1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
3 * fixed some bugs revealed by Paul Dietz' test suite:
4 ** MISC.641: LET-conversion were not supposed to work in late
7 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
8 * new feature: source path information is generated for macro-expansion
9 errors for use in IDE's like Slime (thanks to Helmut Eller)
10 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
11 * bug fix: compiler does not loop forever on an invalid type in
13 * improvement: compilation of most CLOS applications is significantly
15 * optimization: added a limited bytecode compiler for simple toplevel
16 forms, speeding up compilation and FASL loading
18 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
19 * minor incompatible change: in sbcl-0.9.11 (but not earlier
20 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
21 into subclasses of the finalized class. Now user calls to
22 FINALIZE-INHERITANCE finalize just the one class, and calls by the
23 system return before any subclasses are finalized.
24 * minor incompatible change: The reader no longer ignores errors
25 regarding non-existent packages in #+ and #- feature tests.
26 * new feature: command line options --no-sysinit, --no-userinit to
27 inhibit loading the corresponding init files
28 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
29 for sending data through UDP sockets (thanks to François-René Rideau)
30 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
31 error is signalled instead (thanks to François-René Rideau)
32 * minor incompatible change: Error signalling behaviour of lexical
33 operations violating package locks has changed slightly. Refer to
34 documentation on package locks for details.
35 * enhancement: EVAL can process IF-expressions without resorting to the
37 * bug fix: Some operations involving SIN, COS, or TAN weren't being
38 constant-folded properly due to missing out-of-line functions.
39 (reported by Mika Pihlajamäki)
40 * bug fix: LISTEN sometimes returned T even in cases where no data was
41 immediately available from the stream
42 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
43 were reversed. (reported by Levente Mészáros)
44 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
45 when given a vector displaced to an adjustable array. (reported
47 * bug fix: derivation of float boundaries from numbers outside the
48 appropriate float range (reported by John Wiseman)
49 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
51 * bug fix: precomputing cache entries for generic functions with
52 some subclasses of specializers as yet invalid does not attempt to
53 fill a cache line with a negative offset. (reported by Levente
55 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
56 allow condition class objects as arguments to DOCUMENTATION and
57 (SETF DOCUMENTATION); only find and set documentation for
58 structure names for the STRUCTURE doc-type. (suggested by Gary
60 * improvements to the Win32/x86 port:
61 ** (user-homedir-pathname) and default initialization file
62 locations now know about the user's "Documents and Settings"
63 directory (thanks to Yaroslav Kavenchuk)
64 ** run-program is implemented (thanks to Mike Thomas)
65 ** sockets support (thanks to Timothy Ritchey)
66 ** better backtrace support (thanks to Alastair Bridgewater)
67 ** sb-grovel supported
68 ** asdf-install and sb-posix work somewhat
69 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
70 * improvements to the Solaris/x86 port:
71 ** works on Solaris 11/Solaris Express
72 ** floating-point exception handling support
73 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
74 * fixed some bugs revealed by Paul Dietz' test suite:
75 ** REMOVE-METHOD returns its generic function argument even when
76 the method is not one of the generic functions' methods.
77 ** objects declared as MEMBER types can be admissible arguments to
79 ** printing characters should simply be printed by the FORMAT ~:C
81 ** compiler failure when compiling functions with hairy constant
82 defaults for optional parameters.
83 ** compiler produces wrong code when MAYBE-INLINE-expanding a
84 function, which is already optimized.
86 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
87 * new platform: experimental support for SBCL x86/Darwin, including
88 MacOS X 10.4.5 on Intel.
89 * new feature: Unicode character names are now known to the system
90 (through CHAR-NAME and NAME-CHAR).
91 * new feature: the filesystem location of SBCL's core file is
92 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
93 * minor incompatible change: the contrib modules SB-POSIX and
94 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
95 this change is to make it easier to distribute
96 location-independent binaries.
97 * bug fix: as implied by AMOP, standardized classes no longer have
98 slots named by external symbols of public packages. (reported by
100 * optimization: calling functions via constant symbols -- as in
101 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
102 via the function object as in (FUNCALL #'FOO).
103 * optimization: on x86 and x86-64, the calling convention for the
104 case of dealing with an unknown number of values has been altered
105 to be friendlier to the prediction heuristics implemented,
106 particularly with reference to CALL and RET pairing. (thanks to
107 Alastair Bridgewater)
108 * optimization: on x86, the code for access to thread-local symbol
109 values and binding/unbinding of thread-local symbols is smaller.
110 (thanks to Alastair Bridgewater)
111 * enhancement: CONSTANTP is now able to determine constantness of
112 more complex forms, including calls to constant-foldable standardized
113 functions and some special forms beyond QUOTE.
114 * fixed bug: occasional GC crashes on Solaris/x86
115 * optimization: x86-64 supports stack allocation of results of simple
116 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
117 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
118 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
119 (thanks to James Knight)
120 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
121 by the standard (thanks to Stephen Compall)
123 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
124 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
125 be used for bundling the runtime and the core file into one
126 executable binary. This feature is not currently supported on all SBCL
127 platforms. (thanks to James Bielman and NIIMI Satoshi)
128 * new feature: a generational or ephemeral garbage collector is now
129 the default on the PowerPC platform (both Linux and Darwin). The
130 old Cheney (stop and copy) collector is a build-time option.
131 (thanks to Cyrus Harmon, after Raymond Toy)
132 * minor incompatible change: the method by which SBCL finds its
133 contributed modules has changed; it no longer relies on symbolic
134 links from an $SBCL_HOME/systems directory, but searches directly
135 in the subdirectories of $SBCL_HOME.
136 * enhancement: the dynamic heap size on the Linux/PPC platform is
137 markedly larger, even using the older Cheney garbage collector.
138 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
139 certain complicated string types. (reported by Gary King)
140 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
141 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
142 * fixed bug: DOCUMENTATION on structures no longer signals an
143 error if no structure type of the right name exists. (reported by
145 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
146 applicable non-standard primary or :AROUND method on
147 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
148 no longer cause unbound variable errors. (reported by Kevin Reid)
149 * fixed bug: metacircle resolution in cases where methods have slots
150 added before the slots from STANDARD-METHOD. (reported by Jean
152 * fixed bug: the Power PC assembly code for calling into lisp
153 sometimes computed its return address wrongly (depending on the
154 15th bit of the address, fixed at link time). This used to
155 manifest itself in a segmentation violation while building PCL.
156 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
157 many others over the years)
158 * fixed bug: ROOM no longer reports silly numbers for stack usage if
159 the stack pointer (treated as a signed integer) is negative.
160 (thanks to Peter van Eynde)
162 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
163 * new platform: experimental support for the Windows operating
164 system has been added. (thanks to Alastair Bridgewater)
165 * fixed several bugs in and robustified the PPC FFI (including
166 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
167 * bug fix: allow non-simple string symbol names (reported by Paul
169 * bug fix: interrupt handling on NetBSD (thanks to Richard M
171 * bug fix: saving a core corrupted callbacks on x86/x86-64
172 * bug fix: closed a loophole in metacircularity detection and
173 grounding in the PCL implementation of CLOS.
174 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
175 (reported by tomppa on #lisp)
176 * bug fix: clear the direction flag on Lisp -> C transitions, as
177 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
178 (reported by Andrej Grozin and Hendrik Maryns)
179 * optimization: major improvements to GC efficiency on GENCGC platforms
180 * optimization: faster implementation of EQUAL
181 * optimization: emit more efficient opcodes for some common
182 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
183 * optimization: several other minor code-generation improvements on x86-64
184 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
186 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
187 * minor incompatible change: (SETF CLASS-NAME) and (SETF
188 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
189 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
190 by ANSI as a generic function, and for consistency (SETF
191 GENERIC-FUNCTION-NAME) is treated likewise.
192 * fixed bug #233.b: make constraint propagation notice when a variable
193 value is changed after it is referenced but before it is used
194 * fixed bug #296: no more arbitrary behaviour from filenames with
195 odd characters as --load arguments. (reported by Adam Warner)
196 * fixed bug #390: :CHARACTER-SET pathname components now work as
197 expected. (reported by Tim Daly Jr)
198 * fixed bug #391: complicated :TYPE intersections in slot
199 definitions no longer cause an error in PCL internals.
200 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
201 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
202 returning the number of octets which would be written to the
203 file-stream. (thanks to Robert J. Macomber)
204 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
205 arguments to the internal stream functions. (thanks to David
207 * bug fix: stack exhaustion is now handled on the x86/SunOS
208 platform. (thanks to Daisuke Homma)
209 * bug fix: more accurate ROOM results on GENCGC platforms
210 * optimization: improved type inference for arithmetic-for
211 index variables in LOOP
212 * optimization: faster floating-point SQRT on x86-64
213 * fixed some bugs revealed by Paul Dietz' test suite:
214 ** DOCUMENTATION returns NIL instead of "" for method combinations
215 that don't have a docstring
217 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
218 * minor incompatible change: (SETF CLASS-NAME) and (SETF
219 GENERIC-FUNCTION-NAME) are no longer generic functions, and
220 therefore are not customizeable by user code (as seems to be at
221 least permitted and maybe required by AMOP). As a consolation,
222 however, the SBCL implementation of these functions now calls
223 REINITIALIZE-INSTANCE as specified by AMOP.
224 * bug fix: slot-definition documentation is propagated as per ANSI
225 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
226 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
227 argument, as do eslotd initialization methods. (from Pascal
228 Costanza's "Closer" project)
229 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
230 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
232 * bug fix: it is now possible to have more than one subclass of
233 STANDARD-GENERIC-FUNCTION without causing stack overflow.
234 (reported by Bruno Haible, Pascal Costanza and others)
235 * bug fix: the dependent update protocol now works for generic
236 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
238 * bug fix: condition-class instances corresponding to
239 DEFINE-CONDITION forms are now created eagerly. (reported by
240 Kalle Olavi Niemitalo on comp.lang.lisp)
241 * bug fix: floating point printing is more accurate in some
242 circumstances. (thanks to Simon Alexander)
243 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
244 merged with *DEFAULT-PATHNAME-DEFAULTS*.
245 * bug fix: callbacks on OS X now preserve stack-alignment.
246 (thanks to Cyrus Harmon)
247 * enhancement: the x86-64 disassembler is much better at
248 disassembling SSE instructions. (thanks to Lutz Euler)
249 * enhancement: CHECK-TYPE on a variable now allows type-inference
250 (thanks to Vincent Arkesteijn)
251 * optimization: improved performance of EUC-JP external format.
252 (thanks to NIIMI Satoshi)
253 * optimization: performance improvements to IO on file streams of
254 :ELEMENT-TYPE CHARACTER
255 * optimization: much faster memory allocation on x86-64
256 * optimization: faster garbage collections (latency and throughput)
258 * optimization: faster bignum subtraction and division on x86 and x86-64
259 (thanks to Lutz Euler)
260 * optimization: more accurate type-inference for loops that use a
261 floating point index variable or a negative step.
263 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
264 * bug fix: add a workaround to SBCL looping infinitely at startup on
265 Linux kernels with apparently buggy implementations of personality().
266 (thanks to Svein Ove Aas)
267 * bug fix: Unicode symbols are correctly printed in LDB backtraces
268 (thanks to David Lichteblau)
269 * bug fix: local bindings shadow global symbol macros for
270 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
271 * optimization: non-open coded uses of numeric comparison operators
272 (e.g. >) no longer cons when called with more than one parameter
273 on platforms supporting dynamic-extent allocation.
274 * enhancement: saving cores with foreign code loaded is now supported
275 on MIPS/Linux in addition to the previously supported platforms.
276 * bug fix: FIND-RESTART now tests for activity, not applicability when given
277 a restart object as identifier. (reported by Helmut Eller for CMUCL)
278 * bug fix: division by zero in sb-sprof when no samples were collected
279 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
280 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
281 (thanks to Alexander Kjeldaas)
282 * new feature: ignoring the return values of destructive functions
283 like NREVERSE signals a compile-time style-warning.
284 (thanks to Kevin Reid)
285 * new feature: the alignment of alien structure fields can be
286 explicitly specified. (thanks to Cyrus Harmon)
287 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
288 and dump core on SIGQUIT
290 ** incompatible change: threads do not inherit values of specials
291 from their parents (see manual)
292 ** bug fix: threads stacks belonging to dead threads are freed by the
293 next exiting thread, no need to gc to collect thread stacks anymore
294 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
295 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
296 ** bug fix: dynamic variable and thread start related gc lossage
297 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
299 * fixed some bugs revealed by Paul Dietz' test suite:
300 ** SUBTYPEP is slightly more accurate on heinously complicated
301 CONS types where some of the members have uncertain (in the
302 NIL, NIL sense) type relationships to each other.
304 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
305 no longer purified unless :PURIFY T is explicitly specified.
306 ** Non-purified cores are significantly smaller than before
308 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
309 * new feature: timers based on Zach Beane's excellent timer package
310 * added support for the following external formats: koi8-u,
311 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
312 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
313 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
314 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
315 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
316 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
317 * incompatible change: a threaded SBCL will no longer revert to
318 non-threaded mode on non-NPTL systems, but refuse to start entirely.
319 * bug fix: interrupts are disabled until startup is complete; no
320 more sigsegvs when receiving a signal to soon
321 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
323 * bug fix: add a workaround for the memory randomization features in
324 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
325 workaround will only be in effect on systems with the proc filesystem
327 * bug fix: printing objects of type HASH-TABLE signals a
328 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
330 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
331 have been running for over 50 days. (reported by Gilbert Baumann)
332 * bug fix: the logic for getting names of functions gets less
333 confused when confronded with alternate-metaclass
334 funcallable-instances. (reported by Cyrus Harmon)
335 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
336 consistent, even on internal alternate-metaclass objects.
337 * bug fix: flush closure information collected by physical
338 environment analysis prepass before the main pass. (bug reported
340 * bug fix: compiler pack phase does not modify a hashtable, which is
341 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
342 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
343 may be expanded. (reported by Alexander <malishev> on c.l.l)
344 * bug fix: no more style-warnings when compiling code using LAST
345 under high-SPEED low-SPACE settings. (reported by David Wallin
347 * fixed some incompatibilities between SBCL's MOP and the MOP
349 ** the METAOBJECT class is now implemented;
350 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
351 STANDARD-OBJECT, as required;
352 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
353 now compatible (as required by VALIDATE-SUPERCLASS); there
354 remains a consistency requirement over the presence or absence
355 of the FUNCTION class in the superclass list of finalized
356 classes; see the manual for more details;
357 ** the :AROUND method for COMPUTE-SLOTS on
358 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
359 requested slot ordering.
361 ** bug fix: parent thread now can be gc'ed even with a live
363 ** bug fix: binding a special with PROGV to no value is not
364 the same as that symbol not having been bound (thanks to
366 ** bug fix: binding specials is thread safe (thanks to
368 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
369 and SB-SPROF do not die with 'no handler for signal XX in
370 interrupt_handle_now(..)' anymore
371 ** bug fix: WITH-TIMEOUT works with multiple running threads
372 * fixed some bugs revealed by Paul Dietz' test suite:
373 ** ENSURE-GENERIC-FUNCTION should take a method class object for
374 the :method-class keyword argument.
376 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
377 * new port: the Solaris operating system on x86 processors is now
378 mostly supported, though some rough edges in the environment
379 remain. (thanks to Daisuke Homma)
380 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
381 space for its heap. (thanks to Thiemo Seufer)
382 * optimization: SBCL on MIPS platforms now supports dynamic-extent
383 closures. (thanks to Thiemo Seufer)
384 * minor incompatible change: eof selects abort in the debugger.
385 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
386 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
388 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
389 deprecated, and will go away in a future revision of SBCL.
390 * minor incompatible change: GC-ON and GC-OFF are no longer
391 implemented with a counter, it does not matter how many times gc
392 is switched on or off
393 * bug fix: discriminating functions for generic function classes
394 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
395 make invalid assumptions about method precedence order. (reported
397 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
398 platforms. (thanks to Thiemo Seufer)
399 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
400 * bug fix: correct pathname printing with printer escaping is on.
401 (thanks to Kevin Reid)
402 * bug fix: complex VOP definitions in "user-space" no longer trigger
403 package locks. (reported by Zach Beane)
404 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
405 for funcallable-instance functions, and is no different from
407 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
408 characters in the current readtable. (reported by Nicholas Neuss)
409 * bug fix: SUBTYPEP on various CONS types returns more a more
410 accurate acknowledgment of its certainty.
411 * optimizations: REMOVE-DUPLICATES now runs in linear time on
412 lists in some cases. This partially fixes bug 384.
413 * flush all standard streams before prompting in the REPL and the
415 * bug fix: signal handling and triggering gc do not conflict
416 directly anymore, in particular a high frequency sb-sprof does
417 not prevent gc from running
418 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
419 approximation for timezone and DST information between the
420 universal time epoch and the smallest negative 32-bit time_t.
421 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
422 year 1899 so long as the encoded time is non-negative.
423 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
424 the correct answer for non-constant rotate amounts when performing
425 an inline 32-bit rotation.
427 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
428 there is only one thread in the session
429 ** bug fix: memory leak for streams created in one thread and
430 written to in another
431 ** bug fix: lockup when compiled with gcc4
432 ** bug fix: race that allows the gc to be triggered when gc is
434 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
435 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
437 ** bug fix: finished threads are gc'ed properly
438 * fixed some bugs revealed by Paul Dietz' test suite:
439 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
440 has arguments with a different set of applicable methods from
441 the orignal arguments.
442 ** The type error thrown by MAP now has a correct expected-type
444 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
445 name a compiled function.
446 ** Three MISC tests where a large bignum was improperly coerced to
447 a float that couldn't represent that bignum during type
448 derivation were fixed.
449 ** SUBTYPEP can now handle types involving the negation of a
450 list-form FUNCTION type.
451 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
452 as (COMPLEX (AND RATIO (NOT FIXNUM))).
453 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
455 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
456 * New feature: Experimental support for bivalent streams: streams
457 opened with :element-type :default now allow character and binary
458 (unsigned-byte 8) I/O
459 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
460 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
462 * bug fix: no more highly sporadic "couldn't check whether ~S is
463 readable" when reading a stream and an interrupt hits in the middle
464 of a select system call
465 * compiler better recognizes complex arrays (reported by Eduardo
467 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
468 for variables on SBCL built with linkage-tables. (reported by Luis
470 * various error reporting improvements.
471 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
472 (Thanks to Thiemo Seufer)
473 * bug fix: nested reader invokations work correctly
474 * bug fix: it is possible to have simultaneous references to foreign
475 code and foreign data with the same name.
477 ** added x86-64 support
478 ** incompatible change: the threading api now works with thread
479 objects instead of thread ids
480 ** bug fix: threads are protected from signals and interruption when
481 starting up or going down
482 ** bug fix: a race where an exiting thread could lose its stack to gc
483 ** bug fix: don't halt on infinite error in threads if possible
484 ** fixed numerous gc deadlocks introduced in the pthread merge
485 ** bug fix: fixed thread safety issues in read and print
486 ** bug fix: debugger doesn't hang on session lock if interrupted at
487 an inappropriate moment
488 ** bug fix: run-program is now thread safe(r)
489 ** bug fix: inner with-recursive-lock no longer releases the mutex
490 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
491 * fixed some bugs revealed by Paul Dietz' test suite:
492 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
493 EXPECTED-TYPE slots filled.
494 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
496 ** MAKE-CONDITION accepts classes as type-designators.
497 ** COMPILE may never return NIL.
498 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
499 range before calling Unix time functions
501 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
502 * numerous signal handling fixes to increase stability
503 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
504 * minor incompatible change: we now correctly canonize default
505 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
506 as per the MOP, rather than the historical (INITARG INITFUNCTION
507 INITFORM). (reported by Bruno Haible)
508 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
509 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
510 for more information.
511 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
512 pathname is a directory pathname.
513 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
514 * dynamic space size on PPC has been increased to 768Mb. (thanks to
516 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
517 :METACLASS argument in addition to a class name. (reported by
518 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
519 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
520 (reported by Bruno Haible; patch from Gerd Moellmann)
521 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
523 * bug fix: more cleanups to the floating point exception handling on
524 x86-64 (thanks to James Knight)
525 * bug fix: the compiler does not try to constant fold calls of
526 COERCE and BIT-* functions when they return freshly constructed
527 objects. (reported by Nikodemus Siivola and Paul Dietz)
528 * optimization: improved the MIPS versions of generic (in the
529 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
530 * optimization: direct conversion of (unsigned-byte 32) to floats on
531 the PowerPC platform.
532 * optimization: structure instances with raw slots now use less
533 memory, and probably show better memory locality. (thanks to
535 * optimization: DYNAMIC-EXTENT declarations for lists and closures
536 are treated as requests for stack allocation on the x86-64,
537 Alpha-32, PPC and SPARC platforms.
538 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
539 the wrong file descriptor; implementation of SOCKET-OPEN-P.
540 (thanks to Tony Martinez)
542 ** gcing a dead thread can no longer lead to lockups
543 ** threads block signals until they are set up properly
544 ** errno is no longer shared by threads
545 ** interrupt-thread restores the eflags register on x86
546 ** fixed some lockups due to gc/thread interaction
547 * fixed some bugs revealed by Paul Dietz' test suite:
548 ** invalid dotted lists no longer raise a read error when
550 ** signal an error if a symbol that names a declaration is used
551 as the name of a type, or vice versa
552 ** allow using the (declare (typespec var*)) abbreviation for
553 (declare (type typespec var*)) with all type specifiers
554 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
555 ** always return NIL from PPRINT-POP when OBJECT is NIL
556 ** don't signal errors when pretty-printing malformed LABELS,
557 FLET or MACROLET forms
558 ** declarations in a DOLIST body are also in scope for the
560 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
562 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
565 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
566 * fixed cross-compiler leakages that prevented building a 32-bit
567 target with a 64-bit host compiler.
568 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
569 opened with :IF-EXISTS :OVERWRITE.
570 * fixed bug 281: error for an invalid qualifier in a short-form method
571 combination method is not signalled until the faulty method is called.
572 * bug fix: iteration variable type inferrer failed to deal with open
573 intervals. (reported by Alan Shields)
574 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
575 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
576 or more runtime options were provided to the sbcl binary.
577 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
578 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
579 * optimization: there's now a fast-path for fixnum arguments in the
580 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
581 * optimization: the code generated on x86-64 is more compact thanks
582 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
583 * bug fix: floating-point exception handling now partly works on
584 x86-64. (Thanks to James Knight)
585 * improvement to the MIPS backend from Thiemo Seufer: C-style
586 64-bit long long arguments and return values to alien functions
588 * fixed some bugs revealed by Paul Dietz' test suite:
589 ** the type-error signalled from WARN has a filled-in DATUM slot.
590 ** the type-error required when a stream is not associated with
591 a file has the stream as its datum.
592 ** type-errors on single-floats on x86-64 no longer have
593 :INVALID-OBJECT as the datum
594 ** the type-errors signalled for invalid function names now have
595 a correct expected type
596 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
597 for structures defined with a :TYPE. Documentation strings for
598 typed structures are no longer immediately discarded
599 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
600 broadcast streams and synonym streams. FILE-LENGTH now also works
601 on broadcast streams.
603 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
604 * incompatible change: the --noprogrammer option, deprecated since
605 version 0.7.5, has been removed. Please use the equivalent
606 --disable-debugger option instead.
607 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
609 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
610 inoperational for a while now) has been completely removed.
611 * null lexical environments are now printed as #<NULL-LEXENV>,
612 significantly reducing the amount of clutter in typical backtraces.
613 * documentation on weak pointers, finalization, and after GC hooks
614 has been added to the manual.
615 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
616 as other specialized array types, is much faster. SUBSEQ and
617 COPY-SEQ on such arrays have also been sped up.
618 * optimization: EQL is now more efficient when at least other argument
619 is known to be of type (OR FIXNUM (NOT NUMBER)).
620 * fixed bug: the runtime is now less vulnerable to changes in the
621 size of the SBCL object on OS X, and virtual memory is reserved for
622 all spaces that need to be at a fixed address.
623 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
624 * fixed bug: finalizers and after GC hooks that cause consing are now
626 * fixed bug: compiler error messages and summaries are now printed to
627 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
628 * fixed inference of the upper bound of an iteration variable.
629 (reported by Rajat Datta).
630 * fixed bug 211e: calling local functions with duplicated constant
631 keyword argument no longer causes a bogus style warning about an
633 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
634 ftype declaration to be disregarded. (reported by Dave Roberts)
635 * fixed bug 373: caused by erronous compilation of references to alien
636 variables in the runtime on ppc/darwin.
637 * fixed bug 376: CONJUGATE type deriver.
638 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
639 (reported by Baughn on #lisp)
640 * a cross-compiler bug on non-x86 platforms has been identified and
641 fixed. (thanks to Bruno Haible)
642 * improvements to the MIPS runtime code for increased stability.
643 (thanks to Thiemo Seufer)
644 * increased the maximimum compact environment size to allow
645 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
646 * partial workaround for bug 135: don't name the function we're
647 calling for hairy cases of EVAL, so as not to accumulate one environment
648 entry per call to EVAL. (reported by Kevin Reid)
649 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
650 * optimization: type testing for non-vector arrays should be faster.
651 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
653 * bug fix: setting 31st element of a bit vector to zero did not work
655 * bug fix: redefining a class definition which failed due to a
656 previous accessor / function clash now works (but see BUGS entry
657 #380 for more problems in this area). (thanks to Zach Beane)
658 * the long form of DEFINE-METHOD-COMBINATION disables method group
659 checking when given a single method group with pattern *. (thanks
661 * on x86 compiler supports stack allocation of results of simple
662 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
663 * support for building and running on Mac OS X 10.4 "Tiger" has been added
664 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
665 * fixed some bugs related to Unicode integration:
666 ** the restarts for recovering from input and output encoding
667 errors only appear when there is in fact such an error to
669 * fixed some bugs revealed by Paul Dietz' test suite:
670 ** MISC.549 and similar: late transformation of unsafe type
671 assertions into derived types caused unexpected code
673 ** SCALE-FLOAT type deriver is less wrong.
674 ** type derivers for EXP, LOG and similar functions compute result
675 types for complex arguments better.
676 ** (MISC.563) CONJUGATE type deriver works for very restricted
678 ** out-of-line type testers for character strings are available.
679 ** EQUAL compiler transform understands specialness of objects
681 ** accessing double-floats stored on the stack now works on x86-64.
682 ** debugger internals could sometimes create invalid lispobjs,
683 resulting in GC crashes.
684 ** MISC.548: type check weakening can convert required type into
686 ** initialization forms for bindings are not in scope of free special
689 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
690 * incompatible change: thread support for non-NPTL systems has
691 been removed - locking is buggy and unreliable. A threaded
692 SBCL build will now warn at startup and refuse to create
693 new threads, unless futex support is detected in the kernel
694 * incompatible change: the top level REPL now has only an ABORT
695 restart associated with it, not TOPLEVEL and ABORT as it used to.
696 TOP and TOPLEVEL are now available as debugger commands for
697 returning to the top level.
698 * incompatible change: forms evaluated in the REPL now use the
699 global optimization policy.
700 * incompatible change: user- and system-initialization files are
701 no longer processed with LOAD, but by READ and EVAL; hence the
702 global optimization policy, startup package, readtable, etc,
704 * The .fasl file format number has been incremented because of
705 various incompatible changes.
706 * internal entry point details and argument counts no longer appear
707 in backtraces unless explicitly requested by setting
708 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
709 * built-in and standard functions no longer have names like "top
710 level local call to FOO".
711 * fixed bug 32: functions defined in non-null lexical environments
712 now have more legible printed representation
713 * fixed bug 33: functions defined in non-null lexical environemnts
714 are now more amenable to inspection by INSPECT.
715 * workaround for bug 354: XEPs no longer appear in backtraces unless
716 explicitly requested.
717 * fixed bug: receiving the signal which results from attempting to
718 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
719 SIGBUS on other *BSDs) on architectures where the C stack is also
720 the Lisp stack (x86 and x86-64) from foreign code no longer leads
721 to debugger confusion or wild instability. (reported by Cheuksan
723 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
724 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
725 (reported by Lutz Euler)
726 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
727 compiled code. (reported by Paul Dietz)
728 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
729 specialized parameter to a method to something that is not TYPEP
730 the specializer is now possible.
731 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
732 face of package deletion.
733 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
734 pathname. (reported by David Tolpin on c.l.l)
735 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
736 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
737 STRING-STREAM and FILE-STREAM, respectively.
738 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
739 than previously; in addition, its test suite is now run on build.
740 (thanks to Luke Gorrie)
741 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
742 * optimization: added a immediate representation for single-floats
744 * fixed some bugs revealed by Paul Dietz' test suite:
745 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
746 ** attempting to create a package with a colliding nickname causes
747 correctable errors to be signalled.
748 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
749 ** several x86-64 backend bugs related to sign-extension of immediate
752 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
753 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
754 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
756 * fixed loading of multiply forward-referenced layouts.
757 (thanks to Cheuksan Wang)
758 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
759 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
760 (see COMPILE-FILE documentation for details of :PRINT option)
761 * fixed bugs 19 and 317: fixed-format floating point printing is
762 more accurate. This also fixes a bug reported by Adam Warner
763 related to the ~@F format directive.
764 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
766 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
767 dispatch table if the from-char is a dispatch macro character.
768 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
769 by the wordsize no longer ignore the last word. (reported by Lutz
771 * fixed bug in type checking of dynamic-extent variables. (reported
773 * optimization: sequence traversal functions use their freedom to
774 coerce function designators to functions.
775 * optimization: code with many calls to CLOS methods specialized on
776 CLOS classes has had redundant type checks removed.
777 * optimization: added declarations to speed up operations that access
778 the internal character database (for example STRING-UPCASE)
779 * optimization: comparison operations between floats and sufficiently small
780 fixnums no longer create extra rationals
781 * fixed some bugs related to Unicode integration:
782 ** portions of multibyte characters at the end of buffers for
783 character-based file input are correctly transferred to the
784 start of the buffer at the next read.
785 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
786 passing it through to OPEN.
787 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
788 argument given to internal calls to OPEN.
789 ** the built-in comment readers (introduced by character sequences
790 ";" and "#|") are more forgiving to encoding errors; they will
791 STYLE-WARN and then attempt to resync the stream at a character
792 boundary. (thanks to Teemu Kalvas)
793 * fixed some bugs in the x86-64 port:
794 ** Negative short int return values from c-calls are sign-extended
796 ** The stack is aligned to 16-bytes for c-calls, as required by
797 the ABI. (thanks to Cheuksan Wang)
798 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
799 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
800 for enums. (thanks to Vincent Arkesteijn)
801 ** Multiple small optimizations and bugfixes for floating point
803 * fixed some bugs revealed by Paul Dietz' test suite:
804 ** Space, Tab, Linefeed, Return and Page have the invalid
805 secondary constituent character trait.
806 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
808 ** WITH-INPUT-FROM-STRING should only update the index place on
810 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
812 ** Bit-array operations (BIT-AND and similar) worked incorrectly
813 with one-dimensional arrays with fill pointers.
814 ** TYPE-OF failed on a complex with an integer realpart and a
816 ** compiler failure during type inference for the code of form
817 (IF (EQL X (THE ...)) ...) (MISC.535).
819 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
820 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
821 platform. The port supports 61-bit fixnums, large memory spaces
822 and reloading shared object files.
823 * enhancement: saving cores with foreign code loaded is now
824 supported on ppc/Darwin in addition to the previously supported
826 * enhancement: the statistical profiler now walks deeper into the
827 call stack for more meaningful call-graphs and accrued time
828 reports (x86/x86-64 only). It also now reports time spent in
830 * enhancement: it is now possible to trace most individual methods
831 of a generic function in addition to tracing the generic function
833 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
835 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
836 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
837 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
838 produces an error. (thanks to Vincent Arkesteijn)
839 * bug fix: NAMESTRING on pathnames with :WILD components in their
840 directories works correctly. (thanks to Artem V. Andreev)
841 * fixed bug 125: compiler preserves identity of closures. (reported
843 * bug fixed: functions with &REST arguments sometimes failed with
844 "Undefined function" when compiled with (DEBUG 3). (reported by
846 * bug fix: overflow during compiling of setting element of a bit
847 vector with constant index and value. (reported by Timmy Douglas)
848 * build fix: fixed the dependence on *LOAD-PATHNAME* and
849 *COMPILE-FILE-PATHNAME* being absolute pathnames.
850 * on x86 compiler partially supports stack allocation of dynamic-extent
852 * GO and RETURN-FROM do not check the extent of their exit points
853 when compiled with SAFETY 0.
854 * fixed some bugs related to Unicode integration:
855 ** encoding and decoding errors are now much more robustly
856 handled; it should now be possible to recover even from invalid
857 input or output to the terminal. (thanks to Teemu Kalvas)
858 ** provided a first cut at implementing STRING-TO-OCTETS and
859 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
860 ** altered the SB-MD5 contributed module slightly, changing the
861 interface just enough for it to be supportable for builds where
862 lisp characters are not eight bits.
863 * fixed some bugs revealed by Paul Dietz' test suite:
864 ** the FORMATTER-generated functions for ~V[ conditionals require
865 the correct number of arguments.
866 ** READ-FROM-STRING returns the mandated second value when applied
867 to displaced strings.
868 ** the #\Rubout and #\Backspace characters are treated as invalid
869 constituent characters by the tokenizer.
871 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
872 * new feature: reloading changed shared object files with
873 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
874 * new feature: references to foreign variables and functions
875 can now be compiled and loaded before the corresponding shared
876 object file is loaded, as long as the foreign definitions are
877 available at runtime.
878 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
879 supported. (thanks to Dan Debertin)
880 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
881 just keywords. (thanks to Vincent Arkesteijn)
882 * fixed bug #331: structure-class instances corresponding to
883 DEFSTRUCT forms are now created eagerly.
884 * fixed bug #345: backtraces from calls to undefined functions work
885 on x86 as well. Related bug #61 is now also partially fixed on x86
886 (backtraces from throws to unknown catch tags.)
887 * bug fix: lambda-list parsing is now stricter vrt. order and number
888 of lambda-list keywords.
889 * bug fix: as specified by AMOP, an error is signalled if a
890 class-option appears multiple times in a DEFCLASS form. (reported
892 * bug fix: RANDOM can be compiled when the compiler derives the type
893 of its numeric argument as a disjoint set of small integers.
894 (reported by Paul Dietz)
895 * bug fix: starting a core saved with shared objects loaded when
896 those objects are not available no longer causes threaded SBCL to
897 hang. (reported by Sean Ross)
898 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
900 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
901 argument, which must always be NIL. (reported by Kalle Niemitalo)
902 * bug fix: printing 1.0d+23 no longer results in an error.
903 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
904 * bug fix: structure slot setters preserve evaluation order. (thanks
906 * fixed some bugs related to Unicode integration:
907 ** RUN-PROGRAM can allow its child to take input from a Lisp
908 stream. (reported by Stefan Scholl)
909 ** ASDF-INSTALL successfully downloads .tar.gz files in all
910 locales. (reported by Ken Causey)
911 * fixed some bugs revealed by Paul Dietz' test suite:
912 ** INCF, DECF and REMF evaluate their place form as specified in
914 ** FORMATTER expands ~{ ~} iteration directives with V or #
915 parameters correctly.
916 ** FORMATTER deals with the ~@[ ~] conditional directive where the
917 consequent uses no arguments correctly.
918 ** the system has a partial understanding of the (COMPLEX RATIO)
920 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
923 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
924 * new feature: a build-time option (controlled by the :SB-UNICODE
925 keyword feature, enabled by default) for building the system with
926 support for the entire 21-bit character space defined by the
928 * new feature: the system now has rudimentary external-format
929 support; the primary user-visible change at this time is that
930 characters with the high bit set (such as lower-case-e-acute) will
931 print correctly to a terminal in a UTF-8 environment.
932 * minor incompatible change: BASE-CHAR no longer names a class;
933 however, CHARACTER continues to do so, as required by ANSI.
934 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
935 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
936 has been moved to the SB-EXT package (temporarily re-exported from
938 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
939 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
941 * fixed bug #351: better error-handlind and reporting for malformed
943 * fixed bug #350: bignum-printing is now more memory-efficient,
944 allowing printing of very large bignums, eg. (expt 2 10000000).
945 (reported by Bruno Haible)
946 * fixed bug #302: better primitive-type selection for intersection
948 * fixed bug #308: non-graphic characters now all have names, as
949 required. (reported by Bruno Haible)
950 * bug fix: redefining a class with different superclasses now correctly
951 removes it from the direct-subclasses of its previous superclasses.
952 (reported by David Morse)
953 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
954 new value now works. (reported by Bruno Haible)
955 * bug fix: correct canonicalization of multiple non-standard slot
956 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
957 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
958 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
959 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
960 now exists, an signals an error.
961 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
962 during macroexpansion for non-iteration variables as well. (reported
963 by Bruno Haible for CMUCL)
964 * bug fix: Cyclic structures and unprintable objects in compiler
965 messages no longer cause errors. (reported by Bruno Haible)
966 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
967 and READ-FROM-STRING all now return a primary value of NIL if
968 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
969 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
970 (reported by Bruno Haible for CMUCL)
971 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
972 do not designate a condition. (reported by Bruno Haible for
974 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
975 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
976 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
977 permitting resolution in favour of any of the conflicting symbols.
978 (reported by Bruno Haible for CMUCL)
979 * bug fix: EQUAL compiler optimizations is less aggressive on
980 strings which can potentially compare true despite having distinct
981 specialized array element types.
982 * bug fix: unit enumerations can be defined without dividing by
983 zero. (thanks to Vincent Arkesteijn)
984 * FORMAT compile-time argument count checking has been enhanced.
985 (report from Bruno Haible for CMUCL)
986 * a partial workaround for the bug 262: the compiler does not try to
987 inline-expand a local function doing RETURN-FROM from a deleted
988 BLOCK. (thanks to Peter Denno for the bug report and to David
989 Wragg for the simple test case)
990 * fixed some bugs revealed by Paul Dietz' test suite:
991 ** RENAME-PACKAGE allows all package designators as new package
993 ** constraint propagation and lambda variable substitution are
994 more cautious in dealing with partially deleted code.
995 ** compiler sometimes lost reoptimization passes.
996 ** CERROR, given a condition as condition designator, passes its
997 remaining arguments to the continue format control without
999 ** Case-altering FORMAT directives work correctly on non-ASCII
1001 ** The REAL type specifier handles bounds outside the floating
1002 point ranges without signalling FLOATING-POINT-OVERFLOW.
1003 ** Functions with IR1-transformations can create intercomponent
1004 references to global functions.
1005 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1007 ** FORMAT ~R treats a nil value for its first parameter correctly.
1009 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1010 * enhancement: saving cores with foreign code loaded is now
1011 supported on x86/NetBSD and sparc/Linux in addition to the previously
1012 supported platforms.
1013 * bug fix: on some platforms repeated installations caused multiple
1014 copies of HTML documentation to be installed -- should not happen
1015 any more. (reported by Stefan Scholl)
1016 * bug fix: parsing self-recursive alien record types multiple times
1017 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1018 original patch by Helmut Eller for CMUCL)
1019 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1020 (thanks to Richard Kreuter)
1021 * bug fix: defining classes whose accessors are methods on existing
1022 generic functions in other (locked) packages no longer signals
1023 bogus package lock violations. (reported by François-René Rideau)
1024 * bug fix: special variables as DEFMETHOD parameters no longer have
1025 associated bogus type declarations. (reported by David Wragg and
1027 * bug fix: read-write consistency on streams of element-type
1028 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1029 * bug fix: redefiniton of the only method of a generic function with
1030 no DEFGENERIC no longer emits a full WARNING. In addition,
1031 redefinition of generic functions with no DEFGENERIC to an
1032 incompatible lambda list now signals an error. (thanks to Zach
1034 * bug fix: DEFGENERIC now works even when there's a function of the
1035 same name in an enclosing lexical environment. (thanks to Zach
1037 * fixed compiler failure, caused by instrumenting code during
1038 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1039 * optimization: added loop analysis and improved register allocation
1040 * fixed some bugs revealed by Paul Dietz' test suite:
1041 ** POSITION on displaced vectors with non-zero displacement
1042 returns the right answer.
1043 ** (SIMPLE-STRING) is a valid type specifier for sequence
1045 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1047 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1048 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1050 ** The justification version of the FORMAT ~< directive treats
1051 non-zero minpad parameter correctly.
1053 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1054 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1055 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1056 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1057 the supported interface.
1058 * new feature: Single-stepping of code compiled with DEBUG 2 or
1059 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1060 * new feature: saving cores with foreign code loaded is now
1061 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1062 Timothy Moore's work for CMUCL)
1063 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1064 parameters to * instead of NIL if no initform is supplied.
1065 (reported by Johan Bockgaard)
1066 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1067 correctly when FUNCALL appears as the car of the form. Note:
1068 despite this FUNCALL forms are not currently subject to
1069 compiler-macro expansion. (port of Raymond Toy's fix for the
1070 same from CMUCL, reported by Johan Bockgaard)
1071 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1072 (thanks for Teemu Kalvas)
1073 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1074 case instead of terminating the iteration (thanks for Julian
1075 Squires, Sean Champ and Raymond Toy)
1076 * bug fix: incorrect expansion of defgeneric that caused a style
1077 warning. (thanks for Zach Beane)
1078 * bug fix: slot accessor effective method computation works properly
1079 for classes with multiple non-standard applicable methods on
1080 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1081 * on x86 compiler supports stack allocation of results of LIST and
1082 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1083 CMUCL implementation by Gerd Moellmann)
1084 * fixed some bugs revealed by Paul Dietz' test suite:
1085 ** FORMAT strings with both the ~<~:;~> form of the justification
1086 directive and pretty-printing directives cause an error.
1088 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1089 * incompatible change: the internal functions
1090 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1091 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1092 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1093 instead of the old functions.
1094 * new feature: on platforms where "dladdr" is available foreign
1095 function names now appear in backtraces. (based on Helmut Eller's
1097 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1098 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1100 * documentation: ASDF manual is now distributed with SBCL.
1101 * bug fix: SBCL can now load its contributed modules with REQUIRE
1102 even if the system-provided entries have been removed from
1103 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1105 * bug fix: backtraces involving undefined functions or assembly
1106 routines are more informative. (thanks to Brian Downing)
1107 * bug fix: mutually referent alien structures now work correctly.
1108 (reported by Rick Taube)
1109 * bug fix: structures defined by WITH-ALIEN can be referred to
1110 within other definitions in the same WITH-ALIEN.
1111 * bug fix: division operators (MOD, TRUNCATE and the like) with
1112 constant zero divisors and integer dividends no longer generate
1114 * bug fix: provide default methods for INPUT-STREAM-P and
1115 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1116 * bug fix: improve the detection and resolution of MOP metacycles
1117 (where effective method computation depends on the generic function
1118 having its effective method computed). (reported by Bruno Haible)
1119 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1120 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1121 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1122 * bug fix: pathnames with relative directory components can be
1123 represented relative to default pathnames.
1124 * optimization: in taking the GCD of bignums, reduce the two bignums
1125 to approximately the same size (using Euclid's algorithm) before
1126 applying the more sophisticated binary GCD. (thanks to Juho
1128 * optimization: COUNT on bitvectors now operates word-at-a-time.
1129 * optimization: ASH with a positive, but not necessarily constant,
1130 (leftwards) shift, when in a modular context, is compiled to a
1132 * fixed some bugs revealed by Paul Dietz' test suite:
1133 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1134 if the corresponding argument is NIL.
1135 ** FORMAT directives accepting a minpad parameter treat negative
1137 ** literal commas as character parameters to FORMAT directives are
1139 ** literal spaces directly after ~<Newline> directives within a
1140 format-logical-block (~:< ~@:>) do not induce :FILL-style
1141 conditional newlines.
1142 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1143 as specified: it no longer includes conditional newlines).
1144 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1146 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1147 :PER-LINE-PREFIX argument does not evaluate to a string.
1149 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1150 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1151 the manual for details; for now, package locks can be disabled by
1152 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1153 if no major problems are found then it is likely that they will be
1154 compiled in unconditionally.
1155 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1156 unsupported operators on all platforms. To load a shared library
1157 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1158 object file, link it into a shared library outside of SBCL and
1159 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1161 * bug workaround: ROOM T and the GC were not working together
1162 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1163 reporting facility aren't preserved. That reporting has been
1164 disabled, so now until and unless someone figures out how to make it
1165 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1166 * minor incompatible change: as threatened around sbcl-0.8.0, the
1167 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1168 The symbols are also exported from SB-PCL for backwards
1169 compatibility, but more so than before SB-PCL should be treated as
1170 an implementation-internal package.
1171 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1173 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1174 It is known as of this release not to work on the Alpha, however.
1175 * fixed bug #167: errors signalled due to illegal syntax in method
1176 bodies are now more legible.
1177 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1178 designators and can hence be used with TYPEP.
1179 * fixed bug #333: CHECK-TYPE now ensures that the type error
1180 signalled, if any, has the right object to be accessed by
1181 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1182 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1183 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1185 * fixed bug #334: programmatic addition of slots using specialized
1186 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1187 and :ALLOCATION :CLASS slots.
1188 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1189 float exponents. (rereported by Peter Seibel)
1190 * fixed a bug: #\Space (and other whitespace characters) are no
1191 longer considered to be macro characters in standard syntax by
1192 GET-MACRO-CHARACTER.
1193 * fixed bug: initialization of condition class metaobjects no longer
1194 causes an instance of the condition to be created. (reported by
1196 * fixed bug: it is now possible to have slots such that
1197 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1198 description is neither :INSTANCE nor :CLASS.
1199 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1200 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1201 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1202 non-standard SLOT-DEFINITION-ALLOCATION values.
1203 * fixed bugs: various race conditions handling exiting threads.
1204 CL-PPCRE's thread torture test now passes.
1205 * fixed arguably-a-bug: GC time proportional to number of threads in
1206 system even when most of them are idle
1207 * optimization: improved performance of BIT and SBIT on bit-vectors.
1208 * .fasl file incompatibility: The fasl file version number has
1209 been incremented because of changes associated with package locks.
1211 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1212 * minor incompatible change: the system no longer provides
1213 optimization hints (conditions which are TYPEP
1214 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1215 mode; these hints are still emitted when higher SPEED optimization
1217 * new contrib module: a sampling profiler (profiling by statistical
1218 sampling, rather than by instrumenting functions) is available as
1219 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1220 * the behaviour of the standard function ED is now customizeable by
1221 third parties through a hook variable: see ED's documentation
1222 string for information on the protocol.
1223 * the compiler no longer emits efficiency notes for (FUNCALL X)
1224 when the type of X is uncertain under default optimization
1226 * fixed bug 276: mutating a binding of a specialized parameter to a
1227 method to something that is not TYPEP the specializer is now
1229 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1230 exhibit the required behaviour on the x86 platform. (thanks to
1231 Peter van Eynde, Eric Marsden and Bruno Haible)
1232 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1233 even for difficult arguments. (reported by Peter Graves)
1234 * fixed bug 141a: the backquote printer now descends quoted
1236 * fixed another bug in backquote printing: no more destructive
1237 modification of the form's list structure. (reported by Brian
1239 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1240 move between its address being taken and the call to
1241 interrupt_thread, fixing a crashing race condition.
1242 * the SB-POSIX contrib implementation has been adjusted so that it
1243 no longer exhibits ridiculously poor performance when constructing
1244 instances corresponding to C structs.
1246 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1247 * minor incompatible change: the sb-grovel contrib now treats C
1248 structures as alien (in the sense of SB-ALIEN) objects rather than
1249 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1250 has implications for memory management of client code
1251 (sb-grovel-returned objects must now be manually managed) and for
1252 type safety (alien objects now have full types).
1253 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1254 used to control emission of compiler diagnostics, rather than the
1255 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1256 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1257 quality should be considered deprecated.
1258 * install.sh now installs the user manual as well
1259 * (not quite a new documentable feature, but worth considering in
1260 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1261 beginnings of a semantically meaningful condition hierarchy is
1262 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1264 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1265 designator as the defaults argument.
1266 * fixed bug: Displaced arrays whose displaced-to array has become
1267 too small now cause ARRAY-DIMENSION to signal an error, providing
1268 for safer bounds-checking. (reported by Bruno Haible)
1269 * fixed bug: DEFCLASS slot definitions with identical :READER and
1270 :WRITER names now signal a reasonable error. (reported by Thomas
1272 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1274 * fixed bug: Invalid slot specification errors now print correctly.
1275 (thanks to Zach Beane)
1276 * fixed bug 320: Shared to local slot value transfers in class
1277 redefinitions now happen corrently from superclasses as
1278 well. (reported by Bruno Haible)
1279 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1281 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1282 improper lists correctly. (reported by Bruno Haible)
1283 * fixed bug 313: source-transform for <fun-name> was erroneously
1284 applied to a call of a value of a variable with name <fun-name>.
1285 (reported by Antonio Menezes Leitao)
1286 * fixed bug 307: The obsolete instance protocol ensures that
1287 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1288 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1289 remove dead unknown-values globs from the middle of the stack.
1290 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1291 easier to resume long computations after using *BREAK-ON-SIGNALS*
1292 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1293 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1294 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1295 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1296 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1297 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1299 * fixed a (fairly theoretical) bug in string printing: if
1300 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1301 does not have array-element-type equal to the most general string
1303 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1304 function name. (thanks to Nikodemus Siivola)
1305 * fixed bug: compile-time format string checker failed on
1306 non-closed ~{. (reported by Thomas F Burdick)
1307 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1308 don't warn on legal loop code involving more than one aggregate
1309 boolean. (thanks to Nikodemus Siivola)
1310 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1311 in keyword position to MAKE-PACKAGE was not being reported as
1313 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1314 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1316 * fixed bugs 280 and 312: the checking for multiple definitions in a
1317 file is less likely to become confused by uses of inline
1319 * fixed bug: the #S reader macro performs the keyword coercion
1320 specified for slot names. (reported by Kalle Niemitalo)
1321 * fixed bug: lambda lists may contain symbols whose names start with
1322 & but are not lambda-list-keywords; their occurrence triggers a
1324 * fixed bug 321: define-method-combination argument lambda lists do
1325 not cause invalid code to be generated when &optional or &aux
1326 variables have default values. (reported by Bruno Haible)
1327 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1328 have CLOS classes; this appears also to have fixed reported
1329 instability in stack exhaustion detection.
1330 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1331 MUFFLE-WARNING when no associated restart is present is now
1333 * optimization: rearranged the expansion of various defining macros
1334 so that each expands into only one top-level form in a
1335 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1337 * optimization: used a previously little-used slot in symbols to
1338 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1340 * fixed some bugs revealed by Paul Dietz' test suite:
1341 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1343 ** Fixed an optimization bug related to inheritance of initforms
1344 from local to shared slots.
1345 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1346 ** CAST optimizer forgot to flush argument derived type.
1347 ** print/read consistency on floats is now orders of magnitude
1348 more likely. (thanks also to Bruno Haible for a similar report
1350 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1351 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1352 for objects of type REAL. Make it so.
1353 ** SXHASH is specified (infelicitously) to respect similarity,
1354 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1355 it so. (thanks to Markus Ziegler)
1356 ** on the Alpha, the compiler succeeds in compiling functions
1357 returning a known number of arguments greater than 63.
1358 ** fixed handling of invalid NIL arguments in keyword position
1360 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1362 ** provided a readably-printable representation for RANDOM-STATE
1364 ** ensured that pathnames, where they have a namestring, always
1365 print using #P"..." syntax.
1367 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1368 * Support for the forthcoming 2.0 version of the NetBSD kernel
1369 running on x86 hardware has been added. (thanks to Perry
1370 E. Metzger most immediately, and others for their past work)
1371 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1372 current release version; SBCL's OpenBSD support had been broken
1373 since about the time of OpenBSD's switch to ELF binary format.)
1374 * [placeholder for DX summary]
1375 ** user code with &REST lists declared dynamic-extent, under high
1376 speed or space and low safety and debug optimization policy.
1377 * The manual has been converted to Texinfo format and the debugger
1378 chapter from the cmucl manual has been added.
1379 * A facility has been added to extract documentation strings from
1380 sbcl and store them as Texinfo-formatted snippets for inclusion in
1381 the manual (via Texinfo's @include directive)
1382 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1383 the test case to Dave Roberts)
1384 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1385 pointers. (reported by Sean Ross)
1386 * bug fix: PROFILE output is printed nicely even for large numerical
1387 values. (thanks to Zach Beane)
1388 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1389 greater than 32 handle EOF correctly.
1390 * bug fix: on X86 an immediate argument of the IMUL instruction is
1391 correctly printed in disassembly. (thanks to Lutz Euler)
1392 * bug fix: class slots in redefined classes preserve their old
1393 values. (thanks to Bruno Haible and Nikodemus Siivola)
1394 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1395 succeeds. (reported by Marco Baringer)
1396 * fixed some bugs revealed by Paul Dietz' test suite:
1397 ** READ-SEQUENCE now works on ECHO-STREAMs.
1398 ** RATIONALIZE works more according to its specification. (thanks
1401 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1402 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1403 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1404 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1405 should matter to you only if you rebind the printer control
1406 variables and then find you want different bindings in the
1407 debugger than in the ordinary execution of your program.)
1408 * The runtime build system has been tweaked to support building
1409 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1410 assembler and linker. (thanks to Nikodemus Siivola)
1411 * Unbound, undefined, undeclared variables now trigger full
1412 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1413 more often programmer error than deliberate exploitation of undefined
1415 * optimization: the hash algorithm for strings has changed to one
1416 that is less vulnerable to spurious collisions. (thanks to Juho
1418 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1419 needless bounds checking. (thanks to Juho Snellman)
1420 * optimization: implemented multiplication as a modular
1421 (UNSIGNED-BYTE 32) operation on the PPC backend.
1422 * fixed some bugs revealed by Paul Dietz' test suite:
1423 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1425 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1426 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1427 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1429 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1431 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1433 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1434 their output stream on EOF from read.
1435 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1436 have been read to end-of-file.
1437 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1439 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1440 description of determination of which consecutive characters
1442 ** Printing the "Space" character with escaping on now yields "#\\ ",
1443 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1444 ** Reading floating-point numbers with *READ-BASE* set to a number
1445 less than 10 works correctly.
1446 ** Reading floating-point numbers with *READ-BASE* set to a number
1447 more than 10 works correctly.
1448 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1449 the readtable currently in effect.
1451 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1452 * minor incompatible change: parsing of namestrings on a physical
1453 (Unix) host has changed; numbers after the final #\. in a
1454 namestring are no longer interpreted as a version field. This is
1455 intented to be largely invisible to the user, except that the
1456 meaning of the namestring "*.*.*" has changed: it now refers to a
1457 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1458 should usually be replaced by
1459 (make-pathname :name :wild :type :wild :version :wild)
1460 with the added benefit that this is more likely to be portable.
1461 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1462 option to OPEN now signals an error if the file being opened
1463 exists; this may have an impact on existing code.
1464 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1465 signals received, on the PowerPC platforms (both Linux and
1466 Darwin). (thanks to Pierre Mai for pointing out the location of
1468 * several fixes on OS X: The system now builds and runs cleanly on
1469 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1470 (thanks to Brian Mastenbrook)
1471 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1472 second-resolution: integer multiples of 1/3600 between -24 and 24.
1473 (thanks to Vincent Arkesteijn)
1474 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1475 type when called with 1 argument; PEEK-CHAR checked type of
1476 PEEK-TYPE only after having read first character from a
1477 stream. (reported by Peter Graves)
1478 * bug fix: the garbage collector now has much better locality
1479 behaviour, and in particular no longer treats all memory as being
1480 exhausted when many small objects point to each other in a deeply
1482 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1483 recognized as being TYPEP their class.
1484 * bug fix: the PUSHNEW documentation string has been corrected.
1485 (thanks to Vincent Arkesteijn)
1486 * bug fix: defaulting of the value for the last of an atypically
1487 large number of multiple values being bound was not being
1488 performed correctly on the Alpha or PPC platforms
1489 * optimization: implemented multiplication as a modular
1490 (UNSIGNED-BYTE 32) operation on the x86 backend.
1491 * optimization: SEARCH on simple-base-strings can now be open-coded.
1492 (see also contrib/compiler-extras.lisp for inspiration for
1493 teaching the compiler about the Boyer-Moore algorithm).
1494 * value, returned by MAX (and MIN) called with several EQUALP, but
1495 not EQL, arguments now does not depend on compiler settings.
1496 * fixed some bugs revealed by Paul Dietz' test suite:
1497 ** in stack analysis liveness information is propagated from
1498 non-local entry points.
1499 ** pathwise CAST removing failed when the CAST node did not start
1501 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1502 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1504 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1505 host is already defined.
1506 ** RENAME-FILE works on streams instead of signalling an internal
1508 ** PEEK-CHAR uses the current readtable when determining whether
1509 or not a character is whitespace.
1510 ** MERGE-PATHNAMES handles the case when the pathname does not
1511 specify a name while the default-pathname specifies a version.
1512 ** Pathnames now stand a better chance of respecting print/read
1514 ** Attempting to use standardized file system operators with a
1515 pathname with invalid :DIRECTORY components signals a
1517 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1519 ** DIRECTORY on logical pathnames is more correct.
1520 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1521 signal a TYPE-ERROR if their argument is not a stream.
1522 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1523 designator argument does not designate a stream.
1524 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1525 examining the synonym.
1526 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1528 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1529 element-type larger than ([UN]SIGNED-BYTE 32).
1531 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1532 * When built with the :SB-FUTEX feature, threaded builds now take
1533 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1534 for faster/more reliable mutex and condition variable support.
1535 * Incompatible change (but one you probably shouldn't have been using
1536 anyway): the interface and code for arbitrating between multiple
1537 threads in the same user session has been redesigned.
1538 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1539 error when called without an explicit environment argument.
1540 (thanks to Nikodemus Siivola)
1541 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1542 confused about their position. (thanks to Adam Warner and Gerd
1544 * bug fix: Pretty printing backquoted forms with unquotations in the
1545 argument list position of various code constructs such as LAMBDA
1546 now works correctly. (reported by Paul Dietz)
1547 * bug fix: Pretty printing unquotations no longer loses all
1548 stream position information.
1549 * optimization: performance of string output streams is now less
1550 poor for multiple small sequence writes.
1551 * optimization: performance of CSUBTYPEP in the presence of complex
1552 expressions involving CONS and NOT many times has been improved.
1553 (reported by Paul Dietz)
1554 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1556 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1557 streams for the REPL, for the debugger, and so on. (thanks to
1559 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1560 * fixed some bugs revealed by Paul Dietz' test suite:
1561 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1562 is ignored if it is too small, rather than propagating through
1563 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1564 ** extremely complex negations of CONS types were not being
1565 sufficiently canonicalized, leading to inconsistencies in
1567 ** VALUES tranformer lost derived type.
1569 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1570 * fixed a bootstrapping bug: the build process no longer assumes
1571 that the various BOOLE-related constants have the same value in
1572 host and target lisps. (noted by Paul Dietz' test suite on an
1573 SBCL binary built from CLISP)
1574 * The system can now be dynamically linked on the MIPS platform,
1575 which enables dynamic loading of foreign code from Lisp. (thanks
1576 to Ralf Baechle for discussions on the MIPS ABI)
1577 * The system now records debugging information for its own source
1578 files in a filesystem-position-independent manner, relative to
1579 the "SYS" logical host.
1580 * fixed a compiler bug: MV-LET convertion did not check references
1581 to the "max args" entry point. (reported by Brian Downing)
1582 * tweaked disassembly notes to be less confident about proclaiming
1583 some instruction as an LRA. (thanks to Brian Downing)
1584 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1585 now each have their own history, command character, and other
1586 characteristics. (thanks to David Lichteblau)
1587 * fixed some bugs revealed by Paul Dietz' test suite:
1588 ** compiler failure in compiling LOGAND expressions including a
1590 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1591 shift greater than 32.
1592 ** FLUSH-DEST did not mark blocks for type check regeneration.
1593 ** HANDLER-CASE failed to accept declarations in handler clauses
1594 in some circumstances.
1596 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1597 * New code in contrib/sb-introspect (still probably not entirely
1598 stable yet) provides some support for smart Lisp development
1599 environments like SLIME.
1600 * The conditions signalled for errors occurring when loading .fasl
1601 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1602 in a way which should help ASDF recover gracefully.
1603 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1604 stuff has been cleaned up. If you code contrib/ stuff, this might
1605 affect you, and you can look at contrib/README, contrib/STANDARDS,
1606 and/or the 0.8.4.27 diff to check.
1607 * In full calls the compiler now does not generate checks for declared
1608 argument types for all arguments.
1609 * various threading fixes
1610 ** and some experimental patches which didn't make it into
1611 the main tree for this release, but which are shipped in
1612 contrib/experimental-thread.patch as a possible fix for some
1613 failures (deadlock, spinning...) in GC-intensive multithreaded
1615 * fixed PPC build problem (source code incompatibility of different
1616 library versions): added offsetof() hackery which attempts to divine
1617 where glibc maintainers put uc_mcontext today
1618 * fixed bug 282: compiler does not trust type assertions while passing
1619 arguments to a full call.
1620 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1621 &OPTIONAL VALUES type parameter.
1622 * fix bug 214: algorithm for noting rejected templates is now more
1623 similar to that of template seletion. (also reported by rydis on
1625 * fixed bug 141b: printing backquoted information readably and prettily
1626 inserts a space where necessary.
1627 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1628 (CHARACTER 10) are now reported as errors, rather than propagated
1629 as unknown types. (reported by piso on #lisp)
1630 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1631 with values NIL and :ERROR. (thanks to Milan Zamazal)
1632 * fixed bug 191c: CLOS now does proper keyword argument checking as
1633 described in CLHS 7.6.5 and 7.6.5.1.
1634 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1635 longer raise an error; further, using a list as a for-as-arithmetic
1636 counter now raises a meaningful error.
1637 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1638 now understood by sequence creation functions such as MAKE-SEQUENCE
1640 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1641 a STRING-INPUT-STREAM.
1642 * compiler enhancement: SIGNUM is now better able to derive the type
1644 * type declarations inside WITH-SLOTS are checked. (reported by
1646 * fixed some bugs revealed by Paul Dietz' test suite:
1647 ** incorrect optimization of TRUNCATE for a positive first
1648 argument and negative second.
1649 ** compiler failure in let-convertion during flushing dead code.
1650 ** compiler failure while deriving type of TRUNCATE on an
1651 interval, containing 0.
1652 ** ASH of a negative bignum by a negative bignum count now returns
1654 ** intersection of CONS types now canonicalizes properly, fixing
1655 inconsistencies in SUBTYPEP.
1657 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1658 * incompatible change: The --disable-debugger command line
1659 option now clobbers the debugger at a more fundamental
1660 level, by redefining #'INVOKE-DEBUGGER instead of by
1661 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1662 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1663 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1664 enter the debugger REPL and then suffer recursive errors
1665 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1666 while under the new system, BREAK in --disable-debugger mode
1667 terminates the system just as an unhandled error would.
1668 * fixed compiler performance when processing loops with a step >1;
1669 * bug fix: DOCUMENTATION now retrieves generic function
1670 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1671 support has been systematized, and now supports the methods
1672 specified by ANSI, along with a default method and a method for
1673 slot documentation. (reported by Nathan Froyd)
1674 * bug fix: effective methods associated with a generic function are
1675 no longer cached over a change of that generic function's method
1676 combination. (reported by Andreas Fuchs)
1677 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1678 names a non-existent directory. (thanks to Andreas Fuchs)
1679 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1680 the PPC platform, lead to stack corruption; this has been fixed.
1681 (reported by Rainer Joswig)
1682 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1683 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1684 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1685 arithmetic in general, is now much more reliable on the PPC
1687 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1688 platform now returns the right answer.
1689 * optimization: restored some effective method precomputation in
1690 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1691 precomputation is now tunable.
1692 * optimization: compiler-internal data structure use has been
1693 reviewed, and changes have been made that should improve the
1694 performance of the compiler by about 20%.
1695 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1696 simple-base-strings and simple-bit-vectors is improved.
1697 * optimization: the optimization of 32-bit logical and arithmetic
1698 functions introduced in version 0.8.3 on the x86 has been
1699 implemented on the mips, ppc and sparc platforms; an
1700 implementation of the same facility, but for 64-bit arithmetic,
1701 has been added for the alpha.
1702 * microoptimization: the compiler is better able to make use of the
1703 x86 LEA instruction for multiplication by constants.
1704 * bug fix: in some situations compiler did not report usage of
1705 generic arithmetic in (SPEED 3) policy.
1706 * bug 145b fix: compiler used wrong type specifier while converting
1707 MEMBER-types to numeric.
1708 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1710 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1711 index argument now works properly on the Alpha platform.
1712 * bug fix: floating point exception treatment on the Alpha platform
1714 * bug fix: FILE-POSITION works much better on string input and
1715 output streams. (thanks to Nikodemus Siivola)
1716 * bug fix: many threading/garbage collection symptoms sorted.
1717 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1718 might be pseudo-atomic.
1719 * internal change: Stopping for GC is now done with signals not
1720 ptrace. GC is now done in whichever thread wanted it, instead of
1722 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1724 * bug fix: result form in DO is not contained in the implicit
1726 * incompatible change: ICR structure is changed; the value part of
1727 CONTINUATION is now called LVAR; corresponding functions are
1728 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1729 * added type deriver for ISQRT (thanks to Robert E. Brown).
1730 * bug fix: better support for loading from the command line when an
1731 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1733 * fixed some bugs revealed by Paul Dietz' test suite:
1734 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1735 ** ROUND and FROUND now give the right answer when given very
1736 small float arguments.
1737 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1739 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1740 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1741 ** LCM should return a non-negative integer.
1742 ** PARSE-INTEGER returned the index of a terminator instead of the
1743 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1744 ** PARSE-INTEGER returned an incorrect index being applied to a
1746 ** LCM with two arguments of 0 returns 0 rather than signalling
1748 ** unsigned addition of a 32-bit constant with the high bit set no
1749 longer causes an internal compiler error on the x86.
1750 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1751 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1752 with negative last argument.
1753 ** byte specifiers with zero size and position no longer cause
1754 an error during type derivation.
1755 ** bignum multiplication on the Alpha platform now returns the
1757 * porting: The system now builds on SuSE AMD64, although it still
1758 generates a 32-bit binary.
1759 * .fasl file incompatibility: The fasl file version number has
1760 been incremented (because of the changes to internal compiler
1761 data structures referred to above).
1763 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1764 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1765 more accurately, on the Darwin kernel running on PowerPC hardware.
1766 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1767 * Compiler code deletion notes now signal a condition of type
1768 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1769 an associated MUFFLE-WARNING restart.
1770 * The compiler now performs limited argument count validation of
1771 constant format strings in FORMAT, and where appropriate in ERROR,
1772 CERROR and WARN. (thanks to Gerd Moellmann)
1773 * New ASDF-INSTALL contrib can be used for automatic download and
1774 installation of third-party Lisp code from CCLAN or other sites
1776 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1777 which forces another thread to execute a function supplied by the
1779 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1780 now accept and act upon their :ELEMENT-TYPE keyword argument.
1781 (reported by Martin Atzmueller, Edi Weitz)
1782 * bug fix: FILE-POSITION now accepts position designators up to
1783 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1784 is the greater. (thanks to Patrik Nordebo)
1785 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1786 Kalle Olavi Niemitalo)
1787 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1788 printable. (reported by Eric Marsden)
1789 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1790 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1791 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1792 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1793 (thanks to Stig Sandoe)
1794 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1795 UNREAD-CHAR now correctly includes the unread character in the
1796 target sequence. (thanks to Gerd Moellmann)
1797 * bug fix in threaded builds: the system can now be suspended and
1798 resumed by shell job control with minimal disruption.
1799 * bug fixes in times and timezones >2038AD
1800 * better handling of "where is GNU make?" problem in build scripts
1801 (thanks to Nikodemus Siivola)
1802 * new optimization: inside a named function any reference to a
1803 function with the same name is considered to be a self-reference;
1804 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1805 optimization quality.
1806 * new optimization on x86: logical functions and + now have
1807 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1808 used when the result is truncated to 32 bits.
1809 * VALUES declaration is partially enabled.
1810 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1811 * fixed some bugs revealed by Paul Dietz' test suite:
1812 ** The system now obeys the constraint imposed by
1813 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1814 types form a lattice under type intersection.
1815 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1816 ** ASSOC now ignores NIL elements in an alist.
1817 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1818 and (1+ MOST-POSITIVE-FIXNUM) answers.
1819 ** The addition of a method with invalid qualifiers to a generic
1820 function does not cause an error to be signalled immediately;
1821 a warning is signalled, and the error is generated only on
1822 calling the generic function.
1823 * changed .fasl file version number, in part to add type codes for
1824 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1825 obscure ANSI requirements
1827 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1828 * fixed bug 148: failure to inline-expand a local function left
1829 garbage, confusing the compiler.
1830 * fixed bugs 3cd: structure slot readers perform type check if the
1831 slot can have an invalid value (i.e. it is either not initialized
1832 or can be written with a less specific slot writer).
1833 * bug fix: the compiler now traps array references to elements off
1834 the end of an array; previously, the bounds checking in some
1835 circumstances could go off-by-one.
1836 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1838 * type declarations for array element types now obey the description
1839 on the CLHS page "Declaration TYPE", as per discussions on
1840 sbcl-help around 2003-05-08. This means that a declaration
1841 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1842 declaration, all references to BAR will be asserted or assumed
1843 (with THE, so dependent on compiler policy) to involve objects of
1844 type FOO. Note that no such declaration is implied in
1845 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1846 * declared types of functions from the "Conditions"
1847 chapter. (reported by Paul Dietz)
1848 * bug fix: CERROR accepts a function as its first argument.
1849 * bug fix: NTH an NTHCDR accept a bignum as index
1850 arguments. (reported by Adam Warner)
1851 * optimization: character compare routines now optimize comparing
1852 against a constant character. (reported by Gilbert Baumann)
1853 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1854 argument now works properly on the MIPS platform.
1855 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1856 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1857 declared to have a functional type, does not cause an error
1859 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1860 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1861 * fixed bug: if last continuation of a deleted block has a
1862 destination, this destination should be deleted too. (reported by
1864 * fixed a bug in the bootstrap process: the host compiler's values
1865 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1866 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1867 test case from Patrik Nordebo)
1868 * improved the ability of the disassembler on the PPC platform to
1869 provide helpful disassembly notes.
1870 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1871 the class in more cases than previously.
1872 * bug fix: FILE-POSITION now understands :START and :END for
1873 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1874 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1875 8) data. (thanks to David Lichteblau)
1876 * bug fix: it is possible to add a method to a generic function
1877 without lambda list.
1878 * bug fix: reader failed to signal END-OF-FILE inside an
1879 object representation. (reported by Nikodemus Siivola)
1880 * fixed some bugs revealed by Paul Dietz' test suite:
1881 ** LAST and [N]BUTLAST should accept a bignum.
1882 ** condition slot accessors are methods.
1883 ** (VECTOR NIL) is a subtype of STRING.
1885 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1886 * minor incompatible change: some nonsensical specialized lambda
1887 lists (used in DEFMETHOD) which were previously ignored now signal
1889 * minor incompatible change: the system is now aware of the types of
1890 variables in the COMMON-LISP package, and will signal errors for
1891 most violations of these type constraints (where previously they
1892 were silently accepted).
1893 * minor incompatible change: COMPILE-FILE now uses the freedom
1894 afforded (ANSI 3.2.2.3) to use derived function types for
1895 functions defined in the same file. This also permits the system
1896 to warn on static type mismatches and function
1897 redefinition. (Currently it does not work with high DEBUG level.)
1898 * minor incompatible change: VALUES declaration is disabled.
1899 * When issuing notes, the compiler now signals a condition of type
1900 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1901 restart for use in user handlers. It is expected that the
1902 COMPILER-NOTE condition will eventually become a condition
1903 supertype to a hierarchy of note types, which will then be
1904 handleable in a similar fashion. However, at the moment, no such
1905 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1906 but it's an implementation detail, not a classification for the
1908 * Changes in type checking closed the following bugs:
1909 ** type checking of unused values (192b, 194d, 203);
1910 ** template selection based on unsafe type assertions (192c, 236);
1911 ** type checking in branches (194bc).
1912 * A short form of VALUES type specifier has ANSI meaning (it has
1913 increased the number of situations when SBCL cannot perform type
1915 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1916 keywords or constants is permissible.
1917 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1918 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1919 classes. (thanks to Antonio Martinez)
1920 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1921 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1922 argument to a pretty-print stream. Instead, it leaves any such
1923 operations to DESCRIBE-OBJECT methods.
1924 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1926 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1927 lambda lists. (thanks to David Lichteblau)
1928 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1929 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1930 (thanks to Gerd Moellmann)
1931 * fixed bug 246: increased compilation speed of long
1932 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1934 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1936 * DEFCONSTANT now throws a condition of type
1937 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1938 constant to a non-EQL value; CONTINUE and ABORT restarts
1939 respectively change and preserve the value.
1940 * fixed bug 63: The code walker, part of the implementation of CLOS,
1941 is now better at handling symbol macros.
1942 * bug fix: There is no longer an internal implementation type named
1943 CL:LENGTH. (reported by Raymond Toy)
1944 * bug fix: In macro-like defining macros/special operators the
1945 implicit block does not enclose the lambda list.
1946 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1947 enclosed in parentheses) are not suitable as type specifiers, and
1948 their use properly signals an error now.
1949 * bug fix: An argument count mismatch for a type specifier in code
1950 being compiled no longer causes an unhandled error at compile
1951 time, but signals a compile-time warning.
1952 * fixed simple vector readable printing
1953 * bug fix: DESCRIBE takes more care over whether the class
1954 precedence list slot of a class is bound before accessing it.
1955 (reported by Markus Krummenacker)
1956 * bug fix: FORMATTER can successfully compile pretty-printer format
1957 strings which use variants of the ~* directive inside.
1958 * bug fix: SEARCH now applies its TEST predicate to the elements of
1959 the arguments in the correct order. (thanks to Wolfhard Buss)
1960 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1961 in type inference. (thanks to Robert E. Brown)
1962 * bug fix: Reading in symbols with an explicit package name of ""
1963 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1964 * fixed some bugs revealed by Paul Dietz' test suite:
1965 ** NIL is now allowed as a structure slot name.
1966 ** Arbitrary numbers, not just REALs, are allowed in certain
1967 circumstances in LOOP for-as-arithmetic clauses.
1968 ** Multiple class redefinitions before slot access no longer
1969 causes a type error.
1970 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1971 association between the name and a class.
1972 ** Generic functions with non-standard method-combination and over
1973 five methods all of which return constants no longer return NIL
1974 after the first few invocations. (thanks to Gerd Moellmann)
1975 ** CALL-NEXT-METHOD with no arguments now passes the original
1976 values of the arguments, even in the presence of assignment.
1977 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1979 ** DISASSEMBLE works with closures and funcallable instances.
1980 ** ADD-METHOD now returns the generic function, not the new method.
1981 ** FIND-METHOD signals an error if the lengths of the specializers
1982 is incompatible with the generic function, even if the ERRORP
1984 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1985 which its argument is a member.
1986 ** DEFCLASS only redefines the class named by its class-name
1987 argument if that name is the proper name of the class;
1988 otherwise, it creates a new class.
1989 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1990 of the UNBOUND-SLOT condition to the name of the slot.
1991 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1992 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1993 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1994 SLOT-MAKUNBOUND in the specified fashion.
1996 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1997 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1998 cross-compilation host. As a consequence, we can now bootstrap our
1999 way up to SBCL starting with a bare gcc toolchain and human-readable
2000 source code (first the source to CLISP, then the source to SBCL).
2001 * A contributed module containing a partial implementation of the
2002 simple-streams interface has been included. (thanks to Rudi
2004 * A contributed module implementing the RFC1321 Message Digest
2005 Algorithm, known as MD5, has been included.
2006 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2007 no longer has any effect, as the code controlled by this feature
2008 has been deleted. (As far as we know, no-one has ever built using
2009 this feature, and its semantics were confused in any case).
2010 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2011 work on conditions (as required by the ANSI specification),
2012 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2013 expected behaviour on conditions. Users should note, however,
2014 that such behaviour is not required by the ANSI specification,
2015 and so use of this behaviour may render their code unportable.
2016 * fixed some bugs revealed by Paul Dietz' test suite:
2017 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2019 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2020 ** SXHASH on condition objects no longer returns NIL.
2021 ** :ALLOCATION :CLASS slots are better treated; their values are
2022 updated on class redefinition, and initforms inherited from
2023 superclasses are applied.
2024 ** REMOVE-METHOD returns its generic function argument even when
2025 no method was removed.
2026 ** SHARED-INITIALIZE now initializes the values of the requested
2027 slots, including those with :ALLOCATION :CLASS.
2028 ** ALLOCATE-INSTANCE now works on structure classes defined via
2029 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2031 ** SLOT-EXISTS-P now works on conditions, as well as structures
2033 ** MAKE-LOAD-FORM now has the required methods on
2034 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2035 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2036 keyword, and now implements the SLOT-NAMES argument.
2037 ** methods with &OPTIONAL arguments no longer allow too many
2038 arguments to be passed in the call without error.
2039 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2040 option is consistent with the required arguments of the generic
2041 function lambda list.
2042 * bug fix: REQUIRE accepts a string designator. (Thanks to
2044 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2045 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2046 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2047 * bug fix: The compiler now checks for duplicated variables in macro
2049 * bug fix: SETQ on globals returns the correct value.
2050 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2051 not a COMPILER-ERROR (followed by some other strange error on
2052 choosing the CONTINUE restart).
2053 * bug fix: make.sh and friends are now more consistent in the way that
2054 they look for GNU "make".
2056 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2057 * experimental native threads support (on x86 Linux >=2.4 only).
2058 This is not compiled in by default: you need to add :SB-THREAD to
2059 the target features. See the "Beyond ANSI" chapter of the manual
2061 * fix for longstanding nonANSIism: The old distinction between
2062 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2063 The return value from CL:FIND-CLASS is now a CLOS class, and
2064 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2065 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2066 * An interface to the MetaObject Protocol, as described in Kiczales,
2067 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2068 MIT Press, 1991, is available from the SB-MOP package.
2069 * incompatible change: the SB-PCL package should now be considered
2070 a private implementation detail, and no longer a semi-private MOP
2072 * minor incompatible change: due to rearrangement for threads, the
2073 control stack and binding stack are now allocated at arbitrary
2074 addresses instead of being hardcoded per-port. Users affected by
2075 this probably have to be doing advanced things with shared
2076 libraries, and will know who they are.
2077 * minor incompatible change: Previously, all --eval forms used were
2078 processed with READ before any of them were processed with EVAL.
2079 Now each --eval form is processed with both READ and EVAL before
2080 the next --eval form is processed. (Thus package operations like
2081 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2082 work as the user might reasonably expect.)
2083 * minor incompatible change: *STANDARD-INPUT* is now only an
2084 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2086 * minor incompatible change: Y-OR-N-P is now character-oriented,
2087 not line oriented. Also, YES-OR-NO-P now works without errors.
2088 (thanks to Antonio Martinez)
2089 * sb-aclrepl module improvements: an integrated inspector, added
2090 repl features, and a bug fix to :trace command.
2091 * Known functions, which cannot be open coded by the backend, are
2092 considered to be able to check types of their arguments. (fixing
2093 a bug report by Nathan J. Froyd)
2094 * fixed a bug in computing method discriminating functions: It is
2095 now possible to define methods specialized on classes which have
2096 forward-referenced superclasses. (thanks to Gerd Moellmann)
2097 * fixed evaluation order in optional entries (reported by Gilbert
2099 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2100 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2101 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2102 required-by-AMOP NAME argument, as well as CLASS and
2103 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2104 * fixed bug 20: DEFMETHOD can define methods using names that are
2105 not the proper names of classes to designate class specializers.
2106 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2107 Unix file descriptors, instead of blowing up. (thanks to Antonio
2109 * Garbage collection refactoring: user-visible change is that a
2110 call to the GC function during WITHOUT-GCING will not do garbage
2111 collection until the end of the WITHOUT-GCING. If you were doing
2112 this you were probably losing anyway.
2113 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2114 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2115 TYPEP the latter but not the former.
2116 * The compiler issues a full WARNING for calls to undefined functions
2117 with names from the CL package.
2118 * MAP-INTO for a vector destination is open coded. (reported by
2119 Brian Downing on c.l.l)
2120 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2121 documentation string.
2122 * fixed some bugs revealed by Paul Dietz' test suite:
2123 ** COPY-ALIST now signals an error if its argument is a dotted
2125 ** Condition slots are now accessed more correctly in the presence
2126 of multiple initargs for a given slot.
2127 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2128 exclude restarts of the same name associated with a different
2130 ** DEFCLASS of forward-referenced classes with another
2131 forward-referenced class in the superclasses list no longer
2133 ** Condition slots are now initialized once each, not multiple
2134 times. (thanks to Gerd Moellmann)
2135 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2137 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2138 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2139 arguments contain duplicated elements.
2140 ** RESTART-CASE understands local macros.
2141 ** RESTART-CASE associates exactly its own restarts with a condition.
2142 ** ENDP in safe mode checks its argument to be of type LIST.
2143 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2144 in question is unbound.
2145 ** Optimized MAKE-INSTANCE functions no longer cause internal
2146 assertion failures in the presence of duplicate initargs.
2147 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2148 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2149 propagated even in the ctor optimized implementation of
2151 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2153 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2154 * a better implementation of SXHASH on (simple) bit vectors,
2155 measured both in execution speed and in distribution of results
2156 over the positive fixnums, has been installed. Likewise, a better
2157 implementation of EQUAL for simple bit vectors is now available.
2158 * fixed CEILING optimization for a divisor of form 2^k.
2159 * fixed bug 240 (emitting extra style warnings "using the lexical
2160 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2161 by Antonio Martinez)
2162 * fixed SXHASH, giving different results for NIL depending on type
2163 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2164 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2165 variable. (found by Rolf Wester)
2166 * SBCL does not ignore type declarations for special
2167 variables. (reported by rif on c.l.l 2003-03-05)
2168 * some bug fixes in contrib/sb-aclrepl/
2169 * fixed some bugs revealed by Paul Dietz' test suite:
2170 ** a bug in the CONS type specifier, whereby the CAR and CDR
2171 types got intertwined, has been fixed;
2172 ** the type system is now able to reason about the interaction
2173 between INTEGER and RATIO types more completely;
2174 ** APPEND, [N]REVERSE and NRECONC check that those their
2175 arguments, which must be proper lists, are really so;
2176 ** An array specialized to be unable to hold elements has been
2177 implemented, as required -- yes, really -- by ANSI;
2178 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2179 on malformed property lists;
2181 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2182 * incompatible packaging change: in line with Unix convention,
2183 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2184 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2185 * REQUIRE and PROVIDE are now optionally capable of doing something
2186 useful. See the documentation string for REQUIRE.
2187 * infrastructure for a managed SBCL contrib system: contributed
2188 modules in this release include:
2189 ** the ASDF system definition facility;
2190 ** an interface to the BSD Sockets API;
2191 ** an ACL-like convenience interface to the repl;
2192 (thanks to Kevin Rosenberg)
2193 ** an implementation of ROTATE-BYTE, with efficient implementation
2195 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2196 gives a closer approximation to the right answer than previously.
2197 (thanks to Raymond Toy)
2198 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2199 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2200 optional environment arguments, as required by ANSI.
2201 * fixed bugs in other functions taking environment objects, allowing
2202 calls with an explicit NIL environment argument to be compiled
2204 * fixed bug 228: primary return values from
2205 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2206 COMPILE or FUNCTION.
2207 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2208 :TYPE will no longer signal errors on innocuous objects.
2209 * fixed bug 231b: SETQ is better at respecting type declarations in
2210 the lexical environment.
2211 * fixed a bug in DEFCLASS: classes named by symbols with no or
2212 unprintable packages can now be defined.
2213 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2214 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2215 * fixed compiler failure related to checking types of functions.
2216 (reported by Robert E. Brown)
2217 * the compiler is now much more consistent in its error-checking
2218 treatment of bounding index arguments to sequence functions: in
2219 (SAFETY 3) code, errors will be signalled in almost all cases if
2220 invalid sequence bounding indices are passed to functions defined
2221 by ANSI to operate on sequences.
2222 * fixed a bug in the build procedure: documentation of SBCL-specific
2223 packages is now preserved and available in the final Lisp image.
2224 * lifted FDEFINITION lookup out of loops in the implementation of
2225 many list operations. (thanks to Robert E. Brown)
2226 * fixed a bug in the reader: the #n# reader macro now works for
2227 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2228 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2229 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2230 E. Brown and Tony Martinez respectively)
2231 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2232 SLOT-BOUNDP: in particular, it is now able to optimize them much
2233 better, and is now not vulnerable to having packages renamed.
2234 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2235 always called when a slot is not present in an instance. (thanks
2237 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2238 longer signal an error on unreading a character following EOF on
2239 the previous constituent stream. (thanks to Tony Martinez)
2240 * fixed some bugs revealed by Paul Dietz' test suite:
2241 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2242 not just nonnegative fixnums;
2243 ** the logical bit-array operators such as BIT-AND now accept an
2244 explicit NIL for their "opt-arg" argument (to indicate a
2245 freshly-consed result bit-array);
2246 ** ELT now signals an error on an invalid sequence index in safe
2248 ** the type system is now cleverer about negations of numeric
2249 types, and consequently understands the BIGNUM and RATIO types
2251 ** the type system is now cleverer about the interaction between
2252 INTEGER and RATIO types: while bugs still remain, many more
2253 cases are accurately computed;
2254 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2255 if it is in the last clause;
2256 ** CONSTANTLY now correctly returns a side-effect-free function in
2258 ** DECLARE is no longer treated as a special-operator; in
2259 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2260 * incremented fasl file version number due to the change in the
2261 DEFSTRUCT-SLOT-DESCRIPTION structure.
2263 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2264 * minor incompatible change: code processed by the "interpreter" or
2265 EVAL now has a compilation optimization policy of (DEBUG 2)
2266 (changed from (DEBUG 1)) to improve debuggability of interactive
2267 development, and to allow the use of the debug RETURN command in
2269 * an experimental implementation of the RETURN command for the
2270 debugger has been included. (thanks to Frederik Kuivinen)
2271 * fixed bug 62: constraints were not propagated into a loop.
2272 * fixed bug in embedded calls of SORT (reported and investigated by
2274 * fixed some bugs revealed by Paul Dietz' test suite:
2275 ** printing and reading of arrays with some dimensions having
2276 length 0 (thanks to Gerd Moellmann);
2277 ** BOA constructor with &AUX argument without a default value does
2278 not cause a type error;
2279 ** CONSTANTP now returns true for all self-evaluating objects.
2281 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2282 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2283 accessors that are related by inheritance, as specified in the
2284 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2285 Valtteri Vuorikoski)
2286 * The compiler is now able to inline functions that were defined in
2287 a complex lexical environment (e.g. inside a MACROLET).
2288 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2290 * Support for the upcoming FreeBSD-5.0 release has been included.
2291 (thanks to Dag-Erling Smorgrav)
2292 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2293 effect when it is not in a toplevel context.
2294 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2295 stand a better chance of being correct. (thanks to Gerd
2297 * fixed bug in COERCE, which now signals an error on coercing a
2298 rational to a bounded real type which excludes the expected
2300 * The compiler is now able to derive types more accurately from the
2301 COERCE and COMPILE functions.
2302 * fixed bug 223: functional binding is considered to be constant
2303 only for symbols in the CL package.
2304 * fixed bug 231: SETQ did not check the type of a variable being set
2305 (reported by Robert E. Brown)
2306 * A new optimization for MAKE-INSTANCE has been included, fixing
2307 various bugs (including relating to :ALLOCATION :CLASS slots and
2308 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2310 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2311 ** As required by ANSI, LOOP now disallows anonymous collection
2312 clauses such as COLLECT I in conjunction with aggregate boolean
2313 clauses such as THEREIS (= I 1);
2314 ** LOOP now signals an error when any variable is reused in the
2315 same loop (including the potentially useful construct analogous
2316 to WITH A = 1 WITH A = (1+ A);
2317 ** IT is only a special loop symbol within the first clause of a
2318 conditional loop clause;
2319 ** LOOP with a typed iteration variable over a hashtable now
2320 signals a type error iff it should.
2321 * fixed some other bugs revealed by Paul Dietz' test suite:
2322 ** FILE-STREAM now names the class previously known as FD-STREAM;
2323 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2324 argument) no longer signals an error;
2325 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2326 of the slot symbol, rather than using the current package
2327 ((:CONC-NAME "") continues to intern the slot's name in the
2329 * incremented fasl file version number, because of the incompatible
2330 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2331 of the new implementation of DEFINE-COMPILER-MACRO.
2333 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2334 * Support for building SBCL for MIPS platforms running in
2335 little-endian mode has now been checked in, and basic
2336 functionality on said platforms verified.
2337 * minor incompatible change: PCL now records the pathname of a file
2338 in which methods and the like are defined, rather than its
2340 * minor incompatible change: TRUENAME now considers the truename of
2341 a file naming a directory to be the pathname with :DIRECTORY
2342 component indicating that directory.
2343 * minor incompatible change: a NAMED clause in the extended form of
2344 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2345 reason for the previous behaviour is unclear.
2346 * more systematization and improvement of CLOS and MOP conformance
2347 in PCL (thanks to Gerd Moellman and Pierre Mai):
2348 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2350 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2351 primary methods with no specializers;
2352 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2354 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2355 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2356 FORWARD-REFERENCED-CLASSes; error reporting on
2357 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2359 ** SXHASH on CLOS instances now uses a slot internal to the
2360 instance to return different numbers on distinct instances,
2361 while preserving the same return value through invocations of
2363 ** DEFMETHOD signals errors when methods with longer incongruent
2364 lambda lists are added to generic functions;
2365 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2366 CLASS, as specified in AMOP;
2367 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2368 on the order returned by the primary method for classes of
2369 class STANDARD-CLASS;
2370 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2371 * fixed some bugs shown by Paul Dietz' test suite:
2372 ** DOLIST puts its body in TAGBODY;
2373 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2375 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2376 value producing form;
2377 ** if more variables are given to PROGV than values, extra
2378 variables are bound and made to have no value;
2379 ** NSUBSTITUTE on list arguments gets the right answer with
2381 ** ELT signals an error of type TYPE-ERROR when the index argument
2382 is not a valid sequence index;
2383 ** LOOP signals (at macroexpansion time) an error of type
2384 PROGRAM-ERROR when duplicate variable names are found;
2385 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2386 ** FUNCALL of special-operators now cause an error of type
2388 ** PSETQ now works as required in the presence of side-effecting
2389 symbol-macro places;
2390 ** NCONC accepts any object as its last argument;
2391 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2393 ** loop-for-as-package does not require a package to be explicitely
2395 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2397 * fixed bug 166: compiler preserves "there is a way to go"
2398 invariant when deleting code.
2399 * fixed bug 172: macro lambda lists with required arguments after
2400 &REST arguments now cause an error to be signalled. (thanks to
2402 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2403 bug. (thanks to Matthew Danish)
2404 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2406 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2407 arguments are lexically rebound. (thanks to Gerd Moellmann and
2409 * fixed bug 194: error messages are now more informative when there
2410 is no primary method applicable in a call to a generic
2411 function. (thanks to Gerd Moellmann)
2412 * fixed bug in command line argument checking (thanks to Julian
2414 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2415 (thanks to Matthew Danish)
2416 * incremented fasl file version number, because of the
2417 SXHASH-related changes in the layout of CLOS data structures
2419 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2420 * minor incompatible change: The runtime (the Unix executable named
2421 "sbcl") is now much pickier about the .core files it will load.
2422 Essentially it now requires .core files to descend from the same
2423 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2424 as the runtime does. (The intent is to prevent the crashes which
2425 can occur, and which can even be reported as mysterious failures,
2426 when people patch the sources or change the build parameters
2427 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2428 sbcl and .core files.)
2429 * fixed bug: VALUES-LIST is no longer optimized away.
2430 * fixed bug 142: The FFI conversion of C string values to Lisp
2431 string values no longer conses excessively. (thanks to Nathan
2432 Froyd porting Raymond Toy's fix to CMU CL)
2433 * began to systematize and improve MOP conformance in PCL (thanks to
2434 Nathan Froyd, Gerd Moellman and Pierre Mai):
2435 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2437 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2438 ** FINALIZE-INHERITANCE is now called on class finalization;
2439 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2440 argument precedence order.
2441 * fixed bug 202: The compiler no longer fails on functions whose
2442 derived types contradict their declared type.
2443 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2444 so it can be non-toplevel.
2445 * The fasl file version number has changed (because of the new
2446 implementation of DEFMACRO).
2447 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2448 safe code, that any length requirement by their type-specifier
2449 argument is valid. The exceptions to this are described in bug
2451 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2452 functions now check the types of their inputs as required by ANSI.
2453 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2454 introduced symbol is DECLAREd to be SPECIAL.
2455 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2456 * fixed bug: :COUNT argument to sequence functions may be negative
2457 * fixed bug: body of DO-SYMBOLS may contain declarations
2458 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2459 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2460 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2461 symbol macro only once
2462 * fixed printing of call frame when argument list is unavailable
2463 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2464 * compiler no longer signals WARNING on unknown keyword
2467 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2468 * A beta-quality port to the mips architecture running Linux,
2469 based on the old CMUCL backend, has been made. It has been tested
2470 on a big-endian kernel, and works sufficiently well to be able to
2471 rebuild itself; it has not been tested in little-endian mode.
2472 * fixed an inconsistency between gencgc.c and purify.c which made
2473 dumping/loading .core files unreliable
2474 * fixed bug 120a: The compiler now deals correctly with IFs where
2475 the consequent is the same as the alternative, instead of
2476 misderiving the return type. (thanks to Alexey Dejneka)
2477 * fixed bug 113: Logical pathnames are now dumpable (the logical
2478 host is resolved at load-time, throwing an error if it is not
2480 * fixed bug 174: FORMAT's error message is slightly clearer when a
2481 non-printing character is used in a format directive.
2482 * fixed several bugs in compiler checking of type declarations, i.e.
2483 violations of the Python "declarations are assertions" principle
2484 (thanks to Alexey Dejneka)
2485 * fixed several bugs in PCL's error checking (thanks to Gerd
2487 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2489 * fixed bug in compilation of functions as first class values
2490 (thanks to Antonio Martinez-Shotton)
2491 * The compiler's handling TYPE-ERRORs which it can prove will
2492 inevitably happen at runtime has been cleaned up and corrected
2493 in several ways. (thanks to Alexey Dejneka)
2494 * improved argument type checking for various basic arithmetic
2495 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2496 lavished on them in the past that they can be compiled in many
2497 ways in different special cases
2498 * fixed bug 181: compiler checks validity of user supplied type
2500 * cleaned up code flushing in optimization: Function calls which
2501 should signal errors for safety purposes (e.g. which ANSI says
2502 should signal errors when their arguments are of incorrect type)
2503 are no longer optimized away.
2504 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2505 * incremented fasl file version number, because changes in the
2506 implementation of sequence functions like COERCE caused
2507 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2508 in old inline expansions) to become undefined. (Actually these
2509 changes were later undone, so we might very well be binary
2510 compatible with 0.7.7 after all, but leaving the version number
2511 incremented seemed like the simplest and most conservative
2514 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2515 * An alpha-quality port to the parisc architecture running Linux,
2516 based on the old CMUCL backend, has been made. This, even more so
2517 than the other backends, should be considered still a work in
2518 progress; known problems include that the Linux kernel in 64-bit
2519 mode does not propagate the correct sigcontext structure to
2520 userspace, and consequently SBCL on a parisc64 kernel will not
2522 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2523 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2524 suppressed.) Also now that the compiler is looking at declarations
2525 in the environment, it checks optimization declarations as well,
2526 and suppresses inlining when (> DEBUG SPEED).
2527 * More fixes have been made to treatment of floating point exception
2528 treatment and other Unix signals. In particular, floating point
2529 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2530 * The detection and handling of control stack exhaustion (infinite
2531 or very deeply nested recursion) has changed. Stack exhaustion
2532 detection is now done by write-protecting pages at the OS level
2533 and applies at all optimization settings; when found, a
2534 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2535 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2536 be caught using IGNORE-ERRORS.
2537 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2538 correctly when transforming recursive LABELS functions to LETs.
2539 (thanks to Alexey Dejneka)
2540 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2541 that are names of constants or global variables.
2542 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2543 alien routines with docstrings.
2544 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2545 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2547 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2548 (thanks to Eric Marsden)
2549 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2550 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2551 * Bug fix: The SPARC backend can now compile functions involving
2552 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2553 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2554 to a functions expecting a PATHNAME-DESIGNATOR.
2555 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2556 lambda lists. (thanks to Alexey Dejneka)
2557 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2558 functions so that EXPORTing the name of the class doesn't cause
2559 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2560 to Antonio Martinez for reporting this)
2561 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2562 constant-folded without failing an assertion. (thanks to Einar
2563 Floystad Dorum for reporting this)
2564 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2565 types are now dealt with more correctly by the compiler.
2566 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2567 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2568 behaviour analogous to RENAME-FILE. This puts its behaviour more
2569 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2571 * The fasl file version number has changed again. (because of the
2572 bug fix involving the names of PCL MAKE-INSTANCE functions)
2574 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2575 * bug fix: Floating point exceptions are treated much more
2576 consistently on the x86/Linux and PPC/Linux platforms.
2577 * Array initialization with :INITIAL-ELEMENT is now much faster for
2578 cases when the compiler cannot open code the array creation, but
2579 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2580 array accesses have also seen a speed increase.
2581 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2582 specified. (This was at the root of some bad interactions between
2583 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2585 * bug fix: Internal error arguments for undefined functions are now
2586 computed correctly on the PPC/Linux platform.
2587 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2588 Raymond Toy's patch for CMU CL)
2589 * Support for the Solaris 9 operating environment has been included
2590 (thanks to Daniel Merritt)
2591 * A very ugly but hopefully complete draft of the missing FFI chapter
2592 of the manual has been created by reformatting the corresponding
2593 CMU CL manual chapter into (currently very ugly and incoherent)
2594 DocBook and bringing it up to date for SBCL behavior. Thus, the
2595 manual is now essentially complete, at least by my extreme
2596 once-and-only-once standards, whereby it's acceptable to refer to
2597 the doc strings of SB-EXT functions as the primary documentation.
2598 * The fasl file version number has changed again, due to cleanup of
2599 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2600 is no longer a static symbol.)
2602 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2603 * SBCL now builds with OpenMCL (version 0.12) as the
2604 cross-compilation host; also, more progress has been made toward
2605 bootstrapping under CLISP.
2606 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2608 * bug 158 fixed: The compiler can now deal with integer loop
2609 increments different from 1; fixing this turned out also to fix
2611 * bug 169 fixed: no more bogus warnings about using lexical bindings
2612 despite the presence of perfectly good SPECIAL declarations (thanks
2613 to David Lichteblau)
2614 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2615 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2616 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2618 * bug 180 fixed: Method combination specifications no longer ignore
2619 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2620 * bug fix: Structure type predicate functions now check their argument
2621 count as they should.
2622 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2623 correctly. (thanks to Pierre Mai)
2624 * minor incompatible change: The --noprogrammer option is deprecated
2625 in favor of the new --disable-debugger option, which is very similar.
2626 (The major difference is that it takes effect at a slightly different
2627 time at startup, causing handling of errors in --sysinit and
2628 --userinit files will be affected differently.) The
2629 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2630 been added to allow this functionality to be controlled from ordinary
2631 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2632 maintainers, who might want to run non-interactive scripts to
2633 build SBCL cores which will later be used interactively.)
2634 * minor incompatible change: The LOAD function no longer, when given
2635 a wild pathname to load, loads all files matching that pathname.
2636 Instead, an error of type FILE-ERROR is signalled.
2638 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2639 * bug 147 fixed: The compiler preserves its block link/count
2640 invariants more correctly now so that it doesn't crash. (thanks
2642 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2644 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2645 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2646 no-value-for-this-character as NIL (as specified by ANSI).
2647 * HOST-NAMESTRING on physical pathnames now returns a string that is
2648 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2649 (thanks to Christophe Rhodes)
2650 * The Alpha port handles icache flushing more correctly. (thanks to
2652 * More progress has been made toward bootstrapping under CLISP. (thanks
2653 to Christophe Rhodes)
2654 * The fasl file format has changed again, because dynamic loading
2655 on OpenBSD (which has non-ELF object files) motivated some cleanups
2656 in the way that foreign symbols are transformed and passed around.
2657 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2658 is no longer treated as whitespace by the reader, but instead as
2659 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2660 |A
\7fB|, instead of A as it used to.
2662 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2663 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2664 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2665 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2666 this as of 0.7.1.45, but was left out of the previous news section
2667 (thanks to Dan Barlow)
2668 * SBCL now runs on the Solaris operating system on SPARC architectures
2669 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2670 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2671 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2672 * SPARC backend cleanups, allowing builds of cores optimized for V8
2673 and V9 SPARCS, and also emission of code targeted to a particular
2674 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2676 * SBCL is closer to bootstrapping under CLISP, thanks to various
2677 fixes by Christophe Rhodes.
2678 * The fasl file format has changed again, to allow the compiler's
2679 INFO database to support symbol macros.
2680 * The user manual (in doc/) is formatted into HTML more nicely.
2681 (thanks to coreythomas)
2682 * The system is smarter about SUBTYPEP relationships, especially
2683 those involving NOT types (including types such as ATOM which are
2684 represented internally using NOT types). Thus SUBTYPEP is less
2685 likely to return (VALUES NIL NIL) in general, and in particular
2686 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2688 * The fasl file format has changed again, because the internal
2689 representation of types now includes a new slot to support the new
2690 SUBTYPEP-of-NOT-types logic.
2691 * (not a change in the main branch of SBCL, but a related prototype
2692 which can hopefully be merged into the main branch of SBCL in the
2693 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2694 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2695 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2696 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2697 dispatch macro character combinations. (thanks to Alexey Dejneka)
2698 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2699 that superficially look like logical namestrings correctly.
2700 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2702 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2703 default supplied-p) for &optional and &key arguments. (thanks to
2706 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2707 * incompatible change: The compiler is now less aggressive about
2708 tail call optimization, doing it only when (> SPACE DEBUG) or
2709 (> SPEED DEBUG). (This is an incompatible change because there are
2710 programs which relied on the old CMU-CL-style behavior to optimize
2711 away their unbounded recursion which will now die of stack overflow.)
2712 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2713 for non-GENCGC systems has been increased to 20M (since that
2714 seems much closer to the likely performance optimum for modern
2715 systems than the old 4M value was)
2716 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2717 and *DEBUG-PRINT-LEVEL*
2718 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2719 of CMU CL's support for SPARC, and various endianness and other
2720 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2721 * new syntactic sugar for the Unix command line: --load foo.bar is now
2722 an alternate notation for --eval '(load "foo.bar")'.
2724 ** The system now detects stack overflow and handles it gracefully,
2725 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2726 optimization settings. (This is a good thing in general, and
2727 its introduction in this version should be particularly timely
2728 for anyone whose code fails because of suppression of tail
2730 ** The system now hunts for the C variable "environ" in a more
2731 devious way, to avoid segfaults when the C library version differs
2732 between compile time and run time. (thanks to Christophe Rhodes)
2733 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2734 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2735 ** The compiler no longer issues bogus style warnings for undefined
2736 classes in the same source file as the DEFCLASSes which defined
2737 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2739 ** fixes in CONDITION class precedence list for undefined function
2740 errors (thanks to Alexei Dejneka)
2741 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2742 correctly. (thanks to Dan Barlow)
2743 ** portability fixes aiming at bootstrapping under CLISP (thanks
2744 to Dave McDonald and Christophe Rhodes)
2745 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2746 ** fixes in type translation and and type inference (thanks to
2748 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2749 ** various fixes in the new ports (thanks to Dan Barlow)
2750 * several changes related to debugging:
2751 ** suppression of tail recursion, as noted above
2752 ** stack overflow detection, as noted above
2753 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2754 is now the default. (For some time encapsulation has been more
2755 reliable than the breakpoint-based :ENCAPSULATE NIL
2756 implementation, at least on X86 systems; and I just noticed that
2757 encapsulation also seems closer to the spirit of the ANSI
2760 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2762 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2763 up properly again. (There was a packaging bug in 0.7.0 which
2764 left their definitions in SB-SYS::LOAD-FOREIGN and
2765 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2766 vital for most things which interface to C-level interfaces,
2767 like extensions working with sockets or databases or
2768 Perl-compatible regexes or whatever, and the need to fix
2769 this bug is the main reason that 0.7.1 was released so
2771 ** DEFGENERIC is now choosier about the methods it redefines, so that
2772 reLOADing a previously-LOADed file containing DEFGENERICs does
2773 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2774 cycle now works as it should. (thanks to Alexey Dejneka)
2775 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2776 fixed by Christophe Rhodes. (He actually submitted this patch
2777 months ago, and I delayed until after 0.7.0.)
2778 ** Bug 111 (internal compiler confusion about runtime checks on
2779 FUNCTION types) was fixed by Alexey Dejneka.
2780 * Some internal cleanups (getting rid of variables which aren't
2781 needed now that the byte interpreter is gone) caused the fasl
2782 file format number to change again.
2784 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2785 * major incompatible change: The default fasl file extension, i.e. the
2786 default extension for files produced by COMPILE-FILE, has changed
2787 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2789 ** There are many changes in the implementation of the compiler.
2790 SBCL is now essentially a compiler-only implementation of ANSI
2791 Common Lisp. EVAL still "interprets" a few special cases, but
2792 almost all the interesting cases are handled by creating
2793 a LAMBDA expression, calling COMPILE on it, then calling
2794 FUNCALL on the result.
2795 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2796 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2797 Since the code is newer, there might still be some new bugs
2798 (though not as many as before Martin Atzmueller's fixes:-). But
2799 the new code is substantially simpler and clearer, and hopefully
2800 any remaining bugs will be simpler, less fundamental, and more
2801 fixable then the bugs in the old code.
2802 ** The revised compiler is still a little unsteady on its feet.
2804 *** The debugging information it produces (particularly the names
2805 of FUNCTION objects) is sometimes much less useful than what
2806 the old compiler produced.
2807 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2808 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2809 within a MACROLET) has been temporarily weakened.
2810 ** There are new compiler optimizations for various functions:
2811 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2812 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2813 *** the math functions TRUNCATE, FLOOR, and CEILING
2814 *** the function-of-all-trades COERCE
2815 Mostly these should be transparent, but there's one
2816 potentially-annoying problem (bug 117): when the compiler
2817 inline-expands a function and does type analysis on the result,
2818 it can create control paths which have type mismatches, and
2819 when it can't prove that those control paths aren't taken,
2820 it will issue WARNINGs about the type mismatches. This is
2821 a particular problem in practice for the new sequence functions.
2822 It's not clear how this should be fixed, and for now, a
2823 workaround is given in the entry for 117 in the BUGS file.
2824 ** (Because of the interaction between the two previous items --
2825 occasional inlining problems and new inline expansions -- some
2826 of the new sequence function optimizations won't really kick in
2827 completely until debugging information, and then inlining, are
2828 straightened out in some future version.)
2829 * minor incompatible changes:
2830 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2831 DIRECTORY no longer implicitly promotes NIL slots of its
2832 pathname argument to :WILD. In particular, when you ask for the
2833 contents of a directory (which you used to be able to do without
2834 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2835 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2836 ** changes in behavior that ANSI explicitly defines to be
2837 implementation dependent:
2838 *** The new compiler-only implementation still conforms with ANSI,
2839 but acts a little different than before. Besides the obvious
2840 changes in performance tradeoffs (that the cost per form passed
2841 to EVAL has gone up, and the cost per form executed by EVAL
2842 has gone down), the behavior of the system changes a little
2843 because there are no longer any interpreted function objects.
2844 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2845 e.g. doing COMPILE on the output of interactive DEFUN is
2847 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2849 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2850 has changed from (:CL) to NIL.
2851 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2852 e.g. #\Newline and #\Tab, don't have names specified in the
2853 ANSI Common Lisp standard, is now based on their ASCII symbolic
2854 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2855 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2856 are no longer used for output.
2857 ** changes in internal implementation constants:
2858 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2859 4 million. (If your application spends a lot of time GCing and
2860 you have a lot of RAM, you might want to experiment with
2861 increasing it even more.)
2862 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2863 However, almost all old code should still continue to work without
2864 immediate update, as SB-C-CALL is now a (deprecated) nickname
2866 ** Old operator names in the style DEF-FOO are now deprecated in
2867 favor of new corresponding names DEFINE-FOO, for consistency with
2868 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2869 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2870 internal symbols, but a few supported extensions like
2871 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2872 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2873 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2874 etc. as you get deeper into recursive calls to the debugger
2875 command loop, instead of the old "5]", "5]]", "5]]]"
2876 sequence. (I was motivated to do this when squabbles between
2877 ILISP and SBCL left me very deeply nested in the debugger. In the
2878 short term, this change will probably provoke more ILISP/SBCL
2879 squabbles, but hopefully it will be an improvement in the long run.)
2880 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2881 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2882 function, since declaiming return types involving aliens is
2883 (1) annoyingly messy to do by hand and (2) vital to efficient
2884 compilation of code which calls such functions.
2885 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2886 longer reexported by the SB-EXT package. They're solely useful
2887 for alien code, so it seems more logical that you should get
2888 them from the SB-ALIEN package, not in SB-EXT.
2889 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2890 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2891 features. Instead, the code that they used to control is always
2892 built into the system.
2893 * many other bug fixes
2894 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2895 advantage of the new EVAL-WHEN stuff and to clean them up in
2896 general, and they are now more ANSI-compliant in a number of
2897 ways. Martin Atzmueller is responsible for a lot of this.
2898 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2900 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2901 *** correct ERROR type for various file operations
2902 *** some fixes for Lisp streams
2903 *** DEFMETHOD syntax checking
2904 *** changing old weird representation of debug information as
2905 strings (which, among their other deficiencies, don't transform
2906 correctly when you rename packages, and don't change their
2907 print representation when you change things like *PACKAGE*
2908 and *PRINT-LENGTH*) to symbols and lists of symbols
2909 He also made several improvements and fixed several bugs in DESCRIBE.
2910 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2912 *** misbehavior of WRITE-STRING/WRITE-LINE
2913 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2914 and several other LOOP problems as well
2915 *** DIRECTORY when similar filenames are present
2916 *** DEFGENERIC with :METHOD options
2917 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2918 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2919 *** argument ordering in FIND with :TEST option
2920 *** mishandled package designator argument in APROPOS-LIST
2921 *** various problems in the backquote readmacro
2922 *** a bug in APROPOS
2923 *** probably some others that I'm not describing very well here,
2924 since the CVS log documents them by reference to sbcl-devel
2925 messages, and the SourceForge archives aren't working well.:-(
2926 ** Dan Barlow improved the Alpha port (and is making progress on the
2927 PPC port, for those of you who think different).
2928 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2929 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2930 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2931 continues to work on the SPARC port (for those of you in a position
2932 to look down upon our little PC-compatible boxes from a great height).
2933 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2934 and uses that copy, rather than the current dynamic value, when
2935 it's trying to decide whether to truncate output. Thus e.g.
2936 (let ((*print-lines* 50))
2937 (pprint-logical-block (stream nil)
2939 (let ((*print-lines* 8))
2940 (print (aref possiblybigthings i) stream)))))
2941 should now truncate the logical block only at 50 lines, instead of
2942 often truncating it at 8 lines, as it did before.
2943 * The doc/cmucl/ directory, containing old CMU CL documentation
2944 from the time of the fork, is no longer part of the base system.
2945 SourceForge has shut down its anonymous FTP service, and with it
2946 my original plan for distributing the old CMU CL documentation
2947 there. For now, if you need these files you can download an old
2948 SBCL source release and extract them from it.
2949 * The fasl file version number changed again, for dozens of reasons,
2950 some of which are apparent above.
2952 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2953 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2954 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2955 patch, so that DISASSEMBLE output is much nicer.
2956 * The code in the SB-PROFILE package now seems reasonably stable.
2957 I still haven't decided what the final interface should look like
2958 (I'd like PROFILE to interact cleanly with TRACE, since both
2959 facilities use function encapsulation) but if you have a need
2960 for profiling now, you can probably use it successfully with
2961 the current CMU-CL-style interface.
2962 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2963 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2964 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2965 link now returns the dangling link itself, and for similar
2966 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2967 link itself. (In these cases the old code signalled an error and
2968 looped endlessly, respectively.) Thus, DIRECTORY now works even
2969 in the presence of dangling and cyclic symbolic links.
2970 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2971 is now a supported extension again, since the consensus on
2972 sbcl-devel was that it can be useful for ordinary development
2973 work, not just for debugging SBCL itself.
2974 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2975 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2976 that currently-defined functions might be redefined later with
2977 different return types.
2978 * Hash tables can be printed readably, as inspired by CMU CL code
2979 of Eric Marsden and SBCL code of Martin Atzmueller.
2980 * better error handling in CLOS method combination, thanks to
2981 Martin Atzmueller porting Pierre Mai's CMU CL patches
2982 * more overflow fixes for >16Mbyte I/O buffers
2983 * A bug in READ has been fixed, so that now a single Ctrl-D
2984 character suffices to cause end-of-file on character streams.
2985 In particular, now you only need one Ctrl-D at the command
2986 line (not two) to exit SBCL.
2987 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2989 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2990 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2991 does the right thing.
2992 * The implementation of some type tests, especially for CONDITION
2993 types, is now tidier and maybe faster, due to CMU CL code
2994 originally by Douglas Crosher, ported by Martin Atzmueller.
2995 * Some math functions have been fixed, and there are new
2996 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2997 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2998 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2999 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3000 likely to be built into sbcl-0.7.0.)
3001 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3002 CL, ported to SBCL by Martin Atzmueller)
3003 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3004 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3005 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3006 * There's a new slam.sh hack to shorten the edit/compile/debug
3007 cycle for low-level changes to SBCL itself, and a new
3008 :SB-AFTER-XC-CORE target feature to control the generation of
3009 the after-xc.core file needed by slam.sh.
3010 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3011 COMPILE-FILE is no longer supported, so that now every function
3012 gets an entry point, so that block compilation looks a little
3013 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3014 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3015 returns the number of bytes consed since the system started,
3016 rather than the number consed since the first time the function
3017 was called. (The new definition parallels ANSI functions like
3018 CL:GET-INTERNAL-RUN-TIME.)
3019 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3020 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3021 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3022 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3023 * Fasl file version numbers are now independent of the target CPU,
3024 since historically most system changes which required version
3025 number changes have affected all CPUs equally. Similarly,
3026 the byte fasl file version is now equal to the ordinary
3029 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3030 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3031 is no longer recognized. I apologize for this, because it was
3032 listed in SB-EXT as a supported extension, but I found that
3033 its existing behavior was poorly specified, as well as incorrectly
3034 specified, and it looked like too much of a mess to straighten it
3035 out. I have enough on my hands trying to get ANSI stuff to work..
3036 * many patches ported from CMU CL by Martin Atzmueller, with
3037 half a dozen bug fixes in pretty-printing and the debugger, and
3038 half a dozen others elsewhere
3039 * fixed bug 13: Floating point infinities are now supported again.
3040 They might still be a little bit flaky, but thanks to bug reports
3041 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3042 as flaky as they were.
3043 * The --noprogrammer command line option is now supported. (Its
3044 behavior is slightly different in detail from what the old man
3045 page claimed it would do, but it's still appropriate under the
3046 same circumstances that the man page talks about.)
3047 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3048 are now supported, and enabled by default. Thus, the compiler can
3049 handle many floating point and complex operations much less
3050 inefficiently. (Thus e.g. you can implement a complex FFT
3052 * The compiler now detects type mismatches between DECLAIM FTYPE
3053 and DEFUN better, and implements CHECK-TYPE more correctly, and
3054 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3055 thanks to patches from Martin Atzmueller.
3056 * various fixes to make the cross-compiler more portable to
3057 ANSI-conforming-but-different cross-compilation hosts (notably
3058 Lispworks for Windows, following bug reports from Arthur Lemmens)
3059 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3060 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3061 fixed thanks to Pierre Mai's CMU CL patches.
3062 * fixes to keep the system from overflowing internal counters when
3063 it tries to use i/o buffers larger than 16M bytes
3064 * fixed bug 45a: Various internal functions required to support
3065 complex special functions have been merged from CMU CL sources.
3066 (When I was first setting up SBCL, I misunderstood a compile-time
3067 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3068 * improved support for type intersection and union, fixing bug 12
3069 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3070 more obscure bugs as well
3071 * some steps toward byte-compiling non-performance-critical
3072 parts of the system, courtesy of patches from Martin Atzmueller
3073 * Christophe Rhodes has made some debian packages of sbcl at
3074 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3075 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3076 stable, but are nonetheless usable. When he's ready, I'd be happy
3077 to add them to the SourceForge "File Releases" section. (And if
3078 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3079 * new fasl file format version number (because of changes in
3080 internal representation of (OR ..) types to accommodate the new
3081 support for (AND ..) types, among other things)
3083 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3084 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3086 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3088 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3089 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3090 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3091 * DESCRIBE now works on structure objects again.
3092 * Most function call argument type mismatches are now handled as
3093 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3094 know whether the function will be redefined before the call is
3095 executed. (The compiler could flag local calls with full WARNINGs,
3096 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3097 it doesn't keep track of enough information to know whether calls
3098 are local in this sense.)
3099 * Compiler output is now more verbose, with messages truncated
3100 later than before. (There should be some supported way for users
3101 to override the default verbosity, but I haven't decided how to
3102 provide it yet, so this behavior is still controlled by the internal
3103 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3104 src/compiler/ir1util.lisp.)
3105 * Fasl file format version numbers have increased again, because
3106 support for the Gray streams extension changes the layout of the
3107 system's STREAM objects.
3108 * The Gray subclassable streams extension now works, thanks to a
3109 patch from Martin Atzmueller.
3110 * The full LOAD-FOREIGN extension (not just the primitive
3111 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3112 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3113 but like most other programs, it defaults to copying the Unix
3114 environment from the original process instead of starting the
3115 new process in an empty environment.
3116 * Extensions which manipulate the Unix environment now support
3117 an :ENVIRONMENT keyword option which doesn't smash case or
3118 do other bad things. The CMU-CL-style :ENV option is retained
3119 for porting convenience.
3120 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3121 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3123 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3125 * A patch from Martin Atzmueller seems to have solved the SIGINT
3126 problem, and as far as we know, signal-handling now works cleanly.
3127 (If you find any new bugs, please report them!)
3128 * The system no longer defaults Lisp source file names to types
3129 ".l", ".cl", or ".lsp", but only to ".lisp".
3130 * The compiler no longer uses special default file extensions for
3131 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3132 seems to expect a single default extension for all compiled code,
3133 and there's no compelling reason to try to stretch the standard
3134 to allow two different extensions.) Instead, byte-compiled files
3135 default to the same extension as native-compiled files.
3136 * Fasl file format version numbers have increased again, because
3137 a rearrangement of internal implementation packages made some
3138 dumped symbols in old fasl files unreadable in new cores.
3139 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3140 many fewer weird special cases.
3141 * Bug #17 (differing COMPILE-FILE behavior between logical and
3142 physical pathnames) has been fixed, and some related misbehavior too,
3143 thanks to a patch from Martin Atzmueller.
3144 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3145 by Tim Moore, ported to SBCL by Martin Atzmueller.
3146 * Martin Atzmueller fixed several filesystem-related problems,
3147 including bug #36, in part by porting CMU CL patches, which were
3148 written in part by Paul Werkowski.
3149 * More compiler warnings in src/runtime/ are gone, thanks to
3150 more patches from Martin Atzmueller.
3151 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3154 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3156 * DESCRIBE now works on CONDITION objects.
3157 * The debugger now handles errors which arise when trying to print
3158 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3160 * The build system now uses an additional file, customize-target-features.lisp,
3161 to allow local modifications to the target *FEATURES* list. (The point of
3162 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3163 debugging features enabled, without having to worry about propagating it
3164 into everyone's system when I do a "cvs update".) When no
3165 customize-target-features.lisp file exists, the target *FEATURES* list
3166 should be constructed the same way as before.
3167 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3168 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3169 warn about behavior which is undefined under the ANSI spec. Note: This
3170 is specified by ANSI, but it's not very popular with programmers.
3171 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3172 macro in the SBCL sources for an example of a workaround which you
3173 might use to make portable ANSI-standard code which does what you want.
3174 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3175 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3176 when it's used as a non-toplevel form. (This is required in order
3177 to implement the DEFCONSTANT-EQX macro.)
3178 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3179 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3180 funny things, due to bugs in the implementation of EVAL-WHEN
3181 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3182 non-toplevel DEFCONSTANTs are uncommon.)
3183 * The core file version number and fasl file version number have been
3184 incremented, because the old noncompliant DEFCONSTANT behavior involved
3185 calling functions which no longer exist, and because I also took the
3186 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3187 * fixed bug 1 (error handling before read-eval-print loop starts), and
3188 redid debugger restarts and related debugger commands somewhat while
3190 ** The QUIT debugger command is gone, since it did something
3191 rather different than the SB-EXT:QUIT command, and since it never
3192 worked properly outside the main toplevel read/eval/print loop.
3193 Invoking the new TOPLEVEL restart provides the same functionality.
3194 ** The GO debugger command is also gone, since you can just invoke
3195 the CONTINUE restart directly instead.
3196 ** The TOP debugger command is also gone, since it's redundant with the
3197 FRAME 0 command, and since it interfered with abbreviations for the
3199 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3201 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3202 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3203 some porting work by Martin Atzmueller)
3204 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3205 with a boatload of patches which he ported to SBCL. Now that those
3207 ** The system tries to make sure that its low-priority messages
3208 are prefixed by semicolons, to help people who like to use
3209 syntax highlighting in their ILISP buffer. (This patch
3210 was originally due to Raymond Toy.)
3211 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3212 patches originally written by Raymond Toy.
3213 ** The compiler understands coercion between single-value and
3214 multiple-VALUES type expressions better, getting rid of some very
3215 weird behavior, thanks to patches originally by Robert MacLachlan
3216 and Douglas Crosher.
3217 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3218 lambda lists, thanks to a patch originally by Pierre Mai.
3219 ** The system no longer bogusly warns about "abbreviated type
3221 ** The compiler gets less confused by inlining and RETURN-FROM,
3222 thanks to some patches originally by Tim Moore.
3223 ** The system no longer hangs when dumping circular lists to fasl
3224 files, thanks to a patch originally from Douglas Crosher.
3225 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3226 undefined function error.
3227 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3228 in structure slot initforms) for now, documented workaround instead:-|
3229 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3230 * fixed bug 5: added stubs for various Gray stream functions called
3231 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3232 installed, at least appropriate type errors are generated
3233 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3234 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3235 consistently in DEFMETHOD forms.
3236 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3237 it doesn't seem to affect SBCL after all
3238 * The C runtime system now builds with better optimization and many
3239 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3241 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3243 * The system is now under CVS at SourceForge (instead of the
3244 CVS repository on my home machine).
3245 * The new signal handling code has been tweaked to treat register
3246 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3247 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3248 downstream, e.g. in debug-int.lisp, has implicit dependencies
3249 on the unsignedness of integer representation of machine words,
3250 and that caused the system to bomb out with infinite regress
3251 when trying to recover from type errors involving signed values,
3252 e.g. (BUTLAST '(1 2 3) -1).)
3253 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3254 (This was one of the bugs Peter Van Eynde reported back in July.)
3255 * The system now uses code inspired by Colin Walters' O(N)
3256 implementation of MAP (from the cmucl-imp@cons.org mailing
3257 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3258 inline the MAP operation, and there is more than one
3259 sequence argument to the MAP call (so that it can't just
3260 do ETYPECASE once and for all based on the type of the
3261 single sequence argument). (The old non-inline implementation
3262 of the general M-argument sequence-of-length-N case required
3263 O(M*N*N) time when any of the sequence arguments were LISTs.)
3264 * The QUIT :UNIX-CODE keyword argument has been renamed to
3265 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3266 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3268 * Raymond Wiker's patches to port dynamic loading from Linux to
3269 FreeBSD have been added.
3270 * The BUGS file is now more nearly up to date, thanks in large part
3271 to Martin Atzmueller's review of it.
3272 * The debugger now flushes standard output streams before it begins
3273 its output ("debugger invoked" and so forth).
3274 * The core version number and fasl file version number have both
3275 been incremented, because of incompatible changes in the layout
3277 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3278 used to not be called for a saved Lisp image.)
3279 * Martin Atzmueller's version of a patch to fix a compiler crash,
3280 as posted on sbcl-devel 13 September 2000, has been installed.
3281 * Instead of installing Martin Atzmueller's patch for the
3282 compiler transform for SUBSEQ, I deleted the compiler transform,
3283 and transforms for some similar consing operations.
3284 * A bug in signal handling which kept TRACE from working on OpenBSD
3286 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3287 away by constant folding
3288 * The system now defines its address space constants in one place
3289 (in the Lisp sources), and propagates them automatically elsewhere
3290 (through GENESIS and the sbcl.h file). Therefore, patching the
3291 address map is less unnecessarily tedious and error-prone. The
3292 Lisp names of address space constants have also been systematized.
3293 * CVS tags like dollar-Header-dollar have been removed from
3294 the sources, because they have never saved me trouble and
3295 they've been source of trouble working with patches and other
3296 diff-related operations.
3297 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3298 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3300 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3302 * The system has been ported to OpenBSD.
3303 * The system now compiles with a simple "sh make.sh" on the systems
3304 that it's supported on. I.e., now you no longer need to tweak
3305 text in the target-features.lisp-expr and symlinks in src/runtime/
3306 by hand, the make.sh takes care of it for you.
3307 * The system is no longer so grossly inefficient when compiling code
3308 involving vectors implemented as general (not simple) vectors (VECTOR T),
3309 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3310 which dares to use the various sequence functions on non-simple
3311 vectors, takes less of a performance hit.
3312 * There is now a primitive type predicate VECTOR-T-P
3313 to test for the (VECTOR T) type, so that e.g.
3314 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3315 can now be compiled with some semblance of efficiency. (The old code
3316 turned the type declaration into a full call to %TYPEP at runtime!)
3317 * AREF on (VECTOR T) is still not fast, since it's still compiled
3318 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3319 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3320 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3321 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3322 without making full calls to SUBTYPEP at runtime.
3323 (Some analogous efficiency issues for non-simple vectors specialized to
3324 element types other than T, or for non-simple multidimensional arrays,
3325 have not been addressed. They could almost certainly be handled the
3326 same way if anyone is motivated to do so.)
3327 * The changes in array handling break binary compatibility, so
3328 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3329 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3330 instead of (VALUES T T).
3331 * By following the instructions that Dan Barlow posted to sbcl-devel
3332 on 2 July 2000, I was able to enable primitive dynamic object
3333 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3334 functionality is not implemented (since it calls ld to resolve
3335 library references automatically, requiring RUN-PROGRAM for its
3336 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3337 try to resolve library references) is now supported.
3338 * The system now flushes the standard output streams when it terminates,
3339 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3340 them at several other probably-convenient times, e.g. in each pass of
3341 the toplevel read-eval-print loop, and after evaluating a form given
3342 as an "--eval" command-line option. (These changes were motivated by a
3343 discussion of stream flushing issues on cmucl-imp in August 2000.)
3344 * The source transform for TYPEP of array types no longer assumes
3345 that an array whose element type is a not-yet-defined type
3346 is implemented as an array of T, but instead punts, so that the
3347 type will be interpreted at runtime.
3348 * There is now some support for cross-compiling in make.sh: each of
3349 the phases of make.sh has its own script. (This should be transparent
3350 to people doing ordinary, non-cross-compile builds.)
3351 * Since my laptop doesn't have hundreds of megabytes of memory like
3352 my desktop machine, I became more motivated to do some items on
3353 my to-do list in order to reduce the size of the system a little:
3354 ** Arrange for various needed-only-at-cold-init things to be
3355 uninterned after cold init. To support this, those things have
3356 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3357 symbols with such names are now uninterned after cold init).
3358 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3359 fasl files for cold load.
3360 ** Remove the old compiler structure pooling code (which used to
3361 be conditional on the target feature :SB-ALLOC) completely.
3362 ** Redo the representation of some data in cold init to be more compact.
3363 (I also looked into supporting byte compiled code at bootstrap time,
3364 which would probably reduce the size of the system a lot, but that
3365 looked too complicated, so I punted for now.)
3366 * The maximum signal nesting depth in the src/runtime/ support code has
3367 been reduced from 4096 to 256. (I don't know any reason for the very
3368 large old value. If the new smaller value turns out to break something,
3369 I'll probably just bump it back up.)
3370 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3372 * Many, many bugs reported by Peter Van Eynde have been added to
3373 the BUGS list; some have even been fixed.
3374 * While enabling dynamic object file loading, I tried to make the
3375 code easier to understand, renaming various functions and variables
3376 with less ambiguous names, and changing some function calling
3377 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3378 * While trying to figure out how to do the OpenBSD port, I tried to
3379 clean up some of the code in src/runtime/. In particular, I dropped
3380 support for non-POSIX signal handling, added various comments,
3381 tweaked the code to reduce the number of compilation warnings, and
3382 renamed some files to increase consistency.
3383 * To support the new automatic configuration functionality in make.sh,
3384 the source file target-features.lisp-expr has been replaced with the
3385 source file base-target-features.lisp-expr and the machine-generated
3386 file local-target-features.lisp-expr.
3387 * fixed a stupid quoting error in make.sh so that using CMU CL
3388 "lisp -batch" as cross-compilation host works again
3390 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3392 * DESCRIBE no longer tries to call itself recursively to describe
3393 bound/fbound values, so that it no longer fails on symbols which are
3394 bound to themselves (like keywords, T, and NIL).
3395 * DESCRIBE now works on generic functions.
3396 * The printer now prints less-screwed-up representations of closures
3397 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3398 * A private symbol is used instead of the :EMPTY keyword previously
3399 used to mark empty slots in hash tables. Thus
3400 (DEFVAR *HT* (MAKE-HASH-TABLE))
3401 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3402 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3403 now does what ANSI says that it should. (You can still get
3404 similar noncompliant behavior if bang on the hash table
3405 implementation with all the symbols you get back from
3406 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3407 This breaks binary compatibility, since tests for equality to
3408 :EMPTY are wired into things like the macroexpansion of
3409 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3411 * There's now a minimal placeholder implementation for CL:STEP,
3412 as required by ANSI.
3413 * An obscure bug in the interaction of the normal compiler, the byte
3414 compiler, inlining, and structure predicates has been patched
3415 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3416 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3418 * Missing ordinary arguments in a macro call are now detected even
3419 when the macro lambda list contains &KEY or &REST.
3420 * The debugger no longer complains about encountering the top of the
3421 stack when you type "FRAME 0" to explicitly instruct it to go to
3422 the top of the stack. And it now prints the frame you request even
3423 if it's the current frame (instead of saying "You are here.").
3424 * As specified by ANSI, the system now always prints keywords
3425 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3427 * The default initial SIZE of HASH-TABLEs is now smaller.
3428 * Type information from CLOS class dispatch is now propagated
3429 into DEFMETHOD bodies, so that e.g.
3430 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3432 is now basically equivalent to
3433 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3434 (DECLARE (TYPE SINGLE-FLOAT X))
3436 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3437 operation, without having to do run-time type dispatch.
3438 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3439 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3440 * Since I don't seem to be making big file reorganizations very often
3441 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3442 is overflowing my ability to conveniently back them up), I've finally
3443 checked the system into CVS. (The CVS repository is on my home system,
3444 not at SourceForge -- putting it on SourceForge might come later.)
3445 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3446 high-level GC-NOTIFY-FOO functions send their output. (There's
3447 still very little control of where low-level verbose GC functions
3448 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3449 less than it used to -- the GC-NOTIFY-FOO functions are now under
3450 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3451 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3452 in only one place in the source code, and propagates it automatically
3453 everywhere that it's needed. Thus e.g. when I bump the version from
3454 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3455 * The C source files now include boilerplate legalese and documentation
3456 at the head of each file (just as the Lisp source files already did).
3457 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3458 to his page will be replaced with a link to his new CLiki service.
3460 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3462 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3463 * The build process now looks for GNU make under the default name "gmake",
3464 instead of "make" as it used to. If GNU make is not available as "gmake"
3465 on your system, you can change this default behavior by setting the
3466 GNUMAKE environment variable.
3467 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3468 can build without error under CMU CL.
3470 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3472 * There is now a partial SBCL user manual (with some new text and some
3473 text cribbed from the CMU CL manual).
3474 * The beginnings of a profiler have been added (starting with the
3475 CMU CL profiler and simplifying and cleaning up). Eventually the
3476 main interface should be through the TRACE macro, but for now,
3477 it's still accessed through vaguely CMU-CL-style functions and macros
3478 exported from the package SB-PROFILE.
3479 * Some problems left over from porting CMU CL to the new
3480 cross-compilation bootstrap process have been cleaned up:
3481 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3482 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3483 part when getting rid of the compiler *BACKEND* stuff.)
3484 ** The value of *NULL-TYPE* was screwed up, because it was
3485 being initialized before the type system knew the final
3486 definition of the 'NULL type. This screwed up several key
3487 optimizations in the compiler, causing inefficiency in all sorts
3488 of places. (I found it because I wanted to understand why
3489 GET-INTERNAL-RUN-TIME was consing.)
3490 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3491 PROCLAIM FTYPE information. Unfortunately this broke binary
3492 compatibility again, since now the forms output by DEFGENERIC
3493 to refer to functions which didn't exist in 0.6.3.
3494 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3495 can use the new (as of 0.6.3) transform for SOME into MAP into
3497 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3498 values to (INTEGER 0 1000000), so that the time code will no longer
3499 occasionally get blown up by Linux returning 1000000 microseconds
3500 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3501 its output conform to the ANSI spec. (Alas, this makes its output
3502 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3503 * A full call to MAP NIL with a single sequence argument no longer conses.
3504 * fixes to problems pointed out by Martin Atzmueller:
3505 * The manual page no longer talks about multiprocessing as though
3506 it were currently supported.
3507 * The ILISP support patches have been removed from the distribution,
3508 because as of version 5.10.1, ILISP now supports SBCL without us
3509 having to maintain patches.
3510 * added a modified version of Raymond Toy's recent CMU CL patch for
3511 EQUALP comparison of HASH-TABLE
3513 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3515 * The system still can't cross-compile itself with
3516 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3517 entails), but at least it can compile more of itself that way
3518 than it used to be able to, and various buglets which were uncovered
3519 by trying to cross-compile itself that way have now been fixed.
3520 * This release breaks binary compatibility again. This time
3521 at least I've incremented the FASL file format version to 2, so that the
3522 problem can be detected reliably instead of just causing weird errors.
3523 * various new style warnings:
3524 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3525 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3526 ** using the lexical binding of a variable named in the *FOO* style
3527 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3528 as specified by ANSI.
3529 * *RANDOM-STATE* is no longer automatically initialized from
3530 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3531 default behavior of the system is to repeat its behavior every time
3532 it's run. If you'd like to change this behavior, you can always
3533 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3534 old convention there was no comparably easy way to get the system to
3535 repeat its behavior every time it was run.
3536 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3537 FTYPE declarations has been removed, in favor of their ANSI
3538 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3539 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3540 the types of their sequence arguments can be determined at compile time.
3541 This is done through a new open code expansion for MAP which eliminates
3542 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3543 argument types can be determined at compile time.
3544 * The optimizer now transforms COERCE into an identity operation when it
3545 can prove that the coerced object is already of the correct type. (This
3546 can be a win for machine generated code, including the output of other
3547 optimization transforms, such as the MAP transform above.)
3548 * Credit information has been moved from source file headers into CREDITS.
3549 * Source file headers have been made more standard.
3550 * The CASE macro now compiles without complaining even when it has
3553 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3555 * (Note that the way that the PCL macroexpansions were rewritten
3556 to accommodate the change in DEFGENERIC below breaks binary
3557 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3558 not run under sbcl-0.6.2. Once we get out of alpha releases,
3559 i.e. hit release 1.0.0, we'll probably try to maintain binary
3560 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3561 and sbcl-1.4.4. Until then, however, it might be fairly common
3562 for maintenance releases to break binary compatibility.)
3563 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3565 * The reporting of unhandled signals has been changed to print some
3566 explanatory text as well as the report form. (Previously only
3567 the report form was printed.)
3568 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3569 it defines, so that the compiler no longer issues undefined function
3570 warnings for compiled-but-not-yet-loaded generic functions.
3571 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3572 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3573 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3574 section of the ANSI spec are supported.
3575 * The "" nickname for the "KEYWORD" package has been removed.
3576 The reader still handles symbol tokens which begin with a package marker
3577 as keywords, but it doesn't expose its mechanism for doing so in the
3578 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3579 * The system now issues STYLE-WARNINGs for contradictory TYPE
3580 proclamations. (Warnings for contradictory FTYPE proclamations would
3581 be nice too, but those can't be done usefully unless the type system
3582 is made smarter about FUNCTION types.)
3583 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3584 systematized, so that "*target-*.lisp is supposed to exist only on the
3585 target and imply that there's a related file which exists on the
3586 host, and *host-*.lisp is supposed to exist only on the host and imply
3587 that there's a related file which exists on the target. This involves a
3588 lot of renaming. Hopefully the acute confusion caused by the renaming
3589 will be justified by the reduction in chronic confusion..
3590 ** runtime-type.lisp -> early-target-type.lisp
3591 ** target-type.lisp -> late-target-type.lisp
3592 ** early-host-format.lisp -> early-format.lisp
3593 ** late-host-format.lisp -> late-format.lisp
3594 ** host-error.lisp -> misc-error.lisp
3595 ** early-error.lisp -> early-target-error.lisp
3596 ** late-error.lisp -> late-target-error.lisp
3597 ** host-defboot.lisp -> early-defboot.lisp
3598 ** code/misc.lisp -> code/target-misc.lisp
3599 ** code/host-misc.lisp -> code/misc.lisp
3600 ** code/numbers.lisp -> code/target-numbers.lisp
3601 ** code/early-numbers.lisp -> numbers.lisp
3602 ** early-host-type.lisp -> early-type.lisp
3603 ** late-host-type.lisp -> late-type.lisp
3604 ** host-typep.lisp -> typep.lisp
3605 ** load.lisp -> target-load.lisp
3606 ** host-load.lisp -> load.lisp
3607 ** host-disassem.lisp -> disassem.lisp
3608 ** host-insts.lisp -> insts.lisp
3609 ** byte-comp.lisp -> target-byte-comp.lisp
3610 ** host-byte-comp.lisp -> byte-comp.lisp
3611 ** host-signal.lisp -> signal.lisp
3612 ** host-defstruct.lisp -> defstruct.lisp
3613 ** late-target-type.lisp -> deftypes-for-target.lisp
3614 Furthermore, several other previously target-only files foo.lisp (e.g.
3615 hash-table.lisp and random.lisp) have been split into a target-and-host
3616 foo.lisp file and a target-only target-foo.lisp file, with their key type
3617 definitions in the target-and-host part, so that the cross-compiler will
3618 know more about target types.
3619 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3620 gone away. In their place are various *BACKEND-FOO* variables
3621 corresponding to the slots of the old structure.
3622 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3623 into the target SBCL has been fixed.
3624 * The chill.lisp system for loading cold code into a running SBCL
3626 * Support for the CMU CL "scavenger hook" extension has been removed.
3627 (It was undocumented and unused in the CMU CL sources that SBCL was
3628 derived from, and stale in sbcl-0.6.1.)
3629 * Various errors in the cross-compiler type system were detected
3630 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3631 (enabling various consistency checks). Many of them were fixed,
3632 but some hard problems remain, so the compiler is back to
3633 running without *TYPE-SYSTEM-INITIALIZED* for now.
3634 * As part of the cross-compiler type system cleanup, I implemented
3635 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3636 * I have started adding UNCROSS calls throughout the type system
3637 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3638 on cross-compiler input files will be able to go away, and various
3640 * CONSTANTP now returns true for quoted forms (as explicitly required
3643 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3645 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3646 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3647 (MAKE-LIST -1). In the longer term, it ought to become true
3648 that declarations are assertions even at SAFETY 1. For now, it's not
3649 quite true even at SAFETY 3, but it's at least more nearly true..
3650 (Note that this change seems to increases the size of the system by
3651 O(5%) and to decrease the speed of the compiler by 20% or more.)
3652 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3653 for the problem of printing dozens of lines of distracting information
3654 about low-level system machinery as part of the top stack frame
3655 on entry to the debugger when an undefined function was called.
3656 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3657 is not reset to COMMON-LISP-USER.
3658 * Compilation of stuff related to dyncount.lisp has been made conditional
3659 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3660 smaller. The various dyncount-related symbols have been moved into
3661 a new "SB-DYNCOUNT" package.
3662 * tty-inspect.lisp has been renamed to inspect.lisp.
3663 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3664 feature has gone away. (When we eventually port to other flavors of
3665 libc and/or Unix, we'll try to make the differences between flavors
3666 invisible at the user level.)
3667 * Various other *FEATURES* tags, and/or their associated conditionals,
3668 have been removed if obsolescent, or given better documentation, or
3669 sometimes given more-mnemonic names.
3671 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3673 * tidied up "make.sh" script
3674 * tidied up system directory structure
3675 * better "clean.sh" behavior
3676 * added doc/FOR-CMUCL-DEVELOPERS
3677 * many many small tweaks to output format, e.g. removing possibly-confusing
3678 trailing #\. character in DESCRIBE-INSTANCE
3679 * (EQUALP #\A 'A) no longer signals an error.
3680 * new hashing code, including EQUALP hashing
3681 * tidied up Lisp initialization and toplevel
3682 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3683 * command line argument processing
3684 * added POSIX-GETENV function to deal with Unix-ish environment variables
3685 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3686 terminating SBCL on EOF
3687 * non-verbose GC by default
3688 * There is no more "sbcl" shell script; the sbcl file is now the C
3689 runtime executable (just like CMU CL).
3690 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3692 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3693 symbol and package information as Lisp native symbol and package objects
3694 instead of strings naming symbols and strings naming packages. This way,
3695 whenever packages are renamed (as in warm init), debug information is
3696 transformed along with everything else.
3697 * tweaked the optimization policy declarations which control the building
3698 of SBCL itself. Now, among other things, the system no longer saves
3699 source location debugging information. (This helps two problems at once
3700 by reducing SBCL size and by keeping SBCL from trying to look for its
3701 sources -- which may not exist -- when reporting errors.)
3702 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3703 debugging and testing purposes
3704 * cleaned up printing, making the printer call PRINT-OBJECT for
3705 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3706 methods, giving nearly-ANSI behavior
3707 * converted almost all special variables to use *FOO* naming convention
3708 * deleted PARSE-TIME functionality, since it can be done portably
3709 * moved some files out of cold init into warm init
3710 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3712 * regularized formatting of source files
3713 * added an install.sh script
3714 * fixed ridiculous memory usage of cross-compiler by making
3715 compiler/alloc.lisp not try to do pooling unless it can hook
3716 itself into the GC of the cross-compilation host. Now the system
3717 builds nicely on my old laptop.
3718 * added :SB-ALLOC in target-features.lisp-expr
3719 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3720 was not implemented)
3721 * re-did condition handling and note reporting in the compiler. Notes
3722 are no longer handled by signalling conditions. Style warnings
3723 and warnings are handled more correctly and reported in such a way
3724 that it's easy to find one or the other in your output (so that you
3725 can e.g. figure out which of many problems caused COMPILE-FILE to
3727 * changed the severity of several compiler warnings from full WARNING
3728 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3729 compiler note reporting so that it doesn't use the condition system
3730 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3731 COMPILE-FILE command)
3732 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3733 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3734 no longer supported, since their implementation was deeply intertwingled
3735 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3736 * removed bogus "support" for compiler macros named (SETF FOO), and
3737 removed the compiler macro for SETF INFO (but only after making a fool
3738 of myself on the cmucl-imp mailing list by posting a bogus patch for
3739 DEFINE-COMPILER-MACRO..)
3740 * Compiled files containing forms which have side effects on the Lisp
3741 reader (such as DEFPACKAGE forms) are now handled more correctly.
3742 (Compiler queuing of top level lambdas has been suppressed by setting
3743 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3744 * deleted various currently-unused source files, e.g. gengc.lisp. They
3745 may be added back at some point e.g. when porting to other architectures,
3746 but until they are it's distracting to distribute them and to try to
3748 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3749 there were so many of them they're just distractions, and UNCROSS is
3750 known to be able to handle the current sources
3751 * moved PROFILE functionality into TRACE, so that it will be clear
3752 how the wrapping and unwrapping of functions when you profile them
3753 interacts with the wrapping and unwrapping of functions when you
3754 trace them. (Actually, the functionality isn't there yet, but at least
3755 the interface specification is there. Hopefully, the functionality will
3756 arrive with some future maintenance release.)
3757 * removed host-oops.lisp
3758 * changed signature of QUIT function to allow UNIX-CODE argument
3759 * fixed READ-SEQUENCE bug
3760 * tweaked verbose GC output so that it looks more like the progress
3761 output that ANSI specifies for functions like LOAD
3762 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3763 * added <http://sbcl.sourceforge.com> to the banner information printed by