1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
3 * new feature: source path information is generated for macro-expansion
4 errors for use in IDE's like Slime (thanks to Helmut Eller)
5 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
6 * bug fix: compiler does not loop forever on an invalid type in
8 * improvement: compilation of most CLOS applications is significantly
11 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
12 * minor incompatible change: in sbcl-0.9.11 (but not earlier
13 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
14 into subclasses of the finalized class. Now user calls to
15 FINALIZE-INHERITANCE finalize just the one class, and calls by the
16 system return before any subclasses are finalized.
17 * minor incompatible change: The reader no longer ignores errors
18 regarding non-existent packages in #+ and #- feature tests.
19 * new feature: command line options --no-sysinit, --no-userinit to
20 inhibit loading the corresponding init files
21 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
22 for sending data through UDP sockets (thanks to François-René Rideau)
23 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
24 error is signalled instead (thanks to François-René Rideau)
25 * minor incompatible change: Error signalling behaviour of lexical
26 operations violating package locks has changed slightly. Refer to
27 documentation on package locks for details.
28 * enhancement: EVAL can process IF-expressions without resorting to the
30 * bug fix: Some operations involving SIN, COS, or TAN weren't being
31 constant-folded properly due to missing out-of-line functions.
32 (reported by Mika Pihlajamäki)
33 * bug fix: LISTEN sometimes returned T even in cases where no data was
34 immediately available from the stream
35 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
36 were reversed. (reported by Levente Mészáros)
37 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
38 when given a vector displaced to an adjustable array. (reported
40 * bug fix: derivation of float boundaries from numbers outside the
41 appropriate float range (reported by John Wiseman)
42 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
44 * bug fix: precomputing cache entries for generic functions with
45 some subclasses of specializers as yet invalid does not attempt to
46 fill a cache line with a negative offset. (reported by Levente
48 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
49 allow condition class objects as arguments to DOCUMENTATION and
50 (SETF DOCUMENTATION); only find and set documentation for
51 structure names for the STRUCTURE doc-type. (suggested by Gary
53 * improvements to the Win32/x86 port:
54 ** (user-homedir-pathname) and default initialization file
55 locations now know about the user's "Documents and Settings"
56 directory (thanks to Yaroslav Kavenchuk)
57 ** run-program is implemented (thanks to Mike Thomas)
58 ** sockets support (thanks to Timothy Ritchey)
59 ** better backtrace support (thanks to Alastair Bridgewater)
60 ** sb-grovel supported
61 ** asdf-install and sb-posix work somewhat
62 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
63 * improvements to the Solaris/x86 port:
64 ** works on Solaris 11/Solaris Express
65 ** floating-point exception handling support
66 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
67 * fixed some bugs revealed by Paul Dietz' test suite:
68 ** REMOVE-METHOD returns its generic function argument even when
69 the method is not one of the generic functions' methods.
70 ** objects declared as MEMBER types can be admissible arguments to
72 ** printing characters should simply be printed by the FORMAT ~:C
74 ** compiler failure when compiling functions with hairy constant
75 defaults for optional parameters.
76 ** compiler produces wrong code when MAYBE-INLINE-expanding a
77 function, which is already optimized.
79 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
80 * new platform: experimental support for SBCL x86/Darwin, including
81 MacOS X 10.4.5 on Intel.
82 * new feature: Unicode character names are now known to the system
83 (through CHAR-NAME and NAME-CHAR).
84 * new feature: the filesystem location of SBCL's core file is
85 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
86 * minor incompatible change: the contrib modules SB-POSIX and
87 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
88 this change is to make it easier to distribute
89 location-independent binaries.
90 * bug fix: as implied by AMOP, standardized classes no longer have
91 slots named by external symbols of public packages. (reported by
93 * optimization: calling functions via constant symbols -- as in
94 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
95 via the function object as in (FUNCALL #'FOO).
96 * optimization: on x86 and x86-64, the calling convention for the
97 case of dealing with an unknown number of values has been altered
98 to be friendlier to the prediction heuristics implemented,
99 particularly with reference to CALL and RET pairing. (thanks to
100 Alastair Bridgewater)
101 * optimization: on x86, the code for access to thread-local symbol
102 values and binding/unbinding of thread-local symbols is smaller.
103 (thanks to Alastair Bridgewater)
104 * enhancement: CONSTANTP is now able to determine constantness of
105 more complex forms, including calls to constant-foldable standardized
106 functions and some special forms beyond QUOTE.
107 * fixed bug: occasional GC crashes on Solaris/x86
108 * optimization: x86-64 supports stack allocation of results of simple
109 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
110 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
111 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
112 (thanks to James Knight)
113 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
114 by the standard (thanks to Stephen Compall)
116 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
117 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
118 be used for bundling the runtime and the core file into one
119 executable binary. This feature is not currently supported on all SBCL
120 platforms. (thanks to James Bielman and NIIMI Satoshi)
121 * new feature: a generational or ephemeral garbage collector is now
122 the default on the PowerPC platform (both Linux and Darwin). The
123 old Cheney (stop and copy) collector is a build-time option.
124 (thanks to Cyrus Harmon, after Raymond Toy)
125 * minor incompatible change: the method by which SBCL finds its
126 contributed modules has changed; it no longer relies on symbolic
127 links from an $SBCL_HOME/systems directory, but searches directly
128 in the subdirectories of $SBCL_HOME.
129 * enhancement: the dynamic heap size on the Linux/PPC platform is
130 markedly larger, even using the older Cheney garbage collector.
131 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
132 certain complicated string types. (reported by Gary King)
133 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
134 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
135 * fixed bug: DOCUMENTATION on structures no longer signals an
136 error if no structure type of the right name exists. (reported by
138 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
139 applicable non-standard primary or :AROUND method on
140 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
141 no longer cause unbound variable errors. (reported by Kevin Reid)
142 * fixed bug: metacircle resolution in cases where methods have slots
143 added before the slots from STANDARD-METHOD. (reported by Jean
145 * fixed bug: the Power PC assembly code for calling into lisp
146 sometimes computed its return address wrongly (depending on the
147 15th bit of the address, fixed at link time). This used to
148 manifest itself in a segmentation violation while building PCL.
149 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
150 many others over the years)
151 * fixed bug: ROOM no longer reports silly numbers for stack usage if
152 the stack pointer (treated as a signed integer) is negative.
153 (thanks to Peter van Eynde)
155 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
156 * new platform: experimental support for the Windows operating
157 system has been added. (thanks to Alastair Bridgewater)
158 * fixed several bugs in and robustified the PPC FFI (including
159 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
160 * bug fix: allow non-simple string symbol names (reported by Paul
162 * bug fix: interrupt handling on NetBSD (thanks to Richard M
164 * bug fix: saving a core corrupted callbacks on x86/x86-64
165 * bug fix: closed a loophole in metacircularity detection and
166 grounding in the PCL implementation of CLOS.
167 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
168 (reported by tomppa on #lisp)
169 * bug fix: clear the direction flag on Lisp -> C transitions, as
170 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
171 (reported by Andrej Grozin and Hendrik Maryns)
172 * optimization: major improvements to GC efficiency on GENCGC platforms
173 * optimization: faster implementation of EQUAL
174 * optimization: emit more efficient opcodes for some common
175 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
176 * optimization: several other minor code-generation improvements on x86-64
177 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
179 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
180 * minor incompatible change: (SETF CLASS-NAME) and (SETF
181 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
182 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
183 by ANSI as a generic function, and for consistency (SETF
184 GENERIC-FUNCTION-NAME) is treated likewise.
185 * fixed bug #233.b: make constraint propagation notice when a variable
186 value is changed after it is referenced but before it is used
187 * fixed bug #296: no more arbitrary behaviour from filenames with
188 odd characters as --load arguments. (reported by Adam Warner)
189 * fixed bug #390: :CHARACTER-SET pathname components now work as
190 expected. (reported by Tim Daly Jr)
191 * fixed bug #391: complicated :TYPE intersections in slot
192 definitions no longer cause an error in PCL internals.
193 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
194 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
195 returning the number of octets which would be written to the
196 file-stream. (thanks to Robert J. Macomber)
197 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
198 arguments to the internal stream functions. (thanks to David
200 * bug fix: stack exhaustion is now handled on the x86/SunOS
201 platform. (thanks to Daisuke Homma)
202 * bug fix: more accurate ROOM results on GENCGC platforms
203 * optimization: improved type inference for arithmetic-for
204 index variables in LOOP
205 * optimization: faster floating-point SQRT on x86-64
206 * fixed some bugs revealed by Paul Dietz' test suite:
207 ** DOCUMENTATION returns NIL instead of "" for method combinations
208 that don't have a docstring
210 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
211 * minor incompatible change: (SETF CLASS-NAME) and (SETF
212 GENERIC-FUNCTION-NAME) are no longer generic functions, and
213 therefore are not customizeable by user code (as seems to be at
214 least permitted and maybe required by AMOP). As a consolation,
215 however, the SBCL implementation of these functions now calls
216 REINITIALIZE-INSTANCE as specified by AMOP.
217 * bug fix: slot-definition documentation is propagated as per ANSI
218 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
219 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
220 argument, as do eslotd initialization methods. (from Pascal
221 Costanza's "Closer" project)
222 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
223 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
225 * bug fix: it is now possible to have more than one subclass of
226 STANDARD-GENERIC-FUNCTION without causing stack overflow.
227 (reported by Bruno Haible, Pascal Costanza and others)
228 * bug fix: the dependent update protocol now works for generic
229 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
231 * bug fix: condition-class instances corresponding to
232 DEFINE-CONDITION forms are now created eagerly. (reported by
233 Kalle Olavi Niemitalo on comp.lang.lisp)
234 * bug fix: floating point printing is more accurate in some
235 circumstances. (thanks to Simon Alexander)
236 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
237 merged with *DEFAULT-PATHNAME-DEFAULTS*.
238 * bug fix: callbacks on OS X now preserve stack-alignment.
239 (thanks to Cyrus Harmon)
240 * enhancement: the x86-64 disassembler is much better at
241 disassembling SSE instructions. (thanks to Lutz Euler)
242 * enhancement: CHECK-TYPE on a variable now allows type-inference
243 (thanks to Vincent Arkesteijn)
244 * optimization: improved performance of EUC-JP external format.
245 (thanks to NIIMI Satoshi)
246 * optimization: performance improvements to IO on file streams of
247 :ELEMENT-TYPE CHARACTER
248 * optimization: much faster memory allocation on x86-64
249 * optimization: faster garbage collections (latency and throughput)
251 * optimization: faster bignum subtraction and division on x86 and x86-64
252 (thanks to Lutz Euler)
253 * optimization: more accurate type-inference for loops that use a
254 floating point index variable or a negative step.
256 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
257 * bug fix: add a workaround to SBCL looping infinitely at startup on
258 Linux kernels with apparently buggy implementations of personality().
259 (thanks to Svein Ove Aas)
260 * bug fix: Unicode symbols are correctly printed in LDB backtraces
261 (thanks to David Lichteblau)
262 * bug fix: local bindings shadow global symbol macros for
263 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
264 * optimization: non-open coded uses of numeric comparison operators
265 (e.g. >) no longer cons when called with more than one parameter
266 on platforms supporting dynamic-extent allocation.
267 * enhancement: saving cores with foreign code loaded is now supported
268 on MIPS/Linux in addition to the previously supported platforms.
269 * bug fix: FIND-RESTART now tests for activity, not applicability when given
270 a restart object as identifier. (reported by Helmut Eller for CMUCL)
271 * bug fix: division by zero in sb-sprof when no samples were collected
272 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
273 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
274 (thanks to Alexander Kjeldaas)
275 * new feature: ignoring the return values of destructive functions
276 like NREVERSE signals a compile-time style-warning.
277 (thanks to Kevin Reid)
278 * new feature: the alignment of alien structure fields can be
279 explicitly specified. (thanks to Cyrus Harmon)
280 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
281 and dump core on SIGQUIT
283 ** incompatible change: threads do not inherit values of specials
284 from their parents (see manual)
285 ** bug fix: threads stacks belonging to dead threads are freed by the
286 next exiting thread, no need to gc to collect thread stacks anymore
287 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
288 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
289 ** bug fix: dynamic variable and thread start related gc lossage
290 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
292 * fixed some bugs revealed by Paul Dietz' test suite:
293 ** SUBTYPEP is slightly more accurate on heinously complicated
294 CONS types where some of the members have uncertain (in the
295 NIL, NIL sense) type relationships to each other.
297 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
298 no longer purified unless :PURIFY T is explicitly specified.
299 ** Non-purified cores are significantly smaller than before
301 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
302 * new feature: timers based on Zach Beane's excellent timer package
303 * added support for the following external formats: koi8-u,
304 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
305 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
306 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
307 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
308 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
309 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
310 * incompatible change: a threaded SBCL will no longer revert to
311 non-threaded mode on non-NPTL systems, but refuse to start entirely.
312 * bug fix: interrupts are disabled until startup is complete; no
313 more sigsegvs when receiving a signal to soon
314 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
316 * bug fix: add a workaround for the memory randomization features in
317 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
318 workaround will only be in effect on systems with the proc filesystem
320 * bug fix: printing objects of type HASH-TABLE signals a
321 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
323 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
324 have been running for over 50 days. (reported by Gilbert Baumann)
325 * bug fix: the logic for getting names of functions gets less
326 confused when confronded with alternate-metaclass
327 funcallable-instances. (reported by Cyrus Harmon)
328 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
329 consistent, even on internal alternate-metaclass objects.
330 * bug fix: flush closure information collected by physical
331 environment analysis prepass before the main pass. (bug reported
333 * bug fix: compiler pack phase does not modify a hashtable, which is
334 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
335 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
336 may be expanded. (reported by Alexander <malishev> on c.l.l)
337 * bug fix: no more style-warnings when compiling code using LAST
338 under high-SPEED low-SPACE settings. (reported by David Wallin
340 * fixed some incompatibilities between SBCL's MOP and the MOP
342 ** the METAOBJECT class is now implemented;
343 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
344 STANDARD-OBJECT, as required;
345 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
346 now compatible (as required by VALIDATE-SUPERCLASS); there
347 remains a consistency requirement over the presence or absence
348 of the FUNCTION class in the superclass list of finalized
349 classes; see the manual for more details;
350 ** the :AROUND method for COMPUTE-SLOTS on
351 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
352 requested slot ordering.
354 ** bug fix: parent thread now can be gc'ed even with a live
356 ** bug fix: binding a special with PROGV to no value is not
357 the same as that symbol not having been bound (thanks to
359 ** bug fix: binding specials is thread safe (thanks to
361 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
362 and SB-SPROF do not die with 'no handler for signal XX in
363 interrupt_handle_now(..)' anymore
364 ** bug fix: WITH-TIMEOUT works with multiple running threads
365 * fixed some bugs revealed by Paul Dietz' test suite:
366 ** ENSURE-GENERIC-FUNCTION should take a method class object for
367 the :method-class keyword argument.
369 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
370 * new port: the Solaris operating system on x86 processors is now
371 mostly supported, though some rough edges in the environment
372 remain. (thanks to Daisuke Homma)
373 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
374 space for its heap. (thanks to Thiemo Seufer)
375 * optimization: SBCL on MIPS platforms now supports dynamic-extent
376 closures. (thanks to Thiemo Seufer)
377 * minor incompatible change: eof selects abort in the debugger.
378 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
379 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
381 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
382 deprecated, and will go away in a future revision of SBCL.
383 * minor incompatible change: GC-ON and GC-OFF are no longer
384 implemented with a counter, it does not matter how many times gc
385 is switched on or off
386 * bug fix: discriminating functions for generic function classes
387 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
388 make invalid assumptions about method precedence order. (reported
390 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
391 platforms. (thanks to Thiemo Seufer)
392 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
393 * bug fix: correct pathname printing with printer escaping is on.
394 (thanks to Kevin Reid)
395 * bug fix: complex VOP definitions in "user-space" no longer trigger
396 package locks. (reported by Zach Beane)
397 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
398 for funcallable-instance functions, and is no different from
400 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
401 characters in the current readtable. (reported by Nicholas Neuss)
402 * bug fix: SUBTYPEP on various CONS types returns more a more
403 accurate acknowledgment of its certainty.
404 * optimizations: REMOVE-DUPLICATES now runs in linear time on
405 lists in some cases. This partially fixes bug 384.
406 * flush all standard streams before prompting in the REPL and the
408 * bug fix: signal handling and triggering gc do not conflict
409 directly anymore, in particular a high frequency sb-sprof does
410 not prevent gc from running
411 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
412 approximation for timezone and DST information between the
413 universal time epoch and the smallest negative 32-bit time_t.
414 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
415 year 1899 so long as the encoded time is non-negative.
416 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
417 the correct answer for non-constant rotate amounts when performing
418 an inline 32-bit rotation.
420 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
421 there is only one thread in the session
422 ** bug fix: memory leak for streams created in one thread and
423 written to in another
424 ** bug fix: lockup when compiled with gcc4
425 ** bug fix: race that allows the gc to be triggered when gc is
427 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
428 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
430 ** bug fix: finished threads are gc'ed properly
431 * fixed some bugs revealed by Paul Dietz' test suite:
432 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
433 has arguments with a different set of applicable methods from
434 the orignal arguments.
435 ** The type error thrown by MAP now has a correct expected-type
437 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
438 name a compiled function.
439 ** Three MISC tests where a large bignum was improperly coerced to
440 a float that couldn't represent that bignum during type
441 derivation were fixed.
442 ** SUBTYPEP can now handle types involving the negation of a
443 list-form FUNCTION type.
444 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
445 as (COMPLEX (AND RATIO (NOT FIXNUM))).
446 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
448 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
449 * New feature: Experimental support for bivalent streams: streams
450 opened with :element-type :default now allow character and binary
451 (unsigned-byte 8) I/O
452 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
453 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
455 * bug fix: no more highly sporadic "couldn't check whether ~S is
456 readable" when reading a stream and an interrupt hits in the middle
457 of a select system call
458 * compiler better recognizes complex arrays (reported by Eduardo
460 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
461 for variables on SBCL built with linkage-tables. (reported by Luis
463 * various error reporting improvements.
464 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
465 (Thanks to Thiemo Seufer)
466 * bug fix: nested reader invokations work correctly
467 * bug fix: it is possible to have simultaneous references to foreign
468 code and foreign data with the same name.
470 ** added x86-64 support
471 ** incompatible change: the threading api now works with thread
472 objects instead of thread ids
473 ** bug fix: threads are protected from signals and interruption when
474 starting up or going down
475 ** bug fix: a race where an exiting thread could lose its stack to gc
476 ** bug fix: don't halt on infinite error in threads if possible
477 ** fixed numerous gc deadlocks introduced in the pthread merge
478 ** bug fix: fixed thread safety issues in read and print
479 ** bug fix: debugger doesn't hang on session lock if interrupted at
480 an inappropriate moment
481 ** bug fix: run-program is now thread safe(r)
482 ** bug fix: inner with-recursive-lock no longer releases the mutex
483 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
484 * fixed some bugs revealed by Paul Dietz' test suite:
485 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
486 EXPECTED-TYPE slots filled.
487 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
489 ** MAKE-CONDITION accepts classes as type-designators.
490 ** COMPILE may never return NIL.
491 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
492 range before calling Unix time functions
494 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
495 * numerous signal handling fixes to increase stability
496 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
497 * minor incompatible change: we now correctly canonize default
498 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
499 as per the MOP, rather than the historical (INITARG INITFUNCTION
500 INITFORM). (reported by Bruno Haible)
501 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
502 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
503 for more information.
504 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
505 pathname is a directory pathname.
506 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
507 * dynamic space size on PPC has been increased to 768Mb. (thanks to
509 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
510 :METACLASS argument in addition to a class name. (reported by
511 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
512 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
513 (reported by Bruno Haible; patch from Gerd Moellmann)
514 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
516 * bug fix: more cleanups to the floating point exception handling on
517 x86-64 (thanks to James Knight)
518 * bug fix: the compiler does not try to constant fold calls of
519 COERCE and BIT-* functions when they return freshly constructed
520 objects. (reported by Nikodemus Siivola and Paul Dietz)
521 * optimization: improved the MIPS versions of generic (in the
522 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
523 * optimization: direct conversion of (unsigned-byte 32) to floats on
524 the PowerPC platform.
525 * optimization: structure instances with raw slots now use less
526 memory, and probably show better memory locality. (thanks to
528 * optimization: DYNAMIC-EXTENT declarations for lists and closures
529 are treated as requests for stack allocation on the x86-64,
530 Alpha-32, PPC and SPARC platforms.
531 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
532 the wrong file descriptor; implementation of SOCKET-OPEN-P.
533 (thanks to Tony Martinez)
535 ** gcing a dead thread can no longer lead to lockups
536 ** threads block signals until they are set up properly
537 ** errno is no longer shared by threads
538 ** interrupt-thread restores the eflags register on x86
539 ** fixed some lockups due to gc/thread interaction
540 * fixed some bugs revealed by Paul Dietz' test suite:
541 ** invalid dotted lists no longer raise a read error when
543 ** signal an error if a symbol that names a declaration is used
544 as the name of a type, or vice versa
545 ** allow using the (declare (typespec var*)) abbreviation for
546 (declare (type typespec var*)) with all type specifiers
547 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
548 ** always return NIL from PPRINT-POP when OBJECT is NIL
549 ** don't signal errors when pretty-printing malformed LABELS,
550 FLET or MACROLET forms
551 ** declarations in a DOLIST body are also in scope for the
553 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
555 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
558 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
559 * fixed cross-compiler leakages that prevented building a 32-bit
560 target with a 64-bit host compiler.
561 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
562 opened with :IF-EXISTS :OVERWRITE.
563 * fixed bug 281: error for an invalid qualifier in a short-form method
564 combination method is not signalled until the faulty method is called.
565 * bug fix: iteration variable type inferrer failed to deal with open
566 intervals. (reported by Alan Shields)
567 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
568 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
569 or more runtime options were provided to the sbcl binary.
570 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
571 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
572 * optimization: there's now a fast-path for fixnum arguments in the
573 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
574 * optimization: the code generated on x86-64 is more compact thanks
575 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
576 * bug fix: floating-point exception handling now partly works on
577 x86-64. (Thanks to James Knight)
578 * improvement to the MIPS backend from Thiemo Seufer: C-style
579 64-bit long long arguments and return values to alien functions
581 * fixed some bugs revealed by Paul Dietz' test suite:
582 ** the type-error signalled from WARN has a filled-in DATUM slot.
583 ** the type-error required when a stream is not associated with
584 a file has the stream as its datum.
585 ** type-errors on single-floats on x86-64 no longer have
586 :INVALID-OBJECT as the datum
587 ** the type-errors signalled for invalid function names now have
588 a correct expected type
589 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
590 for structures defined with a :TYPE. Documentation strings for
591 typed structures are no longer immediately discarded
592 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
593 broadcast streams and synonym streams. FILE-LENGTH now also works
594 on broadcast streams.
596 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
597 * incompatible change: the --noprogrammer option, deprecated since
598 version 0.7.5, has been removed. Please use the equivalent
599 --disable-debugger option instead.
600 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
602 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
603 inoperational for a while now) has been completely removed.
604 * null lexical environments are now printed as #<NULL-LEXENV>,
605 significantly reducing the amount of clutter in typical backtraces.
606 * documentation on weak pointers, finalization, and after GC hooks
607 has been added to the manual.
608 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
609 as other specialized array types, is much faster. SUBSEQ and
610 COPY-SEQ on such arrays have also been sped up.
611 * optimization: EQL is now more efficient when at least other argument
612 is known to be of type (OR FIXNUM (NOT NUMBER)).
613 * fixed bug: the runtime is now less vulnerable to changes in the
614 size of the SBCL object on OS X, and virtual memory is reserved for
615 all spaces that need to be at a fixed address.
616 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
617 * fixed bug: finalizers and after GC hooks that cause consing are now
619 * fixed bug: compiler error messages and summaries are now printed to
620 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
621 * fixed inference of the upper bound of an iteration variable.
622 (reported by Rajat Datta).
623 * fixed bug 211e: calling local functions with duplicated constant
624 keyword argument no longer causes a bogus style warning about an
626 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
627 ftype declaration to be disregarded. (reported by Dave Roberts)
628 * fixed bug 373: caused by erronous compilation of references to alien
629 variables in the runtime on ppc/darwin.
630 * fixed bug 376: CONJUGATE type deriver.
631 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
632 (reported by Baughn on #lisp)
633 * a cross-compiler bug on non-x86 platforms has been identified and
634 fixed. (thanks to Bruno Haible)
635 * improvements to the MIPS runtime code for increased stability.
636 (thanks to Thiemo Seufer)
637 * increased the maximimum compact environment size to allow
638 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
639 * partial workaround for bug 135: don't name the function we're
640 calling for hairy cases of EVAL, so as not to accumulate one environment
641 entry per call to EVAL. (reported by Kevin Reid)
642 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
643 * optimization: type testing for non-vector arrays should be faster.
644 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
646 * bug fix: setting 31st element of a bit vector to zero did not work
648 * bug fix: redefining a class definition which failed due to a
649 previous accessor / function clash now works (but see BUGS entry
650 #380 for more problems in this area). (thanks to Zach Beane)
651 * the long form of DEFINE-METHOD-COMBINATION disables method group
652 checking when given a single method group with pattern *. (thanks
654 * on x86 compiler supports stack allocation of results of simple
655 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
656 * support for building and running on Mac OS X 10.4 "Tiger" has been added
657 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
658 * fixed some bugs related to Unicode integration:
659 ** the restarts for recovering from input and output encoding
660 errors only appear when there is in fact such an error to
662 * fixed some bugs revealed by Paul Dietz' test suite:
663 ** MISC.549 and similar: late transformation of unsafe type
664 assertions into derived types caused unexpected code
666 ** SCALE-FLOAT type deriver is less wrong.
667 ** type derivers for EXP, LOG and similar functions compute result
668 types for complex arguments better.
669 ** (MISC.563) CONJUGATE type deriver works for very restricted
671 ** out-of-line type testers for character strings are available.
672 ** EQUAL compiler transform understands specialness of objects
674 ** accessing double-floats stored on the stack now works on x86-64.
675 ** debugger internals could sometimes create invalid lispobjs,
676 resulting in GC crashes.
677 ** MISC.548: type check weakening can convert required type into
679 ** initialization forms for bindings are not in scope of free special
682 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
683 * incompatible change: thread support for non-NPTL systems has
684 been removed - locking is buggy and unreliable. A threaded
685 SBCL build will now warn at startup and refuse to create
686 new threads, unless futex support is detected in the kernel
687 * incompatible change: the top level REPL now has only an ABORT
688 restart associated with it, not TOPLEVEL and ABORT as it used to.
689 TOP and TOPLEVEL are now available as debugger commands for
690 returning to the top level.
691 * incompatible change: forms evaluated in the REPL now use the
692 global optimization policy.
693 * incompatible change: user- and system-initialization files are
694 no longer processed with LOAD, but by READ and EVAL; hence the
695 global optimization policy, startup package, readtable, etc,
697 * The .fasl file format number has been incremented because of
698 various incompatible changes.
699 * internal entry point details and argument counts no longer appear
700 in backtraces unless explicitly requested by setting
701 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
702 * built-in and standard functions no longer have names like "top
703 level local call to FOO".
704 * fixed bug 32: functions defined in non-null lexical environments
705 now have more legible printed representation
706 * fixed bug 33: functions defined in non-null lexical environemnts
707 are now more amenable to inspection by INSPECT.
708 * workaround for bug 354: XEPs no longer appear in backtraces unless
709 explicitly requested.
710 * fixed bug: receiving the signal which results from attempting to
711 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
712 SIGBUS on other *BSDs) on architectures where the C stack is also
713 the Lisp stack (x86 and x86-64) from foreign code no longer leads
714 to debugger confusion or wild instability. (reported by Cheuksan
716 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
717 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
718 (reported by Lutz Euler)
719 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
720 compiled code. (reported by Paul Dietz)
721 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
722 specialized parameter to a method to something that is not TYPEP
723 the specializer is now possible.
724 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
725 face of package deletion.
726 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
727 pathname. (reported by David Tolpin on c.l.l)
728 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
729 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
730 STRING-STREAM and FILE-STREAM, respectively.
731 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
732 than previously; in addition, its test suite is now run on build.
733 (thanks to Luke Gorrie)
734 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
735 * optimization: added a immediate representation for single-floats
737 * fixed some bugs revealed by Paul Dietz' test suite:
738 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
739 ** attempting to create a package with a colliding nickname causes
740 correctable errors to be signalled.
741 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
742 ** several x86-64 backend bugs related to sign-extension of immediate
745 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
746 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
747 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
749 * fixed loading of multiply forward-referenced layouts.
750 (thanks to Cheuksan Wang)
751 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
752 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
753 (see COMPILE-FILE documentation for details of :PRINT option)
754 * fixed bugs 19 and 317: fixed-format floating point printing is
755 more accurate. This also fixes a bug reported by Adam Warner
756 related to the ~@F format directive.
757 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
759 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
760 dispatch table if the from-char is a dispatch macro character.
761 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
762 by the wordsize no longer ignore the last word. (reported by Lutz
764 * fixed bug in type checking of dynamic-extent variables. (reported
766 * optimization: sequence traversal functions use their freedom to
767 coerce function designators to functions.
768 * optimization: code with many calls to CLOS methods specialized on
769 CLOS classes has had redundant type checks removed.
770 * optimization: added declarations to speed up operations that access
771 the internal character database (for example STRING-UPCASE)
772 * optimization: comparison operations between floats and sufficiently small
773 fixnums no longer create extra rationals
774 * fixed some bugs related to Unicode integration:
775 ** portions of multibyte characters at the end of buffers for
776 character-based file input are correctly transferred to the
777 start of the buffer at the next read.
778 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
779 passing it through to OPEN.
780 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
781 argument given to internal calls to OPEN.
782 ** the built-in comment readers (introduced by character sequences
783 ";" and "#|") are more forgiving to encoding errors; they will
784 STYLE-WARN and then attempt to resync the stream at a character
785 boundary. (thanks to Teemu Kalvas)
786 * fixed some bugs in the x86-64 port:
787 ** Negative short int return values from c-calls are sign-extended
789 ** The stack is aligned to 16-bytes for c-calls, as required by
790 the ABI. (thanks to Cheuksan Wang)
791 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
792 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
793 for enums. (thanks to Vincent Arkesteijn)
794 ** Multiple small optimizations and bugfixes for floating point
796 * fixed some bugs revealed by Paul Dietz' test suite:
797 ** Space, Tab, Linefeed, Return and Page have the invalid
798 secondary constituent character trait.
799 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
801 ** WITH-INPUT-FROM-STRING should only update the index place on
803 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
805 ** Bit-array operations (BIT-AND and similar) worked incorrectly
806 with one-dimensional arrays with fill pointers.
807 ** TYPE-OF failed on a complex with an integer realpart and a
809 ** compiler failure during type inference for the code of form
810 (IF (EQL X (THE ...)) ...) (MISC.535).
812 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
813 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
814 platform. The port supports 61-bit fixnums, large memory spaces
815 and reloading shared object files.
816 * enhancement: saving cores with foreign code loaded is now
817 supported on ppc/Darwin in addition to the previously supported
819 * enhancement: the statistical profiler now walks deeper into the
820 call stack for more meaningful call-graphs and accrued time
821 reports (x86/x86-64 only). It also now reports time spent in
823 * enhancement: it is now possible to trace most individual methods
824 of a generic function in addition to tracing the generic function
826 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
828 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
829 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
830 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
831 produces an error. (thanks to Vincent Arkesteijn)
832 * bug fix: NAMESTRING on pathnames with :WILD components in their
833 directories works correctly. (thanks to Artem V. Andreev)
834 * fixed bug 125: compiler preserves identity of closures. (reported
836 * bug fixed: functions with &REST arguments sometimes failed with
837 "Undefined function" when compiled with (DEBUG 3). (reported by
839 * bug fix: overflow during compiling of setting element of a bit
840 vector with constant index and value. (reported by Timmy Douglas)
841 * build fix: fixed the dependence on *LOAD-PATHNAME* and
842 *COMPILE-FILE-PATHNAME* being absolute pathnames.
843 * on x86 compiler partially supports stack allocation of dynamic-extent
845 * GO and RETURN-FROM do not check the extent of their exit points
846 when compiled with SAFETY 0.
847 * fixed some bugs related to Unicode integration:
848 ** encoding and decoding errors are now much more robustly
849 handled; it should now be possible to recover even from invalid
850 input or output to the terminal. (thanks to Teemu Kalvas)
851 ** provided a first cut at implementing STRING-TO-OCTETS and
852 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
853 ** altered the SB-MD5 contributed module slightly, changing the
854 interface just enough for it to be supportable for builds where
855 lisp characters are not eight bits.
856 * fixed some bugs revealed by Paul Dietz' test suite:
857 ** the FORMATTER-generated functions for ~V[ conditionals require
858 the correct number of arguments.
859 ** READ-FROM-STRING returns the mandated second value when applied
860 to displaced strings.
861 ** the #\Rubout and #\Backspace characters are treated as invalid
862 constituent characters by the tokenizer.
864 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
865 * new feature: reloading changed shared object files with
866 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
867 * new feature: references to foreign variables and functions
868 can now be compiled and loaded before the corresponding shared
869 object file is loaded, as long as the foreign definitions are
870 available at runtime.
871 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
872 supported. (thanks to Dan Debertin)
873 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
874 just keywords. (thanks to Vincent Arkesteijn)
875 * fixed bug #331: structure-class instances corresponding to
876 DEFSTRUCT forms are now created eagerly.
877 * fixed bug #345: backtraces from calls to undefined functions work
878 on x86 as well. Related bug #61 is now also partially fixed on x86
879 (backtraces from throws to unknown catch tags.)
880 * bug fix: lambda-list parsing is now stricter vrt. order and number
881 of lambda-list keywords.
882 * bug fix: as specified by AMOP, an error is signalled if a
883 class-option appears multiple times in a DEFCLASS form. (reported
885 * bug fix: RANDOM can be compiled when the compiler derives the type
886 of its numeric argument as a disjoint set of small integers.
887 (reported by Paul Dietz)
888 * bug fix: starting a core saved with shared objects loaded when
889 those objects are not available no longer causes threaded SBCL to
890 hang. (reported by Sean Ross)
891 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
893 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
894 argument, which must always be NIL. (reported by Kalle Niemitalo)
895 * bug fix: printing 1.0d+23 no longer results in an error.
896 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
897 * bug fix: structure slot setters preserve evaluation order. (thanks
899 * fixed some bugs related to Unicode integration:
900 ** RUN-PROGRAM can allow its child to take input from a Lisp
901 stream. (reported by Stefan Scholl)
902 ** ASDF-INSTALL successfully downloads .tar.gz files in all
903 locales. (reported by Ken Causey)
904 * fixed some bugs revealed by Paul Dietz' test suite:
905 ** INCF, DECF and REMF evaluate their place form as specified in
907 ** FORMATTER expands ~{ ~} iteration directives with V or #
908 parameters correctly.
909 ** FORMATTER deals with the ~@[ ~] conditional directive where the
910 consequent uses no arguments correctly.
911 ** the system has a partial understanding of the (COMPLEX RATIO)
913 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
916 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
917 * new feature: a build-time option (controlled by the :SB-UNICODE
918 keyword feature, enabled by default) for building the system with
919 support for the entire 21-bit character space defined by the
921 * new feature: the system now has rudimentary external-format
922 support; the primary user-visible change at this time is that
923 characters with the high bit set (such as lower-case-e-acute) will
924 print correctly to a terminal in a UTF-8 environment.
925 * minor incompatible change: BASE-CHAR no longer names a class;
926 however, CHARACTER continues to do so, as required by ANSI.
927 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
928 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
929 has been moved to the SB-EXT package (temporarily re-exported from
931 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
932 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
934 * fixed bug #351: better error-handlind and reporting for malformed
936 * fixed bug #350: bignum-printing is now more memory-efficient,
937 allowing printing of very large bignums, eg. (expt 2 10000000).
938 (reported by Bruno Haible)
939 * fixed bug #302: better primitive-type selection for intersection
941 * fixed bug #308: non-graphic characters now all have names, as
942 required. (reported by Bruno Haible)
943 * bug fix: redefining a class with different superclasses now correctly
944 removes it from the direct-subclasses of its previous superclasses.
945 (reported by David Morse)
946 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
947 new value now works. (reported by Bruno Haible)
948 * bug fix: correct canonicalization of multiple non-standard slot
949 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
950 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
951 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
952 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
953 now exists, an signals an error.
954 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
955 during macroexpansion for non-iteration variables as well. (reported
956 by Bruno Haible for CMUCL)
957 * bug fix: Cyclic structures and unprintable objects in compiler
958 messages no longer cause errors. (reported by Bruno Haible)
959 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
960 and READ-FROM-STRING all now return a primary value of NIL if
961 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
962 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
963 (reported by Bruno Haible for CMUCL)
964 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
965 do not designate a condition. (reported by Bruno Haible for
967 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
968 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
969 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
970 permitting resolution in favour of any of the conflicting symbols.
971 (reported by Bruno Haible for CMUCL)
972 * bug fix: EQUAL compiler optimizations is less aggressive on
973 strings which can potentially compare true despite having distinct
974 specialized array element types.
975 * bug fix: unit enumerations can be defined without dividing by
976 zero. (thanks to Vincent Arkesteijn)
977 * FORMAT compile-time argument count checking has been enhanced.
978 (report from Bruno Haible for CMUCL)
979 * a partial workaround for the bug 262: the compiler does not try to
980 inline-expand a local function doing RETURN-FROM from a deleted
981 BLOCK. (thanks to Peter Denno for the bug report and to David
982 Wragg for the simple test case)
983 * fixed some bugs revealed by Paul Dietz' test suite:
984 ** RENAME-PACKAGE allows all package designators as new package
986 ** constraint propagation and lambda variable substitution are
987 more cautious in dealing with partially deleted code.
988 ** compiler sometimes lost reoptimization passes.
989 ** CERROR, given a condition as condition designator, passes its
990 remaining arguments to the continue format control without
992 ** Case-altering FORMAT directives work correctly on non-ASCII
994 ** The REAL type specifier handles bounds outside the floating
995 point ranges without signalling FLOATING-POINT-OVERFLOW.
996 ** Functions with IR1-transformations can create intercomponent
997 references to global functions.
998 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1000 ** FORMAT ~R treats a nil value for its first parameter correctly.
1002 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1003 * enhancement: saving cores with foreign code loaded is now
1004 supported on x86/NetBSD and sparc/Linux in addition to the previously
1005 supported platforms.
1006 * bug fix: on some platforms repeated installations caused multiple
1007 copies of HTML documentation to be installed -- should not happen
1008 any more. (reported by Stefan Scholl)
1009 * bug fix: parsing self-recursive alien record types multiple times
1010 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1011 original patch by Helmut Eller for CMUCL)
1012 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1013 (thanks to Richard Kreuter)
1014 * bug fix: defining classes whose accessors are methods on existing
1015 generic functions in other (locked) packages no longer signals
1016 bogus package lock violations. (reported by François-René Rideau)
1017 * bug fix: special variables as DEFMETHOD parameters no longer have
1018 associated bogus type declarations. (reported by David Wragg and
1020 * bug fix: read-write consistency on streams of element-type
1021 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1022 * bug fix: redefiniton of the only method of a generic function with
1023 no DEFGENERIC no longer emits a full WARNING. In addition,
1024 redefinition of generic functions with no DEFGENERIC to an
1025 incompatible lambda list now signals an error. (thanks to Zach
1027 * bug fix: DEFGENERIC now works even when there's a function of the
1028 same name in an enclosing lexical environment. (thanks to Zach
1030 * fixed compiler failure, caused by instrumenting code during
1031 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1032 * optimization: added loop analysis and improved register allocation
1033 * fixed some bugs revealed by Paul Dietz' test suite:
1034 ** POSITION on displaced vectors with non-zero displacement
1035 returns the right answer.
1036 ** (SIMPLE-STRING) is a valid type specifier for sequence
1038 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1040 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1041 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1043 ** The justification version of the FORMAT ~< directive treats
1044 non-zero minpad parameter correctly.
1046 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1047 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1048 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1049 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1050 the supported interface.
1051 * new feature: Single-stepping of code compiled with DEBUG 2 or
1052 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1053 * new feature: saving cores with foreign code loaded is now
1054 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1055 Timothy Moore's work for CMUCL)
1056 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1057 parameters to * instead of NIL if no initform is supplied.
1058 (reported by Johan Bockgaard)
1059 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1060 correctly when FUNCALL appears as the car of the form. Note:
1061 despite this FUNCALL forms are not currently subject to
1062 compiler-macro expansion. (port of Raymond Toy's fix for the
1063 same from CMUCL, reported by Johan Bockgaard)
1064 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1065 (thanks for Teemu Kalvas)
1066 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1067 case instead of terminating the iteration (thanks for Julian
1068 Squires, Sean Champ and Raymond Toy)
1069 * bug fix: incorrect expansion of defgeneric that caused a style
1070 warning. (thanks for Zach Beane)
1071 * bug fix: slot accessor effective method computation works properly
1072 for classes with multiple non-standard applicable methods on
1073 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1074 * on x86 compiler supports stack allocation of results of LIST and
1075 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1076 CMUCL implementation by Gerd Moellmann)
1077 * fixed some bugs revealed by Paul Dietz' test suite:
1078 ** FORMAT strings with both the ~<~:;~> form of the justification
1079 directive and pretty-printing directives cause an error.
1081 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1082 * incompatible change: the internal functions
1083 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1084 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1085 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1086 instead of the old functions.
1087 * new feature: on platforms where "dladdr" is available foreign
1088 function names now appear in backtraces. (based on Helmut Eller's
1090 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1091 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1093 * documentation: ASDF manual is now distributed with SBCL.
1094 * bug fix: SBCL can now load its contributed modules with REQUIRE
1095 even if the system-provided entries have been removed from
1096 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1098 * bug fix: backtraces involving undefined functions or assembly
1099 routines are more informative. (thanks to Brian Downing)
1100 * bug fix: mutually referent alien structures now work correctly.
1101 (reported by Rick Taube)
1102 * bug fix: structures defined by WITH-ALIEN can be referred to
1103 within other definitions in the same WITH-ALIEN.
1104 * bug fix: division operators (MOD, TRUNCATE and the like) with
1105 constant zero divisors and integer dividends no longer generate
1107 * bug fix: provide default methods for INPUT-STREAM-P and
1108 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1109 * bug fix: improve the detection and resolution of MOP metacycles
1110 (where effective method computation depends on the generic function
1111 having its effective method computed). (reported by Bruno Haible)
1112 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1113 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1114 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1115 * bug fix: pathnames with relative directory components can be
1116 represented relative to default pathnames.
1117 * optimization: in taking the GCD of bignums, reduce the two bignums
1118 to approximately the same size (using Euclid's algorithm) before
1119 applying the more sophisticated binary GCD. (thanks to Juho
1121 * optimization: COUNT on bitvectors now operates word-at-a-time.
1122 * optimization: ASH with a positive, but not necessarily constant,
1123 (leftwards) shift, when in a modular context, is compiled to a
1125 * fixed some bugs revealed by Paul Dietz' test suite:
1126 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1127 if the corresponding argument is NIL.
1128 ** FORMAT directives accepting a minpad parameter treat negative
1130 ** literal commas as character parameters to FORMAT directives are
1132 ** literal spaces directly after ~<Newline> directives within a
1133 format-logical-block (~:< ~@:>) do not induce :FILL-style
1134 conditional newlines.
1135 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1136 as specified: it no longer includes conditional newlines).
1137 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1139 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1140 :PER-LINE-PREFIX argument does not evaluate to a string.
1142 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1143 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1144 the manual for details; for now, package locks can be disabled by
1145 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1146 if no major problems are found then it is likely that they will be
1147 compiled in unconditionally.
1148 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1149 unsupported operators on all platforms. To load a shared library
1150 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1151 object file, link it into a shared library outside of SBCL and
1152 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1154 * bug workaround: ROOM T and the GC were not working together
1155 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1156 reporting facility aren't preserved. That reporting has been
1157 disabled, so now until and unless someone figures out how to make it
1158 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1159 * minor incompatible change: as threatened around sbcl-0.8.0, the
1160 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1161 The symbols are also exported from SB-PCL for backwards
1162 compatibility, but more so than before SB-PCL should be treated as
1163 an implementation-internal package.
1164 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1166 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1167 It is known as of this release not to work on the Alpha, however.
1168 * fixed bug #167: errors signalled due to illegal syntax in method
1169 bodies are now more legible.
1170 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1171 designators and can hence be used with TYPEP.
1172 * fixed bug #333: CHECK-TYPE now ensures that the type error
1173 signalled, if any, has the right object to be accessed by
1174 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1175 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1176 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1178 * fixed bug #334: programmatic addition of slots using specialized
1179 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1180 and :ALLOCATION :CLASS slots.
1181 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1182 float exponents. (rereported by Peter Seibel)
1183 * fixed a bug: #\Space (and other whitespace characters) are no
1184 longer considered to be macro characters in standard syntax by
1185 GET-MACRO-CHARACTER.
1186 * fixed bug: initialization of condition class metaobjects no longer
1187 causes an instance of the condition to be created. (reported by
1189 * fixed bug: it is now possible to have slots such that
1190 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1191 description is neither :INSTANCE nor :CLASS.
1192 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1193 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1194 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1195 non-standard SLOT-DEFINITION-ALLOCATION values.
1196 * fixed bugs: various race conditions handling exiting threads.
1197 CL-PPCRE's thread torture test now passes.
1198 * fixed arguably-a-bug: GC time proportional to number of threads in
1199 system even when most of them are idle
1200 * optimization: improved performance of BIT and SBIT on bit-vectors.
1201 * .fasl file incompatibility: The fasl file version number has
1202 been incremented because of changes associated with package locks.
1204 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1205 * minor incompatible change: the system no longer provides
1206 optimization hints (conditions which are TYPEP
1207 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1208 mode; these hints are still emitted when higher SPEED optimization
1210 * new contrib module: a sampling profiler (profiling by statistical
1211 sampling, rather than by instrumenting functions) is available as
1212 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1213 * the behaviour of the standard function ED is now customizeable by
1214 third parties through a hook variable: see ED's documentation
1215 string for information on the protocol.
1216 * the compiler no longer emits efficiency notes for (FUNCALL X)
1217 when the type of X is uncertain under default optimization
1219 * fixed bug 276: mutating a binding of a specialized parameter to a
1220 method to something that is not TYPEP the specializer is now
1222 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1223 exhibit the required behaviour on the x86 platform. (thanks to
1224 Peter van Eynde, Eric Marsden and Bruno Haible)
1225 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1226 even for difficult arguments. (reported by Peter Graves)
1227 * fixed bug 141a: the backquote printer now descends quoted
1229 * fixed another bug in backquote printing: no more destructive
1230 modification of the form's list structure. (reported by Brian
1232 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1233 move between its address being taken and the call to
1234 interrupt_thread, fixing a crashing race condition.
1235 * the SB-POSIX contrib implementation has been adjusted so that it
1236 no longer exhibits ridiculously poor performance when constructing
1237 instances corresponding to C structs.
1239 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1240 * minor incompatible change: the sb-grovel contrib now treats C
1241 structures as alien (in the sense of SB-ALIEN) objects rather than
1242 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1243 has implications for memory management of client code
1244 (sb-grovel-returned objects must now be manually managed) and for
1245 type safety (alien objects now have full types).
1246 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1247 used to control emission of compiler diagnostics, rather than the
1248 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1249 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1250 quality should be considered deprecated.
1251 * install.sh now installs the user manual as well
1252 * (not quite a new documentable feature, but worth considering in
1253 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1254 beginnings of a semantically meaningful condition hierarchy is
1255 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1257 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1258 designator as the defaults argument.
1259 * fixed bug: Displaced arrays whose displaced-to array has become
1260 too small now cause ARRAY-DIMENSION to signal an error, providing
1261 for safer bounds-checking. (reported by Bruno Haible)
1262 * fixed bug: DEFCLASS slot definitions with identical :READER and
1263 :WRITER names now signal a reasonable error. (reported by Thomas
1265 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1267 * fixed bug: Invalid slot specification errors now print correctly.
1268 (thanks to Zach Beane)
1269 * fixed bug 320: Shared to local slot value transfers in class
1270 redefinitions now happen corrently from superclasses as
1271 well. (reported by Bruno Haible)
1272 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1274 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1275 improper lists correctly. (reported by Bruno Haible)
1276 * fixed bug 313: source-transform for <fun-name> was erroneously
1277 applied to a call of a value of a variable with name <fun-name>.
1278 (reported by Antonio Menezes Leitao)
1279 * fixed bug 307: The obsolete instance protocol ensures that
1280 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1281 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1282 remove dead unknown-values globs from the middle of the stack.
1283 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1284 easier to resume long computations after using *BREAK-ON-SIGNALS*
1285 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1286 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1287 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1288 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1289 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1290 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1292 * fixed a (fairly theoretical) bug in string printing: if
1293 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1294 does not have array-element-type equal to the most general string
1296 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1297 function name. (thanks to Nikodemus Siivola)
1298 * fixed bug: compile-time format string checker failed on
1299 non-closed ~{. (reported by Thomas F Burdick)
1300 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1301 don't warn on legal loop code involving more than one aggregate
1302 boolean. (thanks to Nikodemus Siivola)
1303 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1304 in keyword position to MAKE-PACKAGE was not being reported as
1306 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1307 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1309 * fixed bugs 280 and 312: the checking for multiple definitions in a
1310 file is less likely to become confused by uses of inline
1312 * fixed bug: the #S reader macro performs the keyword coercion
1313 specified for slot names. (reported by Kalle Niemitalo)
1314 * fixed bug: lambda lists may contain symbols whose names start with
1315 & but are not lambda-list-keywords; their occurrence triggers a
1317 * fixed bug 321: define-method-combination argument lambda lists do
1318 not cause invalid code to be generated when &optional or &aux
1319 variables have default values. (reported by Bruno Haible)
1320 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1321 have CLOS classes; this appears also to have fixed reported
1322 instability in stack exhaustion detection.
1323 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1324 MUFFLE-WARNING when no associated restart is present is now
1326 * optimization: rearranged the expansion of various defining macros
1327 so that each expands into only one top-level form in a
1328 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1330 * optimization: used a previously little-used slot in symbols to
1331 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1333 * fixed some bugs revealed by Paul Dietz' test suite:
1334 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1336 ** Fixed an optimization bug related to inheritance of initforms
1337 from local to shared slots.
1338 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1339 ** CAST optimizer forgot to flush argument derived type.
1340 ** print/read consistency on floats is now orders of magnitude
1341 more likely. (thanks also to Bruno Haible for a similar report
1343 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1344 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1345 for objects of type REAL. Make it so.
1346 ** SXHASH is specified (infelicitously) to respect similarity,
1347 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1348 it so. (thanks to Markus Ziegler)
1349 ** on the Alpha, the compiler succeeds in compiling functions
1350 returning a known number of arguments greater than 63.
1351 ** fixed handling of invalid NIL arguments in keyword position
1353 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1355 ** provided a readably-printable representation for RANDOM-STATE
1357 ** ensured that pathnames, where they have a namestring, always
1358 print using #P"..." syntax.
1360 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1361 * Support for the forthcoming 2.0 version of the NetBSD kernel
1362 running on x86 hardware has been added. (thanks to Perry
1363 E. Metzger most immediately, and others for their past work)
1364 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1365 current release version; SBCL's OpenBSD support had been broken
1366 since about the time of OpenBSD's switch to ELF binary format.)
1367 * [placeholder for DX summary]
1368 ** user code with &REST lists declared dynamic-extent, under high
1369 speed or space and low safety and debug optimization policy.
1370 * The manual has been converted to Texinfo format and the debugger
1371 chapter from the cmucl manual has been added.
1372 * A facility has been added to extract documentation strings from
1373 sbcl and store them as Texinfo-formatted snippets for inclusion in
1374 the manual (via Texinfo's @include directive)
1375 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1376 the test case to Dave Roberts)
1377 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1378 pointers. (reported by Sean Ross)
1379 * bug fix: PROFILE output is printed nicely even for large numerical
1380 values. (thanks to Zach Beane)
1381 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1382 greater than 32 handle EOF correctly.
1383 * bug fix: on X86 an immediate argument of the IMUL instruction is
1384 correctly printed in disassembly. (thanks to Lutz Euler)
1385 * bug fix: class slots in redefined classes preserve their old
1386 values. (thanks to Bruno Haible and Nikodemus Siivola)
1387 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1388 succeeds. (reported by Marco Baringer)
1389 * fixed some bugs revealed by Paul Dietz' test suite:
1390 ** READ-SEQUENCE now works on ECHO-STREAMs.
1391 ** RATIONALIZE works more according to its specification. (thanks
1394 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1395 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1396 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1397 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1398 should matter to you only if you rebind the printer control
1399 variables and then find you want different bindings in the
1400 debugger than in the ordinary execution of your program.)
1401 * The runtime build system has been tweaked to support building
1402 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1403 assembler and linker. (thanks to Nikodemus Siivola)
1404 * Unbound, undefined, undeclared variables now trigger full
1405 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1406 more often programmer error than deliberate exploitation of undefined
1408 * optimization: the hash algorithm for strings has changed to one
1409 that is less vulnerable to spurious collisions. (thanks to Juho
1411 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1412 needless bounds checking. (thanks to Juho Snellman)
1413 * optimization: implemented multiplication as a modular
1414 (UNSIGNED-BYTE 32) operation on the PPC backend.
1415 * fixed some bugs revealed by Paul Dietz' test suite:
1416 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1418 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1419 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1420 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1422 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1424 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1426 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1427 their output stream on EOF from read.
1428 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1429 have been read to end-of-file.
1430 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1432 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1433 description of determination of which consecutive characters
1435 ** Printing the "Space" character with escaping on now yields "#\\ ",
1436 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1437 ** Reading floating-point numbers with *READ-BASE* set to a number
1438 less than 10 works correctly.
1439 ** Reading floating-point numbers with *READ-BASE* set to a number
1440 more than 10 works correctly.
1441 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1442 the readtable currently in effect.
1444 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1445 * minor incompatible change: parsing of namestrings on a physical
1446 (Unix) host has changed; numbers after the final #\. in a
1447 namestring are no longer interpreted as a version field. This is
1448 intented to be largely invisible to the user, except that the
1449 meaning of the namestring "*.*.*" has changed: it now refers to a
1450 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1451 should usually be replaced by
1452 (make-pathname :name :wild :type :wild :version :wild)
1453 with the added benefit that this is more likely to be portable.
1454 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1455 option to OPEN now signals an error if the file being opened
1456 exists; this may have an impact on existing code.
1457 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1458 signals received, on the PowerPC platforms (both Linux and
1459 Darwin). (thanks to Pierre Mai for pointing out the location of
1461 * several fixes on OS X: The system now builds and runs cleanly on
1462 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1463 (thanks to Brian Mastenbrook)
1464 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1465 second-resolution: integer multiples of 1/3600 between -24 and 24.
1466 (thanks to Vincent Arkesteijn)
1467 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1468 type when called with 1 argument; PEEK-CHAR checked type of
1469 PEEK-TYPE only after having read first character from a
1470 stream. (reported by Peter Graves)
1471 * bug fix: the garbage collector now has much better locality
1472 behaviour, and in particular no longer treats all memory as being
1473 exhausted when many small objects point to each other in a deeply
1475 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1476 recognized as being TYPEP their class.
1477 * bug fix: the PUSHNEW documentation string has been corrected.
1478 (thanks to Vincent Arkesteijn)
1479 * bug fix: defaulting of the value for the last of an atypically
1480 large number of multiple values being bound was not being
1481 performed correctly on the Alpha or PPC platforms
1482 * optimization: implemented multiplication as a modular
1483 (UNSIGNED-BYTE 32) operation on the x86 backend.
1484 * optimization: SEARCH on simple-base-strings can now be open-coded.
1485 (see also contrib/compiler-extras.lisp for inspiration for
1486 teaching the compiler about the Boyer-Moore algorithm).
1487 * value, returned by MAX (and MIN) called with several EQUALP, but
1488 not EQL, arguments now does not depend on compiler settings.
1489 * fixed some bugs revealed by Paul Dietz' test suite:
1490 ** in stack analysis liveness information is propagated from
1491 non-local entry points.
1492 ** pathwise CAST removing failed when the CAST node did not start
1494 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1495 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1497 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1498 host is already defined.
1499 ** RENAME-FILE works on streams instead of signalling an internal
1501 ** PEEK-CHAR uses the current readtable when determining whether
1502 or not a character is whitespace.
1503 ** MERGE-PATHNAMES handles the case when the pathname does not
1504 specify a name while the default-pathname specifies a version.
1505 ** Pathnames now stand a better chance of respecting print/read
1507 ** Attempting to use standardized file system operators with a
1508 pathname with invalid :DIRECTORY components signals a
1510 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1512 ** DIRECTORY on logical pathnames is more correct.
1513 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1514 signal a TYPE-ERROR if their argument is not a stream.
1515 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1516 designator argument does not designate a stream.
1517 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1518 examining the synonym.
1519 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1521 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1522 element-type larger than ([UN]SIGNED-BYTE 32).
1524 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1525 * When built with the :SB-FUTEX feature, threaded builds now take
1526 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1527 for faster/more reliable mutex and condition variable support.
1528 * Incompatible change (but one you probably shouldn't have been using
1529 anyway): the interface and code for arbitrating between multiple
1530 threads in the same user session has been redesigned.
1531 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1532 error when called without an explicit environment argument.
1533 (thanks to Nikodemus Siivola)
1534 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1535 confused about their position. (thanks to Adam Warner and Gerd
1537 * bug fix: Pretty printing backquoted forms with unquotations in the
1538 argument list position of various code constructs such as LAMBDA
1539 now works correctly. (reported by Paul Dietz)
1540 * bug fix: Pretty printing unquotations no longer loses all
1541 stream position information.
1542 * optimization: performance of string output streams is now less
1543 poor for multiple small sequence writes.
1544 * optimization: performance of CSUBTYPEP in the presence of complex
1545 expressions involving CONS and NOT many times has been improved.
1546 (reported by Paul Dietz)
1547 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1549 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1550 streams for the REPL, for the debugger, and so on. (thanks to
1552 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1553 * fixed some bugs revealed by Paul Dietz' test suite:
1554 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1555 is ignored if it is too small, rather than propagating through
1556 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1557 ** extremely complex negations of CONS types were not being
1558 sufficiently canonicalized, leading to inconsistencies in
1560 ** VALUES tranformer lost derived type.
1562 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1563 * fixed a bootstrapping bug: the build process no longer assumes
1564 that the various BOOLE-related constants have the same value in
1565 host and target lisps. (noted by Paul Dietz' test suite on an
1566 SBCL binary built from CLISP)
1567 * The system can now be dynamically linked on the MIPS platform,
1568 which enables dynamic loading of foreign code from Lisp. (thanks
1569 to Ralf Baechle for discussions on the MIPS ABI)
1570 * The system now records debugging information for its own source
1571 files in a filesystem-position-independent manner, relative to
1572 the "SYS" logical host.
1573 * fixed a compiler bug: MV-LET convertion did not check references
1574 to the "max args" entry point. (reported by Brian Downing)
1575 * tweaked disassembly notes to be less confident about proclaiming
1576 some instruction as an LRA. (thanks to Brian Downing)
1577 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1578 now each have their own history, command character, and other
1579 characteristics. (thanks to David Lichteblau)
1580 * fixed some bugs revealed by Paul Dietz' test suite:
1581 ** compiler failure in compiling LOGAND expressions including a
1583 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1584 shift greater than 32.
1585 ** FLUSH-DEST did not mark blocks for type check regeneration.
1586 ** HANDLER-CASE failed to accept declarations in handler clauses
1587 in some circumstances.
1589 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1590 * New code in contrib/sb-introspect (still probably not entirely
1591 stable yet) provides some support for smart Lisp development
1592 environments like SLIME.
1593 * The conditions signalled for errors occurring when loading .fasl
1594 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1595 in a way which should help ASDF recover gracefully.
1596 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1597 stuff has been cleaned up. If you code contrib/ stuff, this might
1598 affect you, and you can look at contrib/README, contrib/STANDARDS,
1599 and/or the 0.8.4.27 diff to check.
1600 * In full calls the compiler now does not generate checks for declared
1601 argument types for all arguments.
1602 * various threading fixes
1603 ** and some experimental patches which didn't make it into
1604 the main tree for this release, but which are shipped in
1605 contrib/experimental-thread.patch as a possible fix for some
1606 failures (deadlock, spinning...) in GC-intensive multithreaded
1608 * fixed PPC build problem (source code incompatibility of different
1609 library versions): added offsetof() hackery which attempts to divine
1610 where glibc maintainers put uc_mcontext today
1611 * fixed bug 282: compiler does not trust type assertions while passing
1612 arguments to a full call.
1613 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1614 &OPTIONAL VALUES type parameter.
1615 * fix bug 214: algorithm for noting rejected templates is now more
1616 similar to that of template seletion. (also reported by rydis on
1618 * fixed bug 141b: printing backquoted information readably and prettily
1619 inserts a space where necessary.
1620 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1621 (CHARACTER 10) are now reported as errors, rather than propagated
1622 as unknown types. (reported by piso on #lisp)
1623 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1624 with values NIL and :ERROR. (thanks to Milan Zamazal)
1625 * fixed bug 191c: CLOS now does proper keyword argument checking as
1626 described in CLHS 7.6.5 and 7.6.5.1.
1627 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1628 longer raise an error; further, using a list as a for-as-arithmetic
1629 counter now raises a meaningful error.
1630 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1631 now understood by sequence creation functions such as MAKE-SEQUENCE
1633 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1634 a STRING-INPUT-STREAM.
1635 * compiler enhancement: SIGNUM is now better able to derive the type
1637 * type declarations inside WITH-SLOTS are checked. (reported by
1639 * fixed some bugs revealed by Paul Dietz' test suite:
1640 ** incorrect optimization of TRUNCATE for a positive first
1641 argument and negative second.
1642 ** compiler failure in let-convertion during flushing dead code.
1643 ** compiler failure while deriving type of TRUNCATE on an
1644 interval, containing 0.
1645 ** ASH of a negative bignum by a negative bignum count now returns
1647 ** intersection of CONS types now canonicalizes properly, fixing
1648 inconsistencies in SUBTYPEP.
1650 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1651 * incompatible change: The --disable-debugger command line
1652 option now clobbers the debugger at a more fundamental
1653 level, by redefining #'INVOKE-DEBUGGER instead of by
1654 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1655 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1656 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1657 enter the debugger REPL and then suffer recursive errors
1658 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1659 while under the new system, BREAK in --disable-debugger mode
1660 terminates the system just as an unhandled error would.
1661 * fixed compiler performance when processing loops with a step >1;
1662 * bug fix: DOCUMENTATION now retrieves generic function
1663 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1664 support has been systematized, and now supports the methods
1665 specified by ANSI, along with a default method and a method for
1666 slot documentation. (reported by Nathan Froyd)
1667 * bug fix: effective methods associated with a generic function are
1668 no longer cached over a change of that generic function's method
1669 combination. (reported by Andreas Fuchs)
1670 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1671 names a non-existent directory. (thanks to Andreas Fuchs)
1672 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1673 the PPC platform, lead to stack corruption; this has been fixed.
1674 (reported by Rainer Joswig)
1675 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1676 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1677 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1678 arithmetic in general, is now much more reliable on the PPC
1680 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1681 platform now returns the right answer.
1682 * optimization: restored some effective method precomputation in
1683 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1684 precomputation is now tunable.
1685 * optimization: compiler-internal data structure use has been
1686 reviewed, and changes have been made that should improve the
1687 performance of the compiler by about 20%.
1688 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1689 simple-base-strings and simple-bit-vectors is improved.
1690 * optimization: the optimization of 32-bit logical and arithmetic
1691 functions introduced in version 0.8.3 on the x86 has been
1692 implemented on the mips, ppc and sparc platforms; an
1693 implementation of the same facility, but for 64-bit arithmetic,
1694 has been added for the alpha.
1695 * microoptimization: the compiler is better able to make use of the
1696 x86 LEA instruction for multiplication by constants.
1697 * bug fix: in some situations compiler did not report usage of
1698 generic arithmetic in (SPEED 3) policy.
1699 * bug 145b fix: compiler used wrong type specifier while converting
1700 MEMBER-types to numeric.
1701 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1703 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1704 index argument now works properly on the Alpha platform.
1705 * bug fix: floating point exception treatment on the Alpha platform
1707 * bug fix: FILE-POSITION works much better on string input and
1708 output streams. (thanks to Nikodemus Siivola)
1709 * bug fix: many threading/garbage collection symptoms sorted.
1710 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1711 might be pseudo-atomic.
1712 * internal change: Stopping for GC is now done with signals not
1713 ptrace. GC is now done in whichever thread wanted it, instead of
1715 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1717 * bug fix: result form in DO is not contained in the implicit
1719 * incompatible change: ICR structure is changed; the value part of
1720 CONTINUATION is now called LVAR; corresponding functions are
1721 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1722 * added type deriver for ISQRT (thanks to Robert E. Brown).
1723 * bug fix: better support for loading from the command line when an
1724 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1726 * fixed some bugs revealed by Paul Dietz' test suite:
1727 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1728 ** ROUND and FROUND now give the right answer when given very
1729 small float arguments.
1730 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1732 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1733 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1734 ** LCM should return a non-negative integer.
1735 ** PARSE-INTEGER returned the index of a terminator instead of the
1736 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1737 ** PARSE-INTEGER returned an incorrect index being applied to a
1739 ** LCM with two arguments of 0 returns 0 rather than signalling
1741 ** unsigned addition of a 32-bit constant with the high bit set no
1742 longer causes an internal compiler error on the x86.
1743 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1744 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1745 with negative last argument.
1746 ** byte specifiers with zero size and position no longer cause
1747 an error during type derivation.
1748 ** bignum multiplication on the Alpha platform now returns the
1750 * porting: The system now builds on SuSE AMD64, although it still
1751 generates a 32-bit binary.
1752 * .fasl file incompatibility: The fasl file version number has
1753 been incremented (because of the changes to internal compiler
1754 data structures referred to above).
1756 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1757 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1758 more accurately, on the Darwin kernel running on PowerPC hardware.
1759 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1760 * Compiler code deletion notes now signal a condition of type
1761 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1762 an associated MUFFLE-WARNING restart.
1763 * The compiler now performs limited argument count validation of
1764 constant format strings in FORMAT, and where appropriate in ERROR,
1765 CERROR and WARN. (thanks to Gerd Moellmann)
1766 * New ASDF-INSTALL contrib can be used for automatic download and
1767 installation of third-party Lisp code from CCLAN or other sites
1769 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1770 which forces another thread to execute a function supplied by the
1772 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1773 now accept and act upon their :ELEMENT-TYPE keyword argument.
1774 (reported by Martin Atzmueller, Edi Weitz)
1775 * bug fix: FILE-POSITION now accepts position designators up to
1776 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1777 is the greater. (thanks to Patrik Nordebo)
1778 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1779 Kalle Olavi Niemitalo)
1780 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1781 printable. (reported by Eric Marsden)
1782 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1783 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1784 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1785 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1786 (thanks to Stig Sandoe)
1787 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1788 UNREAD-CHAR now correctly includes the unread character in the
1789 target sequence. (thanks to Gerd Moellmann)
1790 * bug fix in threaded builds: the system can now be suspended and
1791 resumed by shell job control with minimal disruption.
1792 * bug fixes in times and timezones >2038AD
1793 * better handling of "where is GNU make?" problem in build scripts
1794 (thanks to Nikodemus Siivola)
1795 * new optimization: inside a named function any reference to a
1796 function with the same name is considered to be a self-reference;
1797 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1798 optimization quality.
1799 * new optimization on x86: logical functions and + now have
1800 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1801 used when the result is truncated to 32 bits.
1802 * VALUES declaration is partially enabled.
1803 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1804 * fixed some bugs revealed by Paul Dietz' test suite:
1805 ** The system now obeys the constraint imposed by
1806 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1807 types form a lattice under type intersection.
1808 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1809 ** ASSOC now ignores NIL elements in an alist.
1810 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1811 and (1+ MOST-POSITIVE-FIXNUM) answers.
1812 ** The addition of a method with invalid qualifiers to a generic
1813 function does not cause an error to be signalled immediately;
1814 a warning is signalled, and the error is generated only on
1815 calling the generic function.
1816 * changed .fasl file version number, in part to add type codes for
1817 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1818 obscure ANSI requirements
1820 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1821 * fixed bug 148: failure to inline-expand a local function left
1822 garbage, confusing the compiler.
1823 * fixed bugs 3cd: structure slot readers perform type check if the
1824 slot can have an invalid value (i.e. it is either not initialized
1825 or can be written with a less specific slot writer).
1826 * bug fix: the compiler now traps array references to elements off
1827 the end of an array; previously, the bounds checking in some
1828 circumstances could go off-by-one.
1829 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1831 * type declarations for array element types now obey the description
1832 on the CLHS page "Declaration TYPE", as per discussions on
1833 sbcl-help around 2003-05-08. This means that a declaration
1834 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1835 declaration, all references to BAR will be asserted or assumed
1836 (with THE, so dependent on compiler policy) to involve objects of
1837 type FOO. Note that no such declaration is implied in
1838 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1839 * declared types of functions from the "Conditions"
1840 chapter. (reported by Paul Dietz)
1841 * bug fix: CERROR accepts a function as its first argument.
1842 * bug fix: NTH an NTHCDR accept a bignum as index
1843 arguments. (reported by Adam Warner)
1844 * optimization: character compare routines now optimize comparing
1845 against a constant character. (reported by Gilbert Baumann)
1846 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1847 argument now works properly on the MIPS platform.
1848 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1849 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1850 declared to have a functional type, does not cause an error
1852 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1853 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1854 * fixed bug: if last continuation of a deleted block has a
1855 destination, this destination should be deleted too. (reported by
1857 * fixed a bug in the bootstrap process: the host compiler's values
1858 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1859 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1860 test case from Patrik Nordebo)
1861 * improved the ability of the disassembler on the PPC platform to
1862 provide helpful disassembly notes.
1863 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1864 the class in more cases than previously.
1865 * bug fix: FILE-POSITION now understands :START and :END for
1866 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1867 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1868 8) data. (thanks to David Lichteblau)
1869 * bug fix: it is possible to add a method to a generic function
1870 without lambda list.
1871 * bug fix: reader failed to signal END-OF-FILE inside an
1872 object representation. (reported by Nikodemus Siivola)
1873 * fixed some bugs revealed by Paul Dietz' test suite:
1874 ** LAST and [N]BUTLAST should accept a bignum.
1875 ** condition slot accessors are methods.
1876 ** (VECTOR NIL) is a subtype of STRING.
1878 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1879 * minor incompatible change: some nonsensical specialized lambda
1880 lists (used in DEFMETHOD) which were previously ignored now signal
1882 * minor incompatible change: the system is now aware of the types of
1883 variables in the COMMON-LISP package, and will signal errors for
1884 most violations of these type constraints (where previously they
1885 were silently accepted).
1886 * minor incompatible change: COMPILE-FILE now uses the freedom
1887 afforded (ANSI 3.2.2.3) to use derived function types for
1888 functions defined in the same file. This also permits the system
1889 to warn on static type mismatches and function
1890 redefinition. (Currently it does not work with high DEBUG level.)
1891 * minor incompatible change: VALUES declaration is disabled.
1892 * When issuing notes, the compiler now signals a condition of type
1893 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1894 restart for use in user handlers. It is expected that the
1895 COMPILER-NOTE condition will eventually become a condition
1896 supertype to a hierarchy of note types, which will then be
1897 handleable in a similar fashion. However, at the moment, no such
1898 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1899 but it's an implementation detail, not a classification for the
1901 * Changes in type checking closed the following bugs:
1902 ** type checking of unused values (192b, 194d, 203);
1903 ** template selection based on unsafe type assertions (192c, 236);
1904 ** type checking in branches (194bc).
1905 * A short form of VALUES type specifier has ANSI meaning (it has
1906 increased the number of situations when SBCL cannot perform type
1908 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1909 keywords or constants is permissible.
1910 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1911 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1912 classes. (thanks to Antonio Martinez)
1913 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1914 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1915 argument to a pretty-print stream. Instead, it leaves any such
1916 operations to DESCRIBE-OBJECT methods.
1917 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1919 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1920 lambda lists. (thanks to David Lichteblau)
1921 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1922 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1923 (thanks to Gerd Moellmann)
1924 * fixed bug 246: increased compilation speed of long
1925 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1927 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1929 * DEFCONSTANT now throws a condition of type
1930 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1931 constant to a non-EQL value; CONTINUE and ABORT restarts
1932 respectively change and preserve the value.
1933 * fixed bug 63: The code walker, part of the implementation of CLOS,
1934 is now better at handling symbol macros.
1935 * bug fix: There is no longer an internal implementation type named
1936 CL:LENGTH. (reported by Raymond Toy)
1937 * bug fix: In macro-like defining macros/special operators the
1938 implicit block does not enclose the lambda list.
1939 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1940 enclosed in parentheses) are not suitable as type specifiers, and
1941 their use properly signals an error now.
1942 * bug fix: An argument count mismatch for a type specifier in code
1943 being compiled no longer causes an unhandled error at compile
1944 time, but signals a compile-time warning.
1945 * fixed simple vector readable printing
1946 * bug fix: DESCRIBE takes more care over whether the class
1947 precedence list slot of a class is bound before accessing it.
1948 (reported by Markus Krummenacker)
1949 * bug fix: FORMATTER can successfully compile pretty-printer format
1950 strings which use variants of the ~* directive inside.
1951 * bug fix: SEARCH now applies its TEST predicate to the elements of
1952 the arguments in the correct order. (thanks to Wolfhard Buss)
1953 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1954 in type inference. (thanks to Robert E. Brown)
1955 * bug fix: Reading in symbols with an explicit package name of ""
1956 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1957 * fixed some bugs revealed by Paul Dietz' test suite:
1958 ** NIL is now allowed as a structure slot name.
1959 ** Arbitrary numbers, not just REALs, are allowed in certain
1960 circumstances in LOOP for-as-arithmetic clauses.
1961 ** Multiple class redefinitions before slot access no longer
1962 causes a type error.
1963 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1964 association between the name and a class.
1965 ** Generic functions with non-standard method-combination and over
1966 five methods all of which return constants no longer return NIL
1967 after the first few invocations. (thanks to Gerd Moellmann)
1968 ** CALL-NEXT-METHOD with no arguments now passes the original
1969 values of the arguments, even in the presence of assignment.
1970 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1972 ** DISASSEMBLE works with closures and funcallable instances.
1973 ** ADD-METHOD now returns the generic function, not the new method.
1974 ** FIND-METHOD signals an error if the lengths of the specializers
1975 is incompatible with the generic function, even if the ERRORP
1977 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1978 which its argument is a member.
1979 ** DEFCLASS only redefines the class named by its class-name
1980 argument if that name is the proper name of the class;
1981 otherwise, it creates a new class.
1982 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1983 of the UNBOUND-SLOT condition to the name of the slot.
1984 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1985 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1986 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1987 SLOT-MAKUNBOUND in the specified fashion.
1989 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1990 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1991 cross-compilation host. As a consequence, we can now bootstrap our
1992 way up to SBCL starting with a bare gcc toolchain and human-readable
1993 source code (first the source to CLISP, then the source to SBCL).
1994 * A contributed module containing a partial implementation of the
1995 simple-streams interface has been included. (thanks to Rudi
1997 * A contributed module implementing the RFC1321 Message Digest
1998 Algorithm, known as MD5, has been included.
1999 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2000 no longer has any effect, as the code controlled by this feature
2001 has been deleted. (As far as we know, no-one has ever built using
2002 this feature, and its semantics were confused in any case).
2003 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2004 work on conditions (as required by the ANSI specification),
2005 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2006 expected behaviour on conditions. Users should note, however,
2007 that such behaviour is not required by the ANSI specification,
2008 and so use of this behaviour may render their code unportable.
2009 * fixed some bugs revealed by Paul Dietz' test suite:
2010 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2012 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2013 ** SXHASH on condition objects no longer returns NIL.
2014 ** :ALLOCATION :CLASS slots are better treated; their values are
2015 updated on class redefinition, and initforms inherited from
2016 superclasses are applied.
2017 ** REMOVE-METHOD returns its generic function argument even when
2018 no method was removed.
2019 ** SHARED-INITIALIZE now initializes the values of the requested
2020 slots, including those with :ALLOCATION :CLASS.
2021 ** ALLOCATE-INSTANCE now works on structure classes defined via
2022 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2024 ** SLOT-EXISTS-P now works on conditions, as well as structures
2026 ** MAKE-LOAD-FORM now has the required methods on
2027 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2028 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2029 keyword, and now implements the SLOT-NAMES argument.
2030 ** methods with &OPTIONAL arguments no longer allow too many
2031 arguments to be passed in the call without error.
2032 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2033 option is consistent with the required arguments of the generic
2034 function lambda list.
2035 * bug fix: REQUIRE accepts a string designator. (Thanks to
2037 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2038 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2039 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2040 * bug fix: The compiler now checks for duplicated variables in macro
2042 * bug fix: SETQ on globals returns the correct value.
2043 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2044 not a COMPILER-ERROR (followed by some other strange error on
2045 choosing the CONTINUE restart).
2046 * bug fix: make.sh and friends are now more consistent in the way that
2047 they look for GNU "make".
2049 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2050 * experimental native threads support (on x86 Linux >=2.4 only).
2051 This is not compiled in by default: you need to add :SB-THREAD to
2052 the target features. See the "Beyond ANSI" chapter of the manual
2054 * fix for longstanding nonANSIism: The old distinction between
2055 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2056 The return value from CL:FIND-CLASS is now a CLOS class, and
2057 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2058 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2059 * An interface to the MetaObject Protocol, as described in Kiczales,
2060 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2061 MIT Press, 1991, is available from the SB-MOP package.
2062 * incompatible change: the SB-PCL package should now be considered
2063 a private implementation detail, and no longer a semi-private MOP
2065 * minor incompatible change: due to rearrangement for threads, the
2066 control stack and binding stack are now allocated at arbitrary
2067 addresses instead of being hardcoded per-port. Users affected by
2068 this probably have to be doing advanced things with shared
2069 libraries, and will know who they are.
2070 * minor incompatible change: Previously, all --eval forms used were
2071 processed with READ before any of them were processed with EVAL.
2072 Now each --eval form is processed with both READ and EVAL before
2073 the next --eval form is processed. (Thus package operations like
2074 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2075 work as the user might reasonably expect.)
2076 * minor incompatible change: *STANDARD-INPUT* is now only an
2077 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2079 * minor incompatible change: Y-OR-N-P is now character-oriented,
2080 not line oriented. Also, YES-OR-NO-P now works without errors.
2081 (thanks to Antonio Martinez)
2082 * sb-aclrepl module improvements: an integrated inspector, added
2083 repl features, and a bug fix to :trace command.
2084 * Known functions, which cannot be open coded by the backend, are
2085 considered to be able to check types of their arguments. (fixing
2086 a bug report by Nathan J. Froyd)
2087 * fixed a bug in computing method discriminating functions: It is
2088 now possible to define methods specialized on classes which have
2089 forward-referenced superclasses. (thanks to Gerd Moellmann)
2090 * fixed evaluation order in optional entries (reported by Gilbert
2092 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2093 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2094 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2095 required-by-AMOP NAME argument, as well as CLASS and
2096 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2097 * fixed bug 20: DEFMETHOD can define methods using names that are
2098 not the proper names of classes to designate class specializers.
2099 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2100 Unix file descriptors, instead of blowing up. (thanks to Antonio
2102 * Garbage collection refactoring: user-visible change is that a
2103 call to the GC function during WITHOUT-GCING will not do garbage
2104 collection until the end of the WITHOUT-GCING. If you were doing
2105 this you were probably losing anyway.
2106 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2107 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2108 TYPEP the latter but not the former.
2109 * The compiler issues a full WARNING for calls to undefined functions
2110 with names from the CL package.
2111 * MAP-INTO for a vector destination is open coded. (reported by
2112 Brian Downing on c.l.l)
2113 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2114 documentation string.
2115 * fixed some bugs revealed by Paul Dietz' test suite:
2116 ** COPY-ALIST now signals an error if its argument is a dotted
2118 ** Condition slots are now accessed more correctly in the presence
2119 of multiple initargs for a given slot.
2120 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2121 exclude restarts of the same name associated with a different
2123 ** DEFCLASS of forward-referenced classes with another
2124 forward-referenced class in the superclasses list no longer
2126 ** Condition slots are now initialized once each, not multiple
2127 times. (thanks to Gerd Moellmann)
2128 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2130 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2131 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2132 arguments contain duplicated elements.
2133 ** RESTART-CASE understands local macros.
2134 ** RESTART-CASE associates exactly its own restarts with a condition.
2135 ** ENDP in safe mode checks its argument to be of type LIST.
2136 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2137 in question is unbound.
2138 ** Optimized MAKE-INSTANCE functions no longer cause internal
2139 assertion failures in the presence of duplicate initargs.
2140 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2141 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2142 propagated even in the ctor optimized implementation of
2144 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2146 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2147 * a better implementation of SXHASH on (simple) bit vectors,
2148 measured both in execution speed and in distribution of results
2149 over the positive fixnums, has been installed. Likewise, a better
2150 implementation of EQUAL for simple bit vectors is now available.
2151 * fixed CEILING optimization for a divisor of form 2^k.
2152 * fixed bug 240 (emitting extra style warnings "using the lexical
2153 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2154 by Antonio Martinez)
2155 * fixed SXHASH, giving different results for NIL depending on type
2156 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2157 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2158 variable. (found by Rolf Wester)
2159 * SBCL does not ignore type declarations for special
2160 variables. (reported by rif on c.l.l 2003-03-05)
2161 * some bug fixes in contrib/sb-aclrepl/
2162 * fixed some bugs revealed by Paul Dietz' test suite:
2163 ** a bug in the CONS type specifier, whereby the CAR and CDR
2164 types got intertwined, has been fixed;
2165 ** the type system is now able to reason about the interaction
2166 between INTEGER and RATIO types more completely;
2167 ** APPEND, [N]REVERSE and NRECONC check that those their
2168 arguments, which must be proper lists, are really so;
2169 ** An array specialized to be unable to hold elements has been
2170 implemented, as required -- yes, really -- by ANSI;
2171 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2172 on malformed property lists;
2174 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2175 * incompatible packaging change: in line with Unix convention,
2176 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2177 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2178 * REQUIRE and PROVIDE are now optionally capable of doing something
2179 useful. See the documentation string for REQUIRE.
2180 * infrastructure for a managed SBCL contrib system: contributed
2181 modules in this release include:
2182 ** the ASDF system definition facility;
2183 ** an interface to the BSD Sockets API;
2184 ** an ACL-like convenience interface to the repl;
2185 (thanks to Kevin Rosenberg)
2186 ** an implementation of ROTATE-BYTE, with efficient implementation
2188 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2189 gives a closer approximation to the right answer than previously.
2190 (thanks to Raymond Toy)
2191 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2192 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2193 optional environment arguments, as required by ANSI.
2194 * fixed bugs in other functions taking environment objects, allowing
2195 calls with an explicit NIL environment argument to be compiled
2197 * fixed bug 228: primary return values from
2198 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2199 COMPILE or FUNCTION.
2200 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2201 :TYPE will no longer signal errors on innocuous objects.
2202 * fixed bug 231b: SETQ is better at respecting type declarations in
2203 the lexical environment.
2204 * fixed a bug in DEFCLASS: classes named by symbols with no or
2205 unprintable packages can now be defined.
2206 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2207 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2208 * fixed compiler failure related to checking types of functions.
2209 (reported by Robert E. Brown)
2210 * the compiler is now much more consistent in its error-checking
2211 treatment of bounding index arguments to sequence functions: in
2212 (SAFETY 3) code, errors will be signalled in almost all cases if
2213 invalid sequence bounding indices are passed to functions defined
2214 by ANSI to operate on sequences.
2215 * fixed a bug in the build procedure: documentation of SBCL-specific
2216 packages is now preserved and available in the final Lisp image.
2217 * lifted FDEFINITION lookup out of loops in the implementation of
2218 many list operations. (thanks to Robert E. Brown)
2219 * fixed a bug in the reader: the #n# reader macro now works for
2220 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2221 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2222 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2223 E. Brown and Tony Martinez respectively)
2224 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2225 SLOT-BOUNDP: in particular, it is now able to optimize them much
2226 better, and is now not vulnerable to having packages renamed.
2227 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2228 always called when a slot is not present in an instance. (thanks
2230 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2231 longer signal an error on unreading a character following EOF on
2232 the previous constituent stream. (thanks to Tony Martinez)
2233 * fixed some bugs revealed by Paul Dietz' test suite:
2234 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2235 not just nonnegative fixnums;
2236 ** the logical bit-array operators such as BIT-AND now accept an
2237 explicit NIL for their "opt-arg" argument (to indicate a
2238 freshly-consed result bit-array);
2239 ** ELT now signals an error on an invalid sequence index in safe
2241 ** the type system is now cleverer about negations of numeric
2242 types, and consequently understands the BIGNUM and RATIO types
2244 ** the type system is now cleverer about the interaction between
2245 INTEGER and RATIO types: while bugs still remain, many more
2246 cases are accurately computed;
2247 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2248 if it is in the last clause;
2249 ** CONSTANTLY now correctly returns a side-effect-free function in
2251 ** DECLARE is no longer treated as a special-operator; in
2252 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2253 * incremented fasl file version number due to the change in the
2254 DEFSTRUCT-SLOT-DESCRIPTION structure.
2256 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2257 * minor incompatible change: code processed by the "interpreter" or
2258 EVAL now has a compilation optimization policy of (DEBUG 2)
2259 (changed from (DEBUG 1)) to improve debuggability of interactive
2260 development, and to allow the use of the debug RETURN command in
2262 * an experimental implementation of the RETURN command for the
2263 debugger has been included. (thanks to Frederik Kuivinen)
2264 * fixed bug 62: constraints were not propagated into a loop.
2265 * fixed bug in embedded calls of SORT (reported and investigated by
2267 * fixed some bugs revealed by Paul Dietz' test suite:
2268 ** printing and reading of arrays with some dimensions having
2269 length 0 (thanks to Gerd Moellmann);
2270 ** BOA constructor with &AUX argument without a default value does
2271 not cause a type error;
2272 ** CONSTANTP now returns true for all self-evaluating objects.
2274 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2275 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2276 accessors that are related by inheritance, as specified in the
2277 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2278 Valtteri Vuorikoski)
2279 * The compiler is now able to inline functions that were defined in
2280 a complex lexical environment (e.g. inside a MACROLET).
2281 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2283 * Support for the upcoming FreeBSD-5.0 release has been included.
2284 (thanks to Dag-Erling Smorgrav)
2285 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2286 effect when it is not in a toplevel context.
2287 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2288 stand a better chance of being correct. (thanks to Gerd
2290 * fixed bug in COERCE, which now signals an error on coercing a
2291 rational to a bounded real type which excludes the expected
2293 * The compiler is now able to derive types more accurately from the
2294 COERCE and COMPILE functions.
2295 * fixed bug 223: functional binding is considered to be constant
2296 only for symbols in the CL package.
2297 * fixed bug 231: SETQ did not check the type of a variable being set
2298 (reported by Robert E. Brown)
2299 * A new optimization for MAKE-INSTANCE has been included, fixing
2300 various bugs (including relating to :ALLOCATION :CLASS slots and
2301 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2303 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2304 ** As required by ANSI, LOOP now disallows anonymous collection
2305 clauses such as COLLECT I in conjunction with aggregate boolean
2306 clauses such as THEREIS (= I 1);
2307 ** LOOP now signals an error when any variable is reused in the
2308 same loop (including the potentially useful construct analogous
2309 to WITH A = 1 WITH A = (1+ A);
2310 ** IT is only a special loop symbol within the first clause of a
2311 conditional loop clause;
2312 ** LOOP with a typed iteration variable over a hashtable now
2313 signals a type error iff it should.
2314 * fixed some other bugs revealed by Paul Dietz' test suite:
2315 ** FILE-STREAM now names the class previously known as FD-STREAM;
2316 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2317 argument) no longer signals an error;
2318 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2319 of the slot symbol, rather than using the current package
2320 ((:CONC-NAME "") continues to intern the slot's name in the
2322 * incremented fasl file version number, because of the incompatible
2323 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2324 of the new implementation of DEFINE-COMPILER-MACRO.
2326 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2327 * Support for building SBCL for MIPS platforms running in
2328 little-endian mode has now been checked in, and basic
2329 functionality on said platforms verified.
2330 * minor incompatible change: PCL now records the pathname of a file
2331 in which methods and the like are defined, rather than its
2333 * minor incompatible change: TRUENAME now considers the truename of
2334 a file naming a directory to be the pathname with :DIRECTORY
2335 component indicating that directory.
2336 * minor incompatible change: a NAMED clause in the extended form of
2337 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2338 reason for the previous behaviour is unclear.
2339 * more systematization and improvement of CLOS and MOP conformance
2340 in PCL (thanks to Gerd Moellman and Pierre Mai):
2341 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2343 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2344 primary methods with no specializers;
2345 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2347 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2348 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2349 FORWARD-REFERENCED-CLASSes; error reporting on
2350 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2352 ** SXHASH on CLOS instances now uses a slot internal to the
2353 instance to return different numbers on distinct instances,
2354 while preserving the same return value through invocations of
2356 ** DEFMETHOD signals errors when methods with longer incongruent
2357 lambda lists are added to generic functions;
2358 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2359 CLASS, as specified in AMOP;
2360 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2361 on the order returned by the primary method for classes of
2362 class STANDARD-CLASS;
2363 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2364 * fixed some bugs shown by Paul Dietz' test suite:
2365 ** DOLIST puts its body in TAGBODY;
2366 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2368 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2369 value producing form;
2370 ** if more variables are given to PROGV than values, extra
2371 variables are bound and made to have no value;
2372 ** NSUBSTITUTE on list arguments gets the right answer with
2374 ** ELT signals an error of type TYPE-ERROR when the index argument
2375 is not a valid sequence index;
2376 ** LOOP signals (at macroexpansion time) an error of type
2377 PROGRAM-ERROR when duplicate variable names are found;
2378 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2379 ** FUNCALL of special-operators now cause an error of type
2381 ** PSETQ now works as required in the presence of side-effecting
2382 symbol-macro places;
2383 ** NCONC accepts any object as its last argument;
2384 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2386 ** loop-for-as-package does not require a package to be explicitely
2388 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2390 * fixed bug 166: compiler preserves "there is a way to go"
2391 invariant when deleting code.
2392 * fixed bug 172: macro lambda lists with required arguments after
2393 &REST arguments now cause an error to be signalled. (thanks to
2395 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2396 bug. (thanks to Matthew Danish)
2397 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2399 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2400 arguments are lexically rebound. (thanks to Gerd Moellmann and
2402 * fixed bug 194: error messages are now more informative when there
2403 is no primary method applicable in a call to a generic
2404 function. (thanks to Gerd Moellmann)
2405 * fixed bug in command line argument checking (thanks to Julian
2407 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2408 (thanks to Matthew Danish)
2409 * incremented fasl file version number, because of the
2410 SXHASH-related changes in the layout of CLOS data structures
2412 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2413 * minor incompatible change: The runtime (the Unix executable named
2414 "sbcl") is now much pickier about the .core files it will load.
2415 Essentially it now requires .core files to descend from the same
2416 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2417 as the runtime does. (The intent is to prevent the crashes which
2418 can occur, and which can even be reported as mysterious failures,
2419 when people patch the sources or change the build parameters
2420 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2421 sbcl and .core files.)
2422 * fixed bug: VALUES-LIST is no longer optimized away.
2423 * fixed bug 142: The FFI conversion of C string values to Lisp
2424 string values no longer conses excessively. (thanks to Nathan
2425 Froyd porting Raymond Toy's fix to CMU CL)
2426 * began to systematize and improve MOP conformance in PCL (thanks to
2427 Nathan Froyd, Gerd Moellman and Pierre Mai):
2428 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2430 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2431 ** FINALIZE-INHERITANCE is now called on class finalization;
2432 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2433 argument precedence order.
2434 * fixed bug 202: The compiler no longer fails on functions whose
2435 derived types contradict their declared type.
2436 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2437 so it can be non-toplevel.
2438 * The fasl file version number has changed (because of the new
2439 implementation of DEFMACRO).
2440 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2441 safe code, that any length requirement by their type-specifier
2442 argument is valid. The exceptions to this are described in bug
2444 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2445 functions now check the types of their inputs as required by ANSI.
2446 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2447 introduced symbol is DECLAREd to be SPECIAL.
2448 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2449 * fixed bug: :COUNT argument to sequence functions may be negative
2450 * fixed bug: body of DO-SYMBOLS may contain declarations
2451 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2452 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2453 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2454 symbol macro only once
2455 * fixed printing of call frame when argument list is unavailable
2456 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2457 * compiler no longer signals WARNING on unknown keyword
2460 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2461 * A beta-quality port to the mips architecture running Linux,
2462 based on the old CMUCL backend, has been made. It has been tested
2463 on a big-endian kernel, and works sufficiently well to be able to
2464 rebuild itself; it has not been tested in little-endian mode.
2465 * fixed an inconsistency between gencgc.c and purify.c which made
2466 dumping/loading .core files unreliable
2467 * fixed bug 120a: The compiler now deals correctly with IFs where
2468 the consequent is the same as the alternative, instead of
2469 misderiving the return type. (thanks to Alexey Dejneka)
2470 * fixed bug 113: Logical pathnames are now dumpable (the logical
2471 host is resolved at load-time, throwing an error if it is not
2473 * fixed bug 174: FORMAT's error message is slightly clearer when a
2474 non-printing character is used in a format directive.
2475 * fixed several bugs in compiler checking of type declarations, i.e.
2476 violations of the Python "declarations are assertions" principle
2477 (thanks to Alexey Dejneka)
2478 * fixed several bugs in PCL's error checking (thanks to Gerd
2480 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2482 * fixed bug in compilation of functions as first class values
2483 (thanks to Antonio Martinez-Shotton)
2484 * The compiler's handling TYPE-ERRORs which it can prove will
2485 inevitably happen at runtime has been cleaned up and corrected
2486 in several ways. (thanks to Alexey Dejneka)
2487 * improved argument type checking for various basic arithmetic
2488 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2489 lavished on them in the past that they can be compiled in many
2490 ways in different special cases
2491 * fixed bug 181: compiler checks validity of user supplied type
2493 * cleaned up code flushing in optimization: Function calls which
2494 should signal errors for safety purposes (e.g. which ANSI says
2495 should signal errors when their arguments are of incorrect type)
2496 are no longer optimized away.
2497 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2498 * incremented fasl file version number, because changes in the
2499 implementation of sequence functions like COERCE caused
2500 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2501 in old inline expansions) to become undefined. (Actually these
2502 changes were later undone, so we might very well be binary
2503 compatible with 0.7.7 after all, but leaving the version number
2504 incremented seemed like the simplest and most conservative
2507 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2508 * An alpha-quality port to the parisc architecture running Linux,
2509 based on the old CMUCL backend, has been made. This, even more so
2510 than the other backends, should be considered still a work in
2511 progress; known problems include that the Linux kernel in 64-bit
2512 mode does not propagate the correct sigcontext structure to
2513 userspace, and consequently SBCL on a parisc64 kernel will not
2515 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2516 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2517 suppressed.) Also now that the compiler is looking at declarations
2518 in the environment, it checks optimization declarations as well,
2519 and suppresses inlining when (> DEBUG SPEED).
2520 * More fixes have been made to treatment of floating point exception
2521 treatment and other Unix signals. In particular, floating point
2522 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2523 * The detection and handling of control stack exhaustion (infinite
2524 or very deeply nested recursion) has changed. Stack exhaustion
2525 detection is now done by write-protecting pages at the OS level
2526 and applies at all optimization settings; when found, a
2527 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2528 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2529 be caught using IGNORE-ERRORS.
2530 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2531 correctly when transforming recursive LABELS functions to LETs.
2532 (thanks to Alexey Dejneka)
2533 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2534 that are names of constants or global variables.
2535 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2536 alien routines with docstrings.
2537 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2538 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2540 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2541 (thanks to Eric Marsden)
2542 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2543 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2544 * Bug fix: The SPARC backend can now compile functions involving
2545 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2546 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2547 to a functions expecting a PATHNAME-DESIGNATOR.
2548 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2549 lambda lists. (thanks to Alexey Dejneka)
2550 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2551 functions so that EXPORTing the name of the class doesn't cause
2552 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2553 to Antonio Martinez for reporting this)
2554 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2555 constant-folded without failing an assertion. (thanks to Einar
2556 Floystad Dorum for reporting this)
2557 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2558 types are now dealt with more correctly by the compiler.
2559 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2560 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2561 behaviour analogous to RENAME-FILE. This puts its behaviour more
2562 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2564 * The fasl file version number has changed again. (because of the
2565 bug fix involving the names of PCL MAKE-INSTANCE functions)
2567 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2568 * bug fix: Floating point exceptions are treated much more
2569 consistently on the x86/Linux and PPC/Linux platforms.
2570 * Array initialization with :INITIAL-ELEMENT is now much faster for
2571 cases when the compiler cannot open code the array creation, but
2572 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2573 array accesses have also seen a speed increase.
2574 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2575 specified. (This was at the root of some bad interactions between
2576 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2578 * bug fix: Internal error arguments for undefined functions are now
2579 computed correctly on the PPC/Linux platform.
2580 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2581 Raymond Toy's patch for CMU CL)
2582 * Support for the Solaris 9 operating environment has been included
2583 (thanks to Daniel Merritt)
2584 * A very ugly but hopefully complete draft of the missing FFI chapter
2585 of the manual has been created by reformatting the corresponding
2586 CMU CL manual chapter into (currently very ugly and incoherent)
2587 DocBook and bringing it up to date for SBCL behavior. Thus, the
2588 manual is now essentially complete, at least by my extreme
2589 once-and-only-once standards, whereby it's acceptable to refer to
2590 the doc strings of SB-EXT functions as the primary documentation.
2591 * The fasl file version number has changed again, due to cleanup of
2592 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2593 is no longer a static symbol.)
2595 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2596 * SBCL now builds with OpenMCL (version 0.12) as the
2597 cross-compilation host; also, more progress has been made toward
2598 bootstrapping under CLISP.
2599 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2601 * bug 158 fixed: The compiler can now deal with integer loop
2602 increments different from 1; fixing this turned out also to fix
2604 * bug 169 fixed: no more bogus warnings about using lexical bindings
2605 despite the presence of perfectly good SPECIAL declarations (thanks
2606 to David Lichteblau)
2607 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2608 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2609 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2611 * bug 180 fixed: Method combination specifications no longer ignore
2612 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2613 * bug fix: Structure type predicate functions now check their argument
2614 count as they should.
2615 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2616 correctly. (thanks to Pierre Mai)
2617 * minor incompatible change: The --noprogrammer option is deprecated
2618 in favor of the new --disable-debugger option, which is very similar.
2619 (The major difference is that it takes effect at a slightly different
2620 time at startup, causing handling of errors in --sysinit and
2621 --userinit files will be affected differently.) The
2622 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2623 been added to allow this functionality to be controlled from ordinary
2624 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2625 maintainers, who might want to run non-interactive scripts to
2626 build SBCL cores which will later be used interactively.)
2627 * minor incompatible change: The LOAD function no longer, when given
2628 a wild pathname to load, loads all files matching that pathname.
2629 Instead, an error of type FILE-ERROR is signalled.
2631 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2632 * bug 147 fixed: The compiler preserves its block link/count
2633 invariants more correctly now so that it doesn't crash. (thanks
2635 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2637 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2638 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2639 no-value-for-this-character as NIL (as specified by ANSI).
2640 * HOST-NAMESTRING on physical pathnames now returns a string that is
2641 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2642 (thanks to Christophe Rhodes)
2643 * The Alpha port handles icache flushing more correctly. (thanks to
2645 * More progress has been made toward bootstrapping under CLISP. (thanks
2646 to Christophe Rhodes)
2647 * The fasl file format has changed again, because dynamic loading
2648 on OpenBSD (which has non-ELF object files) motivated some cleanups
2649 in the way that foreign symbols are transformed and passed around.
2650 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2651 is no longer treated as whitespace by the reader, but instead as
2652 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2653 |A
\7fB|, instead of A as it used to.
2655 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2656 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2657 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2658 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2659 this as of 0.7.1.45, but was left out of the previous news section
2660 (thanks to Dan Barlow)
2661 * SBCL now runs on the Solaris operating system on SPARC architectures
2662 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2663 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2664 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2665 * SPARC backend cleanups, allowing builds of cores optimized for V8
2666 and V9 SPARCS, and also emission of code targeted to a particular
2667 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2669 * SBCL is closer to bootstrapping under CLISP, thanks to various
2670 fixes by Christophe Rhodes.
2671 * The fasl file format has changed again, to allow the compiler's
2672 INFO database to support symbol macros.
2673 * The user manual (in doc/) is formatted into HTML more nicely.
2674 (thanks to coreythomas)
2675 * The system is smarter about SUBTYPEP relationships, especially
2676 those involving NOT types (including types such as ATOM which are
2677 represented internally using NOT types). Thus SUBTYPEP is less
2678 likely to return (VALUES NIL NIL) in general, and in particular
2679 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2681 * The fasl file format has changed again, because the internal
2682 representation of types now includes a new slot to support the new
2683 SUBTYPEP-of-NOT-types logic.
2684 * (not a change in the main branch of SBCL, but a related prototype
2685 which can hopefully be merged into the main branch of SBCL in the
2686 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2687 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2688 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2689 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2690 dispatch macro character combinations. (thanks to Alexey Dejneka)
2691 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2692 that superficially look like logical namestrings correctly.
2693 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2695 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2696 default supplied-p) for &optional and &key arguments. (thanks to
2699 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2700 * incompatible change: The compiler is now less aggressive about
2701 tail call optimization, doing it only when (> SPACE DEBUG) or
2702 (> SPEED DEBUG). (This is an incompatible change because there are
2703 programs which relied on the old CMU-CL-style behavior to optimize
2704 away their unbounded recursion which will now die of stack overflow.)
2705 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2706 for non-GENCGC systems has been increased to 20M (since that
2707 seems much closer to the likely performance optimum for modern
2708 systems than the old 4M value was)
2709 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2710 and *DEBUG-PRINT-LEVEL*
2711 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2712 of CMU CL's support for SPARC, and various endianness and other
2713 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2714 * new syntactic sugar for the Unix command line: --load foo.bar is now
2715 an alternate notation for --eval '(load "foo.bar")'.
2717 ** The system now detects stack overflow and handles it gracefully,
2718 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2719 optimization settings. (This is a good thing in general, and
2720 its introduction in this version should be particularly timely
2721 for anyone whose code fails because of suppression of tail
2723 ** The system now hunts for the C variable "environ" in a more
2724 devious way, to avoid segfaults when the C library version differs
2725 between compile time and run time. (thanks to Christophe Rhodes)
2726 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2727 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2728 ** The compiler no longer issues bogus style warnings for undefined
2729 classes in the same source file as the DEFCLASSes which defined
2730 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2732 ** fixes in CONDITION class precedence list for undefined function
2733 errors (thanks to Alexei Dejneka)
2734 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2735 correctly. (thanks to Dan Barlow)
2736 ** portability fixes aiming at bootstrapping under CLISP (thanks
2737 to Dave McDonald and Christophe Rhodes)
2738 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2739 ** fixes in type translation and and type inference (thanks to
2741 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2742 ** various fixes in the new ports (thanks to Dan Barlow)
2743 * several changes related to debugging:
2744 ** suppression of tail recursion, as noted above
2745 ** stack overflow detection, as noted above
2746 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2747 is now the default. (For some time encapsulation has been more
2748 reliable than the breakpoint-based :ENCAPSULATE NIL
2749 implementation, at least on X86 systems; and I just noticed that
2750 encapsulation also seems closer to the spirit of the ANSI
2753 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2755 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2756 up properly again. (There was a packaging bug in 0.7.0 which
2757 left their definitions in SB-SYS::LOAD-FOREIGN and
2758 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2759 vital for most things which interface to C-level interfaces,
2760 like extensions working with sockets or databases or
2761 Perl-compatible regexes or whatever, and the need to fix
2762 this bug is the main reason that 0.7.1 was released so
2764 ** DEFGENERIC is now choosier about the methods it redefines, so that
2765 reLOADing a previously-LOADed file containing DEFGENERICs does
2766 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2767 cycle now works as it should. (thanks to Alexey Dejneka)
2768 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2769 fixed by Christophe Rhodes. (He actually submitted this patch
2770 months ago, and I delayed until after 0.7.0.)
2771 ** Bug 111 (internal compiler confusion about runtime checks on
2772 FUNCTION types) was fixed by Alexey Dejneka.
2773 * Some internal cleanups (getting rid of variables which aren't
2774 needed now that the byte interpreter is gone) caused the fasl
2775 file format number to change again.
2777 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2778 * major incompatible change: The default fasl file extension, i.e. the
2779 default extension for files produced by COMPILE-FILE, has changed
2780 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2782 ** There are many changes in the implementation of the compiler.
2783 SBCL is now essentially a compiler-only implementation of ANSI
2784 Common Lisp. EVAL still "interprets" a few special cases, but
2785 almost all the interesting cases are handled by creating
2786 a LAMBDA expression, calling COMPILE on it, then calling
2787 FUNCALL on the result.
2788 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2789 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2790 Since the code is newer, there might still be some new bugs
2791 (though not as many as before Martin Atzmueller's fixes:-). But
2792 the new code is substantially simpler and clearer, and hopefully
2793 any remaining bugs will be simpler, less fundamental, and more
2794 fixable then the bugs in the old code.
2795 ** The revised compiler is still a little unsteady on its feet.
2797 *** The debugging information it produces (particularly the names
2798 of FUNCTION objects) is sometimes much less useful than what
2799 the old compiler produced.
2800 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2801 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2802 within a MACROLET) has been temporarily weakened.
2803 ** There are new compiler optimizations for various functions:
2804 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2805 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2806 *** the math functions TRUNCATE, FLOOR, and CEILING
2807 *** the function-of-all-trades COERCE
2808 Mostly these should be transparent, but there's one
2809 potentially-annoying problem (bug 117): when the compiler
2810 inline-expands a function and does type analysis on the result,
2811 it can create control paths which have type mismatches, and
2812 when it can't prove that those control paths aren't taken,
2813 it will issue WARNINGs about the type mismatches. This is
2814 a particular problem in practice for the new sequence functions.
2815 It's not clear how this should be fixed, and for now, a
2816 workaround is given in the entry for 117 in the BUGS file.
2817 ** (Because of the interaction between the two previous items --
2818 occasional inlining problems and new inline expansions -- some
2819 of the new sequence function optimizations won't really kick in
2820 completely until debugging information, and then inlining, are
2821 straightened out in some future version.)
2822 * minor incompatible changes:
2823 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2824 DIRECTORY no longer implicitly promotes NIL slots of its
2825 pathname argument to :WILD. In particular, when you ask for the
2826 contents of a directory (which you used to be able to do without
2827 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2828 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2829 ** changes in behavior that ANSI explicitly defines to be
2830 implementation dependent:
2831 *** The new compiler-only implementation still conforms with ANSI,
2832 but acts a little different than before. Besides the obvious
2833 changes in performance tradeoffs (that the cost per form passed
2834 to EVAL has gone up, and the cost per form executed by EVAL
2835 has gone down), the behavior of the system changes a little
2836 because there are no longer any interpreted function objects.
2837 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2838 e.g. doing COMPILE on the output of interactive DEFUN is
2840 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2842 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2843 has changed from (:CL) to NIL.
2844 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2845 e.g. #\Newline and #\Tab, don't have names specified in the
2846 ANSI Common Lisp standard, is now based on their ASCII symbolic
2847 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2848 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2849 are no longer used for output.
2850 ** changes in internal implementation constants:
2851 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2852 4 million. (If your application spends a lot of time GCing and
2853 you have a lot of RAM, you might want to experiment with
2854 increasing it even more.)
2855 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2856 However, almost all old code should still continue to work without
2857 immediate update, as SB-C-CALL is now a (deprecated) nickname
2859 ** Old operator names in the style DEF-FOO are now deprecated in
2860 favor of new corresponding names DEFINE-FOO, for consistency with
2861 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2862 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2863 internal symbols, but a few supported extensions like
2864 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2865 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2866 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2867 etc. as you get deeper into recursive calls to the debugger
2868 command loop, instead of the old "5]", "5]]", "5]]]"
2869 sequence. (I was motivated to do this when squabbles between
2870 ILISP and SBCL left me very deeply nested in the debugger. In the
2871 short term, this change will probably provoke more ILISP/SBCL
2872 squabbles, but hopefully it will be an improvement in the long run.)
2873 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2874 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2875 function, since declaiming return types involving aliens is
2876 (1) annoyingly messy to do by hand and (2) vital to efficient
2877 compilation of code which calls such functions.
2878 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2879 longer reexported by the SB-EXT package. They're solely useful
2880 for alien code, so it seems more logical that you should get
2881 them from the SB-ALIEN package, not in SB-EXT.
2882 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2883 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2884 features. Instead, the code that they used to control is always
2885 built into the system.
2886 * many other bug fixes
2887 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2888 advantage of the new EVAL-WHEN stuff and to clean them up in
2889 general, and they are now more ANSI-compliant in a number of
2890 ways. Martin Atzmueller is responsible for a lot of this.
2891 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2893 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2894 *** correct ERROR type for various file operations
2895 *** some fixes for Lisp streams
2896 *** DEFMETHOD syntax checking
2897 *** changing old weird representation of debug information as
2898 strings (which, among their other deficiencies, don't transform
2899 correctly when you rename packages, and don't change their
2900 print representation when you change things like *PACKAGE*
2901 and *PRINT-LENGTH*) to symbols and lists of symbols
2902 He also made several improvements and fixed several bugs in DESCRIBE.
2903 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2905 *** misbehavior of WRITE-STRING/WRITE-LINE
2906 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2907 and several other LOOP problems as well
2908 *** DIRECTORY when similar filenames are present
2909 *** DEFGENERIC with :METHOD options
2910 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2911 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2912 *** argument ordering in FIND with :TEST option
2913 *** mishandled package designator argument in APROPOS-LIST
2914 *** various problems in the backquote readmacro
2915 *** a bug in APROPOS
2916 *** probably some others that I'm not describing very well here,
2917 since the CVS log documents them by reference to sbcl-devel
2918 messages, and the SourceForge archives aren't working well.:-(
2919 ** Dan Barlow improved the Alpha port (and is making progress on the
2920 PPC port, for those of you who think different).
2921 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2922 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2923 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2924 continues to work on the SPARC port (for those of you in a position
2925 to look down upon our little PC-compatible boxes from a great height).
2926 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2927 and uses that copy, rather than the current dynamic value, when
2928 it's trying to decide whether to truncate output. Thus e.g.
2929 (let ((*print-lines* 50))
2930 (pprint-logical-block (stream nil)
2932 (let ((*print-lines* 8))
2933 (print (aref possiblybigthings i) stream)))))
2934 should now truncate the logical block only at 50 lines, instead of
2935 often truncating it at 8 lines, as it did before.
2936 * The doc/cmucl/ directory, containing old CMU CL documentation
2937 from the time of the fork, is no longer part of the base system.
2938 SourceForge has shut down its anonymous FTP service, and with it
2939 my original plan for distributing the old CMU CL documentation
2940 there. For now, if you need these files you can download an old
2941 SBCL source release and extract them from it.
2942 * The fasl file version number changed again, for dozens of reasons,
2943 some of which are apparent above.
2945 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2946 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2947 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2948 patch, so that DISASSEMBLE output is much nicer.
2949 * The code in the SB-PROFILE package now seems reasonably stable.
2950 I still haven't decided what the final interface should look like
2951 (I'd like PROFILE to interact cleanly with TRACE, since both
2952 facilities use function encapsulation) but if you have a need
2953 for profiling now, you can probably use it successfully with
2954 the current CMU-CL-style interface.
2955 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2956 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2957 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2958 link now returns the dangling link itself, and for similar
2959 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2960 link itself. (In these cases the old code signalled an error and
2961 looped endlessly, respectively.) Thus, DIRECTORY now works even
2962 in the presence of dangling and cyclic symbolic links.
2963 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2964 is now a supported extension again, since the consensus on
2965 sbcl-devel was that it can be useful for ordinary development
2966 work, not just for debugging SBCL itself.
2967 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2968 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2969 that currently-defined functions might be redefined later with
2970 different return types.
2971 * Hash tables can be printed readably, as inspired by CMU CL code
2972 of Eric Marsden and SBCL code of Martin Atzmueller.
2973 * better error handling in CLOS method combination, thanks to
2974 Martin Atzmueller porting Pierre Mai's CMU CL patches
2975 * more overflow fixes for >16Mbyte I/O buffers
2976 * A bug in READ has been fixed, so that now a single Ctrl-D
2977 character suffices to cause end-of-file on character streams.
2978 In particular, now you only need one Ctrl-D at the command
2979 line (not two) to exit SBCL.
2980 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2982 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2983 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2984 does the right thing.
2985 * The implementation of some type tests, especially for CONDITION
2986 types, is now tidier and maybe faster, due to CMU CL code
2987 originally by Douglas Crosher, ported by Martin Atzmueller.
2988 * Some math functions have been fixed, and there are new
2989 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2990 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2991 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2992 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2993 likely to be built into sbcl-0.7.0.)
2994 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2995 CL, ported to SBCL by Martin Atzmueller)
2996 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2997 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2998 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2999 * There's a new slam.sh hack to shorten the edit/compile/debug
3000 cycle for low-level changes to SBCL itself, and a new
3001 :SB-AFTER-XC-CORE target feature to control the generation of
3002 the after-xc.core file needed by slam.sh.
3003 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3004 COMPILE-FILE is no longer supported, so that now every function
3005 gets an entry point, so that block compilation looks a little
3006 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3007 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3008 returns the number of bytes consed since the system started,
3009 rather than the number consed since the first time the function
3010 was called. (The new definition parallels ANSI functions like
3011 CL:GET-INTERNAL-RUN-TIME.)
3012 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3013 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3014 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3015 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3016 * Fasl file version numbers are now independent of the target CPU,
3017 since historically most system changes which required version
3018 number changes have affected all CPUs equally. Similarly,
3019 the byte fasl file version is now equal to the ordinary
3022 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3023 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3024 is no longer recognized. I apologize for this, because it was
3025 listed in SB-EXT as a supported extension, but I found that
3026 its existing behavior was poorly specified, as well as incorrectly
3027 specified, and it looked like too much of a mess to straighten it
3028 out. I have enough on my hands trying to get ANSI stuff to work..
3029 * many patches ported from CMU CL by Martin Atzmueller, with
3030 half a dozen bug fixes in pretty-printing and the debugger, and
3031 half a dozen others elsewhere
3032 * fixed bug 13: Floating point infinities are now supported again.
3033 They might still be a little bit flaky, but thanks to bug reports
3034 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3035 as flaky as they were.
3036 * The --noprogrammer command line option is now supported. (Its
3037 behavior is slightly different in detail from what the old man
3038 page claimed it would do, but it's still appropriate under the
3039 same circumstances that the man page talks about.)
3040 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3041 are now supported, and enabled by default. Thus, the compiler can
3042 handle many floating point and complex operations much less
3043 inefficiently. (Thus e.g. you can implement a complex FFT
3045 * The compiler now detects type mismatches between DECLAIM FTYPE
3046 and DEFUN better, and implements CHECK-TYPE more correctly, and
3047 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3048 thanks to patches from Martin Atzmueller.
3049 * various fixes to make the cross-compiler more portable to
3050 ANSI-conforming-but-different cross-compilation hosts (notably
3051 Lispworks for Windows, following bug reports from Arthur Lemmens)
3052 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3053 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3054 fixed thanks to Pierre Mai's CMU CL patches.
3055 * fixes to keep the system from overflowing internal counters when
3056 it tries to use i/o buffers larger than 16M bytes
3057 * fixed bug 45a: Various internal functions required to support
3058 complex special functions have been merged from CMU CL sources.
3059 (When I was first setting up SBCL, I misunderstood a compile-time
3060 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3061 * improved support for type intersection and union, fixing bug 12
3062 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3063 more obscure bugs as well
3064 * some steps toward byte-compiling non-performance-critical
3065 parts of the system, courtesy of patches from Martin Atzmueller
3066 * Christophe Rhodes has made some debian packages of sbcl at
3067 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3068 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3069 stable, but are nonetheless usable. When he's ready, I'd be happy
3070 to add them to the SourceForge "File Releases" section. (And if
3071 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3072 * new fasl file format version number (because of changes in
3073 internal representation of (OR ..) types to accommodate the new
3074 support for (AND ..) types, among other things)
3076 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3077 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3079 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3081 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3082 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3083 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3084 * DESCRIBE now works on structure objects again.
3085 * Most function call argument type mismatches are now handled as
3086 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3087 know whether the function will be redefined before the call is
3088 executed. (The compiler could flag local calls with full WARNINGs,
3089 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3090 it doesn't keep track of enough information to know whether calls
3091 are local in this sense.)
3092 * Compiler output is now more verbose, with messages truncated
3093 later than before. (There should be some supported way for users
3094 to override the default verbosity, but I haven't decided how to
3095 provide it yet, so this behavior is still controlled by the internal
3096 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3097 src/compiler/ir1util.lisp.)
3098 * Fasl file format version numbers have increased again, because
3099 support for the Gray streams extension changes the layout of the
3100 system's STREAM objects.
3101 * The Gray subclassable streams extension now works, thanks to a
3102 patch from Martin Atzmueller.
3103 * The full LOAD-FOREIGN extension (not just the primitive
3104 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3105 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3106 but like most other programs, it defaults to copying the Unix
3107 environment from the original process instead of starting the
3108 new process in an empty environment.
3109 * Extensions which manipulate the Unix environment now support
3110 an :ENVIRONMENT keyword option which doesn't smash case or
3111 do other bad things. The CMU-CL-style :ENV option is retained
3112 for porting convenience.
3113 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3114 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3116 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3118 * A patch from Martin Atzmueller seems to have solved the SIGINT
3119 problem, and as far as we know, signal-handling now works cleanly.
3120 (If you find any new bugs, please report them!)
3121 * The system no longer defaults Lisp source file names to types
3122 ".l", ".cl", or ".lsp", but only to ".lisp".
3123 * The compiler no longer uses special default file extensions for
3124 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3125 seems to expect a single default extension for all compiled code,
3126 and there's no compelling reason to try to stretch the standard
3127 to allow two different extensions.) Instead, byte-compiled files
3128 default to the same extension as native-compiled files.
3129 * Fasl file format version numbers have increased again, because
3130 a rearrangement of internal implementation packages made some
3131 dumped symbols in old fasl files unreadable in new cores.
3132 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3133 many fewer weird special cases.
3134 * Bug #17 (differing COMPILE-FILE behavior between logical and
3135 physical pathnames) has been fixed, and some related misbehavior too,
3136 thanks to a patch from Martin Atzmueller.
3137 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3138 by Tim Moore, ported to SBCL by Martin Atzmueller.
3139 * Martin Atzmueller fixed several filesystem-related problems,
3140 including bug #36, in part by porting CMU CL patches, which were
3141 written in part by Paul Werkowski.
3142 * More compiler warnings in src/runtime/ are gone, thanks to
3143 more patches from Martin Atzmueller.
3144 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3147 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3149 * DESCRIBE now works on CONDITION objects.
3150 * The debugger now handles errors which arise when trying to print
3151 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3153 * The build system now uses an additional file, customize-target-features.lisp,
3154 to allow local modifications to the target *FEATURES* list. (The point of
3155 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3156 debugging features enabled, without having to worry about propagating it
3157 into everyone's system when I do a "cvs update".) When no
3158 customize-target-features.lisp file exists, the target *FEATURES* list
3159 should be constructed the same way as before.
3160 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3161 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3162 warn about behavior which is undefined under the ANSI spec. Note: This
3163 is specified by ANSI, but it's not very popular with programmers.
3164 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3165 macro in the SBCL sources for an example of a workaround which you
3166 might use to make portable ANSI-standard code which does what you want.
3167 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3168 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3169 when it's used as a non-toplevel form. (This is required in order
3170 to implement the DEFCONSTANT-EQX macro.)
3171 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3172 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3173 funny things, due to bugs in the implementation of EVAL-WHEN
3174 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3175 non-toplevel DEFCONSTANTs are uncommon.)
3176 * The core file version number and fasl file version number have been
3177 incremented, because the old noncompliant DEFCONSTANT behavior involved
3178 calling functions which no longer exist, and because I also took the
3179 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3180 * fixed bug 1 (error handling before read-eval-print loop starts), and
3181 redid debugger restarts and related debugger commands somewhat while
3183 ** The QUIT debugger command is gone, since it did something
3184 rather different than the SB-EXT:QUIT command, and since it never
3185 worked properly outside the main toplevel read/eval/print loop.
3186 Invoking the new TOPLEVEL restart provides the same functionality.
3187 ** The GO debugger command is also gone, since you can just invoke
3188 the CONTINUE restart directly instead.
3189 ** The TOP debugger command is also gone, since it's redundant with the
3190 FRAME 0 command, and since it interfered with abbreviations for the
3192 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3194 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3195 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3196 some porting work by Martin Atzmueller)
3197 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3198 with a boatload of patches which he ported to SBCL. Now that those
3200 ** The system tries to make sure that its low-priority messages
3201 are prefixed by semicolons, to help people who like to use
3202 syntax highlighting in their ILISP buffer. (This patch
3203 was originally due to Raymond Toy.)
3204 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3205 patches originally written by Raymond Toy.
3206 ** The compiler understands coercion between single-value and
3207 multiple-VALUES type expressions better, getting rid of some very
3208 weird behavior, thanks to patches originally by Robert MacLachlan
3209 and Douglas Crosher.
3210 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3211 lambda lists, thanks to a patch originally by Pierre Mai.
3212 ** The system no longer bogusly warns about "abbreviated type
3214 ** The compiler gets less confused by inlining and RETURN-FROM,
3215 thanks to some patches originally by Tim Moore.
3216 ** The system no longer hangs when dumping circular lists to fasl
3217 files, thanks to a patch originally from Douglas Crosher.
3218 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3219 undefined function error.
3220 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3221 in structure slot initforms) for now, documented workaround instead:-|
3222 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3223 * fixed bug 5: added stubs for various Gray stream functions called
3224 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3225 installed, at least appropriate type errors are generated
3226 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3227 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3228 consistently in DEFMETHOD forms.
3229 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3230 it doesn't seem to affect SBCL after all
3231 * The C runtime system now builds with better optimization and many
3232 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3234 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3236 * The system is now under CVS at SourceForge (instead of the
3237 CVS repository on my home machine).
3238 * The new signal handling code has been tweaked to treat register
3239 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3240 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3241 downstream, e.g. in debug-int.lisp, has implicit dependencies
3242 on the unsignedness of integer representation of machine words,
3243 and that caused the system to bomb out with infinite regress
3244 when trying to recover from type errors involving signed values,
3245 e.g. (BUTLAST '(1 2 3) -1).)
3246 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3247 (This was one of the bugs Peter Van Eynde reported back in July.)
3248 * The system now uses code inspired by Colin Walters' O(N)
3249 implementation of MAP (from the cmucl-imp@cons.org mailing
3250 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3251 inline the MAP operation, and there is more than one
3252 sequence argument to the MAP call (so that it can't just
3253 do ETYPECASE once and for all based on the type of the
3254 single sequence argument). (The old non-inline implementation
3255 of the general M-argument sequence-of-length-N case required
3256 O(M*N*N) time when any of the sequence arguments were LISTs.)
3257 * The QUIT :UNIX-CODE keyword argument has been renamed to
3258 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3259 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3261 * Raymond Wiker's patches to port dynamic loading from Linux to
3262 FreeBSD have been added.
3263 * The BUGS file is now more nearly up to date, thanks in large part
3264 to Martin Atzmueller's review of it.
3265 * The debugger now flushes standard output streams before it begins
3266 its output ("debugger invoked" and so forth).
3267 * The core version number and fasl file version number have both
3268 been incremented, because of incompatible changes in the layout
3270 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3271 used to not be called for a saved Lisp image.)
3272 * Martin Atzmueller's version of a patch to fix a compiler crash,
3273 as posted on sbcl-devel 13 September 2000, has been installed.
3274 * Instead of installing Martin Atzmueller's patch for the
3275 compiler transform for SUBSEQ, I deleted the compiler transform,
3276 and transforms for some similar consing operations.
3277 * A bug in signal handling which kept TRACE from working on OpenBSD
3279 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3280 away by constant folding
3281 * The system now defines its address space constants in one place
3282 (in the Lisp sources), and propagates them automatically elsewhere
3283 (through GENESIS and the sbcl.h file). Therefore, patching the
3284 address map is less unnecessarily tedious and error-prone. The
3285 Lisp names of address space constants have also been systematized.
3286 * CVS tags like dollar-Header-dollar have been removed from
3287 the sources, because they have never saved me trouble and
3288 they've been source of trouble working with patches and other
3289 diff-related operations.
3290 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3291 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3293 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3295 * The system has been ported to OpenBSD.
3296 * The system now compiles with a simple "sh make.sh" on the systems
3297 that it's supported on. I.e., now you no longer need to tweak
3298 text in the target-features.lisp-expr and symlinks in src/runtime/
3299 by hand, the make.sh takes care of it for you.
3300 * The system is no longer so grossly inefficient when compiling code
3301 involving vectors implemented as general (not simple) vectors (VECTOR T),
3302 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3303 which dares to use the various sequence functions on non-simple
3304 vectors, takes less of a performance hit.
3305 * There is now a primitive type predicate VECTOR-T-P
3306 to test for the (VECTOR T) type, so that e.g.
3307 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3308 can now be compiled with some semblance of efficiency. (The old code
3309 turned the type declaration into a full call to %TYPEP at runtime!)
3310 * AREF on (VECTOR T) is still not fast, since it's still compiled
3311 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3312 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3313 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3314 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3315 without making full calls to SUBTYPEP at runtime.
3316 (Some analogous efficiency issues for non-simple vectors specialized to
3317 element types other than T, or for non-simple multidimensional arrays,
3318 have not been addressed. They could almost certainly be handled the
3319 same way if anyone is motivated to do so.)
3320 * The changes in array handling break binary compatibility, so
3321 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3322 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3323 instead of (VALUES T T).
3324 * By following the instructions that Dan Barlow posted to sbcl-devel
3325 on 2 July 2000, I was able to enable primitive dynamic object
3326 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3327 functionality is not implemented (since it calls ld to resolve
3328 library references automatically, requiring RUN-PROGRAM for its
3329 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3330 try to resolve library references) is now supported.
3331 * The system now flushes the standard output streams when it terminates,
3332 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3333 them at several other probably-convenient times, e.g. in each pass of
3334 the toplevel read-eval-print loop, and after evaluating a form given
3335 as an "--eval" command-line option. (These changes were motivated by a
3336 discussion of stream flushing issues on cmucl-imp in August 2000.)
3337 * The source transform for TYPEP of array types no longer assumes
3338 that an array whose element type is a not-yet-defined type
3339 is implemented as an array of T, but instead punts, so that the
3340 type will be interpreted at runtime.
3341 * There is now some support for cross-compiling in make.sh: each of
3342 the phases of make.sh has its own script. (This should be transparent
3343 to people doing ordinary, non-cross-compile builds.)
3344 * Since my laptop doesn't have hundreds of megabytes of memory like
3345 my desktop machine, I became more motivated to do some items on
3346 my to-do list in order to reduce the size of the system a little:
3347 ** Arrange for various needed-only-at-cold-init things to be
3348 uninterned after cold init. To support this, those things have
3349 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3350 symbols with such names are now uninterned after cold init).
3351 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3352 fasl files for cold load.
3353 ** Remove the old compiler structure pooling code (which used to
3354 be conditional on the target feature :SB-ALLOC) completely.
3355 ** Redo the representation of some data in cold init to be more compact.
3356 (I also looked into supporting byte compiled code at bootstrap time,
3357 which would probably reduce the size of the system a lot, but that
3358 looked too complicated, so I punted for now.)
3359 * The maximum signal nesting depth in the src/runtime/ support code has
3360 been reduced from 4096 to 256. (I don't know any reason for the very
3361 large old value. If the new smaller value turns out to break something,
3362 I'll probably just bump it back up.)
3363 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3365 * Many, many bugs reported by Peter Van Eynde have been added to
3366 the BUGS list; some have even been fixed.
3367 * While enabling dynamic object file loading, I tried to make the
3368 code easier to understand, renaming various functions and variables
3369 with less ambiguous names, and changing some function calling
3370 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3371 * While trying to figure out how to do the OpenBSD port, I tried to
3372 clean up some of the code in src/runtime/. In particular, I dropped
3373 support for non-POSIX signal handling, added various comments,
3374 tweaked the code to reduce the number of compilation warnings, and
3375 renamed some files to increase consistency.
3376 * To support the new automatic configuration functionality in make.sh,
3377 the source file target-features.lisp-expr has been replaced with the
3378 source file base-target-features.lisp-expr and the machine-generated
3379 file local-target-features.lisp-expr.
3380 * fixed a stupid quoting error in make.sh so that using CMU CL
3381 "lisp -batch" as cross-compilation host works again
3383 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3385 * DESCRIBE no longer tries to call itself recursively to describe
3386 bound/fbound values, so that it no longer fails on symbols which are
3387 bound to themselves (like keywords, T, and NIL).
3388 * DESCRIBE now works on generic functions.
3389 * The printer now prints less-screwed-up representations of closures
3390 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3391 * A private symbol is used instead of the :EMPTY keyword previously
3392 used to mark empty slots in hash tables. Thus
3393 (DEFVAR *HT* (MAKE-HASH-TABLE))
3394 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3395 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3396 now does what ANSI says that it should. (You can still get
3397 similar noncompliant behavior if bang on the hash table
3398 implementation with all the symbols you get back from
3399 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3400 This breaks binary compatibility, since tests for equality to
3401 :EMPTY are wired into things like the macroexpansion of
3402 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3404 * There's now a minimal placeholder implementation for CL:STEP,
3405 as required by ANSI.
3406 * An obscure bug in the interaction of the normal compiler, the byte
3407 compiler, inlining, and structure predicates has been patched
3408 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3409 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3411 * Missing ordinary arguments in a macro call are now detected even
3412 when the macro lambda list contains &KEY or &REST.
3413 * The debugger no longer complains about encountering the top of the
3414 stack when you type "FRAME 0" to explicitly instruct it to go to
3415 the top of the stack. And it now prints the frame you request even
3416 if it's the current frame (instead of saying "You are here.").
3417 * As specified by ANSI, the system now always prints keywords
3418 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3420 * The default initial SIZE of HASH-TABLEs is now smaller.
3421 * Type information from CLOS class dispatch is now propagated
3422 into DEFMETHOD bodies, so that e.g.
3423 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3425 is now basically equivalent to
3426 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3427 (DECLARE (TYPE SINGLE-FLOAT X))
3429 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3430 operation, without having to do run-time type dispatch.
3431 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3432 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3433 * Since I don't seem to be making big file reorganizations very often
3434 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3435 is overflowing my ability to conveniently back them up), I've finally
3436 checked the system into CVS. (The CVS repository is on my home system,
3437 not at SourceForge -- putting it on SourceForge might come later.)
3438 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3439 high-level GC-NOTIFY-FOO functions send their output. (There's
3440 still very little control of where low-level verbose GC functions
3441 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3442 less than it used to -- the GC-NOTIFY-FOO functions are now under
3443 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3444 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3445 in only one place in the source code, and propagates it automatically
3446 everywhere that it's needed. Thus e.g. when I bump the version from
3447 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3448 * The C source files now include boilerplate legalese and documentation
3449 at the head of each file (just as the Lisp source files already did).
3450 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3451 to his page will be replaced with a link to his new CLiki service.
3453 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3455 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3456 * The build process now looks for GNU make under the default name "gmake",
3457 instead of "make" as it used to. If GNU make is not available as "gmake"
3458 on your system, you can change this default behavior by setting the
3459 GNUMAKE environment variable.
3460 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3461 can build without error under CMU CL.
3463 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3465 * There is now a partial SBCL user manual (with some new text and some
3466 text cribbed from the CMU CL manual).
3467 * The beginnings of a profiler have been added (starting with the
3468 CMU CL profiler and simplifying and cleaning up). Eventually the
3469 main interface should be through the TRACE macro, but for now,
3470 it's still accessed through vaguely CMU-CL-style functions and macros
3471 exported from the package SB-PROFILE.
3472 * Some problems left over from porting CMU CL to the new
3473 cross-compilation bootstrap process have been cleaned up:
3474 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3475 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3476 part when getting rid of the compiler *BACKEND* stuff.)
3477 ** The value of *NULL-TYPE* was screwed up, because it was
3478 being initialized before the type system knew the final
3479 definition of the 'NULL type. This screwed up several key
3480 optimizations in the compiler, causing inefficiency in all sorts
3481 of places. (I found it because I wanted to understand why
3482 GET-INTERNAL-RUN-TIME was consing.)
3483 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3484 PROCLAIM FTYPE information. Unfortunately this broke binary
3485 compatibility again, since now the forms output by DEFGENERIC
3486 to refer to functions which didn't exist in 0.6.3.
3487 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3488 can use the new (as of 0.6.3) transform for SOME into MAP into
3490 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3491 values to (INTEGER 0 1000000), so that the time code will no longer
3492 occasionally get blown up by Linux returning 1000000 microseconds
3493 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3494 its output conform to the ANSI spec. (Alas, this makes its output
3495 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3496 * A full call to MAP NIL with a single sequence argument no longer conses.
3497 * fixes to problems pointed out by Martin Atzmueller:
3498 * The manual page no longer talks about multiprocessing as though
3499 it were currently supported.
3500 * The ILISP support patches have been removed from the distribution,
3501 because as of version 5.10.1, ILISP now supports SBCL without us
3502 having to maintain patches.
3503 * added a modified version of Raymond Toy's recent CMU CL patch for
3504 EQUALP comparison of HASH-TABLE
3506 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3508 * The system still can't cross-compile itself with
3509 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3510 entails), but at least it can compile more of itself that way
3511 than it used to be able to, and various buglets which were uncovered
3512 by trying to cross-compile itself that way have now been fixed.
3513 * This release breaks binary compatibility again. This time
3514 at least I've incremented the FASL file format version to 2, so that the
3515 problem can be detected reliably instead of just causing weird errors.
3516 * various new style warnings:
3517 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3518 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3519 ** using the lexical binding of a variable named in the *FOO* style
3520 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3521 as specified by ANSI.
3522 * *RANDOM-STATE* is no longer automatically initialized from
3523 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3524 default behavior of the system is to repeat its behavior every time
3525 it's run. If you'd like to change this behavior, you can always
3526 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3527 old convention there was no comparably easy way to get the system to
3528 repeat its behavior every time it was run.
3529 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3530 FTYPE declarations has been removed, in favor of their ANSI
3531 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3532 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3533 the types of their sequence arguments can be determined at compile time.
3534 This is done through a new open code expansion for MAP which eliminates
3535 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3536 argument types can be determined at compile time.
3537 * The optimizer now transforms COERCE into an identity operation when it
3538 can prove that the coerced object is already of the correct type. (This
3539 can be a win for machine generated code, including the output of other
3540 optimization transforms, such as the MAP transform above.)
3541 * Credit information has been moved from source file headers into CREDITS.
3542 * Source file headers have been made more standard.
3543 * The CASE macro now compiles without complaining even when it has
3546 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3548 * (Note that the way that the PCL macroexpansions were rewritten
3549 to accommodate the change in DEFGENERIC below breaks binary
3550 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3551 not run under sbcl-0.6.2. Once we get out of alpha releases,
3552 i.e. hit release 1.0.0, we'll probably try to maintain binary
3553 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3554 and sbcl-1.4.4. Until then, however, it might be fairly common
3555 for maintenance releases to break binary compatibility.)
3556 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3558 * The reporting of unhandled signals has been changed to print some
3559 explanatory text as well as the report form. (Previously only
3560 the report form was printed.)
3561 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3562 it defines, so that the compiler no longer issues undefined function
3563 warnings for compiled-but-not-yet-loaded generic functions.
3564 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3565 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3566 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3567 section of the ANSI spec are supported.
3568 * The "" nickname for the "KEYWORD" package has been removed.
3569 The reader still handles symbol tokens which begin with a package marker
3570 as keywords, but it doesn't expose its mechanism for doing so in the
3571 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3572 * The system now issues STYLE-WARNINGs for contradictory TYPE
3573 proclamations. (Warnings for contradictory FTYPE proclamations would
3574 be nice too, but those can't be done usefully unless the type system
3575 is made smarter about FUNCTION types.)
3576 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3577 systematized, so that "*target-*.lisp is supposed to exist only on the
3578 target and imply that there's a related file which exists on the
3579 host, and *host-*.lisp is supposed to exist only on the host and imply
3580 that there's a related file which exists on the target. This involves a
3581 lot of renaming. Hopefully the acute confusion caused by the renaming
3582 will be justified by the reduction in chronic confusion..
3583 ** runtime-type.lisp -> early-target-type.lisp
3584 ** target-type.lisp -> late-target-type.lisp
3585 ** early-host-format.lisp -> early-format.lisp
3586 ** late-host-format.lisp -> late-format.lisp
3587 ** host-error.lisp -> misc-error.lisp
3588 ** early-error.lisp -> early-target-error.lisp
3589 ** late-error.lisp -> late-target-error.lisp
3590 ** host-defboot.lisp -> early-defboot.lisp
3591 ** code/misc.lisp -> code/target-misc.lisp
3592 ** code/host-misc.lisp -> code/misc.lisp
3593 ** code/numbers.lisp -> code/target-numbers.lisp
3594 ** code/early-numbers.lisp -> numbers.lisp
3595 ** early-host-type.lisp -> early-type.lisp
3596 ** late-host-type.lisp -> late-type.lisp
3597 ** host-typep.lisp -> typep.lisp
3598 ** load.lisp -> target-load.lisp
3599 ** host-load.lisp -> load.lisp
3600 ** host-disassem.lisp -> disassem.lisp
3601 ** host-insts.lisp -> insts.lisp
3602 ** byte-comp.lisp -> target-byte-comp.lisp
3603 ** host-byte-comp.lisp -> byte-comp.lisp
3604 ** host-signal.lisp -> signal.lisp
3605 ** host-defstruct.lisp -> defstruct.lisp
3606 ** late-target-type.lisp -> deftypes-for-target.lisp
3607 Furthermore, several other previously target-only files foo.lisp (e.g.
3608 hash-table.lisp and random.lisp) have been split into a target-and-host
3609 foo.lisp file and a target-only target-foo.lisp file, with their key type
3610 definitions in the target-and-host part, so that the cross-compiler will
3611 know more about target types.
3612 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3613 gone away. In their place are various *BACKEND-FOO* variables
3614 corresponding to the slots of the old structure.
3615 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3616 into the target SBCL has been fixed.
3617 * The chill.lisp system for loading cold code into a running SBCL
3619 * Support for the CMU CL "scavenger hook" extension has been removed.
3620 (It was undocumented and unused in the CMU CL sources that SBCL was
3621 derived from, and stale in sbcl-0.6.1.)
3622 * Various errors in the cross-compiler type system were detected
3623 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3624 (enabling various consistency checks). Many of them were fixed,
3625 but some hard problems remain, so the compiler is back to
3626 running without *TYPE-SYSTEM-INITIALIZED* for now.
3627 * As part of the cross-compiler type system cleanup, I implemented
3628 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3629 * I have started adding UNCROSS calls throughout the type system
3630 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3631 on cross-compiler input files will be able to go away, and various
3633 * CONSTANTP now returns true for quoted forms (as explicitly required
3636 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3638 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3639 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3640 (MAKE-LIST -1). In the longer term, it ought to become true
3641 that declarations are assertions even at SAFETY 1. For now, it's not
3642 quite true even at SAFETY 3, but it's at least more nearly true..
3643 (Note that this change seems to increases the size of the system by
3644 O(5%) and to decrease the speed of the compiler by 20% or more.)
3645 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3646 for the problem of printing dozens of lines of distracting information
3647 about low-level system machinery as part of the top stack frame
3648 on entry to the debugger when an undefined function was called.
3649 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3650 is not reset to COMMON-LISP-USER.
3651 * Compilation of stuff related to dyncount.lisp has been made conditional
3652 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3653 smaller. The various dyncount-related symbols have been moved into
3654 a new "SB-DYNCOUNT" package.
3655 * tty-inspect.lisp has been renamed to inspect.lisp.
3656 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3657 feature has gone away. (When we eventually port to other flavors of
3658 libc and/or Unix, we'll try to make the differences between flavors
3659 invisible at the user level.)
3660 * Various other *FEATURES* tags, and/or their associated conditionals,
3661 have been removed if obsolescent, or given better documentation, or
3662 sometimes given more-mnemonic names.
3664 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3666 * tidied up "make.sh" script
3667 * tidied up system directory structure
3668 * better "clean.sh" behavior
3669 * added doc/FOR-CMUCL-DEVELOPERS
3670 * many many small tweaks to output format, e.g. removing possibly-confusing
3671 trailing #\. character in DESCRIBE-INSTANCE
3672 * (EQUALP #\A 'A) no longer signals an error.
3673 * new hashing code, including EQUALP hashing
3674 * tidied up Lisp initialization and toplevel
3675 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3676 * command line argument processing
3677 * added POSIX-GETENV function to deal with Unix-ish environment variables
3678 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3679 terminating SBCL on EOF
3680 * non-verbose GC by default
3681 * There is no more "sbcl" shell script; the sbcl file is now the C
3682 runtime executable (just like CMU CL).
3683 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3685 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3686 symbol and package information as Lisp native symbol and package objects
3687 instead of strings naming symbols and strings naming packages. This way,
3688 whenever packages are renamed (as in warm init), debug information is
3689 transformed along with everything else.
3690 * tweaked the optimization policy declarations which control the building
3691 of SBCL itself. Now, among other things, the system no longer saves
3692 source location debugging information. (This helps two problems at once
3693 by reducing SBCL size and by keeping SBCL from trying to look for its
3694 sources -- which may not exist -- when reporting errors.)
3695 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3696 debugging and testing purposes
3697 * cleaned up printing, making the printer call PRINT-OBJECT for
3698 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3699 methods, giving nearly-ANSI behavior
3700 * converted almost all special variables to use *FOO* naming convention
3701 * deleted PARSE-TIME functionality, since it can be done portably
3702 * moved some files out of cold init into warm init
3703 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3705 * regularized formatting of source files
3706 * added an install.sh script
3707 * fixed ridiculous memory usage of cross-compiler by making
3708 compiler/alloc.lisp not try to do pooling unless it can hook
3709 itself into the GC of the cross-compilation host. Now the system
3710 builds nicely on my old laptop.
3711 * added :SB-ALLOC in target-features.lisp-expr
3712 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3713 was not implemented)
3714 * re-did condition handling and note reporting in the compiler. Notes
3715 are no longer handled by signalling conditions. Style warnings
3716 and warnings are handled more correctly and reported in such a way
3717 that it's easy to find one or the other in your output (so that you
3718 can e.g. figure out which of many problems caused COMPILE-FILE to
3720 * changed the severity of several compiler warnings from full WARNING
3721 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3722 compiler note reporting so that it doesn't use the condition system
3723 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3724 COMPILE-FILE command)
3725 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3726 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3727 no longer supported, since their implementation was deeply intertwingled
3728 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3729 * removed bogus "support" for compiler macros named (SETF FOO), and
3730 removed the compiler macro for SETF INFO (but only after making a fool
3731 of myself on the cmucl-imp mailing list by posting a bogus patch for
3732 DEFINE-COMPILER-MACRO..)
3733 * Compiled files containing forms which have side effects on the Lisp
3734 reader (such as DEFPACKAGE forms) are now handled more correctly.
3735 (Compiler queuing of top level lambdas has been suppressed by setting
3736 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3737 * deleted various currently-unused source files, e.g. gengc.lisp. They
3738 may be added back at some point e.g. when porting to other architectures,
3739 but until they are it's distracting to distribute them and to try to
3741 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3742 there were so many of them they're just distractions, and UNCROSS is
3743 known to be able to handle the current sources
3744 * moved PROFILE functionality into TRACE, so that it will be clear
3745 how the wrapping and unwrapping of functions when you profile them
3746 interacts with the wrapping and unwrapping of functions when you
3747 trace them. (Actually, the functionality isn't there yet, but at least
3748 the interface specification is there. Hopefully, the functionality will
3749 arrive with some future maintenance release.)
3750 * removed host-oops.lisp
3751 * changed signature of QUIT function to allow UNIX-CODE argument
3752 * fixed READ-SEQUENCE bug
3753 * tweaked verbose GC output so that it looks more like the progress
3754 output that ANSI specifies for functions like LOAD
3755 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3756 * added <http://sbcl.sourceforge.com> to the banner information printed by