3 Bugs can be reported on the help mailing list
4 sbcl-help@lists.sourceforge.net
5 or on the development mailing list
6 sbcl-devel@lists.sourceforge.net
8 Please include enough information in a bug report that someone reading
9 it can reproduce the problem, i.e. don't write
10 Subject: apparent bug in PRINT-OBJECT (or *PRINT-LENGTH*?)
11 PRINT-OBJECT doesn't seem to work with *PRINT-LENGTH*. Is this a bug?
13 Subject: apparent bug in PRINT-OBJECT (or *PRINT-LENGTH*?)
14 In sbcl-1.2.3 running under OpenBSD 4.5 on my Alpha box, when
15 I compile and load the file
16 (DEFSTRUCT (FOO (:PRINT-OBJECT (LAMBDA (X Y)
17 (LET ((*PRINT-LENGTH* 4))
20 then at the command line type
22 the program loops endlessly instead of printing the object.
27 There is also some information on bugs in the manual page and
28 in the TODO file. Eventually more such information may move here.
30 The gaps in the number sequence belong to old bug descriptions which
31 have gone away (typically because they were fixed, but sometimes for
32 other reasons, e.g. because they were moved elsewhere).
35 KNOWN BUGS OF NO SPECIAL CLASS:
38 DEFSTRUCT almost certainly should overwrite the old LAYOUT information
39 instead of just punting when a contradictory structure definition
40 is loaded. As it is, if you redefine DEFSTRUCTs in a way which
41 changes their layout, you probably have to rebuild your entire
42 program, even if you know or guess enough about the internals of
43 SBCL to wager that this (undefined in ANSI) operation would be safe.
46 ANSI specifies that a type mismatch in a structure slot
47 initialization value should not cause a warning.
49 This one might not be fixed for a while because while we're big
50 believers in ANSI compatibility and all, (1) there's no obvious
51 simple way to do it (short of disabling all warnings for type
52 mismatches everywhere), and (2) there's a good portable
53 workaround, and (3) by their own reasoning, it looks as though
54 ANSI may have gotten it wrong. ANSI justifies this specification
56 The restriction against issuing a warning for type mismatches
57 between a slot-initform and the corresponding slot's :TYPE
58 option is necessary because a slot-initform must be specified
59 in order to specify slot options; in some cases, no suitable
61 However, in SBCL (as in CMU CL or, for that matter, any compiler
62 which really understands Common Lisp types) a suitable default
63 does exist, in all cases, because the compiler understands the
64 concept of functions which never return (i.e. has return type NIL).
65 Thus, as a portable workaround, you can use a call to some
66 known-never-to-return function as the default. E.g.
68 (BAR (ERROR "missing :BAR argument")
69 :TYPE SOME-TYPE-TOO-HAIRY-TO-CONSTRUCT-AN-INSTANCE-OF))
71 (DECLAIM (FTYPE (FUNCTION () NIL) MISSING-ARG))
72 (DEFUN REQUIRED-ARG () ; workaround for SBCL non-ANSI slot init typing
73 (ERROR "missing required argument"))
75 (BAR (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
76 (BLETCH (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
77 (N-REFS-SO-FAR 0 :TYPE (INTEGER 0)))
78 Such code should compile without complaint and work correctly either
79 on SBCL or on any other completely compliant Common Lisp system.
82 bogus warnings about undefined functions for magic functions like
83 SB!C::%%DEFUN and SB!C::%DEFCONSTANT when cross-compiling files
84 like src/code/float.lisp. Fixing this will probably require
85 straightening out enough bootstrap consistency issues that
86 the cross-compiler can run with *TYPE-SYSTEM-INITIALIZED*.
87 Instead, the cross-compiler runs in a slightly flaky state
88 which is sane enough to compile SBCL itself, but which is
89 also unstable in several ways, including its inability
90 to really grok function declarations.
93 The "compiling top-level form:" output ought to be condensed.
94 Perhaps any number of such consecutive lines ought to turn into a
95 single "compiling top-level forms:" line.
98 The way that the compiler munges types with arguments together
99 with types with no arguments (in e.g. TYPE-EXPAND) leads to
100 weirdness visible to the user:
101 (DEFTYPE FOO () 'FIXNUM)
103 (TYPEP 11 '(FOO)) => T, which seems weird
104 (TYPEP 11 'FIXNUM) => T
105 (TYPEP 11 '(FIXNUM)) signals an error, as it should
106 The situation is complicated by the presence of Common Lisp types
107 like UNSIGNED-BYTE (which can either be used in list form or alone)
108 so I'm not 100% sure that the behavior above is actually illegal.
109 But I'm 90+% sure, and the following related behavior,
111 treating the bare symbol AND as equivalent to '(AND), is specifically
112 forbidden (by the ANSI specification of the AND type).
115 It would be nice if the
117 (during macroexpansion)
118 said what macroexpansion was at fault, e.g.
120 (during macroexpansion of IN-PACKAGE,
121 during macroexpansion of DEFFOO)
124 (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
125 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
126 (Also, when this is fixed, we can enable the code in PROCLAIM which
127 checks for incompatible FTYPE redeclarations.)
130 (I *think* this is a bug. It certainly seems like strange behavior. But
131 the ANSI spec is scary, dark, and deep.. -- WHN)
132 (FORMAT NIL "~,1G" 1.4) => "1. "
133 (FORMAT NIL "~3,1G" 1.4) => "1. "
136 from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
138 (setf (find-class 'ccc1) (find-class 'ccc))
139 (defmethod zut ((c ccc1)) 123)
140 DTC's recommended workaround from the mailing list 3 Mar 2000:
141 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
144 The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
145 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
146 inside "~<..~>" (without the colon modifier on the closing syntax).
147 However, SBCL doesn't do this:
148 * (FORMAT T "~<munge~wegnum~>" 12)
153 Sometimes (SB-EXT:QUIT) fails with
154 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
155 Process inferior-lisp exited abnormally with code 1
156 I haven't noticed a repeatable case of this yet.
159 In some cases the compiler believes type declarations on array
160 elements without checking them, e.g.
161 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
164 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
165 (WHEN (CONSP (AREF X 0))
167 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
170 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
171 all cases, e.g. the type assumption *is* checked if the array
172 elements are declared to be of some structure type instead of CONS.
175 The printer doesn't report closures very well. This is true in
179 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
180 It would be nice to make closures have a settable name slot,
181 and make things like DEFSTRUCT and FLET, which create closures,
182 set helpful values into this slot.
185 And as long as we're wishing, it would be awfully nice if INSPECT could
186 also report on closures, telling about the values of the bound variables.
189 The compiler assumes that any time a function of declared FTYPE
190 doesn't signal an error, its arguments were of the declared type.
191 E.g. compiling and loading
192 (DECLAIM (OPTIMIZE (SAFETY 3)))
193 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
195 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE)) FACTORIAL))
197 (COND ((> (FACTORIAL X) 1.0E6)
198 (FORMAT T "too big~%"))
200 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
202 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
205 will cause the INTEGERP case to be selected, giving bogus output a la
207 This violates the "declarations are assertions" principle.
208 According to the ANSI spec, in the section "System Class FUNCTION",
209 this is a case of "lying to the compiler", but the lying is done
210 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
211 not by the unexpectedly general definition of FACTORIAL. In any case,
212 "declarations are assertions" means that lying to the compiler should
213 cause an error to be signalled, and should not cause a bogus
214 result to be returned. Thus, the compiler should not assume
215 that arbitrary functions check their argument types. (It might
216 make sense to add another flag (CHECKED?) to DEFKNOWN to
217 identify functions which *do* check their argument types.)
218 (Also, verify that the compiler handles declared function
219 return types as assertions.)
222 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
223 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
225 (DECLARE (TYPE INDEXOID X))
226 (THE (VALUES INDEXOID)
228 where the implementation of the type check in function FOO
229 includes a full call to %TYPEP. There are also some fundamental problems
230 with the interpretation of VALUES types (inherited from CMU CL, and
231 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
232 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
235 The definitions of SIGCONTEXT-FLOAT-REGISTER and
236 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
237 supported on FreeBSD because the floating point state is not saved,
238 but at least as of FreeBSD 4.0, the floating point state *is* saved,
239 so they could be supported after all. Very likely
240 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
243 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
244 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
245 issue with 'union, 'and, 'or etc. So even though according to the
246 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
247 (and now SBCL) interpret them as legal types.
250 ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
251 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
252 list, and it is probably possible to use substantially the same
253 patches to add it to SBCL.
256 a slew of floating-point-related errors reported by Peter Van Eynde
258 a: (fixed in sbcl-0.6.11.25)
259 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
260 should probably be 1.4012985e-45. In SBCL,
261 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
262 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
263 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
264 and LEAST-NEGATIVE-LONG-FLOAT.
265 c: Many expressions generate floating infinity:
270 PVE's regression tests want them to raise errors. SBCL
271 generates the infinities instead, which may or may not be
273 d: (in section12.erg) various forms a la
274 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
275 don't give the right behavior.
278 type safety errors reported by Peter Van Eynde July 25, 2000:
279 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
281 In general lengths of array type specifications aren't
282 checked by COERCE, so it fails when the spec is
283 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
284 b: CONCATENATE has the same problem of not checking the length
285 of specified output array types. MAKE-SEQUENCE and MAP and
286 MERGE also have the same problem.
287 c: (COERCE 'AND 'FUNCTION) returns something related to
288 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
289 f: (FLOAT-RADIX 2/3) should signal an error instead of
291 g: (LOAD "*.lsp") should signal FILE-ERROR.
292 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
293 should signal TYPE-ERROR.
294 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
295 be used for input and output as needed. It should fail with
296 TYPE-ERROR when handed e.g. the results of
297 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
298 the inappropriate positions, but doesn't.
299 j: (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
301 should probably signal an error instead of making a pathname with
303 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
304 not a binary input stream, but instead cheerfully reads from
305 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
308 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
309 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
311 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
312 signal a PROGRAM-ERROR, and doesn't.
313 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
314 and other DEFCLASS forms with duplicate specifications in their
315 slots, should signal a PROGRAM-ERROR, and doesn't.
316 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
317 causes a COMPILER-ERROR.
320 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
321 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
322 PROGRAM-ERROR, but SBCL accepts it instead.
323 b: SYMBOL-MACROLET should refuse to bind something which is
324 declared as a global variable, signalling PROGRAM-ERROR.
325 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
326 it binds is declared SPECIAL inside.
329 type system errors reported by Peter Van Eynde July 25, 2000:
330 c: (SUBTYPEP '(INTEGER (0) (0)) 'NIL) dies with nested errors.
331 d: In general, the system doesn't like '(INTEGER (0) (0)) -- it
332 blows up at the level of SPECIFIER-TYPE with
333 "Lower bound (0) is greater than upper bound (0)." Probably
334 SPECIFIER-TYPE should return NIL instead.
335 e: (TYPEP 0 '(COMPLEX (EQL 0)) fails with
336 "Component type for Complex is not numeric: (EQL 0)."
337 This might be easy to fix; the type system already knows
338 that (SUBTYPEP '(EQL 0) 'NUMBER) is true.
339 g: The type system isn't all that smart about relationships
340 between hairy types, as shown in the type.erg test results,
341 e.g. (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL.
344 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
346 (DEFGENERIC FOO02 (X))
347 (DEFMETHOD FOO02 ((X NUMBER)) T)
348 (LET ((M (FIND-METHOD (FUNCTION FOO02)
350 (LIST (FIND-CLASS (QUOTE NUMBER))))))
351 (REMOVE-METHOD (FUNCTION FOO02) M)
352 (DEFGENERIC FOO03 (X))
353 (ADD-METHOD (FUNCTION FOO03) M)))
354 should give an error, but SBCL allows it.
355 b: READ should probably return READER-ERROR, not the bare
356 arithmetic error, when input a la "1/0" or "1e1000" causes
360 It has been reported (e.g. by Peter Van Eynde) that there are
361 several metaobject protocol "errors". (In order to fix them, we might
362 need to document exactly what metaobject protocol specification
363 we're following -- the current code is just inherited from PCL.)
366 The implementation of #'+ returns its single argument without
367 type checking, e.g. (+ "illegal") => "illegal".
370 Attempting to use COMPILE on something defined by DEFMACRO fails:
371 (DEFMACRO FOO (X) (CONS X X))
373 Error in function C::GET-LAMBDA-TO-COMPILE:
374 #<Closure Over Function "DEFUN (SETF MACRO-FUNCTION)" {480E21B1}> was defined in a non-null environment.
377 (SUBTYPEP '(AND ZILCH INTEGER) 'ZILCH) => NIL, NIL
378 Note: I looked into fixing this in 0.6.11.15, but gave up. The
379 problem seems to be that there are two relevant type methods for
380 the subtypep operation, HAIRY :COMPLEX-SUBTYPEP-ARG2 and
381 INTERSECTION :COMPLEX-SUBTYPEP-ARG1, and only the first is
382 called. This could be fixed, but type dispatch is messy and
383 confusing enough already, I don't want to complicate it further.
384 Perhaps someday we can make CLOS cross-compiled (instead of compiled
385 after bootstrapping) so that we don't need to have the type system
386 available before CLOS, and then we can rewrite the type methods to
387 CLOS methods, and then expressing the solutions to stuff like this
388 should become much more straightforward. -- WHN 2001-03-14
391 The debugger LIST-LOCATIONS command doesn't work properly.
394 Compiling and loading
395 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
397 then requesting a BACKTRACE at the debugger prompt gives no information
398 about where in the user program the problem occurred.
401 The compiler is supposed to do type inference well enough that
404 ((SIMPLE-ARRAY SINGLE-FLOAT)
406 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
409 is redundant. However, as reported by Juan Jose Garcia Ripoll for
410 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
411 workaround for the problem for now, but can't be done by the TYPECASE
412 macros themselves, since it's too hard for the macro to detect
413 assignments to the variable within the clause.
414 Note: The compiler *is* smart enough to do the type inference in
415 many cases. This case, derived from a couple of MACROEXPAND-1
416 calls on Ripoll's original test case,
418 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
419 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
420 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
421 (LET ((I 0) (G2554 LENGTH))
422 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
425 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
426 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
427 (GO SB-LOOP::NEXT-LOOP)
428 SB-LOOP::END-LOOP))))))
429 demonstrates the problem; but the problem goes away if the TAGBODY
430 and GO forms are removed (leaving the SETF in ordinary, non-looping
431 code), or if the TAGBODY and GO forms are retained, but the
432 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
435 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
436 I am looking into this problem that showed up on the cmucl-help
437 list. It seems to me that the "implementation specific environment
438 hacking functions" found in pcl/walker.lisp are completely messed
439 up. The good thing is that they appear to be barely used within
440 PCL and the munged environment object is passed to cmucl only
441 in calls to macroexpand-1, which is probably why this case fails.
442 SBCL uses essentially the same code, so if the environment hacking
443 is screwed up, it affects us too.
446 Using the pretty-printer from the command prompt gives funny
447 results, apparently because the pretty-printer doesn't know
448 about user's command input, including the user's carriage return
449 that the user, and therefore the pretty-printer thinks that
450 the new output block should start indented 2 or more characters
451 rightward of the correct location.
454 (probably related to bug #70; maybe related to bug #109)
455 As reported by Carl Witty on submit@bugs.debian.org 1999-05-08,
457 (in-package "CL-USER")
458 (defun equal-terms (termx termy)
460 ((alpha-equal-bound-term-lists (listx listy)
461 (or (and (null listx) (null listy))
463 (let ((bindings-x (bindings-of-bound-term (car listx)))
464 (bindings-y (bindings-of-bound-term (car listy))))
465 (if (and (null bindings-x) (null bindings-y))
466 (alpha-equal-terms (term-of-bound-term (car listx))
467 (term-of-bound-term (car listy)))
468 (and (= (length bindings-x) (length bindings-y))
470 (enter-binding-pairs (bindings-of-bound-term (car listx))
471 (bindings-of-bound-term (car listy)))
472 (alpha-equal-terms (term-of-bound-term (car listx))
473 (term-of-bound-term (car listy)))
474 (exit-binding-pairs (bindings-of-bound-term (car listx))
475 (bindings-of-bound-term (car listy)))))))
476 (alpha-equal-bound-term-lists (cdr listx) (cdr listy)))))
478 (alpha-equal-terms (termx termy)
479 (if (and (variable-p termx)
481 (equal-bindings (id-of-variable-term termx)
482 (id-of-variable-term termy))
483 (and (equal-operators-p (operator-of-term termx) (operator-of-term termy))
484 (alpha-equal-bound-term-lists (bound-terms-of-term termx)
485 (bound-terms-of-term termy))))))
489 (with-variable-invocation (alpha-equal-terms termx termy))))))
490 causes an assertion failure
491 The assertion (EQ (C::LAMBDA-TAIL-SET C::CALLER)
492 (C::LAMBDA-TAIL-SET (C::LAMBDA-HOME C::CALLEE))) failed.
494 Bob Rogers reports (1999-07-28 on cmucl-imp@cons.org) a smaller test
495 case with the same problem:
496 (defun parse-fssp-alignment ()
497 ;; Given an FSSP alignment file named by the argument . . .
498 (labels ((get-fssp-char ()
502 ;; Stub body, enough to tickle the bug.
503 (list (read-fssp-char)
507 ANSI specifies that the RESULT-TYPE argument of CONCATENATE must be
508 a subtype of SEQUENCE, but CONCATENATE doesn't check this properly:
509 (CONCATENATE 'SIMPLE-ARRAY #(1 2) '(3)) => #(1 2 3)
510 This also leads to funny behavior when derived type specifiers
511 are used, as originally reported by Milan Zamazal for CMU CL (on the
512 Debian bugs mailing list (?) 2000-02-27), then reported by Martin
513 Atzmueller for SBCL (2000-10-01 on sbcl-devel@lists.sourceforge.net):
514 (DEFTYPE FOO () 'SIMPLE-ARRAY)
515 (CONCATENATE 'FOO #(1 2) '(3))
516 => #<ARRAY-TYPE SIMPLE-ARRAY> is a bad type specifier for
518 The derived type specifier FOO should act the same way as the
519 built-in type SIMPLE-ARRAY here, but it doesn't. That problem
520 doesn't seem to exist for sequence types:
521 (DEFTYPE BAR () 'SIMPLE-VECTOR)
522 (CONCATENATE 'BAR #(1 2) '(3)) => #(1 2 3)
525 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
526 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
527 crashes SBCL. In general tracing anything which is used in the
528 implementation of TRACE is likely to have the same problem.
531 As reported by Daniel Solaz on cmucl-help@cons.org 2000-11-23,
532 SXHASH returns the same value for all non-STRUCTURE-OBJECT instances,
533 notably including all PCL instances. There's a limit to how much
534 SXHASH can do to return unique values for instances, but at least
535 it should probably look at the class name, the way that it does
536 for STRUCTURE-OBJECTs.
539 As reported by Martin Atzmueller on the sbcl-devel list 2000-11-22,
540 > There remains one issue, that is a bug in SBCL:
541 > According to my interpretation of the spec, the ":" and "@" modifiers
542 > should appear _after_ the comma-seperated arguments.
543 > Well, SBCL (and CMUCL for that matter) accept
544 > (ASSERT (STRING= (FORMAT NIL "~:8D" 1) " 1"))
545 > where the correct way (IMHO) should be
546 > (ASSERT (STRING= (FORMAT NIL "~8:D" 1) " 1"))
547 Probably SBCL should stop accepting the "~:8D"-style format arguments,
548 or at least issue a warning.
551 (probably related to bug #65; maybe related to bug #109)
552 The compiler doesn't like &OPTIONAL arguments in LABELS and FLET
554 (DEFUN FIND-BEFORE (ITEM SEQUENCE &KEY (TEST #'EQL))
555 (LABELS ((FIND-ITEM (OBJ SEQ TEST &OPTIONAL (VAL NIL))
556 (LET ((ITEM (FIRST SEQ)))
559 ((FUNCALL TEST OBJ ITEM)
562 (FIND-ITEM OBJ (REST SEQ) TEST (NCONC VAL `(,ITEM))))))))
563 (FIND-ITEM ITEM SEQUENCE TEST)))
564 from David Young's bug report on cmucl-help@cons.org 30 Nov 2000
565 causes sbcl-0.6.9 to fail with
566 error in function SB-KERNEL:ASSERT-ERROR:
567 The assertion (EQ (SB-C::LAMBDA-TAIL-SET SB-C::CALLER)
568 (SB-C::LAMBDA-TAIL-SET
569 (SB-C::LAMBDA-HOME SB-C::CALLEE))) failed.
572 (DECLAIM (OPTIMIZE ..)) doesn't work. E.g. even after
573 (DECLAIM (OPTIMIZE (SPEED 3))), things are still optimized with
574 the previous SPEED policy. This bug will probably get fixed in
575 0.6.9.x in a general cleanup of optimization policy.
578 (DECLAIM (OPTIMIZE ..)) doesn't work properly inside LOCALLY forms.
581 As reported by Martin Atzmueller on sbcl-devel 26 Dec 2000,
582 ANSI says that WITH-OUTPUT-TO-STRING should have a keyword
583 :ELEMENT-TYPE, but in sbcl-0.6.9 this is not defined for
584 WITH-OUTPUT-TO-STRING.
587 ANSI says in one place that type declarations can be abbreviated even
588 when the type name is not a symbol, e.g.
589 (DECLAIM ((VECTOR T) *FOOVECTOR*))
590 SBCL doesn't support this. But ANSI says in another place that this
591 isn't allowed. So it's not clear this is a bug after all. (See the
592 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
596 as pointed out by Dan Barlow on sbcl-devel 2000-07-02:
597 The PICK-TEMPORARY-FILE-NAME utility used by LOAD-FOREIGN uses
598 an easily guessable temporary filename in a way which might open
599 applications using LOAD-FOREIGN to hijacking by malicious users
600 on the same machine. Incantations for doing this safely are
601 floating around the net in various "how to write secure programs
602 despite Unix" documents, and it would be good to (1) fix this in
603 LOAD-FOREIGN, and (2) hunt for any other code which uses temporary
604 files and make it share the same new safe logic.
607 Functions are assigned names based on the context in which they're
608 defined. This is less than ideal for the functions which are
609 used to implement CLOS methods. E.g. the output of
610 (DESCRIBE 'PRINT-OBJECT) lists functions like
611 #<FUNCTION "DEF!STRUCT (TRACE-INFO (:MAKE-LOAD-FORM-FUN SB-KERNEL:JUST-DUMP-IT-NORMALLY) (:PRINT-OBJECT #))" {1020E49}>
613 #<FUNCTION "MACROLET ((FORCE-DELAYED-DEF!METHODS NIL #))" {1242871}>
614 It would be better if these functions' names always identified
615 them as methods, and identified their generic functions and
619 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
620 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
621 the mean of the distribution can be systematically O(0.1%) wrong.
622 Just increasing R-I-E-B is probably not a good solution, since
623 it would decrease efficiency more than is probably necessary. Perhaps
624 using some sort of accept/reject method would be better.
627 (SUBTYPEP '(SATISFIES SOME-UNDEFINED-FUN) NIL)=>NIL,T (should be NIL,NIL)
630 Internally the compiler sometimes evaluates
631 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
632 (I stumbled across this when I added an
633 (assert (not (eq type1 *wild-type*)))
634 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
635 in a type context should probably be translated to T, and so it's
636 probably to ask whether it's equal to the T type and then (using the
637 EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
638 (I haven't tried to investigate this bug enough to guess whether
639 there might be any user-level symptoms.)
642 a latent cross-compilation/bootstrapping bug: The cross-compilation
643 host's CL:CHAR-CODE-LIMIT is used in target code in readtable.lisp
644 and possibly elsewhere. Instead, we should use the target system's
645 CHAR-CODE-LIMIT. This will probably cause problems if we try to
646 bootstrap on a system which uses a different value of CHAR-CODE-LIMIT
650 (subtypep '(or (integer -1 1)
654 (integer -1 1))) => NIL,T
655 An analogous problem with SINGLE-FLOAT and REAL types was fixed in
656 sbcl-0.6.11.22, but some peculiarites of the RATIO type make it
657 awkward to generalize the fix to INTEGER and RATIONAL. It's not
658 clear what's the best fix. (See the "bug in type handling" discussion
659 on cmucl-imp ca. 2001-03-22 and ca. 2001-02-12.)
662 In sbcl-0.6.11.26, (COMPILE 'IN-HOST-COMPILATION-MODE) in
663 src/cold/shared.lisp doesn't correctly translate the
665 (defun in-host-compilation-mode (fn)
666 (let ((*features* (cons :sb-xc-host *features*))
667 ;; the CROSS-FLOAT-INFINITY-KLUDGE, as documented in
668 ;; base-target-features.lisp-expr:
669 (*shebang-features* (set-difference *shebang-features*
670 '(:sb-propagate-float-type
671 :sb-propagate-fun-type))))
672 (with-additional-nickname ("SB-XC" "SB!XC")
674 No error is reported by the compiler, but when the function is executed,
676 TYPE-ERROR in SB-KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER:
677 (:LINUX :X86 :IEEE-FLOATING-POINT :SB-CONSTRAIN-FLOAT-TYPE :SB-TEST
678 :SB-INTERPRETER :SB-DOC :UNIX ...) is not of type SYMBOL.
681 Inconsistencies between derived and declared VALUES return types for
682 DEFUN aren't checked very well. E.g. the logic which successfully
683 catches problems like
684 (declaim (ftype (function (fixnum) float) foo))
686 (declare (type integer x))
687 (values x)) ; wrong return type, detected, gives warning, good!
689 (declaim (ftype (function (t) (values t t)) bar))
691 (values x)) ; wrong number of return values, no warning, bad!
692 The cause of this is seems to be that (1) the internal function
693 VALUES-TYPES-EQUAL-OR-INTERSECT used to make the check handles its
694 arguments symmetrically, and (2) when the type checking code was
695 written back when when SBCL's code was still CMU CL, the intent
697 (declaim (ftype (function (t) t) bar))
699 (values x x)) ; wrong number of return values; should give warning?
700 not be warned for, because a two-valued return value is considered
701 to be compatible with callers who expects a single value to be
702 returned. That intent is probably not appropriate for modern ANSI
703 Common Lisp, but fixing this might be complicated because of other
704 divergences between auld-style and new-style handling of
705 multiple-VALUES types. (Some issues related to this were discussed
706 on cmucl-imp at some length sometime in 2000.)
709 The facility for dumping a running Lisp image to disk gets confused
710 when run without the PURIFY option, and creates an unnecessarily large
711 core file (apparently representing memory usage up to the previous
712 high-water mark). Moreover, when the file is loaded, it confuses the
713 GC, so that thereafter memory usage can never be reduced below that
717 The TRACE facility can't be used on some kinds of functions.
718 (Basically, the breakpoint facility was incompletely implemented
719 in the X86 port of CMU CL, and hasn't been fixed in SBCL.)
722 In sbcl-0.6.11.41 (and in all earlier SBCL, and in CMU
723 CL), out-of-line structure slot setters are horribly inefficient
724 whenever the type of the slot is declared, because out-of-line
725 structure slot setters are implemented as closures to save space,
726 so the compiler doesn't compile the type test into code, but
727 instead just saves the type in a lexical closure and interprets it
729 A proper solution involves deciding whether it's really worth
730 saving space by implementing structure slot accessors as closures.
731 (If it's not worth it, the problem vanishes automatically. If it
732 is worth it, there are hacks we could use to force type tests to
733 be compiled anyway, and even shared. E.g. we could implement
734 an EQUAL hash table mapping from types to compiled type tests,
735 and save the appropriate compiled type test as part of each lexical
736 closure; or we could make the lexical closures be placeholders
737 which overwrite their old definition as a lexical closure with
738 a new compiled definition the first time that they're called.)
739 As a workaround for the problem, #'(SETF FOO) expressions can
740 be replaced with (EFFICIENT-SETF-FUNCTION FOO), where
741 (defmacro efficient-setf-function (place-function-name)
742 (or #+sbcl (and (sb-impl::info :function :accessor-for place-function-name)
743 ;; a workaround for the problem, encouraging the
744 ;; inline expansion of the structure accessor, so
745 ;; that the compiler can optimize its type test
746 (let ((new-value (gensym "NEW-VALUE-"))
747 (structure-value (gensym "STRUCTURE-VALUE-")))
748 `(lambda (,new-value ,structure-value)
749 (setf (,place-function-name ,structure-value)
751 ;; no problem, can just use the ordinary expansion
752 `(function (setf ,place-function-name))))
755 There's apparently a bug in CEILING optimization which caused
756 Douglas Crosher to patch the CMU CL version. Martin Atzmueller
757 applied the patches to SBCL and they didn't seem to cause problems
758 (as reported sbcl-devel 2001-05-04). However, since the patches
759 modify nontrivial code which was apparently written incorrectly
760 the first time around, until regression tests are written I'm not
761 comfortable merging the patches in the CVS version of SBCL.
764 As reported by Arthur Lemmens sbcl-devel 2001-05-05, ANSI
765 requires that SYMBOL-MACROLET refuse to rebind special variables,
766 but SBCL doesn't do this. (Also as reported by AL in the same
767 message, SBCL depended on this nonconforming behavior to build
768 itself, because of the way that **CURRENT-SEGMENT** was implemented.
769 As of sbcl-0.6.12.x, this dependence on the nonconforming behavior
770 has been fixed, but the nonconforming behavior remains.)
773 (DESCRIBE 'SB-ALIEN:DEF-ALIEN-TYPE) reports the macro argument list
777 in #<PACKAGE "SB-ALIEN">.
778 Macro-function: #<FUNCTION "DEF!MACRO DEF-ALIEN-TYPE" {19F4A39}>
779 Macro arguments: (#:whole-470 #:environment-471)
780 On Sat, May 26, 2001 09:45:57 AM CDT it was compiled from:
781 /usr/stuff/sbcl/src/code/host-alieneval.lisp
782 Created: Monday, March 12, 2001 07:47:43 AM CST
785 (reported by Eric Marsden on cmucl-imp 2001-06-15)
786 (and APD pointed out on sbcl-devel 2001-12-29 that it's the same
790 (TIME (ROOM T)) reports more than 200 Mbytes consed even for
791 a clean, just-started SBCL system. And it seems to be right:
792 (ROOM T) can bring a small computer to its knees for a *long*
793 time trying to GC afterwards. Surely there's some more economical
794 way to implement (ROOM T).
797 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
799 ;;; This file fails to compile.
800 ;;; Maybe this bug is related to bugs #65, #70 in the BUGS file.
801 (in-package :cl-user)
807 ;; Uncomment and it works
810 In SBCL 0.6.12.42, the problem is
811 internal error, failed AVER:
812 "(COMMON-LISP:EQ (SB!C::LAMBDA-TAIL-SET SB!C::CALLER)
813 (SB!C::LAMBDA-TAIL-SET (SB!C::LAMBDA-HOME SB!C::CALLEE)))"
816 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
818 ;;; The compiler is flushing the argument type test, and the default
819 ;;; case in the cond, so that calling with say a fixnum 0 causes a
821 (declaim (optimize (safety 2) (speed 3)))
823 (declare (type (or string stream) x))
824 (cond ((typep x 'string) 'string)
825 ((typep x 'stream) 'stream)
828 The symptom in sbcl-0.6.12.42 on OpenBSD is actually (TST 0)=>STREAM
829 (not the SIGBUS reported in the comment) but that's broken too;
830 type declarations are supposed to be treated as assertions unless
831 SAFETY 0, so we should be getting a TYPE-ERROR.
834 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
836 (in-package :cl-user)
837 ;;; Produces an assertion failures when compiled.
839 (declare (type (or (function (t) t) null) z))
840 (let ((z (or z #'identity)))
841 (declare (type (function (t) t) z))
843 The error in sbcl-0.6.12.42 is
844 internal error, failed AVER:
845 "(COMMON-LISP:NOT (COMMON-LISP:EQ SB!C::CHECK COMMON-LISP:T))"
848 reported by Martin Atzmueller 2001-06-25; taken from CMU CL bugs
849 collection; apparently originally reported by Bruno Haible
850 (in-package :cl-user)
851 ;;; From: Bruno Haible
852 ;;; Subject: scope of SPECIAL declarations
853 ;;; It seems CMUCL has a bug relating to the scope of SPECIAL
854 ;;; declarations. I observe this with "CMU Common Lisp 18a x86-linux
857 (declare (special x))
860 (declare (special x)) y)))
861 ;;; Gives: 0 (this should return 1 according to CLHS)
863 (declare (special x))
866 (declare (special x)) y)))
867 ;;; Gives: 1 (correct).
868 The reported results match what we get from the interpreter
872 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
874 (in-package :cl-user)
875 ;;; From: David Gadbois <gadbois@cyc.com>
877 ;;; Logical pathnames aren't externalizable.
879 (let ((tempfile "/tmp/test.lisp"))
880 (setf (logical-pathname-translations "XXX")
881 '(("XXX:**;*.*" "/tmp/**/*.*")))
882 (with-open-file (out tempfile :direction :output)
883 (write-string "(defvar *path* #P\"XXX:XXX;FOO.LISP\")" out))
884 (compile-file tempfile))
885 The error message in sbcl-0.6.12.42 is
887 ; (while making load form for #<SB-IMPL::LOGICAL-HOST "XXX">)
888 ; A logical host can't be dumped as a constant: #<SB-IMPL::LOGICAL-HOST "XXX">
891 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
893 (in-package :cl-user)
894 ;;; The following invokes a compiler error.
895 (declaim (optimize (speed 2) (debug 3)))
898 (unwind-protect nil)))
902 The error message in sbcl-0.6.12.42 is
903 internal error, failed AVER:
904 "(COMMON-LISP:EQ (SB!C::TN-ENVIRONMENT SB!C:TN) SB!C::TN-ENV)"
907 When the compiler inline expands functions, it may be that different
908 kinds of return values are generated from different code branches.
909 E.g. an inline expansion of POSITION generates integer results
910 from one branch, and NIL results from another. When that inline
911 expansion is used in a context where only one of those results
914 (aref *a1* (position x *a2*)))
915 and the compiler can't prove that the unacceptable branch is
916 never taken, then bogus type mismatch warnings can be generated.
917 If you need to suppress the type mismatch warnings, you can
918 suppress the inline expansion,
920 #+sbcl (declare (notinline position)) ; to suppress bug 117 bogowarnings
921 (aref *a1* (position x *a2*)))
922 or, sometimes, suppress them by declaring the result to be of an
925 (aref *a1* (the integer (position x *a2*))))
927 This is not a new compiler problem in 0.7.0, but the new compiler
928 transforms for FIND, POSITION, FIND-IF, and POSITION-IF make it
929 more conspicuous. If you don't need performance from these functions,
930 and the bogus warnings are a nuisance for you, you can return to
931 your pre-0.7.0 state of grace with
932 #+sbcl (declaim (notinline find position find-if position-if)) ; bug 117..
935 as reported by Eric Marsden on cmucl-imp@cons.org 2001-08-14:
936 (= (FLOAT 1 DOUBLE-FLOAT-EPSILON)
937 (+ (FLOAT 1 DOUBLE-FLOAT-EPSILON) DOUBLE-FLOAT-EPSILON)) => T
938 when of course it should be NIL. (He says it only fails for X86,
939 not SPARC; dunno about Alpha.)
941 Also, "the same problem exists for LONG-FLOAT-EPSILON,
942 DOUBLE-FLOAT-NEGATIVE-EPSILON, LONG-FLOAT-NEGATIVE-EPSILON (though
943 for the -negative- the + is replaced by a - in the test)."
945 Raymond Toy comments that this is tricky on the X86 since its FPU
946 uses 80-bit precision internally.
949 The compiler incorrectly figures the return type of
950 (DEFUN FOO (FRAME UP-FRAME)
957 This problem exists in CMU CL 18c too. When I reported it on
958 cmucl-imp@cons.org, Raymond Toy replied 23 Aug 2001 with
959 a partial explanation, but no fix has been found yet.
962 Even in sbcl-0.pre7.x, which is supposed to be free of the old
963 non-ANSI behavior of treating the function return type inferred
964 from the current function definition as a declaration of the
965 return type from any function of that name, the return type of NIL
966 is attached to FOO in 120a above, and used to optimize code which
970 There was some sort of screwup in handling of
971 (IF (NOT (IGNORE-ERRORS ..))). E.g.
973 (if (not (ignore-errors
974 (make-pathname :host "foo" :directory "!bla" :name "bar")))
976 (error "notunlessnot")))
977 The (NOT (IGNORE-ERRORS ..)) form evaluates to T, so this should be
978 printing "ok", but instead it's going to the ERROR. This problem
979 seems to've been introduced by MNA's HANDLER-CASE patch (sbcl-devel
980 2001-07-17) and as a workaround (put in sbcl-0.pre7.14.flaky4.12)
981 I reverted back to the old weird HANDLER-CASE code. However, I
982 think the problem looks like a compiler bug in handling RETURN-FROM,
983 so I left the MNA-patched code in HANDLER-CASE (suppressed with
984 #+NIL) and I'd like to go back to see whether this really is
985 a compiler bug before I delete this BUGS entry.
988 The *USE-IMPLEMENTATION-TYPES* hack causes bugs, particularly
989 (IN-PACKAGE :SB-KERNEL)
990 (TYPE= (SPECIFIER-TYPE '(VECTOR T))
991 (SPECIFIER-TYPE '(VECTOR UNDEFTYPE)))
992 Then because of this, the compiler bogusly optimizes
993 (TYPEP #(11) '(SIMPLE-ARRAY UNDEF-TYPE 1))
994 to T. Unfortunately, just setting *USE-IMPLEMENTATION-TYPES* to
995 NIL around sbcl-0.pre7.14.flaky4.12 didn't work: the compiler complained
996 about type mismatches (probably harmlessly, another instance of bug 117);
997 and then cold init died with a segmentation fault.
1000 As of version 0.pre7.14, SBCL's implementation of MACROLET makes
1001 the entire lexical environment at the point of MACROLET available
1002 in the bodies of the macroexpander functions. In particular, it
1003 allows the function bodies (which run at compile time) to try to
1004 access lexical variables (which are only defined at runtime).
1005 It doesn't even issue a warning, which is bad.
1007 The SBCL behavior arguably conforms to the ANSI spec (since the
1008 spec says that the behavior is undefined, ergo anything conforms).
1009 However, it would be better to issue a compile-time error.
1010 Unfortunately I (WHN) don't see any simple way to detect this
1011 condition in order to issue such an error, so for the meantime
1012 SBCL just does this weird broken "conforming" thing.
1014 The ANSI standard says, in the definition of the special operator
1016 The macro-expansion functions defined by MACROLET are defined
1017 in the lexical environment in which the MACROLET form appears.
1018 Declarations and MACROLET and SYMBOL-MACROLET definitions affect
1019 the local macro definitions in a MACROLET, but the consequences
1020 are undefined if the local macro definitions reference any
1021 local variable or function bindings that are visible in that
1022 lexical environment.
1023 Then it seems to contradict itself by giving the example
1025 (macrolet ((fudge (z)
1026 ;The parameters x and flag are not accessible
1027 ; at this point; a reference to flag would be to
1028 ; the global variable of that name.
1029 ` (if flag (* ,z ,z) ,z)))
1030 ;The parameters x and flag are accessible here.
1034 The comment "a reference to flag would be to the global variable
1035 of the same name" sounds like good behavior for the system to have.
1036 but actual specification quoted above says that the actual behavior
1040 (as reported by Gabe Garza on cmucl-help 2001-09-21)
1042 (defun test-pred (x y)
1046 (func (lambda () x)))
1047 (print (eq func func))
1048 (print (test-pred func func))
1049 (delete func (list func))))
1050 Now calling (TEST-CASE) gives output
1053 (#<FUNCTION {500A9EF9}>)
1054 Evidently Python thinks of the lambda as a code transformation so
1055 much that it forgets that it's also an object.
1058 (fixed in 0.pre7.41)
1061 The DEFSTRUCT section of the ANSI spec, in the :CONC-NAME section,
1062 specifies a precedence rule for name collisions between slot accessors of
1063 structure classes related by inheritance. As of 0.7.0, SBCL still
1067 insufficient syntax checking in MACROLET:
1069 (macrolet ((defmacro bar (z) `(+ z z)))
1071 shouldn't compile without error (because of the extra DEFMACRO symbol).
1074 reported by Alexey Dejneka on sbcl-devel 2001-11-03
1076 "Return X if X is a non-negative integer."
1077 (let ((step (lambda (%funcall)
1080 (t (1+ (funcall %funcall (1- n)))))))))
1083 (funcall step (lambda (n)
1084 (funcall (funcall a a) n))))
1086 (funcall step (lambda (n)
1087 (funcall (funcall a a) n)))))
1089 This function returns its argument. But after removing percents it
1090 does not work: "Result of (1- n) is not a function".
1093 As of sbcl-0.pre7.86.flaky7.3, the cross-compiler, and probably
1094 the CL:COMPILE function (which is based on the same %COMPILE
1095 mechanism) get confused by
1097 (labels ((sxhash-number (x)
1099 (fixnum (sxhash x)) ; through DEFTRANSFORM
1100 (integer (sb!bignum:sxhash-bignum x))
1101 (single-float (sxhash x)) ; through DEFTRANSFORM
1102 (double-float (sxhash x)) ; through DEFTRANSFORM
1103 #!+long-float (long-float (error "stub: no LONG-FLOAT"))
1104 (ratio (let ((result 127810327))
1105 (declare (type fixnum result))
1106 (mixf result (sxhash-number (numerator x)))
1107 (mixf result (sxhash-number (denominator x)))
1109 (complex (let ((result 535698211))
1110 (declare (type fixnum result))
1111 (mixf result (sxhash-number (realpart x)))
1112 (mixf result (sxhash-number (imagpart x)))
1114 (sxhash-recurse (x &optional (depthoid +max-hash-depthoid+))
1115 (declare (type index depthoid))
1118 (if (plusp depthoid)
1119 (mix (sxhash-recurse (car x) (1- depthoid))
1120 (sxhash-recurse (cdr x) (1- depthoid)))
1123 (if (typep x 'structure-object)
1125 (sxhash ; through DEFTRANSFORM
1126 (class-name (layout-class (%instance-layout x)))))
1128 (symbol (sxhash x)) ; through DEFTRANSFORM
1129 (number (sxhash-number x))
1132 (simple-string (sxhash x)) ; through DEFTRANSFORM
1133 (string (%sxhash-substring x))
1134 (bit-vector (let ((result 410823708))
1135 (declare (type fixnum result))
1136 (dotimes (i (min depthoid (length x)))
1137 (mixf result (aref x i)))
1139 (t (logxor 191020317 (sxhash (array-rank x))))))
1142 (sxhash (char-code x)))) ; through DEFTRANSFORM
1144 (sxhash-recurse x)))
1145 complaining "function called with two arguments, but wants exactly
1146 one" about SXHASH-RECURSE. (This might not be strictly a new bug,
1147 since IIRC post-fork CMU CL has also had problems with &OPTIONAL
1148 arguments in FLET/LABELS: it might be an old Python bug which is
1149 only exercised by the new arrangement of the SBCL compiler.)
1153 (DEFUN FOO () (CATCH 0 (PRINT 1331)))
1155 #<SB-C:TN '0!1> is not valid as the second argument to VOP:
1156 SB-C:MAKE-CATCH-BLOCK,
1157 since the TN's primitive type SB-VM::POSITIVE-FIXNUM doesn't allow
1158 any of the SCs allowed by the operand restriction:
1159 (SB-VM::DESCRIPTOR-REG)
1160 The (CATCH 0 ...) construct is bad style (because of unportability
1161 of EQ testing of numbers) but it is legal, and shouldn't cause an
1162 internal compiler error. (This error occurs in sbcl-0.6.13 and in
1163 0.pre7.86.flaky7.14.)
1166 Trying to compile something like
1167 (sb!alien:def-alien-routine "breakpoint_remove" sb!c-call:void
1168 (code-obj sb!c-call:unsigned-long)
1169 (pc-offset sb!c-call:int)
1170 (old-inst sb!c-call:unsigned-long))
1171 in SBCL-0.pre7.86.flaky7.22 after warm init fails with an error
1172 cannot use values types here
1173 probably because the SB-C-CALL:VOID type gets translated to (VALUES).
1174 It should be valid to use VOID for a function return type, so perhaps
1175 instead of calling SPECIFIER-TYPE (which excludes all VALUES types
1176 automatically) we should call VALUES-SPECIFIER-TYPE and handle VALUES
1177 types manually, allowing the special case (VALUES) but still excluding
1178 all more-complex VALUES types.
1181 Ideally, uninterning a symbol would allow it, and its associated
1182 FDEFINITION and PROCLAIM data, to be reclaimed by the GC. However,
1183 at least as of sbcl-0.7.0, this isn't the case. Information about
1184 FDEFINITIONs and PROCLAIMed properties is stored in globaldb.lisp
1185 essentially in ordinary (non-weak) hash tables keyed by symbols.
1186 Thus, once a system has an entry in this system, it tends to live
1187 forever, even when it is uninterned and all other references to it
1191 (reported by Arnaud Rouanet on cmucl-imp 2001-12-18)
1192 (defmethod foo ((x integer))
1194 (defmethod foo :around ((x integer))
1196 (call-next-method)))
1197 Now (FOO 3) should return 3, but instead it returns 4.
1200 (SB-DEBUG:BACKTRACE) output should start with something
1201 including the name BACKTRACE, not (as in 0.pre7.88)
1202 just "0: (\"hairy arg processor\" ...)". Until about
1203 sbcl-0.pre7.109, the names in BACKTRACE were all screwed
1204 up compared to the nice useful names in sbcl-0.6.13.
1205 Around sbcl-0.pre7.109, they were mostly fixed by using
1206 NAMED-LAMBDA to implement DEFUN. However, there are still
1207 some screwups left, e.g. as of sbcl-0.pre7.109, there are
1208 still some functions named "hairy arg processor" and
1209 "SB-INT:&MORE processor".
1212 (reported by Alexey Dejneka sbcl-devel 2002-01-03)
1214 SUBTYPEP does not work well with redefined classes:
1216 * (defclass a () ())
1218 * (defclass b () ())
1223 * (defclass b (a) ())
1228 * (defclass b () ())
1237 Pretty-printing nested backquotes doesn't work right, as
1238 reported by Alexey Dejneka sbcl-devel 2002-01-13:
1240 ``(FOO SB-IMPL::BACKQ-COMMA-AT S)
1241 * (lisp-implementation-version)
1245 KNOWN BUGS RELATED TO THE IR1 INTERPRETER
1247 (Now that the IR1 interpreter has gone away, these should be
1248 relatively straightforward to fix.)
1251 The system accepts DECLAIM in most places where DECLARE would be
1252 accepted, without even issuing a warning. ANSI allows this, but since
1253 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
1254 meaning is rather different, and it's unlikely that the user
1255 has a good reason for doing DECLAIM not at top level, it would be
1256 good to issue a STYLE-WARNING when this happens. A possible
1257 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
1258 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
1259 [This is considered an IR1-interpreter-related bug because until
1260 EVAL-WHEN is rewritten, which won't happen until after the IR1
1261 interpreter is gone, the system's notion of what's a top-level form
1262 and what's not will remain too confused to fix this problem.]
1265 (another wishlist thing..) Reimplement DEFMACRO to be basically
1266 like DEFMACRO-MUNDANELY, just using EVAL-WHEN.