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 In sbcl-0.7.1.13, this gives an error,
141 There is no class named CCC1.
142 DTC's recommended workaround from the mailing list 3 Mar 2000:
143 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
146 The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
147 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
148 inside "~<..~>" (without the colon modifier on the closing syntax).
149 However, SBCL doesn't do this:
150 * (FORMAT T "~<munge~wegnum~>" 12)
155 Sometimes (SB-EXT:QUIT) fails with
156 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
157 Process inferior-lisp exited abnormally with code 1
158 I haven't noticed a repeatable case of this yet.
161 In some cases the compiler believes type declarations on array
162 elements without checking them, e.g.
163 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
166 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
167 (WHEN (CONSP (AREF X 0))
169 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
172 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
173 all cases, e.g. the type assumption *is* checked if the array
174 elements are declared to be of some structure type instead of CONS.
177 The printer doesn't report closures very well. This is true in
181 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
182 It would be nice to make closures have a settable name slot,
183 and make things like DEFSTRUCT and FLET, which create closures,
184 set helpful values into this slot.
187 And as long as we're wishing, it would be awfully nice if INSPECT could
188 also report on closures, telling about the values of the bound variables.
191 The compiler assumes that any time a function of declared FTYPE
192 doesn't signal an error, its arguments were of the declared type.
193 E.g. compiling and loading
194 (DECLAIM (OPTIMIZE (SAFETY 3)))
195 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
197 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE)) FACTORIAL))
199 (COND ((> (FACTORIAL X) 1.0E6)
200 (FORMAT T "too big~%"))
202 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
204 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
207 will cause the INTEGERP case to be selected, giving bogus output a la
209 This violates the "declarations are assertions" principle.
210 According to the ANSI spec, in the section "System Class FUNCTION",
211 this is a case of "lying to the compiler", but the lying is done
212 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
213 not by the unexpectedly general definition of FACTORIAL. In any case,
214 "declarations are assertions" means that lying to the compiler should
215 cause an error to be signalled, and should not cause a bogus
216 result to be returned. Thus, the compiler should not assume
217 that arbitrary functions check their argument types. (It might
218 make sense to add another flag (CHECKED?) to DEFKNOWN to
219 identify functions which *do* check their argument types.)
220 (Also, verify that the compiler handles declared function
221 return types as assertions.)
224 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
225 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
227 (DECLARE (TYPE INDEXOID X))
228 (THE (VALUES INDEXOID)
230 where the implementation of the type check in function FOO
231 includes a full call to %TYPEP. There are also some fundamental problems
232 with the interpretation of VALUES types (inherited from CMU CL, and
233 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
234 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
237 The definitions of SIGCONTEXT-FLOAT-REGISTER and
238 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
239 supported on FreeBSD because the floating point state is not saved,
240 but at least as of FreeBSD 4.0, the floating point state *is* saved,
241 so they could be supported after all. Very likely
242 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
245 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
246 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
247 issue with 'union, 'and, 'or etc. So even though according to the
248 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
249 (and now SBCL) interpret them as legal types.
252 ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
253 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
254 list, and it is probably possible to use substantially the same
255 patches to add it to SBCL.
258 a slew of floating-point-related errors reported by Peter Van Eynde
260 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
261 should probably be 1.4012985e-45. In SBCL,
262 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
263 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
264 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
265 and LEAST-NEGATIVE-LONG-FLOAT.
266 c: Many expressions generate floating infinity on x86/Linux:
271 PVE's regression tests want them to raise errors. sbcl-0.7.0.5
272 on x86/Linux generates the infinities instead. That might or
273 might not be conforming behavior, but it's also inconsistent,
274 which is almost certainly wrong. (Inconsistency: (/ 1 0.0)
275 should give the same result as (/ 1.0 0.0), but instead (/ 1 0.0)
276 generates SINGLE-FLOAT-POSITIVE-INFINITY and (/ 1.0 0.0)
278 d: (in section12.erg) various forms a la
279 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
280 don't give the right behavior.
283 type safety errors reported by Peter Van Eynde July 25, 2000:
284 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
286 In general lengths of array type specifications aren't
287 checked by COERCE, so it fails when the spec is
288 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
289 b: CONCATENATE has the same problem of not checking the length
290 of specified output array types. MAKE-SEQUENCE and MAP and
291 MERGE also have the same problem.
292 c: (COERCE 'AND 'FUNCTION) returns something related to
293 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
294 f: (FLOAT-RADIX 2/3) should signal an error instead of
296 g: (LOAD "*.lsp") should signal FILE-ERROR.
297 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
298 should signal TYPE-ERROR.
299 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
300 be used for input and output as needed. It should fail with
301 TYPE-ERROR when handed e.g. the results of
302 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
303 the inappropriate positions, but doesn't.
304 j: (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
306 should probably signal an error instead of making a pathname with
308 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
309 not a binary input stream, but instead cheerfully reads from
310 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
313 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
314 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
316 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
317 signal a PROGRAM-ERROR, and doesn't.
318 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
319 and other DEFCLASS forms with duplicate specifications in their
320 slots, should signal a PROGRAM-ERROR, and doesn't.
321 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
322 causes a COMPILER-ERROR.
325 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
326 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
327 PROGRAM-ERROR, but SBCL accepts it instead.
328 b: SYMBOL-MACROLET should refuse to bind something which is
329 declared as a global variable, signalling PROGRAM-ERROR.
330 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
331 it binds is declared SPECIAL inside.
334 type system errors reported by Peter Van Eynde July 25, 2000:
335 g: The type system [still] isn't all that smart about relationships
336 between hairy types. [The original example from PVE was
337 (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL, which was fixed
338 by CSR in sbcl-0.7.1.28, but there are still
339 plenty of corner cases out there: (SUBTYPEP 'ATOM 'LIST)
340 returns NIL, NIL in sbcl-0.7.1.31.]
343 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
345 (DEFGENERIC FOO02 (X))
346 (DEFMETHOD FOO02 ((X NUMBER)) T)
347 (LET ((M (FIND-METHOD (FUNCTION FOO02)
349 (LIST (FIND-CLASS (QUOTE NUMBER))))))
350 (REMOVE-METHOD (FUNCTION FOO02) M)
351 (DEFGENERIC FOO03 (X))
352 (ADD-METHOD (FUNCTION FOO03) M)))
353 should give an error, but SBCL allows it.
354 b: READ should probably return READER-ERROR, not the bare
355 arithmetic error, when input a la "1/0" or "1e1000" causes
359 It has been reported (e.g. by Peter Van Eynde) that there are
360 several metaobject protocol "errors". (In order to fix them, we might
361 need to document exactly what metaobject protocol specification
362 we're following -- the current code is just inherited from PCL.)
365 The implementation of #'+ returns its single argument without
366 type checking, e.g. (+ "illegal") => "illegal".
369 (SUBTYPEP '(AND ZILCH INTEGER) 'ZILCH) => NIL, NIL
370 Note: I looked into fixing this in 0.6.11.15, but gave up. The
371 problem seems to be that there are two relevant type methods for
372 the subtypep operation, HAIRY :COMPLEX-SUBTYPEP-ARG2 and
373 INTERSECTION :COMPLEX-SUBTYPEP-ARG1, and only the first is
374 called. This could be fixed, but type dispatch is messy and
375 confusing enough already, I don't want to complicate it further.
376 Perhaps someday we can make CLOS cross-compiled (instead of compiled
377 after bootstrapping) so that we don't need to have the type system
378 available before CLOS, and then we can rewrite the type methods to
379 CLOS methods, and then expressing the solutions to stuff like this
380 should become much more straightforward. -- WHN 2001-03-14
383 The debugger LIST-LOCATIONS command doesn't work properly.
386 Compiling and loading
387 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
389 then requesting a BACKTRACE at the debugger prompt gives no information
390 about where in the user program the problem occurred.
393 The compiler is supposed to do type inference well enough that
396 ((SIMPLE-ARRAY SINGLE-FLOAT)
398 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
401 is redundant. However, as reported by Juan Jose Garcia Ripoll for
402 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
403 workaround for the problem for now, but can't be done by the TYPECASE
404 macros themselves, since it's too hard for the macro to detect
405 assignments to the variable within the clause.
406 Note: The compiler *is* smart enough to do the type inference in
407 many cases. This case, derived from a couple of MACROEXPAND-1
408 calls on Ripoll's original test case,
410 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
411 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
412 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
413 (LET ((I 0) (G2554 LENGTH))
414 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
417 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
418 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
419 (GO SB-LOOP::NEXT-LOOP)
420 SB-LOOP::END-LOOP))))))
421 demonstrates the problem; but the problem goes away if the TAGBODY
422 and GO forms are removed (leaving the SETF in ordinary, non-looping
423 code), or if the TAGBODY and GO forms are retained, but the
424 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
427 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
428 I am looking into this problem that showed up on the cmucl-help
429 list. It seems to me that the "implementation specific environment
430 hacking functions" found in pcl/walker.lisp are completely messed
431 up. The good thing is that they appear to be barely used within
432 PCL and the munged environment object is passed to cmucl only
433 in calls to macroexpand-1, which is probably why this case fails.
434 SBCL uses essentially the same code, so if the environment hacking
435 is screwed up, it affects us too.
438 Using the pretty-printer from the command prompt gives funny
439 results, apparently because the pretty-printer doesn't know
440 about user's command input, including the user's carriage return
441 that the user, and therefore the pretty-printer thinks that
442 the new output block should start indented 2 or more characters
443 rightward of the correct location.
446 (probably related to bug #70; maybe related to bug #109)
447 As reported by Carl Witty on submit@bugs.debian.org 1999-05-08,
449 (in-package "CL-USER")
450 (defun equal-terms (termx termy)
452 ((alpha-equal-bound-term-lists (listx listy)
453 (or (and (null listx) (null listy))
455 (let ((bindings-x (bindings-of-bound-term (car listx)))
456 (bindings-y (bindings-of-bound-term (car listy))))
457 (if (and (null bindings-x) (null bindings-y))
458 (alpha-equal-terms (term-of-bound-term (car listx))
459 (term-of-bound-term (car listy)))
460 (and (= (length bindings-x) (length bindings-y))
462 (enter-binding-pairs (bindings-of-bound-term (car listx))
463 (bindings-of-bound-term (car listy)))
464 (alpha-equal-terms (term-of-bound-term (car listx))
465 (term-of-bound-term (car listy)))
466 (exit-binding-pairs (bindings-of-bound-term (car listx))
467 (bindings-of-bound-term (car listy)))))))
468 (alpha-equal-bound-term-lists (cdr listx) (cdr listy)))))
470 (alpha-equal-terms (termx termy)
471 (if (and (variable-p termx)
473 (equal-bindings (id-of-variable-term termx)
474 (id-of-variable-term termy))
475 (and (equal-operators-p (operator-of-term termx) (operator-of-term termy))
476 (alpha-equal-bound-term-lists (bound-terms-of-term termx)
477 (bound-terms-of-term termy))))))
481 (with-variable-invocation (alpha-equal-terms termx termy))))))
482 causes an assertion failure
483 The assertion (EQ (C::LAMBDA-TAIL-SET C::CALLER)
484 (C::LAMBDA-TAIL-SET (C::LAMBDA-HOME C::CALLEE))) failed.
486 Bob Rogers reports (1999-07-28 on cmucl-imp@cons.org) a smaller test
487 case with the same problem:
488 (defun parse-fssp-alignment ()
489 ;; Given an FSSP alignment file named by the argument . . .
490 (labels ((get-fssp-char ()
494 ;; Stub body, enough to tickle the bug.
495 (list (read-fssp-char)
499 ANSI specifies that the RESULT-TYPE argument of CONCATENATE must be
500 a subtype of SEQUENCE, but CONCATENATE doesn't check this properly:
501 (CONCATENATE 'SIMPLE-ARRAY #(1 2) '(3)) => #(1 2 3)
502 This also leads to funny behavior when derived type specifiers
503 are used, as originally reported by Milan Zamazal for CMU CL (on the
504 Debian bugs mailing list (?) 2000-02-27), then reported by Martin
505 Atzmueller for SBCL (2000-10-01 on sbcl-devel@lists.sourceforge.net):
506 (DEFTYPE FOO () 'SIMPLE-ARRAY)
507 (CONCATENATE 'FOO #(1 2) '(3))
508 => #<ARRAY-TYPE SIMPLE-ARRAY> is a bad type specifier for
510 The derived type specifier FOO should act the same way as the
511 built-in type SIMPLE-ARRAY here, but it doesn't. That problem
512 doesn't seem to exist for sequence types:
513 (DEFTYPE BAR () 'SIMPLE-VECTOR)
514 (CONCATENATE 'BAR #(1 2) '(3)) => #(1 2 3)
517 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
518 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
519 crashes SBCL. In general tracing anything which is used in the
520 implementation of TRACE is likely to have the same problem.
523 As reported by Daniel Solaz on cmucl-help@cons.org 2000-11-23,
524 SXHASH returns the same value for all non-STRUCTURE-OBJECT instances,
525 notably including all PCL instances. There's a limit to how much
526 SXHASH can do to return unique values for instances, but at least
527 it should probably look at the class name, the way that it does
528 for STRUCTURE-OBJECTs.
531 (probably related to bug #65; maybe related to bug #109)
532 The compiler doesn't like &OPTIONAL arguments in LABELS and FLET
534 (DEFUN FIND-BEFORE (ITEM SEQUENCE &KEY (TEST #'EQL))
535 (LABELS ((FIND-ITEM (OBJ SEQ TEST &OPTIONAL (VAL NIL))
536 (LET ((ITEM (FIRST SEQ)))
539 ((FUNCALL TEST OBJ ITEM)
542 (FIND-ITEM OBJ (REST SEQ) TEST (NCONC VAL `(,ITEM))))))))
543 (FIND-ITEM ITEM SEQUENCE TEST)))
544 from David Young's bug report on cmucl-help@cons.org 30 Nov 2000
545 causes sbcl-0.6.9 to fail with
546 error in function SB-KERNEL:ASSERT-ERROR:
547 The assertion (EQ (SB-C::LAMBDA-TAIL-SET SB-C::CALLER)
548 (SB-C::LAMBDA-TAIL-SET
549 (SB-C::LAMBDA-HOME SB-C::CALLEE))) failed.
552 (DECLAIM (OPTIMIZE ..)) doesn't work. E.g. even after
553 (DECLAIM (OPTIMIZE (SPEED 3))), things are still optimized with
554 the previous SPEED policy. This bug will probably get fixed in
555 0.6.9.x in a general cleanup of optimization policy.
558 (DECLAIM (OPTIMIZE ..)) doesn't work properly inside LOCALLY forms.
561 As reported by Martin Atzmueller on sbcl-devel 26 Dec 2000,
562 ANSI says that WITH-OUTPUT-TO-STRING should have a keyword
563 :ELEMENT-TYPE, but in sbcl-0.6.9 this is not defined for
564 WITH-OUTPUT-TO-STRING.
567 ANSI says in one place that type declarations can be abbreviated even
568 when the type name is not a symbol, e.g.
569 (DECLAIM ((VECTOR T) *FOOVECTOR*))
570 SBCL doesn't support this. But ANSI says in another place that this
571 isn't allowed. So it's not clear this is a bug after all. (See the
572 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
576 as pointed out by Dan Barlow on sbcl-devel 2000-07-02:
577 The PICK-TEMPORARY-FILE-NAME utility used by LOAD-FOREIGN uses
578 an easily guessable temporary filename in a way which might open
579 applications using LOAD-FOREIGN to hijacking by malicious users
580 on the same machine. Incantations for doing this safely are
581 floating around the net in various "how to write secure programs
582 despite Unix" documents, and it would be good to (1) fix this in
583 LOAD-FOREIGN, and (2) hunt for any other code which uses temporary
584 files and make it share the same new safe logic.
587 Functions are assigned names based on the context in which they're
588 defined. This is less than ideal for the functions which are
589 used to implement CLOS methods. E.g. the output of
590 (DESCRIBE 'PRINT-OBJECT) lists functions like
591 #<FUNCTION "DEF!STRUCT (TRACE-INFO (:MAKE-LOAD-FORM-FUN SB-KERNEL:JUST-DUMP-IT-NORMALLY) (:PRINT-OBJECT #))" {1020E49}>
593 #<FUNCTION "MACROLET ((FORCE-DELAYED-DEF!METHODS NIL #))" {1242871}>
594 It would be better if these functions' names always identified
595 them as methods, and identified their generic functions and
599 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
600 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
601 the mean of the distribution can be systematically O(0.1%) wrong.
602 Just increasing R-I-E-B is probably not a good solution, since
603 it would decrease efficiency more than is probably necessary. Perhaps
604 using some sort of accept/reject method would be better.
607 Internally the compiler sometimes evaluates
608 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
609 (I stumbled across this when I added an
610 (assert (not (eq type1 *wild-type*)))
611 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
612 in a type context should probably be translated to T, and so it's
613 probably wrong to ask whether it's equal to the T type and then (using
614 the EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
615 (I haven't tried to investigate this bug enough to guess whether
616 there might be any user-level symptoms.)
619 a latent cross-compilation/bootstrapping bug: The cross-compilation
620 host's CL:CHAR-CODE-LIMIT is used in target code in readtable.lisp
621 and possibly elsewhere. Instead, we should use the target system's
622 CHAR-CODE-LIMIT. This will probably cause problems if we try to
623 bootstrap on a system which uses a different value of CHAR-CODE-LIMIT
627 (subtypep '(or (integer -1 1)
631 (integer -1 1))) => NIL,T
632 An analogous problem with SINGLE-FLOAT and REAL types was fixed in
633 sbcl-0.6.11.22, but some peculiarites of the RATIO type make it
634 awkward to generalize the fix to INTEGER and RATIONAL. It's not
635 clear what's the best fix. (See the "bug in type handling" discussion
636 on cmucl-imp ca. 2001-03-22 and ca. 2001-02-12.)
639 Inconsistencies between derived and declared VALUES return types for
640 DEFUN aren't checked very well. E.g. the logic which successfully
641 catches problems like
642 (declaim (ftype (function (fixnum) float) foo))
644 (declare (type integer x))
645 (values x)) ; wrong return type, detected, gives warning, good!
647 (declaim (ftype (function (t) (values t t)) bar))
649 (values x)) ; wrong number of return values, no warning, bad!
650 The cause of this is seems to be that (1) the internal function
651 VALUES-TYPES-EQUAL-OR-INTERSECT used to make the check handles its
652 arguments symmetrically, and (2) when the type checking code was
653 written back when when SBCL's code was still CMU CL, the intent
655 (declaim (ftype (function (t) t) bar))
657 (values x x)) ; wrong number of return values; should give warning?
658 not be warned for, because a two-valued return value is considered
659 to be compatible with callers who expects a single value to be
660 returned. That intent is probably not appropriate for modern ANSI
661 Common Lisp, but fixing this might be complicated because of other
662 divergences between auld-style and new-style handling of
663 multiple-VALUES types. (Some issues related to this were discussed
664 on cmucl-imp at some length sometime in 2000.)
667 The facility for dumping a running Lisp image to disk gets confused
668 when run without the PURIFY option, and creates an unnecessarily large
669 core file (apparently representing memory usage up to the previous
670 high-water mark). Moreover, when the file is loaded, it confuses the
671 GC, so that thereafter memory usage can never be reduced below that
675 The TRACE facility can't be used on some kinds of functions.
676 (Basically, the breakpoint facility was incompletely implemented
677 in the X86 port of CMU CL, and hasn't been fixed in SBCL.)
680 In sbcl-0.6.11.41 (and in all earlier SBCL, and in CMU
681 CL), out-of-line structure slot setters are horribly inefficient
682 whenever the type of the slot is declared, because out-of-line
683 structure slot setters are implemented as closures to save space,
684 so the compiler doesn't compile the type test into code, but
685 instead just saves the type in a lexical closure and interprets it
687 A proper solution involves deciding whether it's really worth
688 saving space by implementing structure slot accessors as closures.
689 (If it's not worth it, the problem vanishes automatically. If it
690 is worth it, there are hacks we could use to force type tests to
691 be compiled anyway, and even shared. E.g. we could implement
692 an EQUAL hash table mapping from types to compiled type tests,
693 and save the appropriate compiled type test as part of each lexical
694 closure; or we could make the lexical closures be placeholders
695 which overwrite their old definition as a lexical closure with
696 a new compiled definition the first time that they're called.)
697 As a workaround for the problem, #'(SETF FOO) expressions can
698 be replaced with (EFFICIENT-SETF-FUNCTION FOO), where
699 (defmacro efficient-setf-function (place-function-name)
700 (or #+sbcl (and (sb-impl::info :function :accessor-for place-function-name)
701 ;; a workaround for the problem, encouraging the
702 ;; inline expansion of the structure accessor, so
703 ;; that the compiler can optimize its type test
704 (let ((new-value (gensym "NEW-VALUE-"))
705 (structure-value (gensym "STRUCTURE-VALUE-")))
706 `(lambda (,new-value ,structure-value)
707 (setf (,place-function-name ,structure-value)
709 ;; no problem, can just use the ordinary expansion
710 `(function (setf ,place-function-name))))
713 There's apparently a bug in CEILING optimization which caused
714 Douglas Crosher to patch the CMU CL version. Martin Atzmueller
715 applied the patches to SBCL and they didn't seem to cause problems
716 (as reported sbcl-devel 2001-05-04). However, since the patches
717 modify nontrivial code which was apparently written incorrectly
718 the first time around, until regression tests are written I'm not
719 comfortable merging the patches in the CVS version of SBCL.
722 As reported by Arthur Lemmens sbcl-devel 2001-05-05, ANSI
723 requires that SYMBOL-MACROLET refuse to rebind special variables,
724 but SBCL doesn't do this. (Also as reported by AL in the same
725 message, SBCL depended on this nonconforming behavior to build
726 itself, because of the way that **CURRENT-SEGMENT** was implemented.
727 As of sbcl-0.6.12.x, this dependence on the nonconforming behavior
728 has been fixed, but the nonconforming behavior remains.)
731 (DESCRIBE 'SB-ALIEN:DEF-ALIEN-TYPE) reports the macro argument list
735 in #<PACKAGE "SB-ALIEN">.
736 Macro-function: #<FUNCTION "DEF!MACRO DEF-ALIEN-TYPE" {19F4A39}>
737 Macro arguments: (#:whole-470 #:environment-471)
738 On Sat, May 26, 2001 09:45:57 AM CDT it was compiled from:
739 /usr/stuff/sbcl/src/code/host-alieneval.lisp
740 Created: Monday, March 12, 2001 07:47:43 AM CST
743 (TIME (ROOM T)) reports more than 200 Mbytes consed even for
744 a clean, just-started SBCL system. And it seems to be right:
745 (ROOM T) can bring a small computer to its knees for a *long*
746 time trying to GC afterwards. Surely there's some more economical
747 way to implement (ROOM T).
750 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
752 ;;; This file fails to compile.
753 ;;; Maybe this bug is related to bugs #65, #70 in the BUGS file.
754 (in-package :cl-user)
760 ;; Uncomment and it works
763 In SBCL 0.6.12.42, the problem is
764 internal error, failed AVER:
765 "(COMMON-LISP:EQ (SB!C::LAMBDA-TAIL-SET SB!C::CALLER)
766 (SB!C::LAMBDA-TAIL-SET (SB!C::LAMBDA-HOME SB!C::CALLEE)))"
769 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
771 ;;; The compiler is flushing the argument type test, and the default
772 ;;; case in the cond, so that calling with say a fixnum 0 causes a
774 (declaim (optimize (safety 2) (speed 3)))
776 (declare (type (or string stream) x))
777 (cond ((typep x 'string) 'string)
778 ((typep x 'stream) 'stream)
781 The symptom in sbcl-0.6.12.42 on OpenBSD is actually (TST 0)=>STREAM
782 (not the SIGBUS reported in the comment) but that's broken too;
783 type declarations are supposed to be treated as assertions unless
784 SAFETY 0, so we should be getting a TYPE-ERROR.
787 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
789 (in-package :cl-user)
790 ;;; From: David Gadbois <gadbois@cyc.com>
792 ;;; Logical pathnames aren't externalizable.
794 (let ((tempfile "/tmp/test.lisp"))
795 (setf (logical-pathname-translations "XXX")
796 '(("XXX:**;*.*" "/tmp/**/*.*")))
797 (with-open-file (out tempfile :direction :output)
798 (write-string "(defvar *path* #P\"XXX:XXX;FOO.LISP\")" out))
799 (compile-file tempfile))
800 The error message in sbcl-0.6.12.42 is
802 ; (while making load form for #<SB-IMPL::LOGICAL-HOST "XXX">)
803 ; A logical host can't be dumped as a constant: #<SB-IMPL::LOGICAL-HOST "XXX">
806 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
808 (in-package :cl-user)
809 ;;; The following invokes a compiler error.
810 (declaim (optimize (speed 2) (debug 3)))
813 (unwind-protect nil)))
817 The error message in sbcl-0.6.12.42 is
818 internal error, failed AVER:
819 "(COMMON-LISP:EQ (SB!C::TN-ENVIRONMENT SB!C:TN) SB!C::TN-ENV)"
822 When the compiler inline expands functions, it may be that different
823 kinds of return values are generated from different code branches.
824 E.g. an inline expansion of POSITION generates integer results
825 from one branch, and NIL results from another. When that inline
826 expansion is used in a context where only one of those results
829 (aref *a1* (position x *a2*)))
830 and the compiler can't prove that the unacceptable branch is
831 never taken, then bogus type mismatch warnings can be generated.
832 If you need to suppress the type mismatch warnings, you can
833 suppress the inline expansion,
835 #+sbcl (declare (notinline position)) ; to suppress bug 117 bogowarnings
836 (aref *a1* (position x *a2*)))
837 or, sometimes, suppress them by declaring the result to be of an
840 (aref *a1* (the integer (position x *a2*))))
842 This is not a new compiler problem in 0.7.0, but the new compiler
843 transforms for FIND, POSITION, FIND-IF, and POSITION-IF make it
844 more conspicuous. If you don't need performance from these functions,
845 and the bogus warnings are a nuisance for you, you can return to
846 your pre-0.7.0 state of grace with
847 #+sbcl (declaim (notinline find position find-if position-if)) ; bug 117..
850 as reported by Eric Marsden on cmucl-imp@cons.org 2001-08-14:
851 (= (FLOAT 1 DOUBLE-FLOAT-EPSILON)
852 (+ (FLOAT 1 DOUBLE-FLOAT-EPSILON) DOUBLE-FLOAT-EPSILON)) => T
853 when of course it should be NIL. (He says it only fails for X86,
854 not SPARC; dunno about Alpha.)
856 Also, "the same problem exists for LONG-FLOAT-EPSILON,
857 DOUBLE-FLOAT-NEGATIVE-EPSILON, LONG-FLOAT-NEGATIVE-EPSILON (though
858 for the -negative- the + is replaced by a - in the test)."
860 Raymond Toy comments that this is tricky on the X86 since its FPU
861 uses 80-bit precision internally.
864 The compiler incorrectly figures the return type of
865 (DEFUN FOO (FRAME UP-FRAME)
872 This problem exists in CMU CL 18c too. When I reported it on
873 cmucl-imp@cons.org, Raymond Toy replied 23 Aug 2001 with
874 a partial explanation, but no fix has been found yet.
877 Even in sbcl-0.pre7.x, which is supposed to be free of the old
878 non-ANSI behavior of treating the function return type inferred
879 from the current function definition as a declaration of the
880 return type from any function of that name, the return type of NIL
881 is attached to FOO in 120a above, and used to optimize code which
885 There was some sort of screwup in handling of
886 (IF (NOT (IGNORE-ERRORS ..))). E.g.
888 (if (not (ignore-errors
889 (make-pathname :host "foo" :directory "!bla" :name "bar")))
891 (error "notunlessnot")))
892 The (NOT (IGNORE-ERRORS ..)) form evaluates to T, so this should be
893 printing "ok", but instead it's going to the ERROR. This problem
894 seems to've been introduced by MNA's HANDLER-CASE patch (sbcl-devel
895 2001-07-17) and as a workaround (put in sbcl-0.pre7.14.flaky4.12)
896 I reverted back to the old weird HANDLER-CASE code. However, I
897 think the problem looks like a compiler bug in handling RETURN-FROM,
898 so I left the MNA-patched code in HANDLER-CASE (suppressed with
899 #+NIL) and I'd like to go back to see whether this really is
900 a compiler bug before I delete this BUGS entry.
903 The *USE-IMPLEMENTATION-TYPES* hack causes bugs, particularly
904 (IN-PACKAGE :SB-KERNEL)
905 (TYPE= (SPECIFIER-TYPE '(VECTOR T))
906 (SPECIFIER-TYPE '(VECTOR UNDEFTYPE)))
907 Then because of this, the compiler bogusly optimizes
908 (TYPEP #(11) '(SIMPLE-ARRAY UNDEF-TYPE 1))
909 to T. Unfortunately, just setting *USE-IMPLEMENTATION-TYPES* to
910 NIL around sbcl-0.pre7.14.flaky4.12 didn't work: the compiler complained
911 about type mismatches (probably harmlessly, another instance of bug 117);
912 and then cold init died with a segmentation fault.
915 As of version 0.pre7.14, SBCL's implementation of MACROLET makes
916 the entire lexical environment at the point of MACROLET available
917 in the bodies of the macroexpander functions. In particular, it
918 allows the function bodies (which run at compile time) to try to
919 access lexical variables (which are only defined at runtime).
920 It doesn't even issue a warning, which is bad.
922 The SBCL behavior arguably conforms to the ANSI spec (since the
923 spec says that the behavior is undefined, ergo anything conforms).
924 However, it would be better to issue a compile-time error.
925 Unfortunately I (WHN) don't see any simple way to detect this
926 condition in order to issue such an error, so for the meantime
927 SBCL just does this weird broken "conforming" thing.
929 The ANSI standard says, in the definition of the special operator
931 The macro-expansion functions defined by MACROLET are defined
932 in the lexical environment in which the MACROLET form appears.
933 Declarations and MACROLET and SYMBOL-MACROLET definitions affect
934 the local macro definitions in a MACROLET, but the consequences
935 are undefined if the local macro definitions reference any
936 local variable or function bindings that are visible in that
938 Then it seems to contradict itself by giving the example
940 (macrolet ((fudge (z)
941 ;The parameters x and flag are not accessible
942 ; at this point; a reference to flag would be to
943 ; the global variable of that name.
944 ` (if flag (* ,z ,z) ,z)))
945 ;The parameters x and flag are accessible here.
949 The comment "a reference to flag would be to the global variable
950 of the same name" sounds like good behavior for the system to have.
951 but actual specification quoted above says that the actual behavior
955 (as reported by Gabe Garza on cmucl-help 2001-09-21)
957 (defun test-pred (x y)
961 (func (lambda () x)))
962 (print (eq func func))
963 (print (test-pred func func))
964 (delete func (list func))))
965 Now calling (TEST-CASE) gives output
968 (#<FUNCTION {500A9EF9}>)
969 Evidently Python thinks of the lambda as a code transformation so
970 much that it forgets that it's also an object.
976 The DEFSTRUCT section of the ANSI spec, in the :CONC-NAME section,
977 specifies a precedence rule for name collisions between slot accessors of
978 structure classes related by inheritance. As of 0.7.0, SBCL still
982 insufficient syntax checking in MACROLET:
984 (macrolet ((defmacro bar (z) `(+ z z)))
986 shouldn't compile without error (because of the extra DEFMACRO symbol).
989 As of sbcl-0.pre7.86.flaky7.3, the cross-compiler, and probably
990 the CL:COMPILE function (which is based on the same %COMPILE
991 mechanism) get confused by
993 (labels ((sxhash-number (x)
995 (fixnum (sxhash x)) ; through DEFTRANSFORM
996 (integer (sb!bignum:sxhash-bignum x))
997 (single-float (sxhash x)) ; through DEFTRANSFORM
998 (double-float (sxhash x)) ; through DEFTRANSFORM
999 #!+long-float (long-float (error "stub: no LONG-FLOAT"))
1000 (ratio (let ((result 127810327))
1001 (declare (type fixnum result))
1002 (mixf result (sxhash-number (numerator x)))
1003 (mixf result (sxhash-number (denominator x)))
1005 (complex (let ((result 535698211))
1006 (declare (type fixnum result))
1007 (mixf result (sxhash-number (realpart x)))
1008 (mixf result (sxhash-number (imagpart x)))
1010 (sxhash-recurse (x &optional (depthoid +max-hash-depthoid+))
1011 (declare (type index depthoid))
1014 (if (plusp depthoid)
1015 (mix (sxhash-recurse (car x) (1- depthoid))
1016 (sxhash-recurse (cdr x) (1- depthoid)))
1019 (if (typep x 'structure-object)
1021 (sxhash ; through DEFTRANSFORM
1022 (class-name (layout-class (%instance-layout x)))))
1024 (symbol (sxhash x)) ; through DEFTRANSFORM
1025 (number (sxhash-number x))
1028 (simple-string (sxhash x)) ; through DEFTRANSFORM
1029 (string (%sxhash-substring x))
1030 (bit-vector (let ((result 410823708))
1031 (declare (type fixnum result))
1032 (dotimes (i (min depthoid (length x)))
1033 (mixf result (aref x i)))
1035 (t (logxor 191020317 (sxhash (array-rank x))))))
1038 (sxhash (char-code x)))) ; through DEFTRANSFORM
1040 (sxhash-recurse x)))
1041 complaining "function called with two arguments, but wants exactly
1042 one" about SXHASH-RECURSE. (This might not be strictly a new bug,
1043 since IIRC post-fork CMU CL has also had problems with &OPTIONAL
1044 arguments in FLET/LABELS: it might be an old Python bug which is
1045 only exercised by the new arrangement of the SBCL compiler.)
1048 Ideally, uninterning a symbol would allow it, and its associated
1049 FDEFINITION and PROCLAIM data, to be reclaimed by the GC. However,
1050 at least as of sbcl-0.7.0, this isn't the case. Information about
1051 FDEFINITIONs and PROCLAIMed properties is stored in globaldb.lisp
1052 essentially in ordinary (non-weak) hash tables keyed by symbols.
1053 Thus, once a system has an entry in this system, it tends to live
1054 forever, even when it is uninterned and all other references to it
1058 (reported by Arnaud Rouanet on cmucl-imp 2001-12-18)
1059 (defmethod foo ((x integer))
1061 (defmethod foo :around ((x integer))
1063 (call-next-method)))
1064 Now (FOO 3) should return 3, but instead it returns 4.
1067 (SB-DEBUG:BACKTRACE) output should start with something
1068 including the name BACKTRACE, not (as in 0.pre7.88)
1069 just "0: (\"hairy arg processor\" ...)". Until about
1070 sbcl-0.pre7.109, the names in BACKTRACE were all screwed
1071 up compared to the nice useful names in sbcl-0.6.13.
1072 Around sbcl-0.pre7.109, they were mostly fixed by using
1073 NAMED-LAMBDA to implement DEFUN. However, there are still
1074 some screwups left, e.g. as of sbcl-0.pre7.109, there are
1075 still some functions named "hairy arg processor" and
1076 "SB-INT:&MORE processor".
1079 (reported by Alexey Dejneka sbcl-devel 2002-01-03)
1081 SUBTYPEP does not work well with redefined classes:
1083 * (defclass a () ())
1085 * (defclass b () ())
1090 * (defclass b (a) ())
1095 * (defclass b () ())
1103 This is probably due to underzealous clearing of the type caches; a
1104 brute-force solution in that case would be to make a defclass expand
1105 into something that included a call to SB-KERNEL::CLEAR-TYPE-CACHES,
1106 but there may be a better solution.
1109 Pretty-printing nested backquotes doesn't work right, as
1110 reported by Alexey Dejneka sbcl-devel 2002-01-13:
1112 ``(FOO SB-IMPL::BACKQ-COMMA-AT S)
1113 * (lisp-implementation-version)
1117 (as reported by Lynn Quam on cmucl-imp ca. 2002-01-16)
1118 %NATURALIZE-C-STRING conses a lot, like 16 bytes per byte
1119 of the naturalized string. We could probably port the patches
1120 from the cmucl-imp mailing list.
1123 (reported by Jesse Bouwman 2001-10-24 through the unfortunately
1124 prominent SourceForge web/db bug tracking system, which is
1125 unfortunately not a reliable way to get a timely response from
1126 the SBCL maintainers)
1127 In the course of trying to build a test case for an
1128 application error, I encountered this behavior:
1129 If you start up sbcl, and then lay on CTRL-C for a
1130 minute or two, the lisp process will eventually say:
1131 %PRIMITIVE HALT called; the party is over.
1132 and throw you into the monitor. If I start up lisp,
1133 attach to the process with strace, and then do the same
1134 (abusive) thing, I get instead:
1135 access failure in heap page not marked as write-protected
1136 and the monitor again. I don't know enough to have the
1137 faintest idea of what is going on here.
1138 This is with sbcl 6.12, uname -a reports:
1139 Linux prep 2.2.19 #4 SMP Tue Apr 24 13:59:52 CDT 2001 i686 unknown
1140 I (WHN) have verified that the same thing occurs on sbcl-0.pre7.141
1141 under OpenBSD 2.9 on my X86 laptop. Do be patient when you try it:
1142 it took more than two minutes (but less than five) for me.
1145 (This was once known as IR1-4, but it lived on even after the
1146 IR1 interpreter went to the big bit bucket in the sky.)
1147 The system accepts DECLAIM in most places where DECLARE would be
1148 accepted, without even issuing a warning. ANSI allows this, but since
1149 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
1150 meaning is rather different, and it's unlikely that the user
1151 has a good reason for doing DECLAIM not at top level, it would be
1152 good to issue a STYLE-WARNING when this happens. A possible
1153 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
1154 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
1155 [This is considered an IR1-interpreter-related bug because until
1156 EVAL-WHEN is rewritten, which won't happen until after the IR1
1157 interpreter is gone, the system's notion of what's a top-level form
1158 and what's not will remain too confused to fix this problem.]
1161 ANSI allows types `(COMPLEX ,FOO) to use very hairy values for
1162 FOO, e.g. (COMPLEX (AND REAL (SATISFIES ODDP))). The old CMU CL
1163 COMPLEX implementation didn't deal with this, and hasn't been
1164 upgraded to do so. (This doesn't seem to be a high priority
1165 conformance problem, since seems hard to construct useful code
1169 Floating point errors are reported poorly. E.g. on x86 OpenBSD
1172 debugger invoked on condition of type SB-KERNEL:FLOATING-POINT-EXCEPTION:
1173 An arithmetic error SB-KERNEL:FLOATING-POINT-EXCEPTION was signalled.
1174 No traps are enabled? How can this be?
1175 It should be possible to be much more specific (overflow, division
1176 by zero, etc.) and of course the "How can this be?" should be fixable.
1179 (reported by Alexey Dejneka sbcl-devel 2002-01-28)
1180 Compiling a file containing
1181 (deftype digit () '(member #\1))
1182 (defun parse-num (string ind)
1185 (if (and (< ind ind)
1186 (typep (char string ind) 'digit))
1188 in sbcl-0.7.1 causes the compiler to fail with
1189 internal error, failed AVER: "(= (LENGTH (BLOCK-SUCC CALL-BLOCK)) 1)"
1190 This problem seems to have been introduced by the sbcl-0.pre7.* compiler
1191 changes, since 0.pre7.73 and 0.6.13 don't suffer from it. A related
1193 (defun parse-num (index)
1200 (when (and (digs) (digs)) x))))
1201 In sbcl-0.7.1, this second test case failed with the same
1202 internal error, failed AVER: "(= (LENGTH (BLOCK-SUCC CALL-BLOCK)) 1)"
1203 After the APD patches in sbcl-0.7.1.2 (new consistency check in
1204 TARGET-IF-DESIRABLE, plus a fix in meta-vmdef.lisp to keep the
1205 new consistency check from failing routinely) this second test case
1206 failed in FIND-IN-PHYSENV instead. Fixes in sbcl-0.7.1.3 (not
1207 closing over unreferenced variables) made this second test case
1208 compile without error, but the original test case still fails.
1210 Another way to get rid of the DEFTYPE without changing the symptom
1213 (defun parse-num (string ind)
1216 (if (and (< ind ind)
1217 (sb-int:memq *ch* '(#\1)))
1219 In sbcl-0.7.1.3, this fails with
1220 internal error, failed AVER: "(= (LENGTH (BLOCK-SUCC CALL-BLOCK)) 1)"
1221 The problem occurs while the inline expansion of MEMQ,
1222 #<LAMBDA :%DEBUG-NAME "varargs entry point for SB-C::.ANONYMOUS.">
1223 is being LET-converted after having its second REF deleted, leaving
1224 it with only one entry in LEAF-REFS.
1227 In sbcl-0.7.1.3 on x86, COMPILE-FILE on the file
1228 (in-package :cl-user)
1231 (defstruct foo bar bletch)
1233 (labels ((kidify1 (kid)
1238 (m+ (frobnicate kid)
1241 (declare (inline kid-frob))
1244 (the simple-vector (foo-bar perd)))))
1246 debugger invoked on condition of type TYPE-ERROR:
1247 The value NIL is not of type SB-C::NODE.
1248 The location of this failure has moved around as various related
1249 issues were cleaned up. As of sbcl-0.7.1.9, it occurs in
1250 NODE-BLOCK called by LAMBDA-COMPONENT called by IR2-CONVERT-CLOSURE.
1253 From the ANSI description of GET-DISPATCH-MACRO-CHARACTER, it
1254 should return NIL when there is no definition, e.g.
1255 (GET-DISPATCH-MACRO-CHARACTER #\# #\{) => NIL
1256 Instead, in sbcl-0.7.1.17 it returns
1257 #<FUNCTION "top level local call SB!IMPL::DISPATCH-CHAR-ERROR">
1260 (essentially the same problem as a CMU CL bug reported by Martin
1261 Cracauer on cmucl-imp 2002-02-19)
1262 There is a hole in structure slot type checking. Compiling and LOADing
1263 (declaim (optimize safety))
1265 (bla 0 :type fixnum))
1267 (let ((foo (make-foo)))
1268 (setf (foo-bla foo) '(1 . 1))
1269 (format t "Is ~a of type ~a a cons? => ~a~%"
1271 (type-of (foo-bla foo))
1272 (consp (foo-bla foo)))))
1274 should signal an error, but in sbcl-0.7.1.21 instead gives the output
1275 Is (1 . 1) of type CONS a cons? => NIL
1276 without signalling an error.
1279 There's some sort of problem with aborting back out of the debugger
1280 after a %DETECT-STACK-EXHAUSTION error in sbcl-0.7.1.38. In some cases
1281 telling the debugger to ABORT doesn't get you back to the main REPL,
1282 but instead just gives you another stack exhaustion error. The problem
1283 doesn't occur in the trivial case
1284 * (defun frob () (frob) (frob))
1287 but it has happened in more complicated cases (which I haven't
1288 figured out how to reproduce).
1290 DEFUNCT CATEGORIES OF BUGS
1292 These labels were used for bugs related to the old IR1 interpreter.
1293 The # values reached 6 before the category was closed down.