1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
3 * feature: thread support on Solaris/x86, and experimental thread support
5 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
7 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
8 default method for SLOT-UNBOUND.
9 * minor incompatible change: prevent the user from specializing the
10 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
11 somewhat counter to the intent of the protocol, I (CSR) think, and
12 additionally it just doesn't work in SBCL as currently
13 implemented, thanks to optimizations (that are always valid for
14 the other three SLOT-VALUEish functions, but not for the setter).
15 * minor incompatible change: the :SB-LDB feature is now enabled by
16 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
17 the low-level debugger.
18 * bug fix: global optimization policy was not visible in LOCALLY and
20 * bug fix: class objects can be used as specializers in methods.
21 (reported by Pascal Costanza)
22 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
23 with a type error. (reported by blitz_ on #lisp)
24 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
25 failed with a type error. (reported by Pascal Bourguignon)
26 * bug fix: merging pathnames against defaults with :DIRECTORY
27 starting with '(:RELATIVE :BACK) should preserve the :BACK.
28 (reported by James Y Knight)
29 * improved SB-BSD-SOCKETS support on Windows. (thanks to Timothy
31 * bug fix: saving large (>2GB) cores on x86-64 now works
32 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
33 constant index and a (simple-array (signed-byte 32)) array
34 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
35 error (patch by Robert J. Macomber)
36 * feature: TIME now displays the amount of run-time spent in GC
37 * fixed some bugs revealed by Paul Dietz' test suite:
38 ** MISC.641: LET-conversion were not supposed to work in late
40 * bug fix: The debugger now does a better job of respecting
41 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
42 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
43 workaround for bug 403.)
45 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
46 * new feature: source path information is generated for macro-expansion
47 errors for use in IDE's like Slime (thanks to Helmut Eller)
48 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
49 * bug fix: compiler does not loop forever on an invalid type in
51 * improvement: compilation of most CLOS applications is significantly
53 * optimization: added a limited bytecode compiler for simple toplevel
54 forms, speeding up compilation and FASL loading
56 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
57 * minor incompatible change: in sbcl-0.9.11 (but not earlier
58 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
59 into subclasses of the finalized class. Now user calls to
60 FINALIZE-INHERITANCE finalize just the one class, and calls by the
61 system return before any subclasses are finalized.
62 * minor incompatible change: The reader no longer ignores errors
63 regarding non-existent packages in #+ and #- feature tests.
64 * new feature: command line options --no-sysinit, --no-userinit to
65 inhibit loading the corresponding init files
66 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
67 for sending data through UDP sockets (thanks to François-René Rideau)
68 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
69 error is signalled instead (thanks to François-René Rideau)
70 * minor incompatible change: Error signalling behaviour of lexical
71 operations violating package locks has changed slightly. Refer to
72 documentation on package locks for details.
73 * enhancement: EVAL can process IF-expressions without resorting to the
75 * bug fix: Some operations involving SIN, COS, or TAN weren't being
76 constant-folded properly due to missing out-of-line functions.
77 (reported by Mika Pihlajamäki)
78 * bug fix: LISTEN sometimes returned T even in cases where no data was
79 immediately available from the stream
80 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
81 were reversed. (reported by Levente Mészáros)
82 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
83 when given a vector displaced to an adjustable array. (reported
85 * bug fix: derivation of float boundaries from numbers outside the
86 appropriate float range (reported by John Wiseman)
87 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
89 * bug fix: precomputing cache entries for generic functions with
90 some subclasses of specializers as yet invalid does not attempt to
91 fill a cache line with a negative offset. (reported by Levente
93 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
94 allow condition class objects as arguments to DOCUMENTATION and
95 (SETF DOCUMENTATION); only find and set documentation for
96 structure names for the STRUCTURE doc-type. (suggested by Gary
98 * improvements to the Win32/x86 port:
99 ** (user-homedir-pathname) and default initialization file
100 locations now know about the user's "Documents and Settings"
101 directory (thanks to Yaroslav Kavenchuk)
102 ** run-program is implemented (thanks to Mike Thomas)
103 ** sockets support (thanks to Timothy Ritchey)
104 ** better backtrace support (thanks to Alastair Bridgewater)
105 ** sb-grovel supported
106 ** asdf-install and sb-posix work somewhat
107 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
108 * improvements to the Solaris/x86 port:
109 ** works on Solaris 11/Solaris Express
110 ** floating-point exception handling support
111 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
112 * fixed some bugs revealed by Paul Dietz' test suite:
113 ** REMOVE-METHOD returns its generic function argument even when
114 the method is not one of the generic functions' methods.
115 ** objects declared as MEMBER types can be admissible arguments to
117 ** printing characters should simply be printed by the FORMAT ~:C
119 ** compiler failure when compiling functions with hairy constant
120 defaults for optional parameters.
121 ** compiler produces wrong code when MAYBE-INLINE-expanding a
122 function, which is already optimized.
124 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
125 * new platform: experimental support for SBCL x86/Darwin, including
126 MacOS X 10.4.5 on Intel.
127 * new feature: Unicode character names are now known to the system
128 (through CHAR-NAME and NAME-CHAR).
129 * new feature: the filesystem location of SBCL's core file is
130 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
131 * minor incompatible change: the contrib modules SB-POSIX and
132 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
133 this change is to make it easier to distribute
134 location-independent binaries.
135 * bug fix: as implied by AMOP, standardized classes no longer have
136 slots named by external symbols of public packages. (reported by
138 * optimization: calling functions via constant symbols -- as in
139 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
140 via the function object as in (FUNCALL #'FOO).
141 * optimization: on x86 and x86-64, the calling convention for the
142 case of dealing with an unknown number of values has been altered
143 to be friendlier to the prediction heuristics implemented,
144 particularly with reference to CALL and RET pairing. (thanks to
145 Alastair Bridgewater)
146 * optimization: on x86, the code for access to thread-local symbol
147 values and binding/unbinding of thread-local symbols is smaller.
148 (thanks to Alastair Bridgewater)
149 * enhancement: CONSTANTP is now able to determine constantness of
150 more complex forms, including calls to constant-foldable standardized
151 functions and some special forms beyond QUOTE.
152 * fixed bug: occasional GC crashes on Solaris/x86
153 * optimization: x86-64 supports stack allocation of results of simple
154 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
155 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
156 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
157 (thanks to James Knight)
158 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
159 by the standard (thanks to Stephen Compall)
161 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
162 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
163 be used for bundling the runtime and the core file into one
164 executable binary. This feature is not currently supported on all SBCL
165 platforms. (thanks to James Bielman and NIIMI Satoshi)
166 * new feature: a generational or ephemeral garbage collector is now
167 the default on the PowerPC platform (both Linux and Darwin). The
168 old Cheney (stop and copy) collector is a build-time option.
169 (thanks to Cyrus Harmon, after Raymond Toy)
170 * minor incompatible change: the method by which SBCL finds its
171 contributed modules has changed; it no longer relies on symbolic
172 links from an $SBCL_HOME/systems directory, but searches directly
173 in the subdirectories of $SBCL_HOME.
174 * enhancement: the dynamic heap size on the Linux/PPC platform is
175 markedly larger, even using the older Cheney garbage collector.
176 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
177 certain complicated string types. (reported by Gary King)
178 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
179 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
180 * fixed bug: DOCUMENTATION on structures no longer signals an
181 error if no structure type of the right name exists. (reported by
183 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
184 applicable non-standard primary or :AROUND method on
185 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
186 no longer cause unbound variable errors. (reported by Kevin Reid)
187 * fixed bug: metacircle resolution in cases where methods have slots
188 added before the slots from STANDARD-METHOD. (reported by Jean
190 * fixed bug: the Power PC assembly code for calling into lisp
191 sometimes computed its return address wrongly (depending on the
192 15th bit of the address, fixed at link time). This used to
193 manifest itself in a segmentation violation while building PCL.
194 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
195 many others over the years)
196 * fixed bug: ROOM no longer reports silly numbers for stack usage if
197 the stack pointer (treated as a signed integer) is negative.
198 (thanks to Peter van Eynde)
200 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
201 * new platform: experimental support for the Windows operating
202 system has been added. (thanks to Alastair Bridgewater)
203 * fixed several bugs in and robustified the PPC FFI (including
204 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
205 * bug fix: allow non-simple string symbol names (reported by Paul
207 * bug fix: interrupt handling on NetBSD (thanks to Richard M
209 * bug fix: saving a core corrupted callbacks on x86/x86-64
210 * bug fix: closed a loophole in metacircularity detection and
211 grounding in the PCL implementation of CLOS.
212 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
213 (reported by tomppa on #lisp)
214 * bug fix: clear the direction flag on Lisp -> C transitions, as
215 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
216 (reported by Andrej Grozin and Hendrik Maryns)
217 * optimization: major improvements to GC efficiency on GENCGC platforms
218 * optimization: faster implementation of EQUAL
219 * optimization: emit more efficient opcodes for some common
220 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
221 * optimization: several other minor code-generation improvements on x86-64
222 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
224 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
225 * minor incompatible change: (SETF CLASS-NAME) and (SETF
226 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
227 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
228 by ANSI as a generic function, and for consistency (SETF
229 GENERIC-FUNCTION-NAME) is treated likewise.
230 * fixed bug #233.b: make constraint propagation notice when a variable
231 value is changed after it is referenced but before it is used
232 * fixed bug #296: no more arbitrary behaviour from filenames with
233 odd characters as --load arguments. (reported by Adam Warner)
234 * fixed bug #390: :CHARACTER-SET pathname components now work as
235 expected. (reported by Tim Daly Jr)
236 * fixed bug #391: complicated :TYPE intersections in slot
237 definitions no longer cause an error in PCL internals.
238 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
239 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
240 returning the number of octets which would be written to the
241 file-stream. (thanks to Robert J. Macomber)
242 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
243 arguments to the internal stream functions. (thanks to David
245 * bug fix: stack exhaustion is now handled on the x86/SunOS
246 platform. (thanks to Daisuke Homma)
247 * bug fix: more accurate ROOM results on GENCGC platforms
248 * optimization: improved type inference for arithmetic-for
249 index variables in LOOP
250 * optimization: faster floating-point SQRT on x86-64
251 * fixed some bugs revealed by Paul Dietz' test suite:
252 ** DOCUMENTATION returns NIL instead of "" for method combinations
253 that don't have a docstring
255 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
256 * minor incompatible change: (SETF CLASS-NAME) and (SETF
257 GENERIC-FUNCTION-NAME) are no longer generic functions, and
258 therefore are not customizeable by user code (as seems to be at
259 least permitted and maybe required by AMOP). As a consolation,
260 however, the SBCL implementation of these functions now calls
261 REINITIALIZE-INSTANCE as specified by AMOP.
262 * bug fix: slot-definition documentation is propagated as per ANSI
263 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
264 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
265 argument, as do eslotd initialization methods. (from Pascal
266 Costanza's "Closer" project)
267 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
268 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
270 * bug fix: it is now possible to have more than one subclass of
271 STANDARD-GENERIC-FUNCTION without causing stack overflow.
272 (reported by Bruno Haible, Pascal Costanza and others)
273 * bug fix: the dependent update protocol now works for generic
274 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
276 * bug fix: condition-class instances corresponding to
277 DEFINE-CONDITION forms are now created eagerly. (reported by
278 Kalle Olavi Niemitalo on comp.lang.lisp)
279 * bug fix: floating point printing is more accurate in some
280 circumstances. (thanks to Simon Alexander)
281 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
282 merged with *DEFAULT-PATHNAME-DEFAULTS*.
283 * bug fix: callbacks on OS X now preserve stack-alignment.
284 (thanks to Cyrus Harmon)
285 * enhancement: the x86-64 disassembler is much better at
286 disassembling SSE instructions. (thanks to Lutz Euler)
287 * enhancement: CHECK-TYPE on a variable now allows type-inference
288 (thanks to Vincent Arkesteijn)
289 * optimization: improved performance of EUC-JP external format.
290 (thanks to NIIMI Satoshi)
291 * optimization: performance improvements to IO on file streams of
292 :ELEMENT-TYPE CHARACTER
293 * optimization: much faster memory allocation on x86-64
294 * optimization: faster garbage collections (latency and throughput)
296 * optimization: faster bignum subtraction and division on x86 and x86-64
297 (thanks to Lutz Euler)
298 * optimization: more accurate type-inference for loops that use a
299 floating point index variable or a negative step.
301 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
302 * bug fix: add a workaround to SBCL looping infinitely at startup on
303 Linux kernels with apparently buggy implementations of personality().
304 (thanks to Svein Ove Aas)
305 * bug fix: Unicode symbols are correctly printed in LDB backtraces
306 (thanks to David Lichteblau)
307 * bug fix: local bindings shadow global symbol macros for
308 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
309 * optimization: non-open coded uses of numeric comparison operators
310 (e.g. >) no longer cons when called with more than one parameter
311 on platforms supporting dynamic-extent allocation.
312 * enhancement: saving cores with foreign code loaded is now supported
313 on MIPS/Linux in addition to the previously supported platforms.
314 * bug fix: FIND-RESTART now tests for activity, not applicability when given
315 a restart object as identifier. (reported by Helmut Eller for CMUCL)
316 * bug fix: division by zero in sb-sprof when no samples were collected
317 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
318 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
319 (thanks to Alexander Kjeldaas)
320 * new feature: ignoring the return values of destructive functions
321 like NREVERSE signals a compile-time style-warning.
322 (thanks to Kevin Reid)
323 * new feature: the alignment of alien structure fields can be
324 explicitly specified. (thanks to Cyrus Harmon)
325 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
326 and dump core on SIGQUIT
328 ** incompatible change: threads do not inherit values of specials
329 from their parents (see manual)
330 ** bug fix: threads stacks belonging to dead threads are freed by the
331 next exiting thread, no need to gc to collect thread stacks anymore
332 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
333 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
334 ** bug fix: dynamic variable and thread start related gc lossage
335 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
337 * fixed some bugs revealed by Paul Dietz' test suite:
338 ** SUBTYPEP is slightly more accurate on heinously complicated
339 CONS types where some of the members have uncertain (in the
340 NIL, NIL sense) type relationships to each other.
342 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
343 no longer purified unless :PURIFY T is explicitly specified.
344 ** Non-purified cores are significantly smaller than before
346 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
347 * new feature: timers based on Zach Beane's excellent timer package
348 * added support for the following external formats: koi8-u,
349 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
350 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
351 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
352 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
353 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
354 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
355 * incompatible change: a threaded SBCL will no longer revert to
356 non-threaded mode on non-NPTL systems, but refuse to start entirely.
357 * bug fix: interrupts are disabled until startup is complete; no
358 more sigsegvs when receiving a signal to soon
359 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
361 * bug fix: add a workaround for the memory randomization features in
362 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
363 workaround will only be in effect on systems with the proc filesystem
365 * bug fix: printing objects of type HASH-TABLE signals a
366 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
368 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
369 have been running for over 50 days. (reported by Gilbert Baumann)
370 * bug fix: the logic for getting names of functions gets less
371 confused when confronded with alternate-metaclass
372 funcallable-instances. (reported by Cyrus Harmon)
373 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
374 consistent, even on internal alternate-metaclass objects.
375 * bug fix: flush closure information collected by physical
376 environment analysis prepass before the main pass. (bug reported
378 * bug fix: compiler pack phase does not modify a hashtable, which is
379 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
380 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
381 may be expanded. (reported by Alexander <malishev> on c.l.l)
382 * bug fix: no more style-warnings when compiling code using LAST
383 under high-SPEED low-SPACE settings. (reported by David Wallin
385 * fixed some incompatibilities between SBCL's MOP and the MOP
387 ** the METAOBJECT class is now implemented;
388 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
389 STANDARD-OBJECT, as required;
390 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
391 now compatible (as required by VALIDATE-SUPERCLASS); there
392 remains a consistency requirement over the presence or absence
393 of the FUNCTION class in the superclass list of finalized
394 classes; see the manual for more details;
395 ** the :AROUND method for COMPUTE-SLOTS on
396 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
397 requested slot ordering.
399 ** bug fix: parent thread now can be gc'ed even with a live
401 ** bug fix: binding a special with PROGV to no value is not
402 the same as that symbol not having been bound (thanks to
404 ** bug fix: binding specials is thread safe (thanks to
406 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
407 and SB-SPROF do not die with 'no handler for signal XX in
408 interrupt_handle_now(..)' anymore
409 ** bug fix: WITH-TIMEOUT works with multiple running threads
410 * fixed some bugs revealed by Paul Dietz' test suite:
411 ** ENSURE-GENERIC-FUNCTION should take a method class object for
412 the :method-class keyword argument.
414 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
415 * new port: the Solaris operating system on x86 processors is now
416 mostly supported, though some rough edges in the environment
417 remain. (thanks to Daisuke Homma)
418 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
419 space for its heap. (thanks to Thiemo Seufer)
420 * optimization: SBCL on MIPS platforms now supports dynamic-extent
421 closures. (thanks to Thiemo Seufer)
422 * minor incompatible change: eof selects abort in the debugger.
423 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
424 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
426 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
427 deprecated, and will go away in a future revision of SBCL.
428 * minor incompatible change: GC-ON and GC-OFF are no longer
429 implemented with a counter, it does not matter how many times gc
430 is switched on or off
431 * bug fix: discriminating functions for generic function classes
432 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
433 make invalid assumptions about method precedence order. (reported
435 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
436 platforms. (thanks to Thiemo Seufer)
437 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
438 * bug fix: correct pathname printing with printer escaping is on.
439 (thanks to Kevin Reid)
440 * bug fix: complex VOP definitions in "user-space" no longer trigger
441 package locks. (reported by Zach Beane)
442 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
443 for funcallable-instance functions, and is no different from
445 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
446 characters in the current readtable. (reported by Nicholas Neuss)
447 * bug fix: SUBTYPEP on various CONS types returns more a more
448 accurate acknowledgment of its certainty.
449 * optimizations: REMOVE-DUPLICATES now runs in linear time on
450 lists in some cases. This partially fixes bug 384.
451 * flush all standard streams before prompting in the REPL and the
453 * bug fix: signal handling and triggering gc do not conflict
454 directly anymore, in particular a high frequency sb-sprof does
455 not prevent gc from running
456 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
457 approximation for timezone and DST information between the
458 universal time epoch and the smallest negative 32-bit time_t.
459 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
460 year 1899 so long as the encoded time is non-negative.
461 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
462 the correct answer for non-constant rotate amounts when performing
463 an inline 32-bit rotation.
465 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
466 there is only one thread in the session
467 ** bug fix: memory leak for streams created in one thread and
468 written to in another
469 ** bug fix: lockup when compiled with gcc4
470 ** bug fix: race that allows the gc to be triggered when gc is
472 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
473 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
475 ** bug fix: finished threads are gc'ed properly
476 * fixed some bugs revealed by Paul Dietz' test suite:
477 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
478 has arguments with a different set of applicable methods from
479 the orignal arguments.
480 ** The type error thrown by MAP now has a correct expected-type
482 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
483 name a compiled function.
484 ** Three MISC tests where a large bignum was improperly coerced to
485 a float that couldn't represent that bignum during type
486 derivation were fixed.
487 ** SUBTYPEP can now handle types involving the negation of a
488 list-form FUNCTION type.
489 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
490 as (COMPLEX (AND RATIO (NOT FIXNUM))).
491 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
493 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
494 * New feature: Experimental support for bivalent streams: streams
495 opened with :element-type :default now allow character and binary
496 (unsigned-byte 8) I/O
497 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
498 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
500 * bug fix: no more highly sporadic "couldn't check whether ~S is
501 readable" when reading a stream and an interrupt hits in the middle
502 of a select system call
503 * compiler better recognizes complex arrays (reported by Eduardo
505 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
506 for variables on SBCL built with linkage-tables. (reported by Luis
508 * various error reporting improvements.
509 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
510 (Thanks to Thiemo Seufer)
511 * bug fix: nested reader invokations work correctly
512 * bug fix: it is possible to have simultaneous references to foreign
513 code and foreign data with the same name.
515 ** added x86-64 support
516 ** incompatible change: the threading api now works with thread
517 objects instead of thread ids
518 ** bug fix: threads are protected from signals and interruption when
519 starting up or going down
520 ** bug fix: a race where an exiting thread could lose its stack to gc
521 ** bug fix: don't halt on infinite error in threads if possible
522 ** fixed numerous gc deadlocks introduced in the pthread merge
523 ** bug fix: fixed thread safety issues in read and print
524 ** bug fix: debugger doesn't hang on session lock if interrupted at
525 an inappropriate moment
526 ** bug fix: run-program is now thread safe(r)
527 ** bug fix: inner with-recursive-lock no longer releases the mutex
528 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
529 * fixed some bugs revealed by Paul Dietz' test suite:
530 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
531 EXPECTED-TYPE slots filled.
532 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
534 ** MAKE-CONDITION accepts classes as type-designators.
535 ** COMPILE may never return NIL.
536 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
537 range before calling Unix time functions
539 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
540 * numerous signal handling fixes to increase stability
541 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
542 * minor incompatible change: we now correctly canonize default
543 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
544 as per the MOP, rather than the historical (INITARG INITFUNCTION
545 INITFORM). (reported by Bruno Haible)
546 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
547 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
548 for more information.
549 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
550 pathname is a directory pathname.
551 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
552 * dynamic space size on PPC has been increased to 768Mb. (thanks to
554 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
555 :METACLASS argument in addition to a class name. (reported by
556 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
557 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
558 (reported by Bruno Haible; patch from Gerd Moellmann)
559 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
561 * bug fix: more cleanups to the floating point exception handling on
562 x86-64 (thanks to James Knight)
563 * bug fix: the compiler does not try to constant fold calls of
564 COERCE and BIT-* functions when they return freshly constructed
565 objects. (reported by Nikodemus Siivola and Paul Dietz)
566 * optimization: improved the MIPS versions of generic (in the
567 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
568 * optimization: direct conversion of (unsigned-byte 32) to floats on
569 the PowerPC platform.
570 * optimization: structure instances with raw slots now use less
571 memory, and probably show better memory locality. (thanks to
573 * optimization: DYNAMIC-EXTENT declarations for lists and closures
574 are treated as requests for stack allocation on the x86-64,
575 Alpha-32, PPC and SPARC platforms.
576 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
577 the wrong file descriptor; implementation of SOCKET-OPEN-P.
578 (thanks to Tony Martinez)
580 ** gcing a dead thread can no longer lead to lockups
581 ** threads block signals until they are set up properly
582 ** errno is no longer shared by threads
583 ** interrupt-thread restores the eflags register on x86
584 ** fixed some lockups due to gc/thread interaction
585 * fixed some bugs revealed by Paul Dietz' test suite:
586 ** invalid dotted lists no longer raise a read error when
588 ** signal an error if a symbol that names a declaration is used
589 as the name of a type, or vice versa
590 ** allow using the (declare (typespec var*)) abbreviation for
591 (declare (type typespec var*)) with all type specifiers
592 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
593 ** always return NIL from PPRINT-POP when OBJECT is NIL
594 ** don't signal errors when pretty-printing malformed LABELS,
595 FLET or MACROLET forms
596 ** declarations in a DOLIST body are also in scope for the
598 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
600 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
603 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
604 * fixed cross-compiler leakages that prevented building a 32-bit
605 target with a 64-bit host compiler.
606 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
607 opened with :IF-EXISTS :OVERWRITE.
608 * fixed bug 281: error for an invalid qualifier in a short-form method
609 combination method is not signalled until the faulty method is called.
610 * bug fix: iteration variable type inferrer failed to deal with open
611 intervals. (reported by Alan Shields)
612 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
613 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
614 or more runtime options were provided to the sbcl binary.
615 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
616 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
617 * optimization: there's now a fast-path for fixnum arguments in the
618 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
619 * optimization: the code generated on x86-64 is more compact thanks
620 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
621 * bug fix: floating-point exception handling now partly works on
622 x86-64. (Thanks to James Knight)
623 * improvement to the MIPS backend from Thiemo Seufer: C-style
624 64-bit long long arguments and return values to alien functions
626 * fixed some bugs revealed by Paul Dietz' test suite:
627 ** the type-error signalled from WARN has a filled-in DATUM slot.
628 ** the type-error required when a stream is not associated with
629 a file has the stream as its datum.
630 ** type-errors on single-floats on x86-64 no longer have
631 :INVALID-OBJECT as the datum
632 ** the type-errors signalled for invalid function names now have
633 a correct expected type
634 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
635 for structures defined with a :TYPE. Documentation strings for
636 typed structures are no longer immediately discarded
637 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
638 broadcast streams and synonym streams. FILE-LENGTH now also works
639 on broadcast streams.
641 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
642 * incompatible change: the --noprogrammer option, deprecated since
643 version 0.7.5, has been removed. Please use the equivalent
644 --disable-debugger option instead.
645 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
647 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
648 inoperational for a while now) has been completely removed.
649 * null lexical environments are now printed as #<NULL-LEXENV>,
650 significantly reducing the amount of clutter in typical backtraces.
651 * documentation on weak pointers, finalization, and after GC hooks
652 has been added to the manual.
653 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
654 as other specialized array types, is much faster. SUBSEQ and
655 COPY-SEQ on such arrays have also been sped up.
656 * optimization: EQL is now more efficient when at least other argument
657 is known to be of type (OR FIXNUM (NOT NUMBER)).
658 * fixed bug: the runtime is now less vulnerable to changes in the
659 size of the SBCL object on OS X, and virtual memory is reserved for
660 all spaces that need to be at a fixed address.
661 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
662 * fixed bug: finalizers and after GC hooks that cause consing are now
664 * fixed bug: compiler error messages and summaries are now printed to
665 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
666 * fixed inference of the upper bound of an iteration variable.
667 (reported by Rajat Datta).
668 * fixed bug 211e: calling local functions with duplicated constant
669 keyword argument no longer causes a bogus style warning about an
671 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
672 ftype declaration to be disregarded. (reported by Dave Roberts)
673 * fixed bug 373: caused by erronous compilation of references to alien
674 variables in the runtime on ppc/darwin.
675 * fixed bug 376: CONJUGATE type deriver.
676 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
677 (reported by Baughn on #lisp)
678 * a cross-compiler bug on non-x86 platforms has been identified and
679 fixed. (thanks to Bruno Haible)
680 * improvements to the MIPS runtime code for increased stability.
681 (thanks to Thiemo Seufer)
682 * increased the maximimum compact environment size to allow
683 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
684 * partial workaround for bug 135: don't name the function we're
685 calling for hairy cases of EVAL, so as not to accumulate one environment
686 entry per call to EVAL. (reported by Kevin Reid)
687 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
688 * optimization: type testing for non-vector arrays should be faster.
689 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
691 * bug fix: setting 31st element of a bit vector to zero did not work
693 * bug fix: redefining a class definition which failed due to a
694 previous accessor / function clash now works (but see BUGS entry
695 #380 for more problems in this area). (thanks to Zach Beane)
696 * the long form of DEFINE-METHOD-COMBINATION disables method group
697 checking when given a single method group with pattern *. (thanks
699 * on x86 compiler supports stack allocation of results of simple
700 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
701 * support for building and running on Mac OS X 10.4 "Tiger" has been added
702 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
703 * fixed some bugs related to Unicode integration:
704 ** the restarts for recovering from input and output encoding
705 errors only appear when there is in fact such an error to
707 * fixed some bugs revealed by Paul Dietz' test suite:
708 ** MISC.549 and similar: late transformation of unsafe type
709 assertions into derived types caused unexpected code
711 ** SCALE-FLOAT type deriver is less wrong.
712 ** type derivers for EXP, LOG and similar functions compute result
713 types for complex arguments better.
714 ** (MISC.563) CONJUGATE type deriver works for very restricted
716 ** out-of-line type testers for character strings are available.
717 ** EQUAL compiler transform understands specialness of objects
719 ** accessing double-floats stored on the stack now works on x86-64.
720 ** debugger internals could sometimes create invalid lispobjs,
721 resulting in GC crashes.
722 ** MISC.548: type check weakening can convert required type into
724 ** initialization forms for bindings are not in scope of free special
727 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
728 * incompatible change: thread support for non-NPTL systems has
729 been removed - locking is buggy and unreliable. A threaded
730 SBCL build will now warn at startup and refuse to create
731 new threads, unless futex support is detected in the kernel
732 * incompatible change: the top level REPL now has only an ABORT
733 restart associated with it, not TOPLEVEL and ABORT as it used to.
734 TOP and TOPLEVEL are now available as debugger commands for
735 returning to the top level.
736 * incompatible change: forms evaluated in the REPL now use the
737 global optimization policy.
738 * incompatible change: user- and system-initialization files are
739 no longer processed with LOAD, but by READ and EVAL; hence the
740 global optimization policy, startup package, readtable, etc,
742 * The .fasl file format number has been incremented because of
743 various incompatible changes.
744 * internal entry point details and argument counts no longer appear
745 in backtraces unless explicitly requested by setting
746 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
747 * built-in and standard functions no longer have names like "top
748 level local call to FOO".
749 * fixed bug 32: functions defined in non-null lexical environments
750 now have more legible printed representation
751 * fixed bug 33: functions defined in non-null lexical environemnts
752 are now more amenable to inspection by INSPECT.
753 * workaround for bug 354: XEPs no longer appear in backtraces unless
754 explicitly requested.
755 * fixed bug: receiving the signal which results from attempting to
756 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
757 SIGBUS on other *BSDs) on architectures where the C stack is also
758 the Lisp stack (x86 and x86-64) from foreign code no longer leads
759 to debugger confusion or wild instability. (reported by Cheuksan
761 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
762 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
763 (reported by Lutz Euler)
764 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
765 compiled code. (reported by Paul Dietz)
766 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
767 specialized parameter to a method to something that is not TYPEP
768 the specializer is now possible.
769 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
770 face of package deletion.
771 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
772 pathname. (reported by David Tolpin on c.l.l)
773 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
774 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
775 STRING-STREAM and FILE-STREAM, respectively.
776 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
777 than previously; in addition, its test suite is now run on build.
778 (thanks to Luke Gorrie)
779 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
780 * optimization: added a immediate representation for single-floats
782 * fixed some bugs revealed by Paul Dietz' test suite:
783 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
784 ** attempting to create a package with a colliding nickname causes
785 correctable errors to be signalled.
786 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
787 ** several x86-64 backend bugs related to sign-extension of immediate
790 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
791 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
792 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
794 * fixed loading of multiply forward-referenced layouts.
795 (thanks to Cheuksan Wang)
796 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
797 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
798 (see COMPILE-FILE documentation for details of :PRINT option)
799 * fixed bugs 19 and 317: fixed-format floating point printing is
800 more accurate. This also fixes a bug reported by Adam Warner
801 related to the ~@F format directive.
802 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
804 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
805 dispatch table if the from-char is a dispatch macro character.
806 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
807 by the wordsize no longer ignore the last word. (reported by Lutz
809 * fixed bug in type checking of dynamic-extent variables. (reported
811 * optimization: sequence traversal functions use their freedom to
812 coerce function designators to functions.
813 * optimization: code with many calls to CLOS methods specialized on
814 CLOS classes has had redundant type checks removed.
815 * optimization: added declarations to speed up operations that access
816 the internal character database (for example STRING-UPCASE)
817 * optimization: comparison operations between floats and sufficiently small
818 fixnums no longer create extra rationals
819 * fixed some bugs related to Unicode integration:
820 ** portions of multibyte characters at the end of buffers for
821 character-based file input are correctly transferred to the
822 start of the buffer at the next read.
823 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
824 passing it through to OPEN.
825 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
826 argument given to internal calls to OPEN.
827 ** the built-in comment readers (introduced by character sequences
828 ";" and "#|") are more forgiving to encoding errors; they will
829 STYLE-WARN and then attempt to resync the stream at a character
830 boundary. (thanks to Teemu Kalvas)
831 * fixed some bugs in the x86-64 port:
832 ** Negative short int return values from c-calls are sign-extended
834 ** The stack is aligned to 16-bytes for c-calls, as required by
835 the ABI. (thanks to Cheuksan Wang)
836 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
837 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
838 for enums. (thanks to Vincent Arkesteijn)
839 ** Multiple small optimizations and bugfixes for floating point
841 * fixed some bugs revealed by Paul Dietz' test suite:
842 ** Space, Tab, Linefeed, Return and Page have the invalid
843 secondary constituent character trait.
844 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
846 ** WITH-INPUT-FROM-STRING should only update the index place on
848 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
850 ** Bit-array operations (BIT-AND and similar) worked incorrectly
851 with one-dimensional arrays with fill pointers.
852 ** TYPE-OF failed on a complex with an integer realpart and a
854 ** compiler failure during type inference for the code of form
855 (IF (EQL X (THE ...)) ...) (MISC.535).
857 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
858 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
859 platform. The port supports 61-bit fixnums, large memory spaces
860 and reloading shared object files.
861 * enhancement: saving cores with foreign code loaded is now
862 supported on ppc/Darwin in addition to the previously supported
864 * enhancement: the statistical profiler now walks deeper into the
865 call stack for more meaningful call-graphs and accrued time
866 reports (x86/x86-64 only). It also now reports time spent in
868 * enhancement: it is now possible to trace most individual methods
869 of a generic function in addition to tracing the generic function
871 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
873 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
874 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
875 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
876 produces an error. (thanks to Vincent Arkesteijn)
877 * bug fix: NAMESTRING on pathnames with :WILD components in their
878 directories works correctly. (thanks to Artem V. Andreev)
879 * fixed bug 125: compiler preserves identity of closures. (reported
881 * bug fixed: functions with &REST arguments sometimes failed with
882 "Undefined function" when compiled with (DEBUG 3). (reported by
884 * bug fix: overflow during compiling of setting element of a bit
885 vector with constant index and value. (reported by Timmy Douglas)
886 * build fix: fixed the dependence on *LOAD-PATHNAME* and
887 *COMPILE-FILE-PATHNAME* being absolute pathnames.
888 * on x86 compiler partially supports stack allocation of dynamic-extent
890 * GO and RETURN-FROM do not check the extent of their exit points
891 when compiled with SAFETY 0.
892 * fixed some bugs related to Unicode integration:
893 ** encoding and decoding errors are now much more robustly
894 handled; it should now be possible to recover even from invalid
895 input or output to the terminal. (thanks to Teemu Kalvas)
896 ** provided a first cut at implementing STRING-TO-OCTETS and
897 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
898 ** altered the SB-MD5 contributed module slightly, changing the
899 interface just enough for it to be supportable for builds where
900 lisp characters are not eight bits.
901 * fixed some bugs revealed by Paul Dietz' test suite:
902 ** the FORMATTER-generated functions for ~V[ conditionals require
903 the correct number of arguments.
904 ** READ-FROM-STRING returns the mandated second value when applied
905 to displaced strings.
906 ** the #\Rubout and #\Backspace characters are treated as invalid
907 constituent characters by the tokenizer.
909 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
910 * new feature: reloading changed shared object files with
911 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
912 * new feature: references to foreign variables and functions
913 can now be compiled and loaded before the corresponding shared
914 object file is loaded, as long as the foreign definitions are
915 available at runtime.
916 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
917 supported. (thanks to Dan Debertin)
918 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
919 just keywords. (thanks to Vincent Arkesteijn)
920 * fixed bug #331: structure-class instances corresponding to
921 DEFSTRUCT forms are now created eagerly.
922 * fixed bug #345: backtraces from calls to undefined functions work
923 on x86 as well. Related bug #61 is now also partially fixed on x86
924 (backtraces from throws to unknown catch tags.)
925 * bug fix: lambda-list parsing is now stricter vrt. order and number
926 of lambda-list keywords.
927 * bug fix: as specified by AMOP, an error is signalled if a
928 class-option appears multiple times in a DEFCLASS form. (reported
930 * bug fix: RANDOM can be compiled when the compiler derives the type
931 of its numeric argument as a disjoint set of small integers.
932 (reported by Paul Dietz)
933 * bug fix: starting a core saved with shared objects loaded when
934 those objects are not available no longer causes threaded SBCL to
935 hang. (reported by Sean Ross)
936 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
938 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
939 argument, which must always be NIL. (reported by Kalle Niemitalo)
940 * bug fix: printing 1.0d+23 no longer results in an error.
941 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
942 * bug fix: structure slot setters preserve evaluation order. (thanks
944 * fixed some bugs related to Unicode integration:
945 ** RUN-PROGRAM can allow its child to take input from a Lisp
946 stream. (reported by Stefan Scholl)
947 ** ASDF-INSTALL successfully downloads .tar.gz files in all
948 locales. (reported by Ken Causey)
949 * fixed some bugs revealed by Paul Dietz' test suite:
950 ** INCF, DECF and REMF evaluate their place form as specified in
952 ** FORMATTER expands ~{ ~} iteration directives with V or #
953 parameters correctly.
954 ** FORMATTER deals with the ~@[ ~] conditional directive where the
955 consequent uses no arguments correctly.
956 ** the system has a partial understanding of the (COMPLEX RATIO)
958 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
961 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
962 * new feature: a build-time option (controlled by the :SB-UNICODE
963 keyword feature, enabled by default) for building the system with
964 support for the entire 21-bit character space defined by the
966 * new feature: the system now has rudimentary external-format
967 support; the primary user-visible change at this time is that
968 characters with the high bit set (such as lower-case-e-acute) will
969 print correctly to a terminal in a UTF-8 environment.
970 * minor incompatible change: BASE-CHAR no longer names a class;
971 however, CHARACTER continues to do so, as required by ANSI.
972 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
973 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
974 has been moved to the SB-EXT package (temporarily re-exported from
976 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
977 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
979 * fixed bug #351: better error-handlind and reporting for malformed
981 * fixed bug #350: bignum-printing is now more memory-efficient,
982 allowing printing of very large bignums, eg. (expt 2 10000000).
983 (reported by Bruno Haible)
984 * fixed bug #302: better primitive-type selection for intersection
986 * fixed bug #308: non-graphic characters now all have names, as
987 required. (reported by Bruno Haible)
988 * bug fix: redefining a class with different superclasses now correctly
989 removes it from the direct-subclasses of its previous superclasses.
990 (reported by David Morse)
991 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
992 new value now works. (reported by Bruno Haible)
993 * bug fix: correct canonicalization of multiple non-standard slot
994 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
995 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
996 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
997 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
998 now exists, an signals an error.
999 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1000 during macroexpansion for non-iteration variables as well. (reported
1001 by Bruno Haible for CMUCL)
1002 * bug fix: Cyclic structures and unprintable objects in compiler
1003 messages no longer cause errors. (reported by Bruno Haible)
1004 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1005 and READ-FROM-STRING all now return a primary value of NIL if
1006 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1007 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1008 (reported by Bruno Haible for CMUCL)
1009 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1010 do not designate a condition. (reported by Bruno Haible for
1012 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1013 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1014 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1015 permitting resolution in favour of any of the conflicting symbols.
1016 (reported by Bruno Haible for CMUCL)
1017 * bug fix: EQUAL compiler optimizations is less aggressive on
1018 strings which can potentially compare true despite having distinct
1019 specialized array element types.
1020 * bug fix: unit enumerations can be defined without dividing by
1021 zero. (thanks to Vincent Arkesteijn)
1022 * FORMAT compile-time argument count checking has been enhanced.
1023 (report from Bruno Haible for CMUCL)
1024 * a partial workaround for the bug 262: the compiler does not try to
1025 inline-expand a local function doing RETURN-FROM from a deleted
1026 BLOCK. (thanks to Peter Denno for the bug report and to David
1027 Wragg for the simple test case)
1028 * fixed some bugs revealed by Paul Dietz' test suite:
1029 ** RENAME-PACKAGE allows all package designators as new package
1031 ** constraint propagation and lambda variable substitution are
1032 more cautious in dealing with partially deleted code.
1033 ** compiler sometimes lost reoptimization passes.
1034 ** CERROR, given a condition as condition designator, passes its
1035 remaining arguments to the continue format control without
1037 ** Case-altering FORMAT directives work correctly on non-ASCII
1039 ** The REAL type specifier handles bounds outside the floating
1040 point ranges without signalling FLOATING-POINT-OVERFLOW.
1041 ** Functions with IR1-transformations can create intercomponent
1042 references to global functions.
1043 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1045 ** FORMAT ~R treats a nil value for its first parameter correctly.
1047 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1048 * enhancement: saving cores with foreign code loaded is now
1049 supported on x86/NetBSD and sparc/Linux in addition to the previously
1050 supported platforms.
1051 * bug fix: on some platforms repeated installations caused multiple
1052 copies of HTML documentation to be installed -- should not happen
1053 any more. (reported by Stefan Scholl)
1054 * bug fix: parsing self-recursive alien record types multiple times
1055 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1056 original patch by Helmut Eller for CMUCL)
1057 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1058 (thanks to Richard Kreuter)
1059 * bug fix: defining classes whose accessors are methods on existing
1060 generic functions in other (locked) packages no longer signals
1061 bogus package lock violations. (reported by François-René Rideau)
1062 * bug fix: special variables as DEFMETHOD parameters no longer have
1063 associated bogus type declarations. (reported by David Wragg and
1065 * bug fix: read-write consistency on streams of element-type
1066 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1067 * bug fix: redefiniton of the only method of a generic function with
1068 no DEFGENERIC no longer emits a full WARNING. In addition,
1069 redefinition of generic functions with no DEFGENERIC to an
1070 incompatible lambda list now signals an error. (thanks to Zach
1072 * bug fix: DEFGENERIC now works even when there's a function of the
1073 same name in an enclosing lexical environment. (thanks to Zach
1075 * fixed compiler failure, caused by instrumenting code during
1076 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1077 * optimization: added loop analysis and improved register allocation
1078 * fixed some bugs revealed by Paul Dietz' test suite:
1079 ** POSITION on displaced vectors with non-zero displacement
1080 returns the right answer.
1081 ** (SIMPLE-STRING) is a valid type specifier for sequence
1083 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1085 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1086 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1088 ** The justification version of the FORMAT ~< directive treats
1089 non-zero minpad parameter correctly.
1091 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1092 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1093 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1094 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1095 the supported interface.
1096 * new feature: Single-stepping of code compiled with DEBUG 2 or
1097 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1098 * new feature: saving cores with foreign code loaded is now
1099 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1100 Timothy Moore's work for CMUCL)
1101 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1102 parameters to * instead of NIL if no initform is supplied.
1103 (reported by Johan Bockgaard)
1104 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1105 correctly when FUNCALL appears as the car of the form. Note:
1106 despite this FUNCALL forms are not currently subject to
1107 compiler-macro expansion. (port of Raymond Toy's fix for the
1108 same from CMUCL, reported by Johan Bockgaard)
1109 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1110 (thanks for Teemu Kalvas)
1111 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1112 case instead of terminating the iteration (thanks for Julian
1113 Squires, Sean Champ and Raymond Toy)
1114 * bug fix: incorrect expansion of defgeneric that caused a style
1115 warning. (thanks for Zach Beane)
1116 * bug fix: slot accessor effective method computation works properly
1117 for classes with multiple non-standard applicable methods on
1118 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1119 * on x86 compiler supports stack allocation of results of LIST and
1120 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1121 CMUCL implementation by Gerd Moellmann)
1122 * fixed some bugs revealed by Paul Dietz' test suite:
1123 ** FORMAT strings with both the ~<~:;~> form of the justification
1124 directive and pretty-printing directives cause an error.
1126 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1127 * incompatible change: the internal functions
1128 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1129 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1130 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1131 instead of the old functions.
1132 * new feature: on platforms where "dladdr" is available foreign
1133 function names now appear in backtraces. (based on Helmut Eller's
1135 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1136 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1138 * documentation: ASDF manual is now distributed with SBCL.
1139 * bug fix: SBCL can now load its contributed modules with REQUIRE
1140 even if the system-provided entries have been removed from
1141 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1143 * bug fix: backtraces involving undefined functions or assembly
1144 routines are more informative. (thanks to Brian Downing)
1145 * bug fix: mutually referent alien structures now work correctly.
1146 (reported by Rick Taube)
1147 * bug fix: structures defined by WITH-ALIEN can be referred to
1148 within other definitions in the same WITH-ALIEN.
1149 * bug fix: division operators (MOD, TRUNCATE and the like) with
1150 constant zero divisors and integer dividends no longer generate
1152 * bug fix: provide default methods for INPUT-STREAM-P and
1153 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1154 * bug fix: improve the detection and resolution of MOP metacycles
1155 (where effective method computation depends on the generic function
1156 having its effective method computed). (reported by Bruno Haible)
1157 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1158 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1159 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1160 * bug fix: pathnames with relative directory components can be
1161 represented relative to default pathnames.
1162 * optimization: in taking the GCD of bignums, reduce the two bignums
1163 to approximately the same size (using Euclid's algorithm) before
1164 applying the more sophisticated binary GCD. (thanks to Juho
1166 * optimization: COUNT on bitvectors now operates word-at-a-time.
1167 * optimization: ASH with a positive, but not necessarily constant,
1168 (leftwards) shift, when in a modular context, is compiled to a
1170 * fixed some bugs revealed by Paul Dietz' test suite:
1171 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1172 if the corresponding argument is NIL.
1173 ** FORMAT directives accepting a minpad parameter treat negative
1175 ** literal commas as character parameters to FORMAT directives are
1177 ** literal spaces directly after ~<Newline> directives within a
1178 format-logical-block (~:< ~@:>) do not induce :FILL-style
1179 conditional newlines.
1180 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1181 as specified: it no longer includes conditional newlines).
1182 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1184 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1185 :PER-LINE-PREFIX argument does not evaluate to a string.
1187 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1188 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1189 the manual for details; for now, package locks can be disabled by
1190 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1191 if no major problems are found then it is likely that they will be
1192 compiled in unconditionally.
1193 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1194 unsupported operators on all platforms. To load a shared library
1195 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1196 object file, link it into a shared library outside of SBCL and
1197 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1199 * bug workaround: ROOM T and the GC were not working together
1200 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1201 reporting facility aren't preserved. That reporting has been
1202 disabled, so now until and unless someone figures out how to make it
1203 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1204 * minor incompatible change: as threatened around sbcl-0.8.0, the
1205 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1206 The symbols are also exported from SB-PCL for backwards
1207 compatibility, but more so than before SB-PCL should be treated as
1208 an implementation-internal package.
1209 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1211 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1212 It is known as of this release not to work on the Alpha, however.
1213 * fixed bug #167: errors signalled due to illegal syntax in method
1214 bodies are now more legible.
1215 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1216 designators and can hence be used with TYPEP.
1217 * fixed bug #333: CHECK-TYPE now ensures that the type error
1218 signalled, if any, has the right object to be accessed by
1219 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1220 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1221 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1223 * fixed bug #334: programmatic addition of slots using specialized
1224 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1225 and :ALLOCATION :CLASS slots.
1226 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1227 float exponents. (rereported by Peter Seibel)
1228 * fixed a bug: #\Space (and other whitespace characters) are no
1229 longer considered to be macro characters in standard syntax by
1230 GET-MACRO-CHARACTER.
1231 * fixed bug: initialization of condition class metaobjects no longer
1232 causes an instance of the condition to be created. (reported by
1234 * fixed bug: it is now possible to have slots such that
1235 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1236 description is neither :INSTANCE nor :CLASS.
1237 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1238 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1239 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1240 non-standard SLOT-DEFINITION-ALLOCATION values.
1241 * fixed bugs: various race conditions handling exiting threads.
1242 CL-PPCRE's thread torture test now passes.
1243 * fixed arguably-a-bug: GC time proportional to number of threads in
1244 system even when most of them are idle
1245 * optimization: improved performance of BIT and SBIT on bit-vectors.
1246 * .fasl file incompatibility: The fasl file version number has
1247 been incremented because of changes associated with package locks.
1249 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1250 * minor incompatible change: the system no longer provides
1251 optimization hints (conditions which are TYPEP
1252 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1253 mode; these hints are still emitted when higher SPEED optimization
1255 * new contrib module: a sampling profiler (profiling by statistical
1256 sampling, rather than by instrumenting functions) is available as
1257 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1258 * the behaviour of the standard function ED is now customizeable by
1259 third parties through a hook variable: see ED's documentation
1260 string for information on the protocol.
1261 * the compiler no longer emits efficiency notes for (FUNCALL X)
1262 when the type of X is uncertain under default optimization
1264 * fixed bug 276: mutating a binding of a specialized parameter to a
1265 method to something that is not TYPEP the specializer is now
1267 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1268 exhibit the required behaviour on the x86 platform. (thanks to
1269 Peter van Eynde, Eric Marsden and Bruno Haible)
1270 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1271 even for difficult arguments. (reported by Peter Graves)
1272 * fixed bug 141a: the backquote printer now descends quoted
1274 * fixed another bug in backquote printing: no more destructive
1275 modification of the form's list structure. (reported by Brian
1277 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1278 move between its address being taken and the call to
1279 interrupt_thread, fixing a crashing race condition.
1280 * the SB-POSIX contrib implementation has been adjusted so that it
1281 no longer exhibits ridiculously poor performance when constructing
1282 instances corresponding to C structs.
1284 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1285 * minor incompatible change: the sb-grovel contrib now treats C
1286 structures as alien (in the sense of SB-ALIEN) objects rather than
1287 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1288 has implications for memory management of client code
1289 (sb-grovel-returned objects must now be manually managed) and for
1290 type safety (alien objects now have full types).
1291 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1292 used to control emission of compiler diagnostics, rather than the
1293 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1294 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1295 quality should be considered deprecated.
1296 * install.sh now installs the user manual as well
1297 * (not quite a new documentable feature, but worth considering in
1298 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1299 beginnings of a semantically meaningful condition hierarchy is
1300 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1302 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1303 designator as the defaults argument.
1304 * fixed bug: Displaced arrays whose displaced-to array has become
1305 too small now cause ARRAY-DIMENSION to signal an error, providing
1306 for safer bounds-checking. (reported by Bruno Haible)
1307 * fixed bug: DEFCLASS slot definitions with identical :READER and
1308 :WRITER names now signal a reasonable error. (reported by Thomas
1310 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1312 * fixed bug: Invalid slot specification errors now print correctly.
1313 (thanks to Zach Beane)
1314 * fixed bug 320: Shared to local slot value transfers in class
1315 redefinitions now happen corrently from superclasses as
1316 well. (reported by Bruno Haible)
1317 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1319 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1320 improper lists correctly. (reported by Bruno Haible)
1321 * fixed bug 313: source-transform for <fun-name> was erroneously
1322 applied to a call of a value of a variable with name <fun-name>.
1323 (reported by Antonio Menezes Leitao)
1324 * fixed bug 307: The obsolete instance protocol ensures that
1325 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1326 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1327 remove dead unknown-values globs from the middle of the stack.
1328 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1329 easier to resume long computations after using *BREAK-ON-SIGNALS*
1330 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1331 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1332 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1333 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1334 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1335 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1337 * fixed a (fairly theoretical) bug in string printing: if
1338 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1339 does not have array-element-type equal to the most general string
1341 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1342 function name. (thanks to Nikodemus Siivola)
1343 * fixed bug: compile-time format string checker failed on
1344 non-closed ~{. (reported by Thomas F Burdick)
1345 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1346 don't warn on legal loop code involving more than one aggregate
1347 boolean. (thanks to Nikodemus Siivola)
1348 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1349 in keyword position to MAKE-PACKAGE was not being reported as
1351 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1352 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1354 * fixed bugs 280 and 312: the checking for multiple definitions in a
1355 file is less likely to become confused by uses of inline
1357 * fixed bug: the #S reader macro performs the keyword coercion
1358 specified for slot names. (reported by Kalle Niemitalo)
1359 * fixed bug: lambda lists may contain symbols whose names start with
1360 & but are not lambda-list-keywords; their occurrence triggers a
1362 * fixed bug 321: define-method-combination argument lambda lists do
1363 not cause invalid code to be generated when &optional or &aux
1364 variables have default values. (reported by Bruno Haible)
1365 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1366 have CLOS classes; this appears also to have fixed reported
1367 instability in stack exhaustion detection.
1368 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1369 MUFFLE-WARNING when no associated restart is present is now
1371 * optimization: rearranged the expansion of various defining macros
1372 so that each expands into only one top-level form in a
1373 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1375 * optimization: used a previously little-used slot in symbols to
1376 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1378 * fixed some bugs revealed by Paul Dietz' test suite:
1379 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1381 ** Fixed an optimization bug related to inheritance of initforms
1382 from local to shared slots.
1383 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1384 ** CAST optimizer forgot to flush argument derived type.
1385 ** print/read consistency on floats is now orders of magnitude
1386 more likely. (thanks also to Bruno Haible for a similar report
1388 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1389 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1390 for objects of type REAL. Make it so.
1391 ** SXHASH is specified (infelicitously) to respect similarity,
1392 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1393 it so. (thanks to Markus Ziegler)
1394 ** on the Alpha, the compiler succeeds in compiling functions
1395 returning a known number of arguments greater than 63.
1396 ** fixed handling of invalid NIL arguments in keyword position
1398 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1400 ** provided a readably-printable representation for RANDOM-STATE
1402 ** ensured that pathnames, where they have a namestring, always
1403 print using #P"..." syntax.
1405 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1406 * Support for the forthcoming 2.0 version of the NetBSD kernel
1407 running on x86 hardware has been added. (thanks to Perry
1408 E. Metzger most immediately, and others for their past work)
1409 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1410 current release version; SBCL's OpenBSD support had been broken
1411 since about the time of OpenBSD's switch to ELF binary format.)
1412 * [placeholder for DX summary]
1413 ** user code with &REST lists declared dynamic-extent, under high
1414 speed or space and low safety and debug optimization policy.
1415 * The manual has been converted to Texinfo format and the debugger
1416 chapter from the cmucl manual has been added.
1417 * A facility has been added to extract documentation strings from
1418 sbcl and store them as Texinfo-formatted snippets for inclusion in
1419 the manual (via Texinfo's @include directive)
1420 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1421 the test case to Dave Roberts)
1422 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1423 pointers. (reported by Sean Ross)
1424 * bug fix: PROFILE output is printed nicely even for large numerical
1425 values. (thanks to Zach Beane)
1426 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1427 greater than 32 handle EOF correctly.
1428 * bug fix: on X86 an immediate argument of the IMUL instruction is
1429 correctly printed in disassembly. (thanks to Lutz Euler)
1430 * bug fix: class slots in redefined classes preserve their old
1431 values. (thanks to Bruno Haible and Nikodemus Siivola)
1432 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1433 succeeds. (reported by Marco Baringer)
1434 * fixed some bugs revealed by Paul Dietz' test suite:
1435 ** READ-SEQUENCE now works on ECHO-STREAMs.
1436 ** RATIONALIZE works more according to its specification. (thanks
1439 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1440 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1441 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1442 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1443 should matter to you only if you rebind the printer control
1444 variables and then find you want different bindings in the
1445 debugger than in the ordinary execution of your program.)
1446 * The runtime build system has been tweaked to support building
1447 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1448 assembler and linker. (thanks to Nikodemus Siivola)
1449 * Unbound, undefined, undeclared variables now trigger full
1450 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1451 more often programmer error than deliberate exploitation of undefined
1453 * optimization: the hash algorithm for strings has changed to one
1454 that is less vulnerable to spurious collisions. (thanks to Juho
1456 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1457 needless bounds checking. (thanks to Juho Snellman)
1458 * optimization: implemented multiplication as a modular
1459 (UNSIGNED-BYTE 32) operation on the PPC backend.
1460 * fixed some bugs revealed by Paul Dietz' test suite:
1461 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1463 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1464 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1465 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1467 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1469 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1471 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1472 their output stream on EOF from read.
1473 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1474 have been read to end-of-file.
1475 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1477 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1478 description of determination of which consecutive characters
1480 ** Printing the "Space" character with escaping on now yields "#\\ ",
1481 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1482 ** Reading floating-point numbers with *READ-BASE* set to a number
1483 less than 10 works correctly.
1484 ** Reading floating-point numbers with *READ-BASE* set to a number
1485 more than 10 works correctly.
1486 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1487 the readtable currently in effect.
1489 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1490 * minor incompatible change: parsing of namestrings on a physical
1491 (Unix) host has changed; numbers after the final #\. in a
1492 namestring are no longer interpreted as a version field. This is
1493 intented to be largely invisible to the user, except that the
1494 meaning of the namestring "*.*.*" has changed: it now refers to a
1495 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1496 should usually be replaced by
1497 (make-pathname :name :wild :type :wild :version :wild)
1498 with the added benefit that this is more likely to be portable.
1499 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1500 option to OPEN now signals an error if the file being opened
1501 exists; this may have an impact on existing code.
1502 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1503 signals received, on the PowerPC platforms (both Linux and
1504 Darwin). (thanks to Pierre Mai for pointing out the location of
1506 * several fixes on OS X: The system now builds and runs cleanly on
1507 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1508 (thanks to Brian Mastenbrook)
1509 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1510 second-resolution: integer multiples of 1/3600 between -24 and 24.
1511 (thanks to Vincent Arkesteijn)
1512 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1513 type when called with 1 argument; PEEK-CHAR checked type of
1514 PEEK-TYPE only after having read first character from a
1515 stream. (reported by Peter Graves)
1516 * bug fix: the garbage collector now has much better locality
1517 behaviour, and in particular no longer treats all memory as being
1518 exhausted when many small objects point to each other in a deeply
1520 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1521 recognized as being TYPEP their class.
1522 * bug fix: the PUSHNEW documentation string has been corrected.
1523 (thanks to Vincent Arkesteijn)
1524 * bug fix: defaulting of the value for the last of an atypically
1525 large number of multiple values being bound was not being
1526 performed correctly on the Alpha or PPC platforms
1527 * optimization: implemented multiplication as a modular
1528 (UNSIGNED-BYTE 32) operation on the x86 backend.
1529 * optimization: SEARCH on simple-base-strings can now be open-coded.
1530 (see also contrib/compiler-extras.lisp for inspiration for
1531 teaching the compiler about the Boyer-Moore algorithm).
1532 * value, returned by MAX (and MIN) called with several EQUALP, but
1533 not EQL, arguments now does not depend on compiler settings.
1534 * fixed some bugs revealed by Paul Dietz' test suite:
1535 ** in stack analysis liveness information is propagated from
1536 non-local entry points.
1537 ** pathwise CAST removing failed when the CAST node did not start
1539 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1540 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1542 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1543 host is already defined.
1544 ** RENAME-FILE works on streams instead of signalling an internal
1546 ** PEEK-CHAR uses the current readtable when determining whether
1547 or not a character is whitespace.
1548 ** MERGE-PATHNAMES handles the case when the pathname does not
1549 specify a name while the default-pathname specifies a version.
1550 ** Pathnames now stand a better chance of respecting print/read
1552 ** Attempting to use standardized file system operators with a
1553 pathname with invalid :DIRECTORY components signals a
1555 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1557 ** DIRECTORY on logical pathnames is more correct.
1558 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1559 signal a TYPE-ERROR if their argument is not a stream.
1560 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1561 designator argument does not designate a stream.
1562 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1563 examining the synonym.
1564 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1566 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1567 element-type larger than ([UN]SIGNED-BYTE 32).
1569 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1570 * When built with the :SB-FUTEX feature, threaded builds now take
1571 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1572 for faster/more reliable mutex and condition variable support.
1573 * Incompatible change (but one you probably shouldn't have been using
1574 anyway): the interface and code for arbitrating between multiple
1575 threads in the same user session has been redesigned.
1576 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1577 error when called without an explicit environment argument.
1578 (thanks to Nikodemus Siivola)
1579 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1580 confused about their position. (thanks to Adam Warner and Gerd
1582 * bug fix: Pretty printing backquoted forms with unquotations in the
1583 argument list position of various code constructs such as LAMBDA
1584 now works correctly. (reported by Paul Dietz)
1585 * bug fix: Pretty printing unquotations no longer loses all
1586 stream position information.
1587 * optimization: performance of string output streams is now less
1588 poor for multiple small sequence writes.
1589 * optimization: performance of CSUBTYPEP in the presence of complex
1590 expressions involving CONS and NOT many times has been improved.
1591 (reported by Paul Dietz)
1592 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1594 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1595 streams for the REPL, for the debugger, and so on. (thanks to
1597 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1598 * fixed some bugs revealed by Paul Dietz' test suite:
1599 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1600 is ignored if it is too small, rather than propagating through
1601 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1602 ** extremely complex negations of CONS types were not being
1603 sufficiently canonicalized, leading to inconsistencies in
1605 ** VALUES tranformer lost derived type.
1607 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1608 * fixed a bootstrapping bug: the build process no longer assumes
1609 that the various BOOLE-related constants have the same value in
1610 host and target lisps. (noted by Paul Dietz' test suite on an
1611 SBCL binary built from CLISP)
1612 * The system can now be dynamically linked on the MIPS platform,
1613 which enables dynamic loading of foreign code from Lisp. (thanks
1614 to Ralf Baechle for discussions on the MIPS ABI)
1615 * The system now records debugging information for its own source
1616 files in a filesystem-position-independent manner, relative to
1617 the "SYS" logical host.
1618 * fixed a compiler bug: MV-LET convertion did not check references
1619 to the "max args" entry point. (reported by Brian Downing)
1620 * tweaked disassembly notes to be less confident about proclaiming
1621 some instruction as an LRA. (thanks to Brian Downing)
1622 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1623 now each have their own history, command character, and other
1624 characteristics. (thanks to David Lichteblau)
1625 * fixed some bugs revealed by Paul Dietz' test suite:
1626 ** compiler failure in compiling LOGAND expressions including a
1628 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1629 shift greater than 32.
1630 ** FLUSH-DEST did not mark blocks for type check regeneration.
1631 ** HANDLER-CASE failed to accept declarations in handler clauses
1632 in some circumstances.
1634 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1635 * New code in contrib/sb-introspect (still probably not entirely
1636 stable yet) provides some support for smart Lisp development
1637 environments like SLIME.
1638 * The conditions signalled for errors occurring when loading .fasl
1639 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1640 in a way which should help ASDF recover gracefully.
1641 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1642 stuff has been cleaned up. If you code contrib/ stuff, this might
1643 affect you, and you can look at contrib/README, contrib/STANDARDS,
1644 and/or the 0.8.4.27 diff to check.
1645 * In full calls the compiler now does not generate checks for declared
1646 argument types for all arguments.
1647 * various threading fixes
1648 ** and some experimental patches which didn't make it into
1649 the main tree for this release, but which are shipped in
1650 contrib/experimental-thread.patch as a possible fix for some
1651 failures (deadlock, spinning...) in GC-intensive multithreaded
1653 * fixed PPC build problem (source code incompatibility of different
1654 library versions): added offsetof() hackery which attempts to divine
1655 where glibc maintainers put uc_mcontext today
1656 * fixed bug 282: compiler does not trust type assertions while passing
1657 arguments to a full call.
1658 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1659 &OPTIONAL VALUES type parameter.
1660 * fix bug 214: algorithm for noting rejected templates is now more
1661 similar to that of template seletion. (also reported by rydis on
1663 * fixed bug 141b: printing backquoted information readably and prettily
1664 inserts a space where necessary.
1665 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1666 (CHARACTER 10) are now reported as errors, rather than propagated
1667 as unknown types. (reported by piso on #lisp)
1668 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1669 with values NIL and :ERROR. (thanks to Milan Zamazal)
1670 * fixed bug 191c: CLOS now does proper keyword argument checking as
1671 described in CLHS 7.6.5 and 7.6.5.1.
1672 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1673 longer raise an error; further, using a list as a for-as-arithmetic
1674 counter now raises a meaningful error.
1675 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1676 now understood by sequence creation functions such as MAKE-SEQUENCE
1678 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1679 a STRING-INPUT-STREAM.
1680 * compiler enhancement: SIGNUM is now better able to derive the type
1682 * type declarations inside WITH-SLOTS are checked. (reported by
1684 * fixed some bugs revealed by Paul Dietz' test suite:
1685 ** incorrect optimization of TRUNCATE for a positive first
1686 argument and negative second.
1687 ** compiler failure in let-convertion during flushing dead code.
1688 ** compiler failure while deriving type of TRUNCATE on an
1689 interval, containing 0.
1690 ** ASH of a negative bignum by a negative bignum count now returns
1692 ** intersection of CONS types now canonicalizes properly, fixing
1693 inconsistencies in SUBTYPEP.
1695 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1696 * incompatible change: The --disable-debugger command line
1697 option now clobbers the debugger at a more fundamental
1698 level, by redefining #'INVOKE-DEBUGGER instead of by
1699 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1700 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1701 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1702 enter the debugger REPL and then suffer recursive errors
1703 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1704 while under the new system, BREAK in --disable-debugger mode
1705 terminates the system just as an unhandled error would.
1706 * fixed compiler performance when processing loops with a step >1;
1707 * bug fix: DOCUMENTATION now retrieves generic function
1708 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1709 support has been systematized, and now supports the methods
1710 specified by ANSI, along with a default method and a method for
1711 slot documentation. (reported by Nathan Froyd)
1712 * bug fix: effective methods associated with a generic function are
1713 no longer cached over a change of that generic function's method
1714 combination. (reported by Andreas Fuchs)
1715 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1716 names a non-existent directory. (thanks to Andreas Fuchs)
1717 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1718 the PPC platform, lead to stack corruption; this has been fixed.
1719 (reported by Rainer Joswig)
1720 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1721 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1722 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1723 arithmetic in general, is now much more reliable on the PPC
1725 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1726 platform now returns the right answer.
1727 * optimization: restored some effective method precomputation in
1728 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1729 precomputation is now tunable.
1730 * optimization: compiler-internal data structure use has been
1731 reviewed, and changes have been made that should improve the
1732 performance of the compiler by about 20%.
1733 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1734 simple-base-strings and simple-bit-vectors is improved.
1735 * optimization: the optimization of 32-bit logical and arithmetic
1736 functions introduced in version 0.8.3 on the x86 has been
1737 implemented on the mips, ppc and sparc platforms; an
1738 implementation of the same facility, but for 64-bit arithmetic,
1739 has been added for the alpha.
1740 * microoptimization: the compiler is better able to make use of the
1741 x86 LEA instruction for multiplication by constants.
1742 * bug fix: in some situations compiler did not report usage of
1743 generic arithmetic in (SPEED 3) policy.
1744 * bug 145b fix: compiler used wrong type specifier while converting
1745 MEMBER-types to numeric.
1746 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1748 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1749 index argument now works properly on the Alpha platform.
1750 * bug fix: floating point exception treatment on the Alpha platform
1752 * bug fix: FILE-POSITION works much better on string input and
1753 output streams. (thanks to Nikodemus Siivola)
1754 * bug fix: many threading/garbage collection symptoms sorted.
1755 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1756 might be pseudo-atomic.
1757 * internal change: Stopping for GC is now done with signals not
1758 ptrace. GC is now done in whichever thread wanted it, instead of
1760 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1762 * bug fix: result form in DO is not contained in the implicit
1764 * incompatible change: ICR structure is changed; the value part of
1765 CONTINUATION is now called LVAR; corresponding functions are
1766 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1767 * added type deriver for ISQRT (thanks to Robert E. Brown).
1768 * bug fix: better support for loading from the command line when an
1769 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1771 * fixed some bugs revealed by Paul Dietz' test suite:
1772 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1773 ** ROUND and FROUND now give the right answer when given very
1774 small float arguments.
1775 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1777 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1778 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1779 ** LCM should return a non-negative integer.
1780 ** PARSE-INTEGER returned the index of a terminator instead of the
1781 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1782 ** PARSE-INTEGER returned an incorrect index being applied to a
1784 ** LCM with two arguments of 0 returns 0 rather than signalling
1786 ** unsigned addition of a 32-bit constant with the high bit set no
1787 longer causes an internal compiler error on the x86.
1788 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1789 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1790 with negative last argument.
1791 ** byte specifiers with zero size and position no longer cause
1792 an error during type derivation.
1793 ** bignum multiplication on the Alpha platform now returns the
1795 * porting: The system now builds on SuSE AMD64, although it still
1796 generates a 32-bit binary.
1797 * .fasl file incompatibility: The fasl file version number has
1798 been incremented (because of the changes to internal compiler
1799 data structures referred to above).
1801 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1802 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1803 more accurately, on the Darwin kernel running on PowerPC hardware.
1804 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1805 * Compiler code deletion notes now signal a condition of type
1806 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1807 an associated MUFFLE-WARNING restart.
1808 * The compiler now performs limited argument count validation of
1809 constant format strings in FORMAT, and where appropriate in ERROR,
1810 CERROR and WARN. (thanks to Gerd Moellmann)
1811 * New ASDF-INSTALL contrib can be used for automatic download and
1812 installation of third-party Lisp code from CCLAN or other sites
1814 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1815 which forces another thread to execute a function supplied by the
1817 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1818 now accept and act upon their :ELEMENT-TYPE keyword argument.
1819 (reported by Martin Atzmueller, Edi Weitz)
1820 * bug fix: FILE-POSITION now accepts position designators up to
1821 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1822 is the greater. (thanks to Patrik Nordebo)
1823 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1824 Kalle Olavi Niemitalo)
1825 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1826 printable. (reported by Eric Marsden)
1827 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1828 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1829 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1830 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1831 (thanks to Stig Sandoe)
1832 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1833 UNREAD-CHAR now correctly includes the unread character in the
1834 target sequence. (thanks to Gerd Moellmann)
1835 * bug fix in threaded builds: the system can now be suspended and
1836 resumed by shell job control with minimal disruption.
1837 * bug fixes in times and timezones >2038AD
1838 * better handling of "where is GNU make?" problem in build scripts
1839 (thanks to Nikodemus Siivola)
1840 * new optimization: inside a named function any reference to a
1841 function with the same name is considered to be a self-reference;
1842 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1843 optimization quality.
1844 * new optimization on x86: logical functions and + now have
1845 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1846 used when the result is truncated to 32 bits.
1847 * VALUES declaration is partially enabled.
1848 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1849 * fixed some bugs revealed by Paul Dietz' test suite:
1850 ** The system now obeys the constraint imposed by
1851 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1852 types form a lattice under type intersection.
1853 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1854 ** ASSOC now ignores NIL elements in an alist.
1855 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1856 and (1+ MOST-POSITIVE-FIXNUM) answers.
1857 ** The addition of a method with invalid qualifiers to a generic
1858 function does not cause an error to be signalled immediately;
1859 a warning is signalled, and the error is generated only on
1860 calling the generic function.
1861 * changed .fasl file version number, in part to add type codes for
1862 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1863 obscure ANSI requirements
1865 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1866 * fixed bug 148: failure to inline-expand a local function left
1867 garbage, confusing the compiler.
1868 * fixed bugs 3cd: structure slot readers perform type check if the
1869 slot can have an invalid value (i.e. it is either not initialized
1870 or can be written with a less specific slot writer).
1871 * bug fix: the compiler now traps array references to elements off
1872 the end of an array; previously, the bounds checking in some
1873 circumstances could go off-by-one.
1874 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1876 * type declarations for array element types now obey the description
1877 on the CLHS page "Declaration TYPE", as per discussions on
1878 sbcl-help around 2003-05-08. This means that a declaration
1879 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1880 declaration, all references to BAR will be asserted or assumed
1881 (with THE, so dependent on compiler policy) to involve objects of
1882 type FOO. Note that no such declaration is implied in
1883 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1884 * declared types of functions from the "Conditions"
1885 chapter. (reported by Paul Dietz)
1886 * bug fix: CERROR accepts a function as its first argument.
1887 * bug fix: NTH an NTHCDR accept a bignum as index
1888 arguments. (reported by Adam Warner)
1889 * optimization: character compare routines now optimize comparing
1890 against a constant character. (reported by Gilbert Baumann)
1891 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1892 argument now works properly on the MIPS platform.
1893 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1894 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1895 declared to have a functional type, does not cause an error
1897 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1898 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1899 * fixed bug: if last continuation of a deleted block has a
1900 destination, this destination should be deleted too. (reported by
1902 * fixed a bug in the bootstrap process: the host compiler's values
1903 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1904 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1905 test case from Patrik Nordebo)
1906 * improved the ability of the disassembler on the PPC platform to
1907 provide helpful disassembly notes.
1908 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1909 the class in more cases than previously.
1910 * bug fix: FILE-POSITION now understands :START and :END for
1911 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1912 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1913 8) data. (thanks to David Lichteblau)
1914 * bug fix: it is possible to add a method to a generic function
1915 without lambda list.
1916 * bug fix: reader failed to signal END-OF-FILE inside an
1917 object representation. (reported by Nikodemus Siivola)
1918 * fixed some bugs revealed by Paul Dietz' test suite:
1919 ** LAST and [N]BUTLAST should accept a bignum.
1920 ** condition slot accessors are methods.
1921 ** (VECTOR NIL) is a subtype of STRING.
1923 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1924 * minor incompatible change: some nonsensical specialized lambda
1925 lists (used in DEFMETHOD) which were previously ignored now signal
1927 * minor incompatible change: the system is now aware of the types of
1928 variables in the COMMON-LISP package, and will signal errors for
1929 most violations of these type constraints (where previously they
1930 were silently accepted).
1931 * minor incompatible change: COMPILE-FILE now uses the freedom
1932 afforded (ANSI 3.2.2.3) to use derived function types for
1933 functions defined in the same file. This also permits the system
1934 to warn on static type mismatches and function
1935 redefinition. (Currently it does not work with high DEBUG level.)
1936 * minor incompatible change: VALUES declaration is disabled.
1937 * When issuing notes, the compiler now signals a condition of type
1938 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1939 restart for use in user handlers. It is expected that the
1940 COMPILER-NOTE condition will eventually become a condition
1941 supertype to a hierarchy of note types, which will then be
1942 handleable in a similar fashion. However, at the moment, no such
1943 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1944 but it's an implementation detail, not a classification for the
1946 * Changes in type checking closed the following bugs:
1947 ** type checking of unused values (192b, 194d, 203);
1948 ** template selection based on unsafe type assertions (192c, 236);
1949 ** type checking in branches (194bc).
1950 * A short form of VALUES type specifier has ANSI meaning (it has
1951 increased the number of situations when SBCL cannot perform type
1953 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1954 keywords or constants is permissible.
1955 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1956 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1957 classes. (thanks to Antonio Martinez)
1958 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1959 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1960 argument to a pretty-print stream. Instead, it leaves any such
1961 operations to DESCRIBE-OBJECT methods.
1962 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1964 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1965 lambda lists. (thanks to David Lichteblau)
1966 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1967 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1968 (thanks to Gerd Moellmann)
1969 * fixed bug 246: increased compilation speed of long
1970 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1972 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1974 * DEFCONSTANT now throws a condition of type
1975 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1976 constant to a non-EQL value; CONTINUE and ABORT restarts
1977 respectively change and preserve the value.
1978 * fixed bug 63: The code walker, part of the implementation of CLOS,
1979 is now better at handling symbol macros.
1980 * bug fix: There is no longer an internal implementation type named
1981 CL:LENGTH. (reported by Raymond Toy)
1982 * bug fix: In macro-like defining macros/special operators the
1983 implicit block does not enclose the lambda list.
1984 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1985 enclosed in parentheses) are not suitable as type specifiers, and
1986 their use properly signals an error now.
1987 * bug fix: An argument count mismatch for a type specifier in code
1988 being compiled no longer causes an unhandled error at compile
1989 time, but signals a compile-time warning.
1990 * fixed simple vector readable printing
1991 * bug fix: DESCRIBE takes more care over whether the class
1992 precedence list slot of a class is bound before accessing it.
1993 (reported by Markus Krummenacker)
1994 * bug fix: FORMATTER can successfully compile pretty-printer format
1995 strings which use variants of the ~* directive inside.
1996 * bug fix: SEARCH now applies its TEST predicate to the elements of
1997 the arguments in the correct order. (thanks to Wolfhard Buss)
1998 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1999 in type inference. (thanks to Robert E. Brown)
2000 * bug fix: Reading in symbols with an explicit package name of ""
2001 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2002 * fixed some bugs revealed by Paul Dietz' test suite:
2003 ** NIL is now allowed as a structure slot name.
2004 ** Arbitrary numbers, not just REALs, are allowed in certain
2005 circumstances in LOOP for-as-arithmetic clauses.
2006 ** Multiple class redefinitions before slot access no longer
2007 causes a type error.
2008 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2009 association between the name and a class.
2010 ** Generic functions with non-standard method-combination and over
2011 five methods all of which return constants no longer return NIL
2012 after the first few invocations. (thanks to Gerd Moellmann)
2013 ** CALL-NEXT-METHOD with no arguments now passes the original
2014 values of the arguments, even in the presence of assignment.
2015 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2017 ** DISASSEMBLE works with closures and funcallable instances.
2018 ** ADD-METHOD now returns the generic function, not the new method.
2019 ** FIND-METHOD signals an error if the lengths of the specializers
2020 is incompatible with the generic function, even if the ERRORP
2022 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2023 which its argument is a member.
2024 ** DEFCLASS only redefines the class named by its class-name
2025 argument if that name is the proper name of the class;
2026 otherwise, it creates a new class.
2027 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2028 of the UNBOUND-SLOT condition to the name of the slot.
2029 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2030 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2031 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2032 SLOT-MAKUNBOUND in the specified fashion.
2034 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2035 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2036 cross-compilation host. As a consequence, we can now bootstrap our
2037 way up to SBCL starting with a bare gcc toolchain and human-readable
2038 source code (first the source to CLISP, then the source to SBCL).
2039 * A contributed module containing a partial implementation of the
2040 simple-streams interface has been included. (thanks to Rudi
2042 * A contributed module implementing the RFC1321 Message Digest
2043 Algorithm, known as MD5, has been included.
2044 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2045 no longer has any effect, as the code controlled by this feature
2046 has been deleted. (As far as we know, no-one has ever built using
2047 this feature, and its semantics were confused in any case).
2048 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2049 work on conditions (as required by the ANSI specification),
2050 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2051 expected behaviour on conditions. Users should note, however,
2052 that such behaviour is not required by the ANSI specification,
2053 and so use of this behaviour may render their code unportable.
2054 * fixed some bugs revealed by Paul Dietz' test suite:
2055 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2057 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2058 ** SXHASH on condition objects no longer returns NIL.
2059 ** :ALLOCATION :CLASS slots are better treated; their values are
2060 updated on class redefinition, and initforms inherited from
2061 superclasses are applied.
2062 ** REMOVE-METHOD returns its generic function argument even when
2063 no method was removed.
2064 ** SHARED-INITIALIZE now initializes the values of the requested
2065 slots, including those with :ALLOCATION :CLASS.
2066 ** ALLOCATE-INSTANCE now works on structure classes defined via
2067 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2069 ** SLOT-EXISTS-P now works on conditions, as well as structures
2071 ** MAKE-LOAD-FORM now has the required methods on
2072 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2073 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2074 keyword, and now implements the SLOT-NAMES argument.
2075 ** methods with &OPTIONAL arguments no longer allow too many
2076 arguments to be passed in the call without error.
2077 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2078 option is consistent with the required arguments of the generic
2079 function lambda list.
2080 * bug fix: REQUIRE accepts a string designator. (Thanks to
2082 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2083 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2084 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2085 * bug fix: The compiler now checks for duplicated variables in macro
2087 * bug fix: SETQ on globals returns the correct value.
2088 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2089 not a COMPILER-ERROR (followed by some other strange error on
2090 choosing the CONTINUE restart).
2091 * bug fix: make.sh and friends are now more consistent in the way that
2092 they look for GNU "make".
2094 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2095 * experimental native threads support (on x86 Linux >=2.4 only).
2096 This is not compiled in by default: you need to add :SB-THREAD to
2097 the target features. See the "Beyond ANSI" chapter of the manual
2099 * fix for longstanding nonANSIism: The old distinction between
2100 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2101 The return value from CL:FIND-CLASS is now a CLOS class, and
2102 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2103 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2104 * An interface to the MetaObject Protocol, as described in Kiczales,
2105 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2106 MIT Press, 1991, is available from the SB-MOP package.
2107 * incompatible change: the SB-PCL package should now be considered
2108 a private implementation detail, and no longer a semi-private MOP
2110 * minor incompatible change: due to rearrangement for threads, the
2111 control stack and binding stack are now allocated at arbitrary
2112 addresses instead of being hardcoded per-port. Users affected by
2113 this probably have to be doing advanced things with shared
2114 libraries, and will know who they are.
2115 * minor incompatible change: Previously, all --eval forms used were
2116 processed with READ before any of them were processed with EVAL.
2117 Now each --eval form is processed with both READ and EVAL before
2118 the next --eval form is processed. (Thus package operations like
2119 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2120 work as the user might reasonably expect.)
2121 * minor incompatible change: *STANDARD-INPUT* is now only an
2122 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2124 * minor incompatible change: Y-OR-N-P is now character-oriented,
2125 not line oriented. Also, YES-OR-NO-P now works without errors.
2126 (thanks to Antonio Martinez)
2127 * sb-aclrepl module improvements: an integrated inspector, added
2128 repl features, and a bug fix to :trace command.
2129 * Known functions, which cannot be open coded by the backend, are
2130 considered to be able to check types of their arguments. (fixing
2131 a bug report by Nathan J. Froyd)
2132 * fixed a bug in computing method discriminating functions: It is
2133 now possible to define methods specialized on classes which have
2134 forward-referenced superclasses. (thanks to Gerd Moellmann)
2135 * fixed evaluation order in optional entries (reported by Gilbert
2137 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2138 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2139 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2140 required-by-AMOP NAME argument, as well as CLASS and
2141 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2142 * fixed bug 20: DEFMETHOD can define methods using names that are
2143 not the proper names of classes to designate class specializers.
2144 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2145 Unix file descriptors, instead of blowing up. (thanks to Antonio
2147 * Garbage collection refactoring: user-visible change is that a
2148 call to the GC function during WITHOUT-GCING will not do garbage
2149 collection until the end of the WITHOUT-GCING. If you were doing
2150 this you were probably losing anyway.
2151 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2152 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2153 TYPEP the latter but not the former.
2154 * The compiler issues a full WARNING for calls to undefined functions
2155 with names from the CL package.
2156 * MAP-INTO for a vector destination is open coded. (reported by
2157 Brian Downing on c.l.l)
2158 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2159 documentation string.
2160 * fixed some bugs revealed by Paul Dietz' test suite:
2161 ** COPY-ALIST now signals an error if its argument is a dotted
2163 ** Condition slots are now accessed more correctly in the presence
2164 of multiple initargs for a given slot.
2165 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2166 exclude restarts of the same name associated with a different
2168 ** DEFCLASS of forward-referenced classes with another
2169 forward-referenced class in the superclasses list no longer
2171 ** Condition slots are now initialized once each, not multiple
2172 times. (thanks to Gerd Moellmann)
2173 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2175 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2176 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2177 arguments contain duplicated elements.
2178 ** RESTART-CASE understands local macros.
2179 ** RESTART-CASE associates exactly its own restarts with a condition.
2180 ** ENDP in safe mode checks its argument to be of type LIST.
2181 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2182 in question is unbound.
2183 ** Optimized MAKE-INSTANCE functions no longer cause internal
2184 assertion failures in the presence of duplicate initargs.
2185 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2186 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2187 propagated even in the ctor optimized implementation of
2189 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2191 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2192 * a better implementation of SXHASH on (simple) bit vectors,
2193 measured both in execution speed and in distribution of results
2194 over the positive fixnums, has been installed. Likewise, a better
2195 implementation of EQUAL for simple bit vectors is now available.
2196 * fixed CEILING optimization for a divisor of form 2^k.
2197 * fixed bug 240 (emitting extra style warnings "using the lexical
2198 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2199 by Antonio Martinez)
2200 * fixed SXHASH, giving different results for NIL depending on type
2201 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2202 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2203 variable. (found by Rolf Wester)
2204 * SBCL does not ignore type declarations for special
2205 variables. (reported by rif on c.l.l 2003-03-05)
2206 * some bug fixes in contrib/sb-aclrepl/
2207 * fixed some bugs revealed by Paul Dietz' test suite:
2208 ** a bug in the CONS type specifier, whereby the CAR and CDR
2209 types got intertwined, has been fixed;
2210 ** the type system is now able to reason about the interaction
2211 between INTEGER and RATIO types more completely;
2212 ** APPEND, [N]REVERSE and NRECONC check that those their
2213 arguments, which must be proper lists, are really so;
2214 ** An array specialized to be unable to hold elements has been
2215 implemented, as required -- yes, really -- by ANSI;
2216 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2217 on malformed property lists;
2219 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2220 * incompatible packaging change: in line with Unix convention,
2221 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2222 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2223 * REQUIRE and PROVIDE are now optionally capable of doing something
2224 useful. See the documentation string for REQUIRE.
2225 * infrastructure for a managed SBCL contrib system: contributed
2226 modules in this release include:
2227 ** the ASDF system definition facility;
2228 ** an interface to the BSD Sockets API;
2229 ** an ACL-like convenience interface to the repl;
2230 (thanks to Kevin Rosenberg)
2231 ** an implementation of ROTATE-BYTE, with efficient implementation
2233 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2234 gives a closer approximation to the right answer than previously.
2235 (thanks to Raymond Toy)
2236 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2237 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2238 optional environment arguments, as required by ANSI.
2239 * fixed bugs in other functions taking environment objects, allowing
2240 calls with an explicit NIL environment argument to be compiled
2242 * fixed bug 228: primary return values from
2243 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2244 COMPILE or FUNCTION.
2245 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2246 :TYPE will no longer signal errors on innocuous objects.
2247 * fixed bug 231b: SETQ is better at respecting type declarations in
2248 the lexical environment.
2249 * fixed a bug in DEFCLASS: classes named by symbols with no or
2250 unprintable packages can now be defined.
2251 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2252 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2253 * fixed compiler failure related to checking types of functions.
2254 (reported by Robert E. Brown)
2255 * the compiler is now much more consistent in its error-checking
2256 treatment of bounding index arguments to sequence functions: in
2257 (SAFETY 3) code, errors will be signalled in almost all cases if
2258 invalid sequence bounding indices are passed to functions defined
2259 by ANSI to operate on sequences.
2260 * fixed a bug in the build procedure: documentation of SBCL-specific
2261 packages is now preserved and available in the final Lisp image.
2262 * lifted FDEFINITION lookup out of loops in the implementation of
2263 many list operations. (thanks to Robert E. Brown)
2264 * fixed a bug in the reader: the #n# reader macro now works for
2265 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2266 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2267 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2268 E. Brown and Tony Martinez respectively)
2269 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2270 SLOT-BOUNDP: in particular, it is now able to optimize them much
2271 better, and is now not vulnerable to having packages renamed.
2272 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2273 always called when a slot is not present in an instance. (thanks
2275 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2276 longer signal an error on unreading a character following EOF on
2277 the previous constituent stream. (thanks to Tony Martinez)
2278 * fixed some bugs revealed by Paul Dietz' test suite:
2279 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2280 not just nonnegative fixnums;
2281 ** the logical bit-array operators such as BIT-AND now accept an
2282 explicit NIL for their "opt-arg" argument (to indicate a
2283 freshly-consed result bit-array);
2284 ** ELT now signals an error on an invalid sequence index in safe
2286 ** the type system is now cleverer about negations of numeric
2287 types, and consequently understands the BIGNUM and RATIO types
2289 ** the type system is now cleverer about the interaction between
2290 INTEGER and RATIO types: while bugs still remain, many more
2291 cases are accurately computed;
2292 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2293 if it is in the last clause;
2294 ** CONSTANTLY now correctly returns a side-effect-free function in
2296 ** DECLARE is no longer treated as a special-operator; in
2297 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2298 * incremented fasl file version number due to the change in the
2299 DEFSTRUCT-SLOT-DESCRIPTION structure.
2301 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2302 * minor incompatible change: code processed by the "interpreter" or
2303 EVAL now has a compilation optimization policy of (DEBUG 2)
2304 (changed from (DEBUG 1)) to improve debuggability of interactive
2305 development, and to allow the use of the debug RETURN command in
2307 * an experimental implementation of the RETURN command for the
2308 debugger has been included. (thanks to Frederik Kuivinen)
2309 * fixed bug 62: constraints were not propagated into a loop.
2310 * fixed bug in embedded calls of SORT (reported and investigated by
2312 * fixed some bugs revealed by Paul Dietz' test suite:
2313 ** printing and reading of arrays with some dimensions having
2314 length 0 (thanks to Gerd Moellmann);
2315 ** BOA constructor with &AUX argument without a default value does
2316 not cause a type error;
2317 ** CONSTANTP now returns true for all self-evaluating objects.
2319 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2320 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2321 accessors that are related by inheritance, as specified in the
2322 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2323 Valtteri Vuorikoski)
2324 * The compiler is now able to inline functions that were defined in
2325 a complex lexical environment (e.g. inside a MACROLET).
2326 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2328 * Support for the upcoming FreeBSD-5.0 release has been included.
2329 (thanks to Dag-Erling Smorgrav)
2330 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2331 effect when it is not in a toplevel context.
2332 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2333 stand a better chance of being correct. (thanks to Gerd
2335 * fixed bug in COERCE, which now signals an error on coercing a
2336 rational to a bounded real type which excludes the expected
2338 * The compiler is now able to derive types more accurately from the
2339 COERCE and COMPILE functions.
2340 * fixed bug 223: functional binding is considered to be constant
2341 only for symbols in the CL package.
2342 * fixed bug 231: SETQ did not check the type of a variable being set
2343 (reported by Robert E. Brown)
2344 * A new optimization for MAKE-INSTANCE has been included, fixing
2345 various bugs (including relating to :ALLOCATION :CLASS slots and
2346 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2348 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2349 ** As required by ANSI, LOOP now disallows anonymous collection
2350 clauses such as COLLECT I in conjunction with aggregate boolean
2351 clauses such as THEREIS (= I 1);
2352 ** LOOP now signals an error when any variable is reused in the
2353 same loop (including the potentially useful construct analogous
2354 to WITH A = 1 WITH A = (1+ A);
2355 ** IT is only a special loop symbol within the first clause of a
2356 conditional loop clause;
2357 ** LOOP with a typed iteration variable over a hashtable now
2358 signals a type error iff it should.
2359 * fixed some other bugs revealed by Paul Dietz' test suite:
2360 ** FILE-STREAM now names the class previously known as FD-STREAM;
2361 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2362 argument) no longer signals an error;
2363 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2364 of the slot symbol, rather than using the current package
2365 ((:CONC-NAME "") continues to intern the slot's name in the
2367 * incremented fasl file version number, because of the incompatible
2368 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2369 of the new implementation of DEFINE-COMPILER-MACRO.
2371 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2372 * Support for building SBCL for MIPS platforms running in
2373 little-endian mode has now been checked in, and basic
2374 functionality on said platforms verified.
2375 * minor incompatible change: PCL now records the pathname of a file
2376 in which methods and the like are defined, rather than its
2378 * minor incompatible change: TRUENAME now considers the truename of
2379 a file naming a directory to be the pathname with :DIRECTORY
2380 component indicating that directory.
2381 * minor incompatible change: a NAMED clause in the extended form of
2382 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2383 reason for the previous behaviour is unclear.
2384 * more systematization and improvement of CLOS and MOP conformance
2385 in PCL (thanks to Gerd Moellman and Pierre Mai):
2386 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2388 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2389 primary methods with no specializers;
2390 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2392 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2393 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2394 FORWARD-REFERENCED-CLASSes; error reporting on
2395 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2397 ** SXHASH on CLOS instances now uses a slot internal to the
2398 instance to return different numbers on distinct instances,
2399 while preserving the same return value through invocations of
2401 ** DEFMETHOD signals errors when methods with longer incongruent
2402 lambda lists are added to generic functions;
2403 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2404 CLASS, as specified in AMOP;
2405 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2406 on the order returned by the primary method for classes of
2407 class STANDARD-CLASS;
2408 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2409 * fixed some bugs shown by Paul Dietz' test suite:
2410 ** DOLIST puts its body in TAGBODY;
2411 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2413 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2414 value producing form;
2415 ** if more variables are given to PROGV than values, extra
2416 variables are bound and made to have no value;
2417 ** NSUBSTITUTE on list arguments gets the right answer with
2419 ** ELT signals an error of type TYPE-ERROR when the index argument
2420 is not a valid sequence index;
2421 ** LOOP signals (at macroexpansion time) an error of type
2422 PROGRAM-ERROR when duplicate variable names are found;
2423 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2424 ** FUNCALL of special-operators now cause an error of type
2426 ** PSETQ now works as required in the presence of side-effecting
2427 symbol-macro places;
2428 ** NCONC accepts any object as its last argument;
2429 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2431 ** loop-for-as-package does not require a package to be explicitely
2433 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2435 * fixed bug 166: compiler preserves "there is a way to go"
2436 invariant when deleting code.
2437 * fixed bug 172: macro lambda lists with required arguments after
2438 &REST arguments now cause an error to be signalled. (thanks to
2440 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2441 bug. (thanks to Matthew Danish)
2442 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2444 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2445 arguments are lexically rebound. (thanks to Gerd Moellmann and
2447 * fixed bug 194: error messages are now more informative when there
2448 is no primary method applicable in a call to a generic
2449 function. (thanks to Gerd Moellmann)
2450 * fixed bug in command line argument checking (thanks to Julian
2452 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2453 (thanks to Matthew Danish)
2454 * incremented fasl file version number, because of the
2455 SXHASH-related changes in the layout of CLOS data structures
2457 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2458 * minor incompatible change: The runtime (the Unix executable named
2459 "sbcl") is now much pickier about the .core files it will load.
2460 Essentially it now requires .core files to descend from the same
2461 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2462 as the runtime does. (The intent is to prevent the crashes which
2463 can occur, and which can even be reported as mysterious failures,
2464 when people patch the sources or change the build parameters
2465 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2466 sbcl and .core files.)
2467 * fixed bug: VALUES-LIST is no longer optimized away.
2468 * fixed bug 142: The FFI conversion of C string values to Lisp
2469 string values no longer conses excessively. (thanks to Nathan
2470 Froyd porting Raymond Toy's fix to CMU CL)
2471 * began to systematize and improve MOP conformance in PCL (thanks to
2472 Nathan Froyd, Gerd Moellman and Pierre Mai):
2473 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2475 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2476 ** FINALIZE-INHERITANCE is now called on class finalization;
2477 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2478 argument precedence order.
2479 * fixed bug 202: The compiler no longer fails on functions whose
2480 derived types contradict their declared type.
2481 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2482 so it can be non-toplevel.
2483 * The fasl file version number has changed (because of the new
2484 implementation of DEFMACRO).
2485 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2486 safe code, that any length requirement by their type-specifier
2487 argument is valid. The exceptions to this are described in bug
2489 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2490 functions now check the types of their inputs as required by ANSI.
2491 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2492 introduced symbol is DECLAREd to be SPECIAL.
2493 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2494 * fixed bug: :COUNT argument to sequence functions may be negative
2495 * fixed bug: body of DO-SYMBOLS may contain declarations
2496 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2497 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2498 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2499 symbol macro only once
2500 * fixed printing of call frame when argument list is unavailable
2501 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2502 * compiler no longer signals WARNING on unknown keyword
2505 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2506 * A beta-quality port to the mips architecture running Linux,
2507 based on the old CMUCL backend, has been made. It has been tested
2508 on a big-endian kernel, and works sufficiently well to be able to
2509 rebuild itself; it has not been tested in little-endian mode.
2510 * fixed an inconsistency between gencgc.c and purify.c which made
2511 dumping/loading .core files unreliable
2512 * fixed bug 120a: The compiler now deals correctly with IFs where
2513 the consequent is the same as the alternative, instead of
2514 misderiving the return type. (thanks to Alexey Dejneka)
2515 * fixed bug 113: Logical pathnames are now dumpable (the logical
2516 host is resolved at load-time, throwing an error if it is not
2518 * fixed bug 174: FORMAT's error message is slightly clearer when a
2519 non-printing character is used in a format directive.
2520 * fixed several bugs in compiler checking of type declarations, i.e.
2521 violations of the Python "declarations are assertions" principle
2522 (thanks to Alexey Dejneka)
2523 * fixed several bugs in PCL's error checking (thanks to Gerd
2525 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2527 * fixed bug in compilation of functions as first class values
2528 (thanks to Antonio Martinez-Shotton)
2529 * The compiler's handling TYPE-ERRORs which it can prove will
2530 inevitably happen at runtime has been cleaned up and corrected
2531 in several ways. (thanks to Alexey Dejneka)
2532 * improved argument type checking for various basic arithmetic
2533 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2534 lavished on them in the past that they can be compiled in many
2535 ways in different special cases
2536 * fixed bug 181: compiler checks validity of user supplied type
2538 * cleaned up code flushing in optimization: Function calls which
2539 should signal errors for safety purposes (e.g. which ANSI says
2540 should signal errors when their arguments are of incorrect type)
2541 are no longer optimized away.
2542 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2543 * incremented fasl file version number, because changes in the
2544 implementation of sequence functions like COERCE caused
2545 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2546 in old inline expansions) to become undefined. (Actually these
2547 changes were later undone, so we might very well be binary
2548 compatible with 0.7.7 after all, but leaving the version number
2549 incremented seemed like the simplest and most conservative
2552 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2553 * An alpha-quality port to the parisc architecture running Linux,
2554 based on the old CMUCL backend, has been made. This, even more so
2555 than the other backends, should be considered still a work in
2556 progress; known problems include that the Linux kernel in 64-bit
2557 mode does not propagate the correct sigcontext structure to
2558 userspace, and consequently SBCL on a parisc64 kernel will not
2560 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2561 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2562 suppressed.) Also now that the compiler is looking at declarations
2563 in the environment, it checks optimization declarations as well,
2564 and suppresses inlining when (> DEBUG SPEED).
2565 * More fixes have been made to treatment of floating point exception
2566 treatment and other Unix signals. In particular, floating point
2567 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2568 * The detection and handling of control stack exhaustion (infinite
2569 or very deeply nested recursion) has changed. Stack exhaustion
2570 detection is now done by write-protecting pages at the OS level
2571 and applies at all optimization settings; when found, a
2572 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2573 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2574 be caught using IGNORE-ERRORS.
2575 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2576 correctly when transforming recursive LABELS functions to LETs.
2577 (thanks to Alexey Dejneka)
2578 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2579 that are names of constants or global variables.
2580 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2581 alien routines with docstrings.
2582 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2583 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2585 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2586 (thanks to Eric Marsden)
2587 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2588 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2589 * Bug fix: The SPARC backend can now compile functions involving
2590 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2591 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2592 to a functions expecting a PATHNAME-DESIGNATOR.
2593 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2594 lambda lists. (thanks to Alexey Dejneka)
2595 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2596 functions so that EXPORTing the name of the class doesn't cause
2597 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2598 to Antonio Martinez for reporting this)
2599 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2600 constant-folded without failing an assertion. (thanks to Einar
2601 Floystad Dorum for reporting this)
2602 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2603 types are now dealt with more correctly by the compiler.
2604 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2605 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2606 behaviour analogous to RENAME-FILE. This puts its behaviour more
2607 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2609 * The fasl file version number has changed again. (because of the
2610 bug fix involving the names of PCL MAKE-INSTANCE functions)
2612 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2613 * bug fix: Floating point exceptions are treated much more
2614 consistently on the x86/Linux and PPC/Linux platforms.
2615 * Array initialization with :INITIAL-ELEMENT is now much faster for
2616 cases when the compiler cannot open code the array creation, but
2617 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2618 array accesses have also seen a speed increase.
2619 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2620 specified. (This was at the root of some bad interactions between
2621 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2623 * bug fix: Internal error arguments for undefined functions are now
2624 computed correctly on the PPC/Linux platform.
2625 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2626 Raymond Toy's patch for CMU CL)
2627 * Support for the Solaris 9 operating environment has been included
2628 (thanks to Daniel Merritt)
2629 * A very ugly but hopefully complete draft of the missing FFI chapter
2630 of the manual has been created by reformatting the corresponding
2631 CMU CL manual chapter into (currently very ugly and incoherent)
2632 DocBook and bringing it up to date for SBCL behavior. Thus, the
2633 manual is now essentially complete, at least by my extreme
2634 once-and-only-once standards, whereby it's acceptable to refer to
2635 the doc strings of SB-EXT functions as the primary documentation.
2636 * The fasl file version number has changed again, due to cleanup of
2637 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2638 is no longer a static symbol.)
2640 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2641 * SBCL now builds with OpenMCL (version 0.12) as the
2642 cross-compilation host; also, more progress has been made toward
2643 bootstrapping under CLISP.
2644 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2646 * bug 158 fixed: The compiler can now deal with integer loop
2647 increments different from 1; fixing this turned out also to fix
2649 * bug 169 fixed: no more bogus warnings about using lexical bindings
2650 despite the presence of perfectly good SPECIAL declarations (thanks
2651 to David Lichteblau)
2652 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2653 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2654 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2656 * bug 180 fixed: Method combination specifications no longer ignore
2657 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2658 * bug fix: Structure type predicate functions now check their argument
2659 count as they should.
2660 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2661 correctly. (thanks to Pierre Mai)
2662 * minor incompatible change: The --noprogrammer option is deprecated
2663 in favor of the new --disable-debugger option, which is very similar.
2664 (The major difference is that it takes effect at a slightly different
2665 time at startup, causing handling of errors in --sysinit and
2666 --userinit files will be affected differently.) The
2667 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2668 been added to allow this functionality to be controlled from ordinary
2669 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2670 maintainers, who might want to run non-interactive scripts to
2671 build SBCL cores which will later be used interactively.)
2672 * minor incompatible change: The LOAD function no longer, when given
2673 a wild pathname to load, loads all files matching that pathname.
2674 Instead, an error of type FILE-ERROR is signalled.
2676 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2677 * bug 147 fixed: The compiler preserves its block link/count
2678 invariants more correctly now so that it doesn't crash. (thanks
2680 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2682 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2683 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2684 no-value-for-this-character as NIL (as specified by ANSI).
2685 * HOST-NAMESTRING on physical pathnames now returns a string that is
2686 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2687 (thanks to Christophe Rhodes)
2688 * The Alpha port handles icache flushing more correctly. (thanks to
2690 * More progress has been made toward bootstrapping under CLISP. (thanks
2691 to Christophe Rhodes)
2692 * The fasl file format has changed again, because dynamic loading
2693 on OpenBSD (which has non-ELF object files) motivated some cleanups
2694 in the way that foreign symbols are transformed and passed around.
2695 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2696 is no longer treated as whitespace by the reader, but instead as
2697 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2698 |A
\7fB|, instead of A as it used to.
2700 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2701 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2702 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2703 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2704 this as of 0.7.1.45, but was left out of the previous news section
2705 (thanks to Dan Barlow)
2706 * SBCL now runs on the Solaris operating system on SPARC architectures
2707 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2708 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2709 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2710 * SPARC backend cleanups, allowing builds of cores optimized for V8
2711 and V9 SPARCS, and also emission of code targeted to a particular
2712 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2714 * SBCL is closer to bootstrapping under CLISP, thanks to various
2715 fixes by Christophe Rhodes.
2716 * The fasl file format has changed again, to allow the compiler's
2717 INFO database to support symbol macros.
2718 * The user manual (in doc/) is formatted into HTML more nicely.
2719 (thanks to coreythomas)
2720 * The system is smarter about SUBTYPEP relationships, especially
2721 those involving NOT types (including types such as ATOM which are
2722 represented internally using NOT types). Thus SUBTYPEP is less
2723 likely to return (VALUES NIL NIL) in general, and in particular
2724 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2726 * The fasl file format has changed again, because the internal
2727 representation of types now includes a new slot to support the new
2728 SUBTYPEP-of-NOT-types logic.
2729 * (not a change in the main branch of SBCL, but a related prototype
2730 which can hopefully be merged into the main branch of SBCL in the
2731 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2732 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2733 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2734 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2735 dispatch macro character combinations. (thanks to Alexey Dejneka)
2736 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2737 that superficially look like logical namestrings correctly.
2738 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2740 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2741 default supplied-p) for &optional and &key arguments. (thanks to
2744 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2745 * incompatible change: The compiler is now less aggressive about
2746 tail call optimization, doing it only when (> SPACE DEBUG) or
2747 (> SPEED DEBUG). (This is an incompatible change because there are
2748 programs which relied on the old CMU-CL-style behavior to optimize
2749 away their unbounded recursion which will now die of stack overflow.)
2750 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2751 for non-GENCGC systems has been increased to 20M (since that
2752 seems much closer to the likely performance optimum for modern
2753 systems than the old 4M value was)
2754 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2755 and *DEBUG-PRINT-LEVEL*
2756 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2757 of CMU CL's support for SPARC, and various endianness and other
2758 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2759 * new syntactic sugar for the Unix command line: --load foo.bar is now
2760 an alternate notation for --eval '(load "foo.bar")'.
2762 ** The system now detects stack overflow and handles it gracefully,
2763 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2764 optimization settings. (This is a good thing in general, and
2765 its introduction in this version should be particularly timely
2766 for anyone whose code fails because of suppression of tail
2768 ** The system now hunts for the C variable "environ" in a more
2769 devious way, to avoid segfaults when the C library version differs
2770 between compile time and run time. (thanks to Christophe Rhodes)
2771 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2772 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2773 ** The compiler no longer issues bogus style warnings for undefined
2774 classes in the same source file as the DEFCLASSes which defined
2775 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2777 ** fixes in CONDITION class precedence list for undefined function
2778 errors (thanks to Alexei Dejneka)
2779 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2780 correctly. (thanks to Dan Barlow)
2781 ** portability fixes aiming at bootstrapping under CLISP (thanks
2782 to Dave McDonald and Christophe Rhodes)
2783 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2784 ** fixes in type translation and and type inference (thanks to
2786 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2787 ** various fixes in the new ports (thanks to Dan Barlow)
2788 * several changes related to debugging:
2789 ** suppression of tail recursion, as noted above
2790 ** stack overflow detection, as noted above
2791 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2792 is now the default. (For some time encapsulation has been more
2793 reliable than the breakpoint-based :ENCAPSULATE NIL
2794 implementation, at least on X86 systems; and I just noticed that
2795 encapsulation also seems closer to the spirit of the ANSI
2798 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2800 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2801 up properly again. (There was a packaging bug in 0.7.0 which
2802 left their definitions in SB-SYS::LOAD-FOREIGN and
2803 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2804 vital for most things which interface to C-level interfaces,
2805 like extensions working with sockets or databases or
2806 Perl-compatible regexes or whatever, and the need to fix
2807 this bug is the main reason that 0.7.1 was released so
2809 ** DEFGENERIC is now choosier about the methods it redefines, so that
2810 reLOADing a previously-LOADed file containing DEFGENERICs does
2811 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2812 cycle now works as it should. (thanks to Alexey Dejneka)
2813 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2814 fixed by Christophe Rhodes. (He actually submitted this patch
2815 months ago, and I delayed until after 0.7.0.)
2816 ** Bug 111 (internal compiler confusion about runtime checks on
2817 FUNCTION types) was fixed by Alexey Dejneka.
2818 * Some internal cleanups (getting rid of variables which aren't
2819 needed now that the byte interpreter is gone) caused the fasl
2820 file format number to change again.
2822 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2823 * major incompatible change: The default fasl file extension, i.e. the
2824 default extension for files produced by COMPILE-FILE, has changed
2825 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2827 ** There are many changes in the implementation of the compiler.
2828 SBCL is now essentially a compiler-only implementation of ANSI
2829 Common Lisp. EVAL still "interprets" a few special cases, but
2830 almost all the interesting cases are handled by creating
2831 a LAMBDA expression, calling COMPILE on it, then calling
2832 FUNCALL on the result.
2833 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2834 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2835 Since the code is newer, there might still be some new bugs
2836 (though not as many as before Martin Atzmueller's fixes:-). But
2837 the new code is substantially simpler and clearer, and hopefully
2838 any remaining bugs will be simpler, less fundamental, and more
2839 fixable then the bugs in the old code.
2840 ** The revised compiler is still a little unsteady on its feet.
2842 *** The debugging information it produces (particularly the names
2843 of FUNCTION objects) is sometimes much less useful than what
2844 the old compiler produced.
2845 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2846 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2847 within a MACROLET) has been temporarily weakened.
2848 ** There are new compiler optimizations for various functions:
2849 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2850 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2851 *** the math functions TRUNCATE, FLOOR, and CEILING
2852 *** the function-of-all-trades COERCE
2853 Mostly these should be transparent, but there's one
2854 potentially-annoying problem (bug 117): when the compiler
2855 inline-expands a function and does type analysis on the result,
2856 it can create control paths which have type mismatches, and
2857 when it can't prove that those control paths aren't taken,
2858 it will issue WARNINGs about the type mismatches. This is
2859 a particular problem in practice for the new sequence functions.
2860 It's not clear how this should be fixed, and for now, a
2861 workaround is given in the entry for 117 in the BUGS file.
2862 ** (Because of the interaction between the two previous items --
2863 occasional inlining problems and new inline expansions -- some
2864 of the new sequence function optimizations won't really kick in
2865 completely until debugging information, and then inlining, are
2866 straightened out in some future version.)
2867 * minor incompatible changes:
2868 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2869 DIRECTORY no longer implicitly promotes NIL slots of its
2870 pathname argument to :WILD. In particular, when you ask for the
2871 contents of a directory (which you used to be able to do without
2872 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2873 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2874 ** changes in behavior that ANSI explicitly defines to be
2875 implementation dependent:
2876 *** The new compiler-only implementation still conforms with ANSI,
2877 but acts a little different than before. Besides the obvious
2878 changes in performance tradeoffs (that the cost per form passed
2879 to EVAL has gone up, and the cost per form executed by EVAL
2880 has gone down), the behavior of the system changes a little
2881 because there are no longer any interpreted function objects.
2882 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2883 e.g. doing COMPILE on the output of interactive DEFUN is
2885 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2887 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2888 has changed from (:CL) to NIL.
2889 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2890 e.g. #\Newline and #\Tab, don't have names specified in the
2891 ANSI Common Lisp standard, is now based on their ASCII symbolic
2892 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2893 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2894 are no longer used for output.
2895 ** changes in internal implementation constants:
2896 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2897 4 million. (If your application spends a lot of time GCing and
2898 you have a lot of RAM, you might want to experiment with
2899 increasing it even more.)
2900 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2901 However, almost all old code should still continue to work without
2902 immediate update, as SB-C-CALL is now a (deprecated) nickname
2904 ** Old operator names in the style DEF-FOO are now deprecated in
2905 favor of new corresponding names DEFINE-FOO, for consistency with
2906 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2907 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2908 internal symbols, but a few supported extensions like
2909 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2910 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2911 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2912 etc. as you get deeper into recursive calls to the debugger
2913 command loop, instead of the old "5]", "5]]", "5]]]"
2914 sequence. (I was motivated to do this when squabbles between
2915 ILISP and SBCL left me very deeply nested in the debugger. In the
2916 short term, this change will probably provoke more ILISP/SBCL
2917 squabbles, but hopefully it will be an improvement in the long run.)
2918 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2919 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2920 function, since declaiming return types involving aliens is
2921 (1) annoyingly messy to do by hand and (2) vital to efficient
2922 compilation of code which calls such functions.
2923 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2924 longer reexported by the SB-EXT package. They're solely useful
2925 for alien code, so it seems more logical that you should get
2926 them from the SB-ALIEN package, not in SB-EXT.
2927 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2928 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2929 features. Instead, the code that they used to control is always
2930 built into the system.
2931 * many other bug fixes
2932 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2933 advantage of the new EVAL-WHEN stuff and to clean them up in
2934 general, and they are now more ANSI-compliant in a number of
2935 ways. Martin Atzmueller is responsible for a lot of this.
2936 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2938 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2939 *** correct ERROR type for various file operations
2940 *** some fixes for Lisp streams
2941 *** DEFMETHOD syntax checking
2942 *** changing old weird representation of debug information as
2943 strings (which, among their other deficiencies, don't transform
2944 correctly when you rename packages, and don't change their
2945 print representation when you change things like *PACKAGE*
2946 and *PRINT-LENGTH*) to symbols and lists of symbols
2947 He also made several improvements and fixed several bugs in DESCRIBE.
2948 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2950 *** misbehavior of WRITE-STRING/WRITE-LINE
2951 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2952 and several other LOOP problems as well
2953 *** DIRECTORY when similar filenames are present
2954 *** DEFGENERIC with :METHOD options
2955 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2956 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2957 *** argument ordering in FIND with :TEST option
2958 *** mishandled package designator argument in APROPOS-LIST
2959 *** various problems in the backquote readmacro
2960 *** a bug in APROPOS
2961 *** probably some others that I'm not describing very well here,
2962 since the CVS log documents them by reference to sbcl-devel
2963 messages, and the SourceForge archives aren't working well.:-(
2964 ** Dan Barlow improved the Alpha port (and is making progress on the
2965 PPC port, for those of you who think different).
2966 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2967 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2968 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2969 continues to work on the SPARC port (for those of you in a position
2970 to look down upon our little PC-compatible boxes from a great height).
2971 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2972 and uses that copy, rather than the current dynamic value, when
2973 it's trying to decide whether to truncate output. Thus e.g.
2974 (let ((*print-lines* 50))
2975 (pprint-logical-block (stream nil)
2977 (let ((*print-lines* 8))
2978 (print (aref possiblybigthings i) stream)))))
2979 should now truncate the logical block only at 50 lines, instead of
2980 often truncating it at 8 lines, as it did before.
2981 * The doc/cmucl/ directory, containing old CMU CL documentation
2982 from the time of the fork, is no longer part of the base system.
2983 SourceForge has shut down its anonymous FTP service, and with it
2984 my original plan for distributing the old CMU CL documentation
2985 there. For now, if you need these files you can download an old
2986 SBCL source release and extract them from it.
2987 * The fasl file version number changed again, for dozens of reasons,
2988 some of which are apparent above.
2990 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2991 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2992 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2993 patch, so that DISASSEMBLE output is much nicer.
2994 * The code in the SB-PROFILE package now seems reasonably stable.
2995 I still haven't decided what the final interface should look like
2996 (I'd like PROFILE to interact cleanly with TRACE, since both
2997 facilities use function encapsulation) but if you have a need
2998 for profiling now, you can probably use it successfully with
2999 the current CMU-CL-style interface.
3000 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3001 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3002 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3003 link now returns the dangling link itself, and for similar
3004 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3005 link itself. (In these cases the old code signalled an error and
3006 looped endlessly, respectively.) Thus, DIRECTORY now works even
3007 in the presence of dangling and cyclic symbolic links.
3008 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3009 is now a supported extension again, since the consensus on
3010 sbcl-devel was that it can be useful for ordinary development
3011 work, not just for debugging SBCL itself.
3012 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3013 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3014 that currently-defined functions might be redefined later with
3015 different return types.
3016 * Hash tables can be printed readably, as inspired by CMU CL code
3017 of Eric Marsden and SBCL code of Martin Atzmueller.
3018 * better error handling in CLOS method combination, thanks to
3019 Martin Atzmueller porting Pierre Mai's CMU CL patches
3020 * more overflow fixes for >16Mbyte I/O buffers
3021 * A bug in READ has been fixed, so that now a single Ctrl-D
3022 character suffices to cause end-of-file on character streams.
3023 In particular, now you only need one Ctrl-D at the command
3024 line (not two) to exit SBCL.
3025 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3027 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3028 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3029 does the right thing.
3030 * The implementation of some type tests, especially for CONDITION
3031 types, is now tidier and maybe faster, due to CMU CL code
3032 originally by Douglas Crosher, ported by Martin Atzmueller.
3033 * Some math functions have been fixed, and there are new
3034 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3035 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3036 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3037 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3038 likely to be built into sbcl-0.7.0.)
3039 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3040 CL, ported to SBCL by Martin Atzmueller)
3041 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3042 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3043 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3044 * There's a new slam.sh hack to shorten the edit/compile/debug
3045 cycle for low-level changes to SBCL itself, and a new
3046 :SB-AFTER-XC-CORE target feature to control the generation of
3047 the after-xc.core file needed by slam.sh.
3048 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3049 COMPILE-FILE is no longer supported, so that now every function
3050 gets an entry point, so that block compilation looks a little
3051 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3052 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3053 returns the number of bytes consed since the system started,
3054 rather than the number consed since the first time the function
3055 was called. (The new definition parallels ANSI functions like
3056 CL:GET-INTERNAL-RUN-TIME.)
3057 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3058 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3059 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3060 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3061 * Fasl file version numbers are now independent of the target CPU,
3062 since historically most system changes which required version
3063 number changes have affected all CPUs equally. Similarly,
3064 the byte fasl file version is now equal to the ordinary
3067 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3068 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3069 is no longer recognized. I apologize for this, because it was
3070 listed in SB-EXT as a supported extension, but I found that
3071 its existing behavior was poorly specified, as well as incorrectly
3072 specified, and it looked like too much of a mess to straighten it
3073 out. I have enough on my hands trying to get ANSI stuff to work..
3074 * many patches ported from CMU CL by Martin Atzmueller, with
3075 half a dozen bug fixes in pretty-printing and the debugger, and
3076 half a dozen others elsewhere
3077 * fixed bug 13: Floating point infinities are now supported again.
3078 They might still be a little bit flaky, but thanks to bug reports
3079 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3080 as flaky as they were.
3081 * The --noprogrammer command line option is now supported. (Its
3082 behavior is slightly different in detail from what the old man
3083 page claimed it would do, but it's still appropriate under the
3084 same circumstances that the man page talks about.)
3085 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3086 are now supported, and enabled by default. Thus, the compiler can
3087 handle many floating point and complex operations much less
3088 inefficiently. (Thus e.g. you can implement a complex FFT
3090 * The compiler now detects type mismatches between DECLAIM FTYPE
3091 and DEFUN better, and implements CHECK-TYPE more correctly, and
3092 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3093 thanks to patches from Martin Atzmueller.
3094 * various fixes to make the cross-compiler more portable to
3095 ANSI-conforming-but-different cross-compilation hosts (notably
3096 Lispworks for Windows, following bug reports from Arthur Lemmens)
3097 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3098 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3099 fixed thanks to Pierre Mai's CMU CL patches.
3100 * fixes to keep the system from overflowing internal counters when
3101 it tries to use i/o buffers larger than 16M bytes
3102 * fixed bug 45a: Various internal functions required to support
3103 complex special functions have been merged from CMU CL sources.
3104 (When I was first setting up SBCL, I misunderstood a compile-time
3105 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3106 * improved support for type intersection and union, fixing bug 12
3107 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3108 more obscure bugs as well
3109 * some steps toward byte-compiling non-performance-critical
3110 parts of the system, courtesy of patches from Martin Atzmueller
3111 * Christophe Rhodes has made some debian packages of sbcl at
3112 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3113 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3114 stable, but are nonetheless usable. When he's ready, I'd be happy
3115 to add them to the SourceForge "File Releases" section. (And if
3116 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3117 * new fasl file format version number (because of changes in
3118 internal representation of (OR ..) types to accommodate the new
3119 support for (AND ..) types, among other things)
3121 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3122 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3124 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3126 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3127 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3128 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3129 * DESCRIBE now works on structure objects again.
3130 * Most function call argument type mismatches are now handled as
3131 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3132 know whether the function will be redefined before the call is
3133 executed. (The compiler could flag local calls with full WARNINGs,
3134 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3135 it doesn't keep track of enough information to know whether calls
3136 are local in this sense.)
3137 * Compiler output is now more verbose, with messages truncated
3138 later than before. (There should be some supported way for users
3139 to override the default verbosity, but I haven't decided how to
3140 provide it yet, so this behavior is still controlled by the internal
3141 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3142 src/compiler/ir1util.lisp.)
3143 * Fasl file format version numbers have increased again, because
3144 support for the Gray streams extension changes the layout of the
3145 system's STREAM objects.
3146 * The Gray subclassable streams extension now works, thanks to a
3147 patch from Martin Atzmueller.
3148 * The full LOAD-FOREIGN extension (not just the primitive
3149 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3150 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3151 but like most other programs, it defaults to copying the Unix
3152 environment from the original process instead of starting the
3153 new process in an empty environment.
3154 * Extensions which manipulate the Unix environment now support
3155 an :ENVIRONMENT keyword option which doesn't smash case or
3156 do other bad things. The CMU-CL-style :ENV option is retained
3157 for porting convenience.
3158 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3159 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3161 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3163 * A patch from Martin Atzmueller seems to have solved the SIGINT
3164 problem, and as far as we know, signal-handling now works cleanly.
3165 (If you find any new bugs, please report them!)
3166 * The system no longer defaults Lisp source file names to types
3167 ".l", ".cl", or ".lsp", but only to ".lisp".
3168 * The compiler no longer uses special default file extensions for
3169 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3170 seems to expect a single default extension for all compiled code,
3171 and there's no compelling reason to try to stretch the standard
3172 to allow two different extensions.) Instead, byte-compiled files
3173 default to the same extension as native-compiled files.
3174 * Fasl file format version numbers have increased again, because
3175 a rearrangement of internal implementation packages made some
3176 dumped symbols in old fasl files unreadable in new cores.
3177 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3178 many fewer weird special cases.
3179 * Bug #17 (differing COMPILE-FILE behavior between logical and
3180 physical pathnames) has been fixed, and some related misbehavior too,
3181 thanks to a patch from Martin Atzmueller.
3182 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3183 by Tim Moore, ported to SBCL by Martin Atzmueller.
3184 * Martin Atzmueller fixed several filesystem-related problems,
3185 including bug #36, in part by porting CMU CL patches, which were
3186 written in part by Paul Werkowski.
3187 * More compiler warnings in src/runtime/ are gone, thanks to
3188 more patches from Martin Atzmueller.
3189 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3192 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3194 * DESCRIBE now works on CONDITION objects.
3195 * The debugger now handles errors which arise when trying to print
3196 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3198 * The build system now uses an additional file, customize-target-features.lisp,
3199 to allow local modifications to the target *FEATURES* list. (The point of
3200 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3201 debugging features enabled, without having to worry about propagating it
3202 into everyone's system when I do a "cvs update".) When no
3203 customize-target-features.lisp file exists, the target *FEATURES* list
3204 should be constructed the same way as before.
3205 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3206 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3207 warn about behavior which is undefined under the ANSI spec. Note: This
3208 is specified by ANSI, but it's not very popular with programmers.
3209 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3210 macro in the SBCL sources for an example of a workaround which you
3211 might use to make portable ANSI-standard code which does what you want.
3212 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3213 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3214 when it's used as a non-toplevel form. (This is required in order
3215 to implement the DEFCONSTANT-EQX macro.)
3216 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3217 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3218 funny things, due to bugs in the implementation of EVAL-WHEN
3219 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3220 non-toplevel DEFCONSTANTs are uncommon.)
3221 * The core file version number and fasl file version number have been
3222 incremented, because the old noncompliant DEFCONSTANT behavior involved
3223 calling functions which no longer exist, and because I also took the
3224 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3225 * fixed bug 1 (error handling before read-eval-print loop starts), and
3226 redid debugger restarts and related debugger commands somewhat while
3228 ** The QUIT debugger command is gone, since it did something
3229 rather different than the SB-EXT:QUIT command, and since it never
3230 worked properly outside the main toplevel read/eval/print loop.
3231 Invoking the new TOPLEVEL restart provides the same functionality.
3232 ** The GO debugger command is also gone, since you can just invoke
3233 the CONTINUE restart directly instead.
3234 ** The TOP debugger command is also gone, since it's redundant with the
3235 FRAME 0 command, and since it interfered with abbreviations for the
3237 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3239 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3240 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3241 some porting work by Martin Atzmueller)
3242 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3243 with a boatload of patches which he ported to SBCL. Now that those
3245 ** The system tries to make sure that its low-priority messages
3246 are prefixed by semicolons, to help people who like to use
3247 syntax highlighting in their ILISP buffer. (This patch
3248 was originally due to Raymond Toy.)
3249 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3250 patches originally written by Raymond Toy.
3251 ** The compiler understands coercion between single-value and
3252 multiple-VALUES type expressions better, getting rid of some very
3253 weird behavior, thanks to patches originally by Robert MacLachlan
3254 and Douglas Crosher.
3255 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3256 lambda lists, thanks to a patch originally by Pierre Mai.
3257 ** The system no longer bogusly warns about "abbreviated type
3259 ** The compiler gets less confused by inlining and RETURN-FROM,
3260 thanks to some patches originally by Tim Moore.
3261 ** The system no longer hangs when dumping circular lists to fasl
3262 files, thanks to a patch originally from Douglas Crosher.
3263 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3264 undefined function error.
3265 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3266 in structure slot initforms) for now, documented workaround instead:-|
3267 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3268 * fixed bug 5: added stubs for various Gray stream functions called
3269 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3270 installed, at least appropriate type errors are generated
3271 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3272 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3273 consistently in DEFMETHOD forms.
3274 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3275 it doesn't seem to affect SBCL after all
3276 * The C runtime system now builds with better optimization and many
3277 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3279 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3281 * The system is now under CVS at SourceForge (instead of the
3282 CVS repository on my home machine).
3283 * The new signal handling code has been tweaked to treat register
3284 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3285 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3286 downstream, e.g. in debug-int.lisp, has implicit dependencies
3287 on the unsignedness of integer representation of machine words,
3288 and that caused the system to bomb out with infinite regress
3289 when trying to recover from type errors involving signed values,
3290 e.g. (BUTLAST '(1 2 3) -1).)
3291 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3292 (This was one of the bugs Peter Van Eynde reported back in July.)
3293 * The system now uses code inspired by Colin Walters' O(N)
3294 implementation of MAP (from the cmucl-imp@cons.org mailing
3295 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3296 inline the MAP operation, and there is more than one
3297 sequence argument to the MAP call (so that it can't just
3298 do ETYPECASE once and for all based on the type of the
3299 single sequence argument). (The old non-inline implementation
3300 of the general M-argument sequence-of-length-N case required
3301 O(M*N*N) time when any of the sequence arguments were LISTs.)
3302 * The QUIT :UNIX-CODE keyword argument has been renamed to
3303 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3304 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3306 * Raymond Wiker's patches to port dynamic loading from Linux to
3307 FreeBSD have been added.
3308 * The BUGS file is now more nearly up to date, thanks in large part
3309 to Martin Atzmueller's review of it.
3310 * The debugger now flushes standard output streams before it begins
3311 its output ("debugger invoked" and so forth).
3312 * The core version number and fasl file version number have both
3313 been incremented, because of incompatible changes in the layout
3315 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3316 used to not be called for a saved Lisp image.)
3317 * Martin Atzmueller's version of a patch to fix a compiler crash,
3318 as posted on sbcl-devel 13 September 2000, has been installed.
3319 * Instead of installing Martin Atzmueller's patch for the
3320 compiler transform for SUBSEQ, I deleted the compiler transform,
3321 and transforms for some similar consing operations.
3322 * A bug in signal handling which kept TRACE from working on OpenBSD
3324 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3325 away by constant folding
3326 * The system now defines its address space constants in one place
3327 (in the Lisp sources), and propagates them automatically elsewhere
3328 (through GENESIS and the sbcl.h file). Therefore, patching the
3329 address map is less unnecessarily tedious and error-prone. The
3330 Lisp names of address space constants have also been systematized.
3331 * CVS tags like dollar-Header-dollar have been removed from
3332 the sources, because they have never saved me trouble and
3333 they've been source of trouble working with patches and other
3334 diff-related operations.
3335 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3336 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3338 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3340 * The system has been ported to OpenBSD.
3341 * The system now compiles with a simple "sh make.sh" on the systems
3342 that it's supported on. I.e., now you no longer need to tweak
3343 text in the target-features.lisp-expr and symlinks in src/runtime/
3344 by hand, the make.sh takes care of it for you.
3345 * The system is no longer so grossly inefficient when compiling code
3346 involving vectors implemented as general (not simple) vectors (VECTOR T),
3347 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3348 which dares to use the various sequence functions on non-simple
3349 vectors, takes less of a performance hit.
3350 * There is now a primitive type predicate VECTOR-T-P
3351 to test for the (VECTOR T) type, so that e.g.
3352 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3353 can now be compiled with some semblance of efficiency. (The old code
3354 turned the type declaration into a full call to %TYPEP at runtime!)
3355 * AREF on (VECTOR T) is still not fast, since it's still compiled
3356 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3357 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3358 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3359 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3360 without making full calls to SUBTYPEP at runtime.
3361 (Some analogous efficiency issues for non-simple vectors specialized to
3362 element types other than T, or for non-simple multidimensional arrays,
3363 have not been addressed. They could almost certainly be handled the
3364 same way if anyone is motivated to do so.)
3365 * The changes in array handling break binary compatibility, so
3366 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3367 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3368 instead of (VALUES T T).
3369 * By following the instructions that Dan Barlow posted to sbcl-devel
3370 on 2 July 2000, I was able to enable primitive dynamic object
3371 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3372 functionality is not implemented (since it calls ld to resolve
3373 library references automatically, requiring RUN-PROGRAM for its
3374 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3375 try to resolve library references) is now supported.
3376 * The system now flushes the standard output streams when it terminates,
3377 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3378 them at several other probably-convenient times, e.g. in each pass of
3379 the toplevel read-eval-print loop, and after evaluating a form given
3380 as an "--eval" command-line option. (These changes were motivated by a
3381 discussion of stream flushing issues on cmucl-imp in August 2000.)
3382 * The source transform for TYPEP of array types no longer assumes
3383 that an array whose element type is a not-yet-defined type
3384 is implemented as an array of T, but instead punts, so that the
3385 type will be interpreted at runtime.
3386 * There is now some support for cross-compiling in make.sh: each of
3387 the phases of make.sh has its own script. (This should be transparent
3388 to people doing ordinary, non-cross-compile builds.)
3389 * Since my laptop doesn't have hundreds of megabytes of memory like
3390 my desktop machine, I became more motivated to do some items on
3391 my to-do list in order to reduce the size of the system a little:
3392 ** Arrange for various needed-only-at-cold-init things to be
3393 uninterned after cold init. To support this, those things have
3394 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3395 symbols with such names are now uninterned after cold init).
3396 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3397 fasl files for cold load.
3398 ** Remove the old compiler structure pooling code (which used to
3399 be conditional on the target feature :SB-ALLOC) completely.
3400 ** Redo the representation of some data in cold init to be more compact.
3401 (I also looked into supporting byte compiled code at bootstrap time,
3402 which would probably reduce the size of the system a lot, but that
3403 looked too complicated, so I punted for now.)
3404 * The maximum signal nesting depth in the src/runtime/ support code has
3405 been reduced from 4096 to 256. (I don't know any reason for the very
3406 large old value. If the new smaller value turns out to break something,
3407 I'll probably just bump it back up.)
3408 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3410 * Many, many bugs reported by Peter Van Eynde have been added to
3411 the BUGS list; some have even been fixed.
3412 * While enabling dynamic object file loading, I tried to make the
3413 code easier to understand, renaming various functions and variables
3414 with less ambiguous names, and changing some function calling
3415 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3416 * While trying to figure out how to do the OpenBSD port, I tried to
3417 clean up some of the code in src/runtime/. In particular, I dropped
3418 support for non-POSIX signal handling, added various comments,
3419 tweaked the code to reduce the number of compilation warnings, and
3420 renamed some files to increase consistency.
3421 * To support the new automatic configuration functionality in make.sh,
3422 the source file target-features.lisp-expr has been replaced with the
3423 source file base-target-features.lisp-expr and the machine-generated
3424 file local-target-features.lisp-expr.
3425 * fixed a stupid quoting error in make.sh so that using CMU CL
3426 "lisp -batch" as cross-compilation host works again
3428 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3430 * DESCRIBE no longer tries to call itself recursively to describe
3431 bound/fbound values, so that it no longer fails on symbols which are
3432 bound to themselves (like keywords, T, and NIL).
3433 * DESCRIBE now works on generic functions.
3434 * The printer now prints less-screwed-up representations of closures
3435 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3436 * A private symbol is used instead of the :EMPTY keyword previously
3437 used to mark empty slots in hash tables. Thus
3438 (DEFVAR *HT* (MAKE-HASH-TABLE))
3439 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3440 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3441 now does what ANSI says that it should. (You can still get
3442 similar noncompliant behavior if bang on the hash table
3443 implementation with all the symbols you get back from
3444 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3445 This breaks binary compatibility, since tests for equality to
3446 :EMPTY are wired into things like the macroexpansion of
3447 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3449 * There's now a minimal placeholder implementation for CL:STEP,
3450 as required by ANSI.
3451 * An obscure bug in the interaction of the normal compiler, the byte
3452 compiler, inlining, and structure predicates has been patched
3453 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3454 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3456 * Missing ordinary arguments in a macro call are now detected even
3457 when the macro lambda list contains &KEY or &REST.
3458 * The debugger no longer complains about encountering the top of the
3459 stack when you type "FRAME 0" to explicitly instruct it to go to
3460 the top of the stack. And it now prints the frame you request even
3461 if it's the current frame (instead of saying "You are here.").
3462 * As specified by ANSI, the system now always prints keywords
3463 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3465 * The default initial SIZE of HASH-TABLEs is now smaller.
3466 * Type information from CLOS class dispatch is now propagated
3467 into DEFMETHOD bodies, so that e.g.
3468 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3470 is now basically equivalent to
3471 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3472 (DECLARE (TYPE SINGLE-FLOAT X))
3474 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3475 operation, without having to do run-time type dispatch.
3476 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3477 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3478 * Since I don't seem to be making big file reorganizations very often
3479 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3480 is overflowing my ability to conveniently back them up), I've finally
3481 checked the system into CVS. (The CVS repository is on my home system,
3482 not at SourceForge -- putting it on SourceForge might come later.)
3483 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3484 high-level GC-NOTIFY-FOO functions send their output. (There's
3485 still very little control of where low-level verbose GC functions
3486 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3487 less than it used to -- the GC-NOTIFY-FOO functions are now under
3488 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3489 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3490 in only one place in the source code, and propagates it automatically
3491 everywhere that it's needed. Thus e.g. when I bump the version from
3492 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3493 * The C source files now include boilerplate legalese and documentation
3494 at the head of each file (just as the Lisp source files already did).
3495 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3496 to his page will be replaced with a link to his new CLiki service.
3498 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3500 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3501 * The build process now looks for GNU make under the default name "gmake",
3502 instead of "make" as it used to. If GNU make is not available as "gmake"
3503 on your system, you can change this default behavior by setting the
3504 GNUMAKE environment variable.
3505 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3506 can build without error under CMU CL.
3508 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3510 * There is now a partial SBCL user manual (with some new text and some
3511 text cribbed from the CMU CL manual).
3512 * The beginnings of a profiler have been added (starting with the
3513 CMU CL profiler and simplifying and cleaning up). Eventually the
3514 main interface should be through the TRACE macro, but for now,
3515 it's still accessed through vaguely CMU-CL-style functions and macros
3516 exported from the package SB-PROFILE.
3517 * Some problems left over from porting CMU CL to the new
3518 cross-compilation bootstrap process have been cleaned up:
3519 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3520 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3521 part when getting rid of the compiler *BACKEND* stuff.)
3522 ** The value of *NULL-TYPE* was screwed up, because it was
3523 being initialized before the type system knew the final
3524 definition of the 'NULL type. This screwed up several key
3525 optimizations in the compiler, causing inefficiency in all sorts
3526 of places. (I found it because I wanted to understand why
3527 GET-INTERNAL-RUN-TIME was consing.)
3528 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3529 PROCLAIM FTYPE information. Unfortunately this broke binary
3530 compatibility again, since now the forms output by DEFGENERIC
3531 to refer to functions which didn't exist in 0.6.3.
3532 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3533 can use the new (as of 0.6.3) transform for SOME into MAP into
3535 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3536 values to (INTEGER 0 1000000), so that the time code will no longer
3537 occasionally get blown up by Linux returning 1000000 microseconds
3538 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3539 its output conform to the ANSI spec. (Alas, this makes its output
3540 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3541 * A full call to MAP NIL with a single sequence argument no longer conses.
3542 * fixes to problems pointed out by Martin Atzmueller:
3543 * The manual page no longer talks about multiprocessing as though
3544 it were currently supported.
3545 * The ILISP support patches have been removed from the distribution,
3546 because as of version 5.10.1, ILISP now supports SBCL without us
3547 having to maintain patches.
3548 * added a modified version of Raymond Toy's recent CMU CL patch for
3549 EQUALP comparison of HASH-TABLE
3551 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3553 * The system still can't cross-compile itself with
3554 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3555 entails), but at least it can compile more of itself that way
3556 than it used to be able to, and various buglets which were uncovered
3557 by trying to cross-compile itself that way have now been fixed.
3558 * This release breaks binary compatibility again. This time
3559 at least I've incremented the FASL file format version to 2, so that the
3560 problem can be detected reliably instead of just causing weird errors.
3561 * various new style warnings:
3562 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3563 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3564 ** using the lexical binding of a variable named in the *FOO* style
3565 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3566 as specified by ANSI.
3567 * *RANDOM-STATE* is no longer automatically initialized from
3568 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3569 default behavior of the system is to repeat its behavior every time
3570 it's run. If you'd like to change this behavior, you can always
3571 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3572 old convention there was no comparably easy way to get the system to
3573 repeat its behavior every time it was run.
3574 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3575 FTYPE declarations has been removed, in favor of their ANSI
3576 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3577 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3578 the types of their sequence arguments can be determined at compile time.
3579 This is done through a new open code expansion for MAP which eliminates
3580 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3581 argument types can be determined at compile time.
3582 * The optimizer now transforms COERCE into an identity operation when it
3583 can prove that the coerced object is already of the correct type. (This
3584 can be a win for machine generated code, including the output of other
3585 optimization transforms, such as the MAP transform above.)
3586 * Credit information has been moved from source file headers into CREDITS.
3587 * Source file headers have been made more standard.
3588 * The CASE macro now compiles without complaining even when it has
3591 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3593 * (Note that the way that the PCL macroexpansions were rewritten
3594 to accommodate the change in DEFGENERIC below breaks binary
3595 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3596 not run under sbcl-0.6.2. Once we get out of alpha releases,
3597 i.e. hit release 1.0.0, we'll probably try to maintain binary
3598 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3599 and sbcl-1.4.4. Until then, however, it might be fairly common
3600 for maintenance releases to break binary compatibility.)
3601 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3603 * The reporting of unhandled signals has been changed to print some
3604 explanatory text as well as the report form. (Previously only
3605 the report form was printed.)
3606 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3607 it defines, so that the compiler no longer issues undefined function
3608 warnings for compiled-but-not-yet-loaded generic functions.
3609 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3610 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3611 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3612 section of the ANSI spec are supported.
3613 * The "" nickname for the "KEYWORD" package has been removed.
3614 The reader still handles symbol tokens which begin with a package marker
3615 as keywords, but it doesn't expose its mechanism for doing so in the
3616 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3617 * The system now issues STYLE-WARNINGs for contradictory TYPE
3618 proclamations. (Warnings for contradictory FTYPE proclamations would
3619 be nice too, but those can't be done usefully unless the type system
3620 is made smarter about FUNCTION types.)
3621 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3622 systematized, so that "*target-*.lisp is supposed to exist only on the
3623 target and imply that there's a related file which exists on the
3624 host, and *host-*.lisp is supposed to exist only on the host and imply
3625 that there's a related file which exists on the target. This involves a
3626 lot of renaming. Hopefully the acute confusion caused by the renaming
3627 will be justified by the reduction in chronic confusion..
3628 ** runtime-type.lisp -> early-target-type.lisp
3629 ** target-type.lisp -> late-target-type.lisp
3630 ** early-host-format.lisp -> early-format.lisp
3631 ** late-host-format.lisp -> late-format.lisp
3632 ** host-error.lisp -> misc-error.lisp
3633 ** early-error.lisp -> early-target-error.lisp
3634 ** late-error.lisp -> late-target-error.lisp
3635 ** host-defboot.lisp -> early-defboot.lisp
3636 ** code/misc.lisp -> code/target-misc.lisp
3637 ** code/host-misc.lisp -> code/misc.lisp
3638 ** code/numbers.lisp -> code/target-numbers.lisp
3639 ** code/early-numbers.lisp -> numbers.lisp
3640 ** early-host-type.lisp -> early-type.lisp
3641 ** late-host-type.lisp -> late-type.lisp
3642 ** host-typep.lisp -> typep.lisp
3643 ** load.lisp -> target-load.lisp
3644 ** host-load.lisp -> load.lisp
3645 ** host-disassem.lisp -> disassem.lisp
3646 ** host-insts.lisp -> insts.lisp
3647 ** byte-comp.lisp -> target-byte-comp.lisp
3648 ** host-byte-comp.lisp -> byte-comp.lisp
3649 ** host-signal.lisp -> signal.lisp
3650 ** host-defstruct.lisp -> defstruct.lisp
3651 ** late-target-type.lisp -> deftypes-for-target.lisp
3652 Furthermore, several other previously target-only files foo.lisp (e.g.
3653 hash-table.lisp and random.lisp) have been split into a target-and-host
3654 foo.lisp file and a target-only target-foo.lisp file, with their key type
3655 definitions in the target-and-host part, so that the cross-compiler will
3656 know more about target types.
3657 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3658 gone away. In their place are various *BACKEND-FOO* variables
3659 corresponding to the slots of the old structure.
3660 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3661 into the target SBCL has been fixed.
3662 * The chill.lisp system for loading cold code into a running SBCL
3664 * Support for the CMU CL "scavenger hook" extension has been removed.
3665 (It was undocumented and unused in the CMU CL sources that SBCL was
3666 derived from, and stale in sbcl-0.6.1.)
3667 * Various errors in the cross-compiler type system were detected
3668 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3669 (enabling various consistency checks). Many of them were fixed,
3670 but some hard problems remain, so the compiler is back to
3671 running without *TYPE-SYSTEM-INITIALIZED* for now.
3672 * As part of the cross-compiler type system cleanup, I implemented
3673 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3674 * I have started adding UNCROSS calls throughout the type system
3675 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3676 on cross-compiler input files will be able to go away, and various
3678 * CONSTANTP now returns true for quoted forms (as explicitly required
3681 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3683 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3684 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3685 (MAKE-LIST -1). In the longer term, it ought to become true
3686 that declarations are assertions even at SAFETY 1. For now, it's not
3687 quite true even at SAFETY 3, but it's at least more nearly true..
3688 (Note that this change seems to increases the size of the system by
3689 O(5%) and to decrease the speed of the compiler by 20% or more.)
3690 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3691 for the problem of printing dozens of lines of distracting information
3692 about low-level system machinery as part of the top stack frame
3693 on entry to the debugger when an undefined function was called.
3694 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3695 is not reset to COMMON-LISP-USER.
3696 * Compilation of stuff related to dyncount.lisp has been made conditional
3697 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3698 smaller. The various dyncount-related symbols have been moved into
3699 a new "SB-DYNCOUNT" package.
3700 * tty-inspect.lisp has been renamed to inspect.lisp.
3701 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3702 feature has gone away. (When we eventually port to other flavors of
3703 libc and/or Unix, we'll try to make the differences between flavors
3704 invisible at the user level.)
3705 * Various other *FEATURES* tags, and/or their associated conditionals,
3706 have been removed if obsolescent, or given better documentation, or
3707 sometimes given more-mnemonic names.
3709 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3711 * tidied up "make.sh" script
3712 * tidied up system directory structure
3713 * better "clean.sh" behavior
3714 * added doc/FOR-CMUCL-DEVELOPERS
3715 * many many small tweaks to output format, e.g. removing possibly-confusing
3716 trailing #\. character in DESCRIBE-INSTANCE
3717 * (EQUALP #\A 'A) no longer signals an error.
3718 * new hashing code, including EQUALP hashing
3719 * tidied up Lisp initialization and toplevel
3720 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3721 * command line argument processing
3722 * added POSIX-GETENV function to deal with Unix-ish environment variables
3723 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3724 terminating SBCL on EOF
3725 * non-verbose GC by default
3726 * There is no more "sbcl" shell script; the sbcl file is now the C
3727 runtime executable (just like CMU CL).
3728 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3730 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3731 symbol and package information as Lisp native symbol and package objects
3732 instead of strings naming symbols and strings naming packages. This way,
3733 whenever packages are renamed (as in warm init), debug information is
3734 transformed along with everything else.
3735 * tweaked the optimization policy declarations which control the building
3736 of SBCL itself. Now, among other things, the system no longer saves
3737 source location debugging information. (This helps two problems at once
3738 by reducing SBCL size and by keeping SBCL from trying to look for its
3739 sources -- which may not exist -- when reporting errors.)
3740 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3741 debugging and testing purposes
3742 * cleaned up printing, making the printer call PRINT-OBJECT for
3743 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3744 methods, giving nearly-ANSI behavior
3745 * converted almost all special variables to use *FOO* naming convention
3746 * deleted PARSE-TIME functionality, since it can be done portably
3747 * moved some files out of cold init into warm init
3748 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3750 * regularized formatting of source files
3751 * added an install.sh script
3752 * fixed ridiculous memory usage of cross-compiler by making
3753 compiler/alloc.lisp not try to do pooling unless it can hook
3754 itself into the GC of the cross-compilation host. Now the system
3755 builds nicely on my old laptop.
3756 * added :SB-ALLOC in target-features.lisp-expr
3757 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3758 was not implemented)
3759 * re-did condition handling and note reporting in the compiler. Notes
3760 are no longer handled by signalling conditions. Style warnings
3761 and warnings are handled more correctly and reported in such a way
3762 that it's easy to find one or the other in your output (so that you
3763 can e.g. figure out which of many problems caused COMPILE-FILE to
3765 * changed the severity of several compiler warnings from full WARNING
3766 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3767 compiler note reporting so that it doesn't use the condition system
3768 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3769 COMPILE-FILE command)
3770 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3771 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3772 no longer supported, since their implementation was deeply intertwingled
3773 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3774 * removed bogus "support" for compiler macros named (SETF FOO), and
3775 removed the compiler macro for SETF INFO (but only after making a fool
3776 of myself on the cmucl-imp mailing list by posting a bogus patch for
3777 DEFINE-COMPILER-MACRO..)
3778 * Compiled files containing forms which have side effects on the Lisp
3779 reader (such as DEFPACKAGE forms) are now handled more correctly.
3780 (Compiler queuing of top level lambdas has been suppressed by setting
3781 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3782 * deleted various currently-unused source files, e.g. gengc.lisp. They
3783 may be added back at some point e.g. when porting to other architectures,
3784 but until they are it's distracting to distribute them and to try to
3786 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3787 there were so many of them they're just distractions, and UNCROSS is
3788 known to be able to handle the current sources
3789 * moved PROFILE functionality into TRACE, so that it will be clear
3790 how the wrapping and unwrapping of functions when you profile them
3791 interacts with the wrapping and unwrapping of functions when you
3792 trace them. (Actually, the functionality isn't there yet, but at least
3793 the interface specification is there. Hopefully, the functionality will
3794 arrive with some future maintenance release.)
3795 * removed host-oops.lisp
3796 * changed signature of QUIT function to allow UNIX-CODE argument
3797 * fixed READ-SEQUENCE bug
3798 * tweaked verbose GC output so that it looks more like the progress
3799 output that ANSI specifies for functions like LOAD
3800 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3801 * added <http://sbcl.sourceforge.com> to the banner information printed by