1 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3 * tidied up "make.sh" script
4 * tidied up system directory structure
5 * better "clean.sh" behavior
6 * added doc/FOR-CMUCL-DEVELOPERS
7 * many many small tweaks to output format, e.g. removing possibly-confusing
8 trailing #\. character in DESCRIBE-INSTANCE
9 * (EQUALP #\A 'A) no longer signals an error.
10 * new hashing code, including EQUALP hashing
11 * tidied up Lisp initialization and toplevel
12 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
13 * command line argument processing
14 * added POSIX-GETENV function to deal with Unix-ish environment variables
15 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
16 terminating SBCL on EOF
17 * non-verbose GC by default
18 * There is no more "sbcl" shell script; the sbcl file is now the C
19 runtime executable (just like CMU CL).
20 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
22 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
23 symbol and package information as Lisp native symbol and package objects
24 instead of strings naming symbols and strings naming packages. This way,
25 whenever packages are renamed (as in warm init), debug information is
26 transformed along with everything else.
27 * tweaked the optimization policy declarations which control the building
28 of SBCL itself. Now, among other things, the system no longer saves
29 source location debugging information. (This helps two problems at once
30 by reducing SBCL size and by keeping SBCL from trying to look for its
31 sources -- which may not exist -- when reporting errors.)
32 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
33 debugging and testing purposes
34 * cleaned up printing, making the printer call PRINT-OBJECT for
35 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
36 methods, giving nearly-ANSI behavior
37 * converted almost all special variables to use *FOO* naming convention
38 * deleted PARSE-TIME functionality, since it can be done portably
39 * moved some files out of cold init into warm init
40 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
42 * regularized formatting of source files
43 * added an install.sh script
44 * fixed ridiculous memory usage of cross-compiler by making
45 compiler/alloc.lisp not try to do pooling unless it can hook
46 itself into the GC of the cross-compilation host. Now the system
47 builds nicely on my old laptop.
48 * added :SB-ALLOC in target-features.lisp-expr
49 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
51 * re-did condition handling and note reporting in the compiler. Notes
52 are no longer handled by signalling conditions. Style warnings
53 and warnings are handled more correctly and reported in such a way
54 that it's easy to find one or the other in your output (so that you
55 can e.g. figure out which of many problems caused COMPILE-FILE to
57 * changed the severity of several compiler warnings from full WARNING
58 to STYLE-WARNING in order to conform with the ANSI spec; also changed
59 compiler note reporting so that it doesn't use the condition system
60 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
62 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
63 function. As a consequence, START-BLOCK and END-BLOCK declarations are
64 no longer supported, since their implementation was deeply intertwingled
65 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
66 * removed bogus "support" for compiler macros named (SETF FOO), and
67 removed the compiler macro for SETF INFO (but only after making a fool
68 of myself on the cmucl-imp mailing list by posting a bogus patch for
69 DEFINE-COMPILER-MACRO..)
70 * Compiled files containing forms which have side effects on the Lisp
71 reader (such as DEFPACKAGE forms) are now handled more correctly.
72 (Compiler queuing of top level lambdas has been suppressed by setting
73 *TOP-LEVEL-LAMBDA-MAX* to 0. )
74 * deleted various currently-unused source files, e.g. gengc.lisp. They
75 may be added back at some point e.g. when porting to other architectures,
76 but until they are it's distracting to distribute them and to try to
78 * deleted "UNCROSS couldn't recurse through.." style warnings, since
79 there were so many of them they're just distractions, and UNCROSS is
80 known to be able to handle the current sources
81 * moved PROFILE functionality into TRACE, so that it will be clear
82 how the wrapping and unwrapping of functions when you profile them
83 interacts with the wrapping and unwrapping of functions when you
84 trace them. (Actually, the functionality isn't there yet, but at least
85 the interface specification is there. Hopefully, the functionality will
86 arrive with some future maintenance release.)
87 * removed host-oops.lisp
88 * changed signature of QUIT function to allow UNIX-CODE argument
89 * fixed READ-SEQUENCE bug
90 * tweaked verbose GC output so that it looks more like the progress
91 output that ANSI specifies for functions like LOAD
92 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
93 * added <http://sbcl.sourceforge.com> to the banner information printed by
96 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
98 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
99 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
100 (MAKE-LIST -1). In the longer term, it ought to become true
101 that declarations are assertions even at SAFETY 1. For now, it's not
102 quite true even at SAFETY 3, but it's at least more nearly true..
103 (Note that this change seems to increases the size of the system by
104 O(5%) and to decrease the speed of the compiler by 20% or more.)
105 * changed ALIEN printing to be much more abbreviated, as a short-term fix
106 for the problem of printing dozens of lines of distracting information
107 about low-level system machinery as part of the top stack frame
108 on entry to the debugger when an undefined function was called.
109 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
110 is not reset to COMMON-LISP-USER.
111 * Compilation of stuff related to dyncount.lisp has been made conditional
112 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
113 smaller. The various dyncount-related symbols have been moved into
114 a new "SB-DYNCOUNT" package.
115 * tty-inspect.lisp has been renamed to inspect.lisp.
116 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
117 feature has gone away. (When we eventually port to other flavors of
118 libc and/or Unix, we'll try to make the differences between flavors
119 invisible at the user level.)
120 * Various other *FEATURES* tags, and/or their associated conditionals,
121 have been removed if obsolescent, or given better documentation, or
122 sometimes given more-mnemonic names.
124 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
126 * (Note that the way that the PCL macroexpansions were rewritten
127 to accommodate the change in DEFGENERIC below breaks binary
128 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
129 not run under sbcl-0.6.2. Once we get out of alpha releases,
130 i.e. hit release 1.0.0, we'll probably try to maintain binary
131 compatibility between maintenance releases, e.g. between sbcl-1.4.3
132 and sbcl-1.4.4. Until then, however, it might be fairly common
133 for maintenance releases to break binary compatibility.)
134 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
136 * The reporting of unhandled signals has been changed to print some
137 explanatory text as well as the report form. (Previously only
138 the report form was printed.)
139 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
140 it defines, so that the compiler no longer issues undefined function
141 warnings for compiled-but-not-yet-loaded generic functions.
142 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
143 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
144 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
145 section of the ANSI spec are supported.
146 * The "" nickname for the "KEYWORD" package has been removed.
147 The reader still handles symbol tokens which begin with a package marker
148 as keywords, but it doesn't expose its mechanism for doing so in the
149 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
150 * The system now issues STYLE-WARNINGs for contradictory TYPE
151 proclamations. (Warnings for contradictory FTYPE proclamations would
152 be nice too, but those can't be done usefully unless the type system
153 is made smarter about FUNCTION types.)
154 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
155 systematized, so that "*target-*.lisp is supposed to exist only on the
156 target and imply that there's a related file which exists on the
157 host, and *host-*.lisp is supposed to exist only on the host and imply
158 that there's a related file which exists on the target. This involves a
159 lot of renaming. Hopefully the acute confusion caused by the renaming
160 will be justified by the reduction in chronic confusion..
161 ** runtime-type.lisp -> early-target-type.lisp
162 ** target-type.lisp -> late-target-type.lisp
163 ** early-host-format.lisp -> early-format.lisp
164 ** late-host-format.lisp -> late-format.lisp
165 ** host-error.lisp -> misc-error.lisp
166 ** early-error.lisp -> early-target-error.lisp
167 ** late-error.lisp -> late-target-error.lisp
168 ** host-defboot.lisp -> early-defboot.lisp
169 ** code/misc.lisp -> code/target-misc.lisp
170 ** code/host-misc.lisp -> code/misc.lisp
171 ** code/numbers.lisp -> code/target-numbers.lisp
172 ** code/early-numbers.lisp -> numbers.lisp
173 ** early-host-type.lisp -> early-type.lisp
174 ** late-host-type.lisp -> late-type.lisp
175 ** host-typep.lisp -> typep.lisp
176 ** load.lisp -> target-load.lisp
177 ** host-load.lisp -> load.lisp
178 ** host-disassem.lisp -> disassem.lisp
179 ** host-insts.lisp -> insts.lisp
180 ** byte-comp.lisp -> target-byte-comp.lisp
181 ** host-byte-comp.lisp -> byte-comp.lisp
182 ** host-signal.lisp -> signal.lisp
183 ** host-defstruct.lisp -> defstruct.lisp
184 ** late-target-type.lisp -> deftypes-for-target.lisp
185 Furthermore, several other previously target-only files foo.lisp (e.g.
186 hash-table.lisp and random.lisp) have been split into a target-and-host
187 foo.lisp file and a target-only target-foo.lisp file, with their key type
188 definitions in the target-and-host part, so that the cross-compiler will
189 know more about target types.
190 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
191 gone away. In their place are various *BACKEND-FOO* variables
192 corresponding to the slots of the old structure.
193 * A bug which caused the SB-COLD bootstrap-time package to be propagated
194 into the target SBCL has been fixed.
195 * The chill.lisp system for loading cold code into a running SBCL
197 * Support for the CMU CL "scavenger hook" extension has been removed.
198 (It was undocumented and unused in the CMU CL sources that SBCL was
199 derived from, and stale in sbcl-0.6.1.)
200 * Various errors in the cross-compiler type system were detected
201 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
202 (enabling various consistency checks). Many of them were fixed,
203 but some hard problems remain, so the compiler is back to
204 running without *TYPE-SYSTEM-INITIALIZED* for now.
205 * As part of the cross-compiler type system cleanup, I implemented
206 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
207 * I have started adding UNCROSS calls throughout the type system
208 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
209 on cross-compiler input files will be able to go away, and various
211 * CONSTANTP now returns true for quoted forms (as explicitly required
214 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
216 * The system still can't cross-compile itself with
217 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
218 entails), but at least it can compile more of itself that way
219 than it used to be able to, and various buglets which were uncovered
220 by trying to cross-compile itself that way have now been fixed.
221 * This release breaks binary compatibility again. This time
222 at least I've incremented the FASL file format version to 2, so that the
223 problem can be detected reliably instead of just causing weird errors.
224 * various new style warnings:
225 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
226 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
227 ** using the lexical binding of a variable named in the *FOO* style
228 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
229 as specified by ANSI.
230 * *RANDOM-STATE* is no longer automatically initialized from
231 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
232 default behavior of the system is to repeat its behavior every time
233 it's run. If you'd like to change this behavior, you can always
234 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
235 old convention there was no comparably easy way to get the system to
236 repeat its behavior every time it was run.
237 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
238 FTYPE declarations has been removed, in favor of their ANSI
239 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
240 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
241 the types of their sequence arguments can be determined at compile time.
242 This is done through a new open code expansion for MAP which eliminates
243 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
244 argument types can be determined at compile time.
245 * The optimizer now transforms COERCE into an identity operation when it
246 can prove that the coerced object is already of the correct type. (This
247 can be a win for machine generated code, including the output of other
248 optimization transforms, such as the MAP transform above.)
249 * Credit information has been moved from source file headers into CREDITS.
250 * Source file headers have been made more standard.
251 * The CASE macro now compiles without complaining even when it has
254 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
256 * There is now a partial SBCL user manual (with some new text and some
257 text cribbed from the CMU CL manual).
258 * The beginnings of a profiler have been added (starting with the
259 CMU CL profiler and simplifying and cleaning up). Eventually the
260 main interface should be through the TRACE macro, but for now,
261 it's still accessed through vaguely CMU-CL-style functions and macros
262 exported from the package SB-PROFILE.
263 * Some problems left over from porting CMU CL to the new
264 cross-compilation bootstrap process have been cleaned up:
265 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
266 instead of SB!XC:DEFMACRO, compounded by an oversight on my
267 part when getting rid of the compiler *BACKEND* stuff.)
268 ** The value of *NULL-TYPE* was screwed up, because it was
269 being initialized before the type system knew the final
270 definition of the 'NULL type. This screwed up several key
271 optimizations in the compiler, causing inefficiency in all sorts
272 of places. (I found it because I wanted to understand why
273 GET-INTERNAL-RUN-TIME was consing.)
274 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
275 PROCLAIM FTYPE information. Unfortunately this broke binary
276 compatibility again, since now the forms output by DEFGENERIC
277 to refer to functions which didn't exist in 0.6.3.
278 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
279 can use the new (as of 0.6.3) transform for SOME into MAP into
281 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
282 values to (INTEGER 0 1000000), so that the time code will no longer
283 occasionally get blown up by Linux returning 1000000 microseconds
284 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
285 its output conform to the ANSI spec. (Alas, this makes its output
286 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
287 * A full call to MAP NIL with a single sequence argument no longer conses.
288 * fixes to problems pointed out by Martin Atzmueller:
289 * The manual page no longer talks about multiprocessing as though
290 it were currently supported.
291 * The ILISP support patches have been removed from the distribution,
292 because as of version 5.10.1, ILISP now supports SBCL without us
293 having to maintain patches.
294 * added a modified version of Raymond Toy's recent CMU CL patch for
295 EQUALP comparison of HASH-TABLE
297 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
299 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
300 * The build process now looks for GNU make under the default name "gmake",
301 instead of "make" as it used to. If GNU make is not available as "gmake"
302 on your system, you can change this default behavior by setting the
303 GNUMAKE environment variable.
304 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
305 can build without error under CMU CL.
307 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
309 * DESCRIBE no longer tries to call itself recursively to describe
310 bound/fbound values, so that it no longer fails on symbols which are
311 bound to themselves (like keywords, T, and NIL).
312 * DESCRIBE now works on generic functions.
313 * The printer now prints less-screwed-up representations of closures
314 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
315 * A private symbol is used instead of the :EMPTY keyword previously
316 used to mark empty slots in hash tables. Thus
317 (DEFVAR *HT* (MAKE-HASH-TABLE))
318 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
319 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
320 now does what ANSI says that it should. (You can still get
321 similar noncompliant behavior if bang on the hash table
322 implementation with all the symbols you get back from
323 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
324 This breaks binary compatibility, since tests for equality to
325 :EMPTY are wired into things like the macroexpansion of
326 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
328 * There's now a minimal placeholder implementation for CL:STEP,
330 * An obscure bug in the interaction of the normal compiler, the byte
331 compiler, inlining, and structure predicates has been patched
332 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
333 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
335 * Missing ordinary arguments in a macro call are now detected even
336 when the macro lambda list contains &KEY or &REST.
337 * The debugger no longer complains about encountering the top of the
338 stack when you type "FRAME 0" to explicitly instruct it to go to
339 the top of the stack. And it now prints the frame you request even
340 if it's the current frame (instead of saying "You are here.").
341 * As specified by ANSI, the system now always prints keywords
342 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
344 * The default initial SIZE of HASH-TABLEs is now smaller.
345 * Type information from CLOS class dispatch is now propagated
346 into DEFMETHOD bodies, so that e.g.
347 (DEFMETHOD FOO ((X SINGLE-FLOAT))
349 is now basically equivalent to
350 (DEFMETHOD FOO ((X SINGLE-FLOAT))
351 (DECLARE (TYPE SINGLE-FLOAT X))
353 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
354 operation, without having to do run-time type dispatch.
355 * The macroexpansion of DEFMETHOD has been tweaked so that it has
356 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
357 * Since I don't seem to be making big file reorganizations very often
358 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
359 is overflowing my ability to conveniently back them up), I've finally
360 checked the system into CVS. (The CVS repository is on my home system,
361 not at SourceForge -- putting it on SourceForge might come later.)
362 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
363 high-level GC-NOTIFY-FOO functions send their output. (There's
364 still very little control of where low-level verbose GC functions
365 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
366 less than it used to -- the GC-NOTIFY-FOO functions are now under
367 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
368 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
369 in only one place in the source code, and propagates it automatically
370 everywhere that it's needed. Thus e.g. when I bump the version from
371 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
372 * The C source files now include boilerplate legalese and documentation
373 at the head of each file (just as the Lisp source files already did).
374 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
375 to his page will be replaced with a link to his new CLiki service.
377 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
379 * The system has been ported to OpenBSD.
380 * The system now compiles with a simple "sh make.sh" on the systems
381 that it's supported on. I.e., now you no longer need to tweak
382 text in the target-features.lisp-expr and symlinks in src/runtime/
383 by hand, the make.sh takes care of it for you.
384 * The system is no longer so grossly inefficient when compiling code
385 involving vectors implemented as general (not simple) vectors (VECTOR T),
386 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
387 which dares to use the various sequence functions on non-simple
388 vectors, takes less of a performance hit.
389 * There is now a primitive type predicate VECTOR-T-P
390 to test for the (VECTOR T) type, so that e.g.
391 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
392 can now be compiled with some semblance of efficiency. (The old code
393 turned the type declaration into a full call to %TYPEP at runtime!)
394 * AREF on (VECTOR T) is still not fast, since it's still compiled
395 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
396 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
397 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
398 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
399 without making full calls to SUBTYPEP at runtime.
400 (Some analogous efficiency issues for non-simple vectors specialized to
401 element types other than T, or for non-simple multidimensional arrays,
402 have not been addressed. They could almost certainly be handled the
403 same way if anyone is motivated to do so.)
404 * The changes in array handling break binary compatibility, so
405 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
406 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
407 instead of (VALUES T T).
408 * By following the instructions that Dan Barlow posted to sbcl-devel
409 on 2 July 2000, I was able to enable primitive dynamic object
410 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
411 functionality is not implemented (since it calls ld to resolve
412 library references automatically, requiring RUN-PROGRAM for its
413 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
414 try to resolve library references) is now supported.
415 * The system now flushes the standard output streams when it terminates,
416 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
417 them at several other probably-convenient times, e.g. in each pass of
418 the toplevel read-eval-print loop, and after evaluating a form given
419 as an "--eval" command-line option. (These changes were motivated by a
420 discussion of stream flushing issues on cmucl-imp in August 2000.)
421 * The source transform for TYPEP of array types no longer assumes
422 that an array whose element type is a not-yet-defined type
423 is implemented as an array of T, but instead punts, so that the
424 type will be interpreted at runtime.
425 * There is now some support for cross-compiling in make.sh: each of
426 the phases of make.sh has its own script. (This should be transparent
427 to people doing ordinary, non-cross-compile builds.)
428 * Since my laptop doesn't have hundreds of megabytes of memory like
429 my desktop machine, I became more motivated to do some items on
430 my to-do list in order to reduce the size of the system a little:
431 ** Arrange for various needed-only-at-cold-init things to be
432 uninterned after cold init. To support this, those things have
433 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
434 symbols with such names are now uninterned after cold init).
435 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
436 fasl files for cold load.
437 ** Remove the old compiler structure pooling code (which used to
438 be conditional on the target feature :SB-ALLOC) completely.
439 ** Redo the representation of some data in cold init to be more compact.
440 (I also looked into supporting byte compiled code at bootstrap time,
441 which would probably reduce the size of the system a lot, but that
442 looked too complicated, so I punted for now.)
443 * The maximum signal nesting depth in the src/runtime/ support code has
444 been reduced from 4096 to 256. (I don't know any reason for the very
445 large old value. If the new smaller value turns out to break something,
446 I'll probably just bump it back up.)
447 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
449 * Many, many bugs reported by Peter Van Eynde have been added to
450 the BUGS list; some have even been fixed.
451 * While enabling dynamic object file loading, I tried to make the
452 code easier to understand, renaming various functions and variables
453 with less ambiguous names, and changing some function calling
454 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
455 * While trying to figure out how to do the OpenBSD port, I tried to
456 clean up some of the code in src/runtime/. In particular, I dropped
457 support for non-POSIX signal handling, added various comments,
458 tweaked the code to reduce the number of compilation warnings, and
459 renamed some files to increase consistency.
460 * To support the new automatic configuration functionality in make.sh,
461 the source file target-features.lisp-expr has been replaced with the
462 source file base-target-features.lisp-expr and the machine-generated
463 file local-target-features.lisp-expr.
464 * fixed a stupid quoting error in make.sh so that using CMU CL
465 "lisp -batch" as cross-compilation host works again
467 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
469 * The system is now under CVS at SourceForge (instead of the
470 CVS repository on my home machine).
471 * The new signal handling code has been tweaked to treat register
472 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
473 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
474 downstream, e.g. in debug-int.lisp, has implicit dependencies
475 on the unsignedness of integer representation of machine words,
476 and that caused the system to bomb out with infinite regress
477 when trying to recover from type errors involving signed values,
478 e.g. (BUTLAST '(1 2 3) -1).)
479 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
480 (This was one of the bugs Peter Van Eynde reported back in July.)
481 * The system now uses code inspired by Colin Walters' O(N)
482 implementation of MAP (from the cmucl-imp@cons.org mailing
483 list, 2 September 2000) when it can't use a DEFTRANSFORM to
484 inline the MAP operation, and there is more than one
485 sequence argument to the MAP call (so that it can't just
486 do ETYPECASE once and for all based on the type of the
487 single sequence argument). (The old non-inline implementation
488 of the general M-argument sequence-of-length-N case required
489 O(M*N*N) time when any of the sequence arguments were LISTs.)
490 * The QUIT :UNIX-CODE keyword argument has been renamed to
491 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
492 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
494 * Raymond Wiker's patches to port dynamic loading from Linux to
495 FreeBSD have been added.
496 * The BUGS file is now more nearly up to date, thanks in large part
497 to Martin Atzmueller's review of it.
498 * The debugger now flushes standard output streams before it begins
499 its output ("debugger invoked" and so forth).
500 * The core version number and fasl file version number have both
501 been incremented, because of incompatible changes in the layout
503 * FINISH-OUTPUT is now called more consistently on QUIT. (It
504 used to not be called for a saved Lisp image.)
505 * Martin Atzmueller's version of a patch to fix a compiler crash,
506 as posted on sbcl-devel 13 September 2000, has been installed.
507 * Instead of installing Martin Atzmueller's patch for the
508 compiler transform for SUBSEQ, I deleted the compiler transform,
509 and transforms for some similar consing operations.
510 * A bug in signal handling which kept TRACE from working on OpenBSD
512 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
513 away by constant folding
514 * The system now defines its address space constants in one place
515 (in the Lisp sources), and propagates them automatically elsewhere
516 (through GENESIS and the sbcl.h file). Therefore, patching the
517 address map is less unnecessarily tedious and error-prone. The
518 Lisp names of address space constants have also been systematized.
519 * CVS tags like dollar-Header-dollar have been removed from
520 the sources, because they have never saved me trouble and
521 they've been source of trouble working with patches and other
522 diff-related operations.
523 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
524 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
526 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
528 * DESCRIBE now works on CONDITION objects.
529 * The debugger now handles errors which arise when trying to print
530 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
532 * The build system now uses an additional file, customize-target-features.lisp,
533 to allow local modifications to the target *FEATURES* list. (The point of
534 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
535 debugging features enabled, without having to worry about propagating it
536 into everyone's system when I do a "cvs update".) When no
537 customize-target-features.lisp file exists, the target *FEATURES* list
538 should be constructed the same way as before.
539 * fixed bugs in DEFCONSTANT ANSI-compatibility:
540 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
541 warn about behavior which is undefined under the ANSI spec. Note: This
542 is specified by ANSI, but it's not very popular with programmers.
543 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
544 macro in the SBCL sources for an example of a workaround which you
545 might use to make portable ANSI-standard code which does what you want.
546 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
547 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
548 when it's used as a non-toplevel form. (This is required in order
549 to implement the DEFCONSTANT-EQX macro.)
550 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
551 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
552 funny things, due to bugs in the implementation of EVAL-WHEN
553 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
554 non-toplevel DEFCONSTANTs are uncommon.)
555 * The core file version number and fasl file version number have been
556 incremented, because the old noncompliant DEFCONSTANT behavior involved
557 calling functions which no longer exist, and because I also took the
558 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
559 * fixed bug 1 (error handling before read-eval-print loop starts), and
560 redid debugger restarts and related debugger commands somewhat while
562 ** The QUIT debugger command is gone, since it did something
563 rather different than the SB-EXT:QUIT command, and since it never
564 worked properly outside the main toplevel read/eval/print loop.
565 Invoking the new TOPLEVEL restart provides the same functionality.
566 ** The GO debugger command is also gone, since you can just invoke
567 the CONTINUE restart directly instead.
568 ** The TOP debugger command is also gone, since it's redundant with the
569 FRAME 0 command, and since it interfered with abbreviations for the
571 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
573 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
574 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
575 some porting work by Martin Atzmueller)
576 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
577 with a boatload of patches which he ported to SBCL. Now that those
579 ** The system tries to make sure that its low-priority messages
580 are prefixed by semicolons, to help people who like to use
581 syntax highlighting in their ILISP buffer. (This patch
582 was originally due to Raymond Toy.)
583 ** The system now optimizes INTEGER-LENGTH better, thanks to more
584 patches originally written by Raymond Toy.
585 ** The compiler understands coercion between single-value and
586 multiple-VALUES type expressions better, getting rid of some very
587 weird behavior, thanks to patches originally by Robert MacLachlan
589 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
590 lambda lists, thanks to a patch originally by Pierre Mai.
591 ** The system no longer bogusly warns about "abbreviated type
593 ** The compiler gets less confused by inlining and RETURN-FROM,
594 thanks to some patches originally by Tim Moore.
595 ** The system no longer hangs when dumping circular lists to fasl
596 files, thanks to a patch originally from Douglas Crosher.
597 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
598 undefined function error.
599 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
600 in structure slot initforms) for now, documented workaround instead:-|
601 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
602 * fixed bug 5: added stubs for various Gray stream functions called
603 in the not-a-CL:STREAM case, so that even when Gray streams aren't
604 installed, at least appropriate type errors are generated
605 * fixed bug 8: better reporting of various PROGRAM-ERRORs
606 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
607 consistently in DEFMETHOD forms.
608 * removed bug 21 from BUGS, since Martin Atzmueller points out that
609 it doesn't seem to affect SBCL after all
610 * The C runtime system now builds with better optimization and many
611 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
613 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
615 * A patch from Martin Atzmueller seems to have solved the SIGINT
616 problem, and as far as we know, signal-handling now works cleanly.
617 (If you find any new bugs, please report them!)
618 * The system no longer defaults Lisp source file names to types
619 ".l", ".cl", or ".lsp", but only to ".lisp".
620 * The compiler no longer uses special default file extensions for
621 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
622 seems to expect a single default extension for all compiled code,
623 and there's no compelling reason to try to stretch the standard
624 to allow two different extensions.) Instead, byte-compiled files
625 default to the same extension as native-compiled files.
626 * Fasl file format version numbers have increased again, because
627 a rearrangement of internal implementation packages made some
628 dumped symbols in old fasl files unreadable in new cores.
629 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
630 many fewer weird special cases.
631 * Bug #17 (differing COMPILE-FILE behavior between logical and
632 physical pathnames) has been fixed, and some related misbehavior too,
633 thanks to a patch from Martin Atzmueller.
634 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
635 by Tim Moore, ported to SBCL by Martin Atzmueller.
636 * Martin Atzmueller fixed several filesystem-related problems,
637 including bug #36, in part by porting CMU CL patches, which were
638 written in part by Paul Werkowski.
639 * More compiler warnings in src/runtime/ are gone, thanks to
640 more patches from Martin Atzmueller.
641 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
644 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
645 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
647 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
649 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
650 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
651 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
652 * DESCRIBE now works on structure objects again.
653 * Most function call argument type mismatches are now handled as
654 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
655 know whether the function will be redefined before the call is
656 executed. (The compiler could flag local calls with full WARNINGs,
657 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
658 it doesn't keep track of enough information to know whether calls
659 are local in this sense.)
660 * Compiler output is now more verbose, with messages truncated
661 later than before. (There should be some supported way for users
662 to override the default verbosity, but I haven't decided how to
663 provide it yet, so this behavior is still controlled by the internal
664 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
665 src/compiler/ir1util.lisp.)
666 * Fasl file format version numbers have increased again, because
667 support for the Gray streams extension changes the layout of the
668 system's STREAM objects.
669 * The Gray subclassable streams extension now works, thanks to a
670 patch from Martin Atzmueller.
671 * The full LOAD-FOREIGN extension (not just the primitive
672 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
673 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
674 but like most other programs, it defaults to copying the Unix
675 environment from the original process instead of starting the
676 new process in an empty environment.
677 * Extensions which manipulate the Unix environment now support
678 an :ENVIRONMENT keyword option which doesn't smash case or
679 do other bad things. The CMU-CL-style :ENV option is retained
680 for porting convenience.
681 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
682 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
684 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
685 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
686 is no longer recognized. I apologize for this, because it was
687 listed in SB-EXT as a supported extension, but I found that
688 its existing behavior was poorly specified, as well as incorrectly
689 specified, and it looked like too much of a mess to straighten it
690 out. I have enough on my hands trying to get ANSI stuff to work..
691 * many patches ported from CMU CL by Martin Atzmueller, with
692 half a dozen bug fixes in pretty-printing and the debugger, and
693 half a dozen others elsewhere
694 * fixed bug 13: Floating point infinities are now supported again.
695 They might still be a little bit flaky, but thanks to bug reports
696 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
697 as flaky as they were.
698 * The --noprogrammer command line option is now supported. (Its
699 behavior is slightly different in detail from what the old man
700 page claimed it would do, but it's still appropriate under the
701 same circumstances that the man page talks about.)
702 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
703 are now supported, and enabled by default. Thus, the compiler can
704 handle many floating point and complex operations much less
705 inefficiently. (Thus e.g. you can implement a complex FFT
707 * The compiler now detects type mismatches between DECLAIM FTYPE
708 and DEFUN better, and implements CHECK-TYPE more correctly, and
709 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
710 thanks to patches from Martin Atzmueller.
711 * various fixes to make the cross-compiler more portable to
712 ANSI-conforming-but-different cross-compilation hosts (notably
713 Lispworks for Windows, following bug reports from Arthur Lemmens)
714 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
715 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
716 fixed thanks to Pierre Mai's CMU CL patches.
717 * fixes to keep the system from overflowing internal counters when
718 it tries to use i/o buffers larger than 16M bytes
719 * fixed bug 45a: Various internal functions required to support
720 complex special functions have been merged from CMU CL sources.
721 (When I was first setting up SBCL, I misunderstood a compile-time
722 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
723 * improved support for type intersection and union, fixing bug 12
724 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
725 more obscure bugs as well
726 * some steps toward byte-compiling non-performance-critical
727 parts of the system, courtesy of patches from Martin Atzmueller
728 * Christophe Rhodes has made some debian packages of sbcl at
729 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
730 From his sbcl-devel e-mail of 2001-04-08 they're not completely
731 stable, but are nonetheless usable. When he's ready, I'd be happy
732 to add them to the SourceForge "File Releases" section. (And if
733 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
734 * new fasl file format version number (because of changes in
735 internal representation of (OR ..) types to accommodate the new
736 support for (AND ..) types, among other things)
738 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
739 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
740 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
741 patch, so that DISASSEMBLE output is much nicer.
742 * The code in the SB-PROFILE package now seems reasonably stable.
743 I still haven't decided what the final interface should look like
744 (I'd like PROFILE to interact cleanly with TRACE, since both
745 facilities use function encapsulation) but if you have a need
746 for profiling now, you can probably use it successfully with
747 the current CMU-CL-style interface.
748 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
749 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
750 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
751 link now returns the dangling link itself, and for similar
752 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
753 link itself. (In these cases the old code signalled an error and
754 looped endlessly, respectively.) Thus, DIRECTORY now works even
755 in the presence of dangling and cyclic symbolic links.
756 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
757 is now a supported extension again, since the consensus on
758 sbcl-devel was that it can be useful for ordinary development
759 work, not just for debugging SBCL itself.
760 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
761 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
762 that currently-defined functions might be redefined later with
763 different return types.
764 * Hash tables can be printed readably, as inspired by CMU CL code
765 of Eric Marsden and SBCL code of Martin Atzmueller.
766 * better error handling in CLOS method combination, thanks to
767 Martin Atzmueller porting Pierre Mai's CMU CL patches
768 * more overflow fixes for >16Mbyte I/O buffers
769 * A bug in READ has been fixed, so that now a single Ctrl-D
770 character suffices to cause end-of-file on character streams.
771 In particular, now you only need one Ctrl-D at the command
772 line (not two) to exit SBCL.
773 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
775 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
776 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
777 does the right thing.
778 * The implementation of some type tests, especially for CONDITION
779 types, is now tidier and maybe faster, due to CMU CL code
780 originally by Douglas Crosher, ported by Martin Atzmueller.
781 * Some math functions have been fixed, and there are new
782 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
783 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
784 * (There are also some new optimizers in contrib/*-extras.lisp. Those
785 aren't built into sbcl-0.6.13, but are a sneak preview of what's
786 likely to be built into sbcl-0.7.0.)
787 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
788 CL, ported to SBCL by Martin Atzmueller)
789 * DESCRIBE now gives more information in some cases. (Pierre Mai's
790 patch to CMU CL, ported to SBCL by Martin Atzmueller)
791 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
792 * There's a new slam.sh hack to shorten the edit/compile/debug
793 cycle for low-level changes to SBCL itself, and a new
794 :SB-AFTER-XC-CORE target feature to control the generation of
795 the after-xc.core file needed by slam.sh.
796 * minor incompatible change: The ENTRY-POINTS &KEY argument to
797 COMPILE-FILE is no longer supported, so that now every function
798 gets an entry point, so that block compilation looks a little
799 more like the plain vanilla ANSI section 3.2.2.3 scheme.
800 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
801 returns the number of bytes consed since the system started,
802 rather than the number consed since the first time the function
803 was called. (The new definition parallels ANSI functions like
804 CL:GET-INTERNAL-RUN-TIME.)
805 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
806 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
807 supported. Now DIRECTORY always does the abstract Common-Lisp-y
808 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
809 * Fasl file version numbers are now independent of the target CPU,
810 since historically most system changes which required version
811 number changes have affected all CPUs equally. Similarly,
812 the byte fasl file version is now equal to the ordinary
815 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
816 * major incompatible change: The default fasl file extension, i.e. the
817 default extension for files produced by COMPILE-FILE, has changed
818 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
820 ** There are many changes in the implementation of the compiler.
821 SBCL is now essentially a compiler-only implementation of ANSI
822 Common Lisp. EVAL still "interprets" a few special cases, but
823 almost all the interesting cases are handled by creating
824 a LAMBDA expression, calling COMPILE on it, then calling
825 FUNCALL on the result.
826 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
827 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
828 Since the code is newer, there might still be some new bugs
829 (though not as many as before Martin Atzmueller's fixes:-). But
830 the new code is substantially simpler and clearer, and hopefully
831 any remaining bugs will be simpler, less fundamental, and more
832 fixable then the bugs in the old code.
833 ** The revised compiler is still a little unsteady on its feet.
835 *** The debugging information it produces (particularly the names
836 of FUNCTION objects) is sometimes much less useful than what
837 the old compiler produced.
838 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
839 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
840 within a MACROLET) has been temporarily weakened.
841 ** There are new compiler optimizations for various functions:
842 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
843 FIND-IF-NOT, POSITION-IF-NOT, and FILL
844 *** the math functions TRUNCATE, FLOOR, and CEILING
845 *** the function-of-all-trades COERCE
846 Mostly these should be transparent, but there's one
847 potentially-annoying problem (bug 117): when the compiler
848 inline-expands a function and does type analysis on the result,
849 it can create control paths which have type mismatches, and
850 when it can't prove that those control paths aren't taken,
851 it will issue WARNINGs about the type mismatches. This is
852 a particular problem in practice for the new sequence functions.
853 It's not clear how this should be fixed, and for now, a
854 workaround is given in the entry for 117 in the BUGS file.
855 ** (Because of the interaction between the two previous items --
856 occasional inlining problems and new inline expansions -- some
857 of the new sequence function optimizations won't really kick in
858 completely until debugging information, and then inlining, are
859 straightened out in some future version.)
860 * minor incompatible changes:
861 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
862 DIRECTORY no longer implicitly promotes NIL slots of its
863 pathname argument to :WILD. In particular, when you ask for the
864 contents of a directory (which you used to be able to do without
865 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
866 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
867 ** changes in behavior that ANSI explicitly defines to be
868 implementation dependent:
869 *** The new compiler-only implementation still conforms with ANSI,
870 but acts a little different than before. Besides the obvious
871 changes in performance tradeoffs (that the cost per form passed
872 to EVAL has gone up, and the cost per form executed by EVAL
873 has gone down), the behavior of the system changes a little
874 because there are no longer any interpreted function objects.
875 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
876 e.g. doing COMPILE on the output of interactive DEFUN is
878 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
880 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
881 has changed from (:CL) to NIL.
882 *** The CHAR-NAME of unprintable ASCII characters which, unlike
883 e.g. #\Newline and #\Tab, don't have names specified in the
884 ANSI Common Lisp standard, is now based on their ASCII symbolic
885 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
886 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
887 are no longer used for output.
888 ** changes in internal implementation constants:
889 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
890 4 million. (If your application spends a lot of time GCing and
891 you have a lot of RAM, you might want to experiment with
892 increasing it even more.)
893 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
894 However, almost all old code should still continue to work without
895 immediate update, as SB-C-CALL is now a (deprecated) nickname
897 ** Old operator names in the style DEF-FOO are now deprecated in
898 favor of new corresponding names DEFINE-FOO, for consistency with
899 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
900 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
901 internal symbols, but a few supported extensions like
902 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
903 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
904 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
905 etc. as you get deeper into recursive calls to the debugger
906 command loop, instead of the old "5]", "5]]", "5]]]"
907 sequence. (I was motivated to do this when squabbles between
908 ILISP and SBCL left me very deeply nested in the debugger. In the
909 short term, this change will probably provoke more ILISP/SBCL
910 squabbles, but hopefully it will be an improvement in the long run.)
911 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
912 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
913 function, since declaiming return types involving aliens is
914 (1) annoyingly messy to do by hand and (2) vital to efficient
915 compilation of code which calls such functions.
916 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
917 longer reexported by the SB-EXT package. They're solely useful
918 for alien code, so it seems more logical that you should get
919 them from the SB-ALIEN package, not in SB-EXT.
920 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
921 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
922 features. Instead, the code that they used to control is always
923 built into the system.
924 * many other bug fixes
925 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
926 advantage of the new EVAL-WHEN stuff and to clean them up in
927 general, and they are now more ANSI-compliant in a number of
928 ways. Martin Atzmueller is responsible for a lot of this.
929 ** Besides the cleanups discussed above, Martin Atzmueller fixed
931 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
932 *** correct ERROR type for various file operations
933 *** some fixes for Lisp streams
934 *** DEFMETHOD syntax checking
935 *** changing old weird representation of debug information as
936 strings (which, among their other deficiencies, don't transform
937 correctly when you rename packages, and don't change their
938 print representation when you change things like *PACKAGE*
939 and *PRINT-LENGTH*) to symbols and lists of symbols
940 He also made several improvements and fixed several bugs in DESCRIBE.
941 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
943 *** misbehavior of WRITE-STRING/WRITE-LINE
944 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
945 and several other LOOP problems as well
946 *** DIRECTORY when similar filenames are present
947 *** DEFGENERIC with :METHOD options
948 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
949 *** bug in the optimization of ARRAY-ELEMENT-TYPE
950 *** argument ordering in FIND with :TEST option
951 *** mishandled package designator argument in APROPOS-LIST
952 *** various problems in the backquote readmacro
954 *** probably some others that I'm not describing very well here,
955 since the CVS log documents them by reference to sbcl-devel
956 messages, and the SourceForge archives aren't working well.:-(
957 ** Dan Barlow improved the Alpha port (and is making progress on the
958 PPC port, for those of you who think different).
959 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
960 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
961 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
962 continues to work on the SPARC port (for those of you in a position
963 to look down upon our little PC-compatible boxes from a great height).
964 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
965 and uses that copy, rather than the current dynamic value, when
966 it's trying to decide whether to truncate output. Thus e.g.
967 (let ((*print-lines* 50))
968 (pprint-logical-block (stream nil)
970 (let ((*print-lines* 8))
971 (print (aref possiblybigthings i) stream)))))
972 should now truncate the logical block only at 50 lines, instead of
973 often truncating it at 8 lines, as it did before.
974 * The doc/cmucl/ directory, containing old CMU CL documentation
975 from the time of the fork, is no longer part of the base system.
976 SourceForge has shut down its anonymous FTP service, and with it
977 my original plan for distributing the old CMU CL documentation
978 there. For now, if you need these files you can download an old
979 SBCL source release and extract them from it.
980 * The fasl file version number changed again, for dozens of reasons,
981 some of which are apparent above.
983 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
985 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
986 up properly again. (There was a packaging bug in 0.7.0 which
987 left their definitions in SB-SYS::LOAD-FOREIGN and
988 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
989 vital for most things which interface to C-level interfaces,
990 like extensions working with sockets or databases or
991 Perl-compatible regexes or whatever, and the need to fix
992 this bug is the main reason that 0.7.1 was released so
994 ** DEFGENERIC is now choosier about the methods it redefines, so that
995 reLOADing a previously-LOADed file containing DEFGENERICs does
996 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
997 cycle now works as it should. (thanks to Alexey Dejneka)
998 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
999 fixed by Christophe Rhodes. (He actually submitted this patch
1000 months ago, and I delayed until after 0.7.0.)
1001 ** Bug 111 (internal compiler confusion about runtime checks on
1002 FUNCTION types) was fixed by Alexey Dejneka.
1003 * Some internal cleanups (getting rid of variables which aren't
1004 needed now that the byte interpreter is gone) caused the fasl
1005 file format number to change again.
1007 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
1008 * incompatible change: The compiler is now less aggressive about
1009 tail call optimization, doing it only when (> SPACE DEBUG) or
1010 (> SPEED DEBUG). (This is an incompatible change because there are
1011 programs which relied on the old CMU-CL-style behavior to optimize
1012 away their unbounded recursion which will now die of stack overflow.)
1013 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
1014 for non-GENCGC systems has been increased to 20M (since that
1015 seems much closer to the likely performance optimum for modern
1016 systems than the old 4M value was)
1017 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
1018 and *DEBUG-PRINT-LEVEL*
1019 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
1020 of CMU CL's support for SPARC, and various endianness and other
1021 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
1022 * new syntactic sugar for the Unix command line: --load foo.bar is now
1023 an alternate notation for --eval '(load "foo.bar")'.
1025 ** The system now detects stack overflow and handles it gracefully,
1026 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
1027 optimization settings. (This is a good thing in general, and
1028 its introduction in this version should be particularly timely
1029 for anyone whose code fails because of suppression of tail
1031 ** The system now hunts for the C variable "environ" in a more
1032 devious way, to avoid segfaults when the C library version differs
1033 between compile time and run time. (thanks to Christophe Rhodes)
1034 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
1035 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
1036 ** The compiler no longer issues bogus style warnings for undefined
1037 classes in the same source file as the DEFCLASSes which defined
1038 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
1040 ** fixes in CONDITION class precedence list for undefined function
1041 errors (thanks to Alexei Dejneka)
1042 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
1043 correctly. (thanks to Dan Barlow)
1044 ** portability fixes aiming at bootstrapping under CLISP (thanks
1045 to Dave McDonald and Christophe Rhodes)
1046 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
1047 ** fixes in type translation and and type inference (thanks to
1049 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
1050 ** various fixes in the new ports (thanks to Dan Barlow)
1051 * several changes related to debugging:
1052 ** suppression of tail recursion, as noted above
1053 ** stack overflow detection, as noted above
1054 ** The default implementation of TRACE has changed. :ENCAPSULATE T
1055 is now the default. (For some time encapsulation has been more
1056 reliable than the breakpoint-based :ENCAPSULATE NIL
1057 implementation, at least on X86 systems; and I just noticed that
1058 encapsulation also seems closer to the spirit of the ANSI
1061 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
1062 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
1063 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
1064 * SBCL now runs on the PPC archtiecture under Linux. It actually did
1065 this as of 0.7.1.45, but was left out of the previous news section
1066 (thanks to Dan Barlow)
1067 * SBCL now runs on the Solaris operating system on SPARC architectures
1068 (thanks to Christophe Rhodes's port of the CMUCL runtime)
1069 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
1070 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
1071 * SPARC backend cleanups, allowing builds of cores optimized for V8
1072 and V9 SPARCS, and also emission of code targeted to a particular
1073 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
1075 * SBCL is closer to bootstrapping under CLISP, thanks to various
1076 fixes by Christophe Rhodes.
1077 * The fasl file format has changed again, to allow the compiler's
1078 INFO database to support symbol macros.
1079 * The user manual (in doc/) is formatted into HTML more nicely.
1080 (thanks to coreythomas)
1081 * The system is smarter about SUBTYPEP relationships, especially
1082 those involving NOT types (including types such as ATOM which are
1083 represented internally using NOT types). Thus SUBTYPEP is less
1084 likely to return (VALUES NIL NIL) in general, and in particular
1085 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
1087 * The fasl file format has changed again, because the internal
1088 representation of types now includes a new slot to support the new
1089 SUBTYPEP-of-NOT-types logic.
1090 * (not a change in the main branch of SBCL, but a related prototype
1091 which can hopefully be merged into the main branch of SBCL in the
1092 future:) Brian Spilsbury has produced a Unicode-enabled variant of
1093 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
1094 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
1095 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
1096 dispatch macro character combinations. (thanks to Alexey Dejneka)
1097 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
1098 that superficially look like logical namestrings correctly.
1099 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
1101 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
1102 default supplied-p) for &optional and &key arguments. (thanks to
1105 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
1106 * bug 147 fixed: The compiler preserves its block link/count
1107 invariants more correctly now so that it doesn't crash. (thanks
1109 * Dynamic loading of object files in OpenBSD is now supported. (thanks
1111 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
1112 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
1113 no-value-for-this-character as NIL (as specified by ANSI).
1114 * HOST-NAMESTRING on physical pathnames now returns a string that is
1115 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
1116 (thanks to Christophe Rhodes)
1117 * The Alpha port handles icache flushing more correctly. (thanks to
1119 * More progress has been made toward bootstrapping under CLISP. (thanks
1120 to Christophe Rhodes)
1121 * The fasl file format has changed again, because dynamic loading
1122 on OpenBSD (which has non-ELF object files) motivated some cleanups
1123 in the way that foreign symbols are transformed and passed around.
1124 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
1125 is no longer treated as whitespace by the reader, but instead as
1126 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
1127 |A
\x7fB|, instead of A as it used to.
1129 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
1130 * SBCL now builds with OpenMCL (version 0.12) as the
1131 cross-compilation host; also, more progress has been made toward
1132 bootstrapping under CLISP.
1133 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
1135 * bug 158 fixed: The compiler can now deal with integer loop
1136 increments different from 1; fixing this turned out also to fix
1138 * bug 169 fixed: no more bogus warnings about using lexical bindings
1139 despite the presence of perfectly good SPECIAL declarations (thanks
1140 to David Lichteblau)
1141 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
1142 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
1143 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
1145 * bug 180 fixed: Method combination specifications no longer ignore
1146 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
1147 * bug fix: Structure type predicate functions now check their argument
1148 count as they should.
1149 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
1150 correctly. (thanks to Pierre Mai)
1151 * minor incompatible change: The --noprogrammer option is deprecated
1152 in favor of the new --disable-debugger option, which is very similar.
1153 (The major difference is that it takes effect at a slightly different
1154 time at startup, causing handling of errors in --sysinit and
1155 --userinit files will be affected differently.) The
1156 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
1157 been added to allow this functionality to be controlled from ordinary
1158 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
1159 maintainers, who might want to run non-interactive scripts to
1160 build SBCL cores which will later be used interactively.)
1161 * minor incompatible change: The LOAD function no longer, when given
1162 a wild pathname to load, loads all files matching that pathname.
1163 Instead, an error of type FILE-ERROR is signalled.
1165 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
1166 * bug fix: Floating point exceptions are treated much more
1167 consistently on the x86/Linux and PPC/Linux platforms.
1168 * Array initialization with :INITIAL-ELEMENT is now much faster for
1169 cases when the compiler cannot open code the array creation, but
1170 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
1171 array accesses have also seen a speed increase.
1172 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
1173 specified. (This was at the root of some bad interactions between
1174 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
1176 * bug fix: Internal error arguments for undefined functions are now
1177 computed correctly on the PPC/Linux platform.
1178 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
1179 Raymond Toy's patch for CMU CL)
1180 * Support for the Solaris 9 operating environment has been included
1181 (thanks to Daniel Merritt)
1182 * A very ugly but hopefully complete draft of the missing FFI chapter
1183 of the manual has been created by reformatting the corresponding
1184 CMU CL manual chapter into (currently very ugly and incoherent)
1185 DocBook and bringing it up to date for SBCL behavior. Thus, the
1186 manual is now essentially complete, at least by my extreme
1187 once-and-only-once standards, whereby it's acceptable to refer to
1188 the doc strings of SB-EXT functions as the primary documentation.
1189 * The fasl file version number has changed again, due to cleanup of
1190 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
1191 is no longer a static symbol.)
1193 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
1194 * An alpha-quality port to the parisc architecture running Linux,
1195 based on the old CMUCL backend, has been made. This, even more so
1196 than the other backends, should be considered still a work in
1197 progress; known problems include that the Linux kernel in 64-bit
1198 mode does not propagate the correct sigcontext structure to
1199 userspace, and consequently SBCL on a parisc64 kernel will not
1201 * fixed bug 189: The compiler now respects NOTINLINE declarations for
1202 functions declared in FLET and LABELS. (I.e. "LET conversion" is
1203 suppressed.) Also now that the compiler is looking at declarations
1204 in the environment, it checks optimization declarations as well,
1205 and suppresses inlining when (> DEBUG SPEED).
1206 * More fixes have been made to treatment of floating point exception
1207 treatment and other Unix signals. In particular, floating point
1208 exceptions no longer cause Bus errors on the SPARC/Linux platform.
1209 * The detection and handling of control stack exhaustion (infinite
1210 or very deeply nested recursion) has changed. Stack exhaustion
1211 detection is now done by write-protecting pages at the OS level
1212 and applies at all optimization settings; when found, a
1213 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
1214 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
1215 be caught using IGNORE-ERRORS.
1216 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
1217 correctly when transforming recursive LABELS functions to LETs.
1218 (thanks to Alexey Dejneka)
1219 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
1220 that are names of constants or global variables.
1221 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
1222 alien routines with docstrings.
1223 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
1224 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
1226 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
1227 (thanks to Eric Marsden)
1228 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
1229 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
1230 * Bug fix: The SPARC backend can now compile functions involving
1231 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
1232 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
1233 to a functions expecting a PATHNAME-DESIGNATOR.
1234 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
1235 lambda lists. (thanks to Alexey Dejneka)
1236 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
1237 functions so that EXPORTing the name of the class doesn't cause
1238 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
1239 to Antonio Martinez for reporting this)
1240 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
1241 constant-folded without failing an assertion. (thanks to Einar
1242 Floystad Dorum for reporting this)
1243 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
1244 types are now dealt with more correctly by the compiler.
1245 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
1246 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
1247 behaviour analogous to RENAME-FILE. This puts its behaviour more
1248 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
1250 * The fasl file version number has changed again. (because of the
1251 bug fix involving the names of PCL MAKE-INSTANCE functions)
1253 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
1254 * A beta-quality port to the mips architecture running Linux,
1255 based on the old CMUCL backend, has been made. It has been tested
1256 on a big-endian kernel, and works sufficiently well to be able to
1257 rebuild itself; it has not been tested in little-endian mode.
1258 * fixed an inconsistency between gencgc.c and purify.c which made
1259 dumping/loading .core files unreliable
1260 * fixed bug 120a: The compiler now deals correctly with IFs where
1261 the consequent is the same as the alternative, instead of
1262 misderiving the return type. (thanks to Alexey Dejneka)
1263 * fixed bug 113: Logical pathnames are now dumpable (the logical
1264 host is resolved at load-time, throwing an error if it is not
1266 * fixed bug 174: FORMAT's error message is slightly clearer when a
1267 non-printing character is used in a format directive.
1268 * fixed several bugs in compiler checking of type declarations, i.e.
1269 violations of the Python "declarations are assertions" principle
1270 (thanks to Alexey Dejneka)
1271 * fixed several bugs in PCL's error checking (thanks to Gerd
1273 * fixed bug: printing of FILE-ERROR (thanks to Antonio
1275 * fixed bug in compilation of functions as first class values
1276 (thanks to Antonio Martinez-Shotton)
1277 * The compiler's handling TYPE-ERRORs which it can prove will
1278 inevitably happen at runtime has been cleaned up and corrected
1279 in several ways. (thanks to Alexey Dejneka)
1280 * improved argument type checking for various basic arithmetic
1281 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
1282 lavished on them in the past that they can be compiled in many
1283 ways in different special cases
1284 * fixed bug 181: compiler checks validity of user supplied type
1286 * cleaned up code flushing in optimization: Function calls which
1287 should signal errors for safety purposes (e.g. which ANSI says
1288 should signal errors when their arguments are of incorrect type)
1289 are no longer optimized away.
1290 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
1291 * incremented fasl file version number, because changes in the
1292 implementation of sequence functions like COERCE caused
1293 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
1294 in old inline expansions) to become undefined. (Actually these
1295 changes were later undone, so we might very well be binary
1296 compatible with 0.7.7 after all, but leaving the version number
1297 incremented seemed like the simplest and most conservative
1300 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
1301 * minor incompatible change: The runtime (the Unix executable named
1302 "sbcl") is now much pickier about the .core files it will load.
1303 Essentially it now requires .core files to descend from the same
1304 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
1305 as the runtime does. (The intent is to prevent the crashes which
1306 can occur, and which can even be reported as mysterious failures,
1307 when people patch the sources or change the build parameters
1308 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
1309 sbcl and .core files.)
1310 * fixed bug: VALUES-LIST is no longer optimized away.
1311 * fixed bug 142: The FFI conversion of C string values to Lisp
1312 string values no longer conses excessively. (thanks to Nathan
1313 Froyd porting Raymond Toy's fix to CMU CL)
1314 * began to systematize and improve MOP conformance in PCL (thanks to
1315 Nathan Froyd, Gerd Moellman and Pierre Mai):
1316 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
1318 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
1319 ** FINALIZE-INHERITANCE is now called on class finalization;
1320 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
1321 argument precedence order.
1322 * fixed bug 202: The compiler no longer fails on functions whose
1323 derived types contradict their declared type.
1324 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
1325 so it can be non-toplevel.
1326 * The fasl file version number has changed (because of the new
1327 implementation of DEFMACRO).
1328 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
1329 safe code, that any length requirement by their type-specifier
1330 argument is valid. The exceptions to this are described in bug
1332 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
1333 functions now check the types of their inputs as required by ANSI.
1334 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
1335 introduced symbol is DECLAREd to be SPECIAL.
1336 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
1337 * fixed bug: :COUNT argument to sequence functions may be negative
1338 * fixed bug: body of DO-SYMBOLS may contain declarations
1339 * fixed bug: PUSHNEW now evaluates its arguments from left to right
1340 (reported by Paul F. Dietz, fixed by Gerd Moellman)
1341 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
1342 symbol macro only once
1343 * fixed printing of call frame when argument list is unavailable
1344 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
1345 * compiler no longer signals WARNING on unknown keyword
1348 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
1349 * Support for building SBCL for MIPS platforms running in
1350 little-endian mode has now been checked in, and basic
1351 functionality on said platforms verified.
1352 * minor incompatible change: PCL now records the pathname of a file
1353 in which methods and the like are defined, rather than its
1355 * minor incompatible change: TRUENAME now considers the truename of
1356 a file naming a directory to be the pathname with :DIRECTORY
1357 component indicating that directory.
1358 * minor incompatible change: a NAMED clause in the extended form of
1359 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
1360 reason for the previous behaviour is unclear.
1361 * more systematization and improvement of CLOS and MOP conformance
1362 in PCL (thanks to Gerd Moellman and Pierre Mai):
1363 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
1365 ** DEFINE-METHOD-COMBINATION no longer signals an error for
1366 primary methods with no specializers;
1367 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
1369 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
1370 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
1371 FORWARD-REFERENCED-CLASSes; error reporting on
1372 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
1374 ** SXHASH on CLOS instances now uses a slot internal to the
1375 instance to return different numbers on distinct instances,
1376 while preserving the same return value through invocations of
1378 ** DEFMETHOD signals errors when methods with longer incongruent
1379 lambda lists are added to generic functions;
1380 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
1381 CLASS, as specified in AMOP;
1382 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
1383 on the order returned by the primary method for classes of
1384 class STANDARD-CLASS;
1385 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
1386 * fixed some bugs shown by Paul Dietz' test suite:
1387 ** DOLIST puts its body in TAGBODY;
1388 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
1390 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
1391 value producing form;
1392 ** if more variables are given to PROGV than values, extra
1393 variables are bound and made to have no value;
1394 ** NSUBSTITUTE on list arguments gets the right answer with
1396 ** ELT signals an error of type TYPE-ERROR when the index argument
1397 is not a valid sequence index;
1398 ** LOOP signals (at macroexpansion time) an error of type
1399 PROGRAM-ERROR when duplicate variable names are found;
1400 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
1401 ** FUNCALL of special-operators now cause an error of type
1403 ** PSETQ now works as required in the presence of side-effecting
1404 symbol-macro places;
1405 ** NCONC accepts any object as its last argument;
1406 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
1408 ** loop-for-as-package does not require a package to be explicitely
1410 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
1412 * fixed bug 166: compiler preserves "there is a way to go"
1413 invariant when deleting code.
1414 * fixed bug 172: macro lambda lists with required arguments after
1415 &REST arguments now cause an error to be signalled. (thanks to
1417 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
1418 bug. (thanks to Matthew Danish)
1419 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
1421 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
1422 arguments are lexically rebound. (thanks to Gerd Moellmann and
1424 * fixed bug 194: error messages are now more informative when there
1425 is no primary method applicable in a call to a generic
1426 function. (thanks to Gerd Moellmann)
1427 * fixed bug in command line argument checking (thanks to Julian
1429 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
1430 (thanks to Matthew Danish)
1431 * incremented fasl file version number, because of the
1432 SXHASH-related changes in the layout of CLOS data structures
1434 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1435 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1436 accessors that are related by inheritance, as specified in the
1437 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1438 Valtteri Vuorikoski)
1439 * The compiler is now able to inline functions that were defined in
1440 a complex lexical environment (e.g. inside a MACROLET).
1441 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1443 * Support for the upcoming FreeBSD-5.0 release has been included.
1444 (thanks to Dag-Erling Smorgrav)
1445 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1446 effect when it is not in a toplevel context.
1447 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1448 stand a better chance of being correct. (thanks to Gerd
1450 * fixed bug in COERCE, which now signals an error on coercing a
1451 rational to a bounded real type which excludes the expected
1453 * The compiler is now able to derive types more accurately from the
1454 COERCE and COMPILE functions.
1455 * fixed bug 223: functional binding is considered to be constant
1456 only for symbols in the CL package.
1457 * fixed bug 231: SETQ did not check the type of a variable being set
1458 (reported by Robert E. Brown)
1459 * A new optimization for MAKE-INSTANCE has been included, fixing
1460 various bugs (including relating to :ALLOCATION :CLASS slots and
1461 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1463 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1464 ** As required by ANSI, LOOP now disallows anonymous collection
1465 clauses such as COLLECT I in conjunction with aggregate boolean
1466 clauses such as THEREIS (= I 1);
1467 ** LOOP now signals an error when any variable is reused in the
1468 same loop (including the potentially useful construct analogous
1469 to WITH A = 1 WITH A = (1+ A);
1470 ** IT is only a special loop symbol within the first clause of a
1471 conditional loop clause;
1472 ** LOOP with a typed iteration variable over a hashtable now
1473 signals a type error iff it should.
1474 * fixed some other bugs revealed by Paul Dietz' test suite:
1475 ** FILE-STREAM now names the class previously known as FD-STREAM;
1476 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
1477 argument) no longer signals an error;
1478 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
1479 of the slot symbol, rather than using the current package
1480 ((:CONC-NAME "") continues to intern the slot's name in the
1482 * incremented fasl file version number, because of the incompatible
1483 change to the DEFSTRUCT-DESCRIPTION structure, and again because
1484 of the new implementation of DEFINE-COMPILER-MACRO.
1486 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1487 * minor incompatible change: code processed by the "interpreter" or
1488 EVAL now has a compilation optimization policy of (DEBUG 2)
1489 (changed from (DEBUG 1)) to improve debuggability of interactive
1490 development, and to allow the use of the debug RETURN command in
1492 * an experimental implementation of the RETURN command for the
1493 debugger has been included. (thanks to Frederik Kuivinen)
1494 * fixed bug 62: constraints were not propagated into a loop.
1495 * fixed bug in embedded calls of SORT (reported and investigated by
1497 * fixed some bugs revealed by Paul Dietz' test suite:
1498 ** printing and reading of arrays with some dimensions having
1499 length 0 (thanks to Gerd Moellmann);
1500 ** BOA constructor with &AUX argument without a default value does
1501 not cause a type error;
1502 ** CONSTANTP now returns true for all self-evaluating objects.
1504 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1505 * incompatible packaging change: in line with Unix convention,
1506 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1507 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1508 * REQUIRE and PROVIDE are now optionally capable of doing something
1509 useful: see the documentation string for REQUIRE
1510 * infrastructure for a managed SBCL contrib system: contributed
1511 modules in this release include:
1512 ** the ASDF system definition facility;
1513 ** an interface to the BSD Sockets API;
1514 ** an ACL-like convenience interface to the repl;
1515 (thanks to Kevin Rosenberg)
1516 ** an implementation of ROTATE-BYTE, with efficient implementation
1518 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1519 gives a closer approximation to the right answer than previously.
1520 (thanks to Raymond Toy)
1521 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1522 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1523 optional environment arguments, as required by ANSI.
1524 * fixed bugs in other functions taking environment objects, allowing
1525 calls with an explicit NIL environment argument to be compiled
1527 * fixed bug 228: primary return values from
1528 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1529 COMPILE or FUNCTION.
1530 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1531 :TYPE will no longer signal errors on innocuous objects.
1532 * fixed bug 231b: SETQ is better at respecting type declarations in
1533 the lexical environment.
1534 * fixed a bug in DEFCLASS: classes named by symbols with no or
1535 unprintable packages can now be defined.
1536 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1537 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1538 * fixed compiler failure related to checking types of functions.
1539 (reported by Robert E. Brown)
1540 * the compiler is now much more consistent in its error-checking
1541 treatment of bounding index arguments to sequence functions: in
1542 (SAFETY 3) code, errors will be signalled in almost all cases if
1543 invalid sequence bounding indices are passed to functions defined
1544 by ANSI to operate on sequences.
1545 * fixed a bug in the build procedure: documentation of SBCL-specific
1546 packages is now preserved and available in the final Lisp image.
1547 * lifted FDEFINITION lookup out of loops in the implementation of
1548 many list operations. (thanks to Robert E. Brown)
1549 * fixed a bug in the reader: the #n# reader macro now works for
1550 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1551 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1552 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1553 E. Brown and Tony Martinez respectively)
1554 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1555 SLOT-BOUNDP: in particular, it is now able to optimize them much
1556 better, and is now not vulnerable to having packages renamed.
1557 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1558 always called when a slot is not present in an instance. (thanks
1560 * fixed some bugs revealed by Paul Dietz' test suite:
1561 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1562 not just nonnegative fixnums;
1563 ** the logical bit-array operators such as BIT-AND now accept an
1564 explicit NIL for their "opt-arg" argument (to indicate a
1565 freshly-consed result bit-array);
1566 ** ELT now signals an error on an invalid sequence index in safe
1568 ** the type system is now cleverer about negations of numeric
1569 types, and consequently understands the BIGNUM and RATIO types
1571 ** the type system is now cleverer about the interaction between
1572 INTEGER and RATIO types: while bugs still remain, many more
1573 cases are accurately computed;
1574 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1575 if it is in the last clause;
1576 ** CONSTANTLY now correctly returns a side-effect-free function in
1579 planned incompatible changes in 0.7.x:
1580 * (not done yet, but planned:) When the profiling interface settles
1581 down, maybe in 0.7.x, maybe later, it might impact TRACE. They both
1582 encapsulate functions, and it's not clear yet how e.g. UNPROFILE
1583 will interact with TRACE
1584 and UNTRACE. (This shouldn't matter, though, unless you are
1585 using profiling. If you never profile anything, TRACE should
1586 continue to behave as before.)
1587 * (not done yet, but planned:) Inlining can now be controlled the
1588 ANSI way, without MAYBE-INLINE, since the idiom
1589 (DECLAIM (INLINE FOO))
1591 (DECLAIM (NOTINLINE FOO))
1592 (DEFUN BAR (..) (FOO ..))
1593 (DEFUN BLETCH (..) (DECLARE (INLINE FOO)) (FOO ..))
1594 now does what ANSI says it should. The CMU-CL-style
1595 SB-EXT:MAYBE-INLINE declaration is now deprecated and ignored.