DESCRIBE no longer tries to call itself recursively to describe
3381 bound/fbound values, so that it no longer fails on symbols which are
3382 bound to themselves (like keywords, T, and NIL).
3383 * DESCRIBE now works on generic functions.
3384 * The printer now prints less-screwed-up representations of closures
3385 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3386 * A private symbol is used instead of the :EMPTY keyword previously
3387 used to mark empty slots in hash tables. Thus
3388 (DEFVAR *HT* (MAKE-HASH-TABLE))
3389 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3390 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3391 now does what ANSI says that it should. (You can still get
3392 similar noncompliant behavior if bang on the hash table
3393 implementation with all the symbols you get back from
3394 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3395 This breaks binary compatibility, since tests for equality to
3396 :EMPTY are wired into things like the macroexpansion of
3397 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3399 * There's now a minimal placeholder implementation for CL:STEP,
3400 as required by ANSI.
3401 * An obscure bug in the interaction of the normal compiler, the byte
3402 compiler, inlining, and structure predicates has been patched
3403 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3404 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3406 * Missing ordinary arguments in a macro call are now detected even
3407 when the macro lambda list contains &KEY or &REST.
3408 * The debugger no longer complains about encountering the top of the
3409 stack when you type "FRAME 0" to explicitly instruct it to go to
3410 the top of the stack. And it now prints the frame you request even
3411 if it's the current frame (instead of saying "You are here.").
3412 * As specified by ANSI, the system now always prints keywords
3413 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3415 * The default initial SIZE of HASH-TABLEs is now smaller.
3416 * Type information from CLOS class dispatch is now propagated
3417 into DEFMETHOD bodies, so that e.g.
3418 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3420 is now basically equivalent to
3421 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3422 (DECLARE (TYPE SINGLE-FLOAT X))
3424 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3425 operation, without having to do run-time type dispatch.
3426 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3427 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3428 * Since I don't seem to be making big file reorganizations very often
3429 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3430 is overflowing my ability to conveniently back them up), I've finally
3431 checked the system into CVS. (The CVS repository is on my home system,
3432 not at SourceForge -- putting it on SourceForge might come later.)
3433 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3434 high-level GC-NOTIFY-FOO functions send their output. (There's
3435 still very little control of where low-level verbose GC functions
3436 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3437 less than it used to -- the GC-NOTIFY-FOO functions are now under
3438 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3439 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3440 in only one place in the source code, and propagates it automatically
3441 everywhere that it's needed. Thus e.g. when I bump the version from
3442 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3443 * The C source files now include boilerplate legalese and documentation
3444 at the head of each file (just as the Lisp source files already did).
3445 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3446 to his page will be replaced with a link to his new CLiki service.
3448 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3450 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3451 * The build process now looks for GNU make under the default name "gmake",
3452 instead of "make" as it used to. If GNU make is not available as "gmake"
3453 on your system, you can change this default behavior by setting the
3454 GNUMAKE environment variable.
3455 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3456 can build without error under CMU CL.
3458 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3460 * There is now a partial SBCL user manual (with some new text and some
3461 text cribbed from the CMU CL manual).
3462 * The beginnings of a profiler have been added (starting with the
3463 CMU CL profiler and simplifying and cleaning up). Eventually the
3464 main interface should be through the TRACE macro, but for now,
3465 it's still accessed through vaguely CMU-CL-style functions and macros
3466 exported from the package SB-PROFILE.
3467 * Some problems left over from porting CMU CL to the new
3468 cross-compilation bootstrap process have been cleaned up:
3469 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3470 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3471 part when getting rid of the compiler *BACKEND* stuff.)
3472 ** The value of *NULL-TYPE* was screwed up, because it was
3473 being initialized before the type system knew the final
3474 definition of the 'NULL type. This screwed up several key
3475 optimizations in the compiler, causing inefficiency in all sorts
3476 of places. (I found it because I wanted to understand why
3477 GET-INTERNAL-RUN-TIME was consing.)
3478 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3479 PROCLAIM FTYPE information. Unfortunately this broke binary
3480 compatibility again, since now the forms output by DEFGENERIC
3481 to refer to functions which didn't exist in 0.6.3.
3482 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3483 can use the new (as of 0.6.3) transform for SOME into MAP into
3485 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3486 values to (INTEGER 0 1000000), so that the time code will no longer
3487 occasionally get blown up by Linux returning 1000000 microseconds
3488 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3489 its output conform to the ANSI spec. (Alas, this makes its output
3490 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3491 * A full call to MAP NIL with a single sequence argument no longer conses.
3492 * fixes to problems pointed out by Martin Atzmueller:
3493 * The manual page no longer talks about multiprocessing as though
3494 it were currently supported.
3495 * The ILISP support patches have been removed from the distribution,
3496 because as of version 5.10.1, ILISP now supports SBCL without us
3497 having to maintain patches.
3498 * added a modified version of Raymond Toy's recent CMU CL patch for
3499 EQUALP comparison of HASH-TABLE
3501 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3503 * The system still can't cross-compile itself with
3504 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3505 entails), but at least it can compile more of itself that way
3506 than it used to be able to, and various buglets which were uncovered
3507 by trying to cross-compile itself that way have now been fixed.
3508 * This release breaks binary compatibility again. This time
3509 at least I've incremented the FASL file format version to 2, so that the
3510 problem can be detected reliably instead of just causing weird errors.
3511 * various new style warnings:
3512 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3513 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3514 ** using the lexical binding of a variable named in the *FOO* style
3515 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3516 as specified by ANSI.
3517 * *RANDOM-STATE* is no longer automatically initialized from
3518 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3519 default behavior of the system is to repeat its behavior every time
3520 it's run. If you'd like to change this behavior, you can always
3521 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3522 old convention there was no comparably easy way to get the system to
3523 repeat its behavior every time it was run.
3524 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3525 FTYPE declarations has been removed, in favor of their ANSI
3526 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3527 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3528 the types of their sequence arguments can be determined at compile time.
3529 This is done through a new open code expansion for MAP which eliminates
3530 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3531 argument types can be determined at compile time.
3532 * The optimizer now transforms COERCE into an identity operation when it
3533 can prove that the coerced object is already of the correct type. (This
3534 can be a win for machine generated code, including the output of other
3535 optimization transforms, such as the MAP transform above.)
3536 * Credit information has been moved from source file headers into CREDITS.
3537 * Source file headers have been made more standard.
3538 * The CASE macro now compiles without complaining even when it has
3541 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3543 * (Note that the way that the PCL macroexpansions were rewritten
3544 to accommodate the change in DEFGENERIC below breaks binary
3545 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3546 not run under sbcl-0.6.2. Once we get out of alpha releases,
3547 i.e. hit release 1.0.0, we'll probably try to maintain binary
3548 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3549 and sbcl-1.4.4. Until then, however, it might be fairly common
3550 for maintenance releases to break binary compatibility.)
3551 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3553 * The reporting of unhandled signals has been changed to print some
3554 explanatory text as well as the report form. (Previously only
3555 the report form was printed.)
3556 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3557 it defines, so that the compiler no longer issues undefined function
3558 warnings for compiled-but-not-yet-loaded generic functions.
3559 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3560 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3561 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3562 section of the ANSI spec are supported.
3563 * The "" nickname for the "KEYWORD" package has been removed.
3564 The reader still handles symbol tokens which begin with a package marker
3565 as keywords, but it doesn't expose its mechanism for doing so in the
3566 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3567 * The system now issues STYLE-WARNINGs for contradictory TYPE
3568 proclamations. (Warnings for contradictory FTYPE proclamations would
3569 be nice too, but those can't be done usefully unless the type system
3570 is made smarter about FUNCTION types.)
3571 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3572 systematized, so that "*target-*.lisp is supposed to exist only on the
3573 target and imply that there's a related file which exists on the
3574 host, and *host-*.lisp is supposed to exist only on the host and imply
3575 that there's a related file which exists on the target. This involves a
3576 lot of renaming. Hopefully the acute confusion caused by the renaming
3577 will be justified by the reduction in chronic confusion..
3578 ** runtime-type.lisp -> early-target-type.lisp
3579 ** target-type.lisp -> late-target-type.lisp
3580 ** early-host-format.lisp -> early-format.lisp
3581 ** late-host-format.lisp -> late-format.lisp
3582 ** host-error.lisp -> misc-error.lisp
3583 ** early-error.lisp -> early-target-error.lisp
3584 ** late-error.lisp -> late-target-error.lisp
3585 ** host-defboot.lisp -> early-defboot.lisp
3586 ** code/misc.lisp -> code/target-misc.lisp
3587 ** code/host-misc.lisp -> code/misc.lisp
3588 ** code/numbers.lisp -> code/target-numbers.lisp
3589 ** code/early-numbers.lisp -> numbers.lisp
3590 ** early-host-type.lisp -> early-type.lisp
3591 ** late-host-type.lisp -> late-type.lisp
3592 ** host-typep.lisp -> typep.lisp
3593 ** load.lisp -> target-load.lisp
3594 ** host-load.lisp -> load.lisp
3595 ** host-disassem.lisp -> disassem.lisp
3596 ** host-insts.lisp -> insts.lisp
3597 ** byte-comp.lisp -> target-byte-comp.lisp
3598 ** host-byte-comp.lisp -> byte-comp.lisp
3599 ** host-signal.lisp -> signal.lisp
3600 ** host-defstruct.lisp -> defstruct.lisp
3601 ** late-target-type.lisp -> deftypes-for-target.lisp
3602 Furthermore, several other previously target-only files foo.lisp (e.g.
3603 hash-table.lisp and random.lisp) have been split into a target-and-host
3604 foo.lisp file and a target-only target-foo.lisp file, with their key type
3605 definitions in the target-and-host part, so that the cross-compiler will
3606 know more about target types.
3607 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3608 gone away. In their place are various *BACKEND-FOO* variables
3609 corresponding to the slots of the old structure.
3610 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3611 into the target SBCL has been fixed.
3612 * The chill.lisp system for loading cold code into a running SBCL
3614 * Support for the CMU CL "scavenger hook" extension has been removed.
3615 (It was undocumented and unused in the CMU CL sources that SBCL was
3616 derived from, and stale in sbcl-0.6.1.)
3617 * Various errors in the cross-compiler type system were detected
3618 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3619 (enabling various consistency checks). Many of them were fixed,
3620 but some hard problems remain, so the compiler is back to
3621 running without *TYPE-SYSTEM-INITIALIZED* for now.
3622 * As part of the cross-compiler type system cleanup, I implemented
3623 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3624 * I have started adding UNCROSS calls throughout the type system
3625 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3626 on cross-compiler input files will be able to go away, and various
3628 * CONSTANTP now returns true for quoted forms (as explicitly required
3631 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3633 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3634 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3635 (MAKE-LIST -1). In the longer term, it ought to become true
3636 that declarations are assertions even at SAFETY 1. For now, it's not
3637 quite true even at SAFETY 3, but it's at least more nearly true..
3638 (Note that this change seems to increases the size of the system by
3639 O(5%) and to decrease the speed of the compiler by 20% or more.)
3640 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3641 for the problem of printing dozens of lines of distracting information
3642 about low-level system machinery as part of the top stack frame
3643 on entry to the debugger when an undefined function was called.
3644 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3645 is not reset to COMMON-LISP-USER.
3646 * Compilation of stuff related to dyncount.lisp has been made conditional
3647 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3648 smaller. The various dyncount-related symbols have been moved into
3649 a new "SB-DYNCOUNT" package.
3650 * tty-inspect.lisp has been renamed to inspect.lisp.
3651 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3652 feature has gone away. (When we eventually port to other flavors of
3653 libc and/or Unix, we'll try to make the differences between flavors
3654 invisible at the user level.)
3655 * Various other *FEATURES* tags, and/or their associated conditionals,
3656 have been removed if obsolescent, or given better documentation, or
3657 sometimes given more-mnemonic names.
3659 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3661 * tidied up "make.sh" script
3662 * tidied up system directory structure
3663 * better "clean.sh" behavior
3664 * added doc/FOR-CMUCL-DEVELOPERS
3665 * many many small tweaks to output format, e.g. removing possibly-confusing
3666 trailing #\. character in DESCRIBE-INSTANCE
3667 * (EQUALP #\A 'A) no longer signals an error.
3668 * new hashing code, including EQUALP hashing
3669 * tidied up Lisp initialization and toplevel
3670 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3671 * command line argument processing
3672 * added POSIX-GETENV function to deal with Unix-ish environment variables
3673 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3674 terminating SBCL on EOF
3675 * non-verbose GC by default
3676 * There is no more "sbcl" shell script; the sbcl file is now the C
3677 runtime executable (just like CMU CL).
3678 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3680 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3681 symbol and package information as Lisp native symbol and package objects
3682 instead of strings naming symbols and strings naming packages. This way,
3683 whenever packages are renamed (as in warm init), debug information is
3684 transformed along with everything else.
3685 * tweaked the optimization policy declarations which control the building
3686 of SBCL itself. Now, among other things, the system no longer saves
3687 source location debugging information. (This helps two problems at once
3688 by reducing SBCL size and by keeping SBCL from trying to look for its
3689 sources -- which may not exist -- when reporting errors.)
3690 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3691 debugging and testing purposes
3692 * cleaned up printing, making the printer call PRINT-OBJECT for
3693 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3694 methods, giving nearly-ANSI behavior
3695 * converted almost all special variables to use *FOO* naming convention
3696 * deleted PARSE-TIME functionality, since it can be done portably
3697 * moved some files out of cold init into warm init
3698 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3700 * regularized formatting of source files
3701 * added an install.sh script
3702 * fixed ridiculous memory usage of cross-compiler by making
3703 compiler/alloc.lisp not try to do pooling unless it can hook
3704 itself into the GC of the cross-compilation host. Now the system
3705 builds nicely on my old laptop.
3706 * added :SB-ALLOC in target-features.lisp-expr
3707 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3708 was not implemented)
3709 * re-did condition handling and note reporting in the compiler. Notes
3710 are no longer handled by signalling conditions. Style warnings
3711 and warnings are handled more correctly and reported in such a way
3712 that it's easy to find one or the other in your output (so that you
3713 can e.g. figure out which of many problems caused COMPILE-FILE to
3715 * changed the severity of several compiler warnings from full WARNING
3716 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3717 compiler note reporting so that it doesn't use the condition system
3718 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3719 COMPILE-FILE command)
3720 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3721 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3722 no longer supported, since their implementation was deeply intertwingled
3723 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3724 * removed bogus "support" for compiler macros named (SETF FOO), and
3725 removed the compiler macro for SETF INFO (but only after making a fool
3726 of myself on the cmucl-imp mailing list by posting a bogus patch for
3727 DEFINE-COMPILER-MACRO..)
3728 * Compiled files containing forms which have side effects on the Lisp
3729 reader (such as DEFPACKAGE forms) are now handled more correctly.
3730 (Compiler queuing of top level lambdas has been suppressed by setting
3731 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3732 * deleted various currently-unused source files, e.g. gengc.lisp. They
3733 may be added back at some point e.g. when porting to other architectures,
3734 but until they are it's distracting to distribute them and to try to
3736 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3737 there were so many of them they're just distractions, and UNCROSS is
3738 known to be able to handle the current sources
3739 * moved PROFILE functionality into TRACE, so that it will be clear
3740 how the wrapping and unwrapping of functions when you profile them
3741 interacts with the wrapping and unwrapping of functions when you
3742 trace them. (Actually, the functionality isn't there yet, but at least
3743 the interface specification is there. Hopefully, the functionality will
3744 arrive with some future maintenance release.)
3745 * removed host-oops.lisp
3746 * changed signature of QUIT function to allow UNIX-CODE argument
3747 * fixed READ-SEQUENCE bug
3748 * tweaked verbose GC output so that it looks more like the progress
3749 output that ANSI specifies for functions like LOAD
3750 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3751 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob