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. ANSI justifies this specification by saying
54 The restriction against issuing a warning for type mismatches
55 between a slot-initform and the corresponding slot's :TYPE
56 option is necessary because a slot-initform must be specified
57 in order to specify slot options; in some cases, no suitable
59 In SBCL, as in CMU CL (or, for that matter, any compiler which
60 really understands Common Lisp types) a suitable default does
61 exist, in all cases, because the compiler understands the concept
62 of functions which never return (i.e. has return type NIL, e.g.
63 ERROR). Thus, as a portable workaround, you can use a call to
64 some known-never-to-return function as the default. E.g.
66 (BAR (ERROR "missing :BAR argument")
67 :TYPE SOME-TYPE-TOO-HAIRY-TO-CONSTRUCT-AN-INSTANCE-OF))
69 (DECLAIM (FTYPE () NIL) MISSING-ARG)
70 (DEFUN REQUIRED-ARG () ; workaround for SBCL non-ANSI slot init typing
71 (ERROR "missing required argument"))
73 (BAR (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
74 (BLETCH (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
75 (N-REFS-SO-FAR 0 :TYPE (INTEGER 0)))
76 Such code will compile without complaint and work correctly either
77 on SBCL or on a completely compliant Common Lisp system.
80 bogus warnings about undefined functions for magic functions like
81 SB!C::%%DEFUN and SB!C::%DEFCONSTANT when cross-compiling files
82 like src/code/float.lisp. Fixing this will probably require
83 straightening out enough bootstrap consistency issues that
84 the cross-compiler can run with *TYPE-SYSTEM-INITIALIZED*.
85 Instead, the cross-compiler runs in a slightly flaky state
86 which is sane enough to compile SBCL itself, but which is
87 also unstable in several ways, including its inability
88 to really grok function declarations.
91 The "byte compiling top-level form:" output ought to be condensed.
92 Perhaps any number of such consecutive lines ought to turn into a
93 single "byte compiling top-level forms:" line.
96 The way that the compiler munges types with arguments together
97 with types with no arguments (in e.g. TYPE-EXPAND) leads to
98 weirdness visible to the user:
99 (DEFTYPE FOO () 'FIXNUM)
101 (TYPEP 11 '(FOO)) => T, which seems weird
102 (TYPEP 11 'FIXNUM) => T
103 (TYPEP 11 '(FIXNUM)) signals an error, as it should
104 The situation is complicated by the presence of Common Lisp types
105 like UNSIGNED-BYTE (which can either be used in list form or alone)
106 so I'm not 100% sure that the behavior above is actually illegal.
107 But I'm 90+% sure, and the following related behavior,
109 treating the bare symbol AND as equivalent to '(AND), is specifically
110 forbidden (by the ANSI specification of the AND type).
113 It would be nice if the
115 (during macroexpansion)
116 said what macroexpansion was at fault, e.g.
118 (during macroexpansion of IN-PACKAGE,
119 during macroexpansion of DEFFOO)
122 The ANSI syntax for non-STANDARD method combination types in CLOS is
123 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
124 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
125 If you mess this up, omitting the PROGN qualifier in in DEFMETHOD,
126 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
127 (DEFMETHOD FOO ((X BAR)) (PRINT 'NUMBER))
128 the error mesage is not easy to understand:
129 INVALID-METHOD-ERROR was called outside the dynamic scope
130 of a method combination function (inside the body of
131 DEFINE-METHOD-COMBINATION or a method on the generic
132 function COMPUTE-EFFECTIVE-METHOD).
133 It would be better if it were more informative, a la
134 The method combination type for this method (STANDARD) does
135 not match the method combination type for the generic function
137 Also, after you make the mistake of omitting the PROGN qualifier
138 on a DEFMETHOD, doing a new DEFMETHOD with the correct qualifier
140 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
142 INVALID-METHOD-ERROR was called outside the dynamic scope
143 of a method combination function (inside the body of
144 DEFINE-METHOD-COMBINATION or a method on the generic
145 function COMPUTE-EFFECTIVE-METHOD).
146 This is not very helpful..
149 (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
150 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
151 (Also, when this is fixed, we can enable the code in PROCLAIM which
152 checks for incompatible FTYPE redeclarations.)
155 from DTC on the CMU CL mailing list 25 Feb 2000:
156 ;;; Compiler fails when this file is compiled.
158 ;;; Problem shows up in delete-block within ir1util.lisp. The assertion
159 ;;; (assert (member (functional-kind lambda) '(:let :mv-let :assignment)))
160 ;;; fails within bind node branch.
162 ;;; Note that if c::*check-consistency* is enabled then an un-reached
163 ;;; entry is also reported.
166 (declare (values nil))
183 (let ((ttt #'(lambda () (go cccc))))
184 (declare (special ttt))
185 (return-from bbbb nil))
188 (return-from bbbb nil))))))
191 (I *think* this is a bug. It certainly seems like strange behavior. But
192 the ANSI spec is scary, dark, and deep..)
193 (FORMAT NIL "~,1G" 1.4) => "1. "
194 (FORMAT NIL "~3,1G" 1.4) => "1. "
197 from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
199 (setf (find-class 'ccc1) (find-class 'ccc))
200 (defmethod zut ((c ccc1)) 123)
201 DTC's recommended workaround from the mailing list 3 Mar 2000:
202 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
205 The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
206 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
207 inside "~<..~>" (without the colon modifier on the closing syntax).
208 However, SBCL doesn't do this:
209 * (FORMAT T "~<munge~wegnum~>" 12)
214 When too many files are opened, OPEN will fail with an
215 uninformative error message
216 error in function OPEN: error opening #P"/tmp/foo.lisp": NIL
217 instead of saying that too many files are open.
220 Right now, when COMPILE-FILE has a read error, it actually pops
221 you into the debugger before giving up on the file. It should
222 instead handle the error, perhaps issuing (and handling)
223 a secondary error "caught ERROR: unrecoverable error during compilation"
224 and then return with FAILURE-P true,
227 reported by Sam Steingold on the cmucl-imp mailing list 12 May 2000:
228 Also, there is another bug: `array-displacement' should return an
229 array or nil as first value (as per ANSI CL), while CMUCL declares
230 it as returning an array as first value always.
231 (Actually, I think the old CMU CL version in SBCL never returns NIL,
232 i.e. it's not just a declaration problem, but the definition doesn't
236 Sometimes (SB-EXT:QUIT) fails with
237 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
238 Process inferior-lisp exited abnormally with code 1
239 I haven't noticed a repeatable case of this yet.
242 some sort of bug in inlining and RETURN-FROM in sbcl-0.6.5: Compiling
245 (BLOCK USED-BY-SOME-Y?
248 (UNLESS (REJECTED? Y)
249 (RETURN-FROM USED-BY-SOME-Y? T)))))
250 (DECLARE (INLINE FROB))
255 error in function SB-KERNEL:ASSERT-ERROR:
256 The assertion (EQ (SB-C::CONTINUATION-KIND SB-C::CONT) :BLOCK-START) failed.
257 This is still present in sbcl-0.6.8.
260 In some cases the compiler believes type declarations on array
261 elements without checking them, e.g.
262 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
265 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
266 (WHEN (CONSP (AREF X 0))
268 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
271 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
272 all cases, e.g. the type assumption *is* checked if the array
273 elements are declared to be of some structure type instead of CONS.
276 The printer doesn't report closures very well. This is true in
280 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
281 It would be nice to make closures have a settable name slot,
282 and make things like DEFSTRUCT and FLET, which create closures,
283 set helpful values into this slot.
286 And as long as we're wishing, it would be awfully nice if INSPECT could
287 also report on closures, telling about the values of the bound variables.
290 The compiler assumes that any time a function of declared FTYPE
291 doesn't signal an error, its arguments were of the declared type.
292 E.g. compiling and loading
293 (DECLAIM (OPTIMIZE (SAFETY 3)))
294 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
295 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE) FACTORIAL)))
297 (COND ((> (FACTORIAL X) 1.0E6)
298 (FORMAT T "too big~%"))
300 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
302 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
305 will cause the INTEGERP case to be selected, giving bogus output a la
307 This violates the "declarations are assertions" principle.
308 According to the ANSI spec, in the section "System Class FUNCTION",
309 this is a case of "lying to the compiler", but the lying is done
310 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
311 not by the unexpectedly general definition of FACTORIAL. In any case,
312 "declarations are assertions" means that lying to the compiler should
313 cause an error to be signalled, and should not cause a bogus
314 result to be returned. Thus, the compiler should not assume
315 that arbitrary functions check their argument types. (It might
316 make sense to add another flag (CHECKED?) to DEFKNOWN to
317 identify functions which *do* check their argument types.)
320 DEFMETHOD doesn't check the syntax of &REST argument lists properly,
321 accepting &REST even when it's not followed by an argument name:
322 (DEFMETHOD FOO ((X T) &REST) NIL)
325 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
326 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
328 (DECLARE (TYPE INDEXOID X))
329 (THE (VALUES INDEXOID)
331 where the implementation of the type check in function FOO
332 includes a full call to %TYPEP. There are also some fundamental problems
333 with the interpretation of VALUES types (inherited from CMU CL, and
334 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
335 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
338 The definitions of SIGCONTEXT-FLOAT-REGISTER and
339 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
340 supported on FreeBSD because the floating point state is not saved,
341 but at least as of FreeBSD 4.0, the floating point state *is* saved,
342 so they could be supported after all. Very likely
343 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
346 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
347 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
348 issue with 'union, 'and, 'or etc. So even though according to the
349 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
350 (and now SBCL) interpret them as legal types.
353 ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
354 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
355 list, and it is probably possible to use substantially the same
356 patches to add it to SBCL.
359 a slew of floating-point-related errors reported by Peter Van Eynde
361 a: (fixed in sbcl-0.6.11.25)
362 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
363 should probably be 1.4012985e-45. In SBCL,
364 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
365 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
366 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
367 and LEAST-NEGATIVE-LONG-FLOAT.
368 c: Many expressions generate floating infinity:
373 PVE's regression tests want them to raise errors. SBCL
374 generates the infinities instead, which may or may not be
376 d: (in section12.erg) various forms a la
377 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
378 don't give the right behavior.
381 type safety errors reported by Peter Van Eynde July 25, 2000:
382 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
384 In general lengths of array type specifications aren't
385 checked by COERCE, so it fails when the spec is
386 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
387 b: CONCATENATE has the same problem of not checking the length
388 of specified output array types. MAKE-SEQUENCE and MAP and
389 MERGE also have the same problem.
390 c: (COERCE 'AND 'FUNCTION) returns something related to
391 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
392 d: ELT signals SIMPLE-ERROR if its index argument
393 isn't a valid index for its sequence argument, but should
394 signal TYPE-ERROR instead.
395 e: FILE-LENGTH is supposed to signal a type error when its
396 argument is not a stream associated with a file, but doesn't.
397 f: (FLOAT-RADIX 2/3) should signal an error instead of
399 g: (LOAD "*.lsp") should signal FILE-ERROR.
400 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
401 should signal TYPE-ERROR.
402 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
403 be used for input and output as needed. It should fail with
404 TYPE-ERROR when handed e.g. the results of
405 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
406 the inappropriate positions, but doesn't.
407 j: (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
409 should probably signal an error instead of making a pathname with
411 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
412 not a binary input stream, but instead cheerfully reads from
413 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
416 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
417 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
419 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
420 signal a PROGRAM-ERROR, and doesn't.
421 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
422 and other DEFCLASS forms with duplicate specifications in their
423 slots, should signal a PROGRAM-ERROR, and doesn't.
424 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
425 causes a COMPILER-ERROR.
428 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
429 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
430 PROGRAM-ERROR, but SBCL accepts it instead.
431 b: SYMBOL-MACROLET should refuse to bind something which is
432 declared as a global variable, signalling PROGRAM-ERROR.
433 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
434 it binds is declared SPECIAL inside.
437 LOOP bugs reported by Peter Van Eynde July 25, 2000:
438 a: (LOOP WITH (A B) DO (PRINT 1)) is a syntax error according to
439 the definition of WITH clauses given in the ANSI spec, but
440 compiles and runs happily in SBCL.
441 b: a messy one involving package iteration:
442 interpreted Form: (LET ((PACKAGE (MAKE-PACKAGE "LOOP-TEST"))) (INTERN "blah" PACKAGE) (LET ((BLAH2 (INTERN "blah2" PACKAGE))) (EXPORT BLAH2 PACKAGE)) (LIST (SORT (LOOP FOR SYM BEING EACH PRESENT-SYMBOL OF PACKAGE FOR SYM-NAME = (SYMBOL-NAME SYM) COLLECT SYM-NAME) (FUNCTION STRING<)) (SORT (LOOP FOR SYM BEING EACH EXTERNAL-SYMBOL OF PACKAGE FOR SYM-NAME = (SYMBOL-NAME SYM) COLLECT SYM-NAME) (FUNCTION STRING<))))
443 Should be: (("blah" "blah2") ("blah2"))
444 SBCL: (("blah") ("blah2"))
445 * (LET ((X 1)) (LOOP FOR I BY (INCF X) FROM X TO 10 COLLECT I))
446 doesn't work -- SBCL's LOOP says BY isn't allowed in a FOR clause.
449 type system errors reported by Peter Van Eynde July 25, 2000:
450 a: (SUBTYPEP 'BIGNUM 'INTEGER) => NIL, NIL
451 but should be (VALUES T T) instead.
452 b: (SUBTYPEP 'EXTENDED-CHAR 'CHARACTER) => NIL, NIL
453 but should be (VALUES T T) instead.
454 c: (SUBTYPEP '(INTEGER (0) (0)) 'NIL) dies with nested errors.
455 d: In general, the system doesn't like '(INTEGER (0) (0)) -- it
456 blows up at the level of SPECIFIER-TYPE with
457 "Lower bound (0) is greater than upper bound (0)." Probably
458 SPECIFIER-TYPE should return NIL instead.
459 e: (TYPEP 0 '(COMPLEX (EQL 0)) fails with
460 "Component type for Complex is not numeric: (EQL 0)."
461 This might be easy to fix; the type system already knows
462 that (SUBTYPEP '(EQL 0) 'NUMBER) is true.
463 f: The type system doesn't know about the condition system,
464 so that e.g. (TYPEP 'SIMPLE-ERROR 'ERROR)=>NIL.
465 g: The type system isn't all that smart about relationships
466 between hairy types, as shown in the type.erg test results,
467 e.g. (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL.
470 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
472 (DEFGENERIC FOO02 (X))
473 (DEFMETHOD FOO02 ((X NUMBER)) T)
474 (LET ((M (FIND-METHOD (FUNCTION FOO02)
476 (LIST (FIND-CLASS (QUOTE NUMBER))))))
477 (REMOVE-METHOD (FUNCTION FOO02) M)
478 (DEFGENERIC FOO03 (X))
479 (ADD-METHOD (FUNCTION FOO03) M)))
480 should give an error, but SBCL allows it.
481 b: READ should probably return READER-ERROR, not the bare
482 arithmetic error, when input a la "1/0" or "1e1000" causes
486 It has been reported (e.g. by Peter Van Eynde) that there are
487 several metaobject protocol "errors". (In order to fix them, we might
488 need to document exactly what metaobject protocol specification
489 we're following -- the current code is just inherited from PCL.)
492 another error from Peter Van Eynde 5 September 2000:
493 (FORMAT NIL "~F" "FOO") should work, but instead reports an error.
494 PVE submitted a patch to deal with this bug, but it exposes other
495 comparably serious bugs, so I didn't apply it. It looks as though
496 the FORMAT code needs a fair amount of rewriting in order to comply
497 with the various details of the ANSI spec.
500 The implementation of #'+ returns its single argument without
501 type checking, e.g. (+ "illegal") => "illegal".
504 In sbcl-0.6.7, there is no doc string for CL:PUSH, probably
505 because it's defined with the DEFMACRO-MUNDANELY macro and something
506 is wrong with doc string setting in that macro.
509 Attempting to use COMPILE on something defined by DEFMACRO fails:
510 (DEFMACRO FOO (X) (CONS X X))
512 Error in function C::GET-LAMBDA-TO-COMPILE:
513 #<Closure Over Function "DEFUN (SETF MACRO-FUNCTION)" {480E21B1}> was defined in a non-null environment.
516 (SUBTYPEP '(AND ZILCH INTEGER) 'ZILCH) => NIL, NIL
517 Note: I looked into fixing this in 0.6.11.15, but gave up. The
518 problem seems to be that there are two relevant type methods for
519 the subtypep operation, HAIRY :COMPLEX-SUBTYPEP-ARG2 and
520 INTERSECTION :COMPLEX-SUBTYPEP-ARG1, and only the first is
521 called. This could be fixed, but type dispatch is messy and
522 confusing enough already, I don't want to complicate it further.
523 Perhaps someday we can make CLOS cross-compiled (instead of compiled
524 after bootstrapping) so that we don't need to have the type system
525 available before CLOS, and then we can rewrite the type methods to
526 CLOS methods, and then expressing the solutions to stuff like this
527 should become much more straightforward. -- WHN 2001-03-14
530 CL:*DEFAULT-PATHNAME-DEFAULTS* doesn't behave as ANSI suggests (reflecting
531 current working directory). And there's no supported way to update
532 or query the current working directory (a la Unix "chdir" and "pwd"),
533 which is functionality that ILISP needs (and currently gets with low-level
537 The debugger LIST-LOCATIONS command doesn't work properly.
540 Compiling and loading
541 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
543 then requesting a BACKTRACE at the debugger prompt gives no information
544 about where in the user program the problem occurred.
547 The compiler is supposed to do type inference well enough that
550 ((SIMPLE-ARRAY SINGLE-FLOAT)
552 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
555 is redundant. However, as reported by Juan Jose Garcia Ripoll for
556 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
557 workaround for the problem for now, but can't be done by the TYPECASE
558 macros themselves, since it's too hard for the macro to detect
559 assignments to the variable within the clause.
560 Note: The compiler *is* smart enough to do the type inference in
561 many cases. This case, derived from a couple of MACROEXPAND-1
562 calls on Ripoll's original test case,
564 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
565 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
566 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
567 (LET ((I 0) (G2554 LENGTH))
568 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
571 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
572 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
573 (GO SB-LOOP::NEXT-LOOP)
574 SB-LOOP::END-LOOP))))))
575 demonstrates the problem; but the problem goes away if the TAGBODY
576 and GO forms are removed (leaving the SETF in ordinary, non-looping
577 code), or if the TAGBODY and GO forms are retained, but the
578 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
581 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
582 I am looking into this problem that showed up on the cmucl-help
583 list. It seems to me that the "implementation specific environment
584 hacking functions" found in pcl/walker.lisp are completely messed
585 up. The good thing is that they appear to be barely used within
586 PCL and the munged environment object is passed to cmucl only
587 in calls to macroexpand-1, which is probably why this case fails.
588 SBCL uses essentially the same code, so if the environment hacking
589 is screwed up, it affects us too.
592 Using the pretty-printer from the command prompt gives funny
593 results, apparently because the pretty-printer doesn't know
594 about user's command input, including the user's carriage return
595 that the user, and therefore the pretty-printer thinks that
596 the new output block should start indented 2 or more characters
597 rightward of the correct location.
600 (probably related to bug #70)
601 As reported by Carl Witty on submit@bugs.debian.org 1999-05-08,
603 (in-package "CL-USER")
604 (defun equal-terms (termx termy)
606 ((alpha-equal-bound-term-lists (listx listy)
607 (or (and (null listx) (null listy))
609 (let ((bindings-x (bindings-of-bound-term (car listx)))
610 (bindings-y (bindings-of-bound-term (car listy))))
611 (if (and (null bindings-x) (null bindings-y))
612 (alpha-equal-terms (term-of-bound-term (car listx))
613 (term-of-bound-term (car listy)))
614 (and (= (length bindings-x) (length bindings-y))
616 (enter-binding-pairs (bindings-of-bound-term (car listx))
617 (bindings-of-bound-term (car listy)))
618 (alpha-equal-terms (term-of-bound-term (car listx))
619 (term-of-bound-term (car listy)))
620 (exit-binding-pairs (bindings-of-bound-term (car listx))
621 (bindings-of-bound-term (car listy)))))))
622 (alpha-equal-bound-term-lists (cdr listx) (cdr listy)))))
624 (alpha-equal-terms (termx termy)
625 (if (and (variable-p termx)
627 (equal-bindings (id-of-variable-term termx)
628 (id-of-variable-term termy))
629 (and (equal-operators-p (operator-of-term termx) (operator-of-term termy))
630 (alpha-equal-bound-term-lists (bound-terms-of-term termx)
631 (bound-terms-of-term termy))))))
635 (with-variable-invocation (alpha-equal-terms termx termy))))))
636 causes an assertion failure
637 The assertion (EQ (C::LAMBDA-TAIL-SET C::CALLER)
638 (C::LAMBDA-TAIL-SET (C::LAMBDA-HOME C::CALLEE))) failed.
640 Bob Rogers reports (1999-07-28 on cmucl-imp@cons.org) a smaller test
641 case with the same problem:
642 (defun parse-fssp-alignment ()
643 ;; Given an FSSP alignment file named by the argument . . .
644 (labels ((get-fssp-char ()
648 ;; Stub body, enough to tickle the bug.
649 (list (read-fssp-char)
653 ANSI specifies that the RESULT-TYPE argument of CONCATENATE must be
654 a subtype of SEQUENCE, but CONCATENATE doesn't check this properly:
655 (CONCATENATE 'SIMPLE-ARRAY #(1 2) '(3)) => #(1 2 3)
656 This also leads to funny behavior when derived type specifiers
657 are used, as originally reported by Milan Zamazal for CMU CL (on the
658 Debian bugs mailing list (?) 2000-02-27), then reported by Martin
659 Atzmueller for SBCL (2000-10-01 on sbcl-devel@lists.sourceforge.net):
660 (DEFTYPE FOO () 'SIMPLE-ARRAY)
661 (CONCATENATE 'FOO #(1 2) '(3))
662 => #<ARRAY-TYPE SIMPLE-ARRAY> is a bad type specifier for
664 The derived type specifier FOO should act the same way as the
665 built-in type SIMPLE-ARRAY here, but it doesn't. That problem
666 doesn't seem to exist for sequence types:
667 (DEFTYPE BAR () 'SIMPLE-VECTOR)
668 (CONCATENATE 'BAR #(1 2) '(3)) => #(1 2 3)
671 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
672 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
673 crashes SBCL. In general tracing anything which is used in the
674 implementation of TRACE is likely to have the same problem.
677 As reported by Daniel Solaz on cmucl-help@cons.org 2000-11-23,
678 SXHASH returns the same value for all non-STRUCTURE-OBJECT instances,
679 notably including all PCL instances. There's a limit to how much
680 SXHASH can do to return unique values for instances, but at least
681 it should probably look at the class name, the way that it does
682 for STRUCTURE-OBJECTs.
685 As reported by Martin Atzmueller on the sbcl-devel list 2000-11-22,
686 > There remains one issue, that is a bug in SBCL:
687 > According to my interpretation of the spec, the ":" and "@" modifiers
688 > should appear _after_ the comma-seperated arguments.
689 > Well, SBCL (and CMUCL for that matter) accept
690 > (ASSERT (STRING= (FORMAT NIL "~:8D" 1) " 1"))
691 > where the correct way (IMHO) should be
692 > (ASSERT (STRING= (FORMAT NIL "~8:D" 1) " 1"))
693 Probably SBCL should stop accepting the "~:8D"-style format arguments,
694 or at least issue a warning.
697 (probably related to bug #65)
698 The compiler doesn't like &OPTIONAL arguments in LABELS and FLET
700 (DEFUN FIND-BEFORE (ITEM SEQUENCE &KEY (TEST #'EQL))
701 (LABELS ((FIND-ITEM (OBJ SEQ TEST &OPTIONAL (VAL NIL))
702 (LET ((ITEM (FIRST SEQ)))
705 ((FUNCALL TEST OBJ ITEM)
708 (FIND-ITEM OBJ (REST SEQ) TEST (NCONC VAL `(,ITEM))))))))
709 (FIND-ITEM ITEM SEQUENCE TEST)))
710 from David Young's bug report on cmucl-help@cons.org 30 Nov 2000
711 causes sbcl-0.6.9 to fail with
712 error in function SB-KERNEL:ASSERT-ERROR:
713 The assertion (EQ (SB-C::LAMBDA-TAIL-SET SB-C::CALLER)
714 (SB-C::LAMBDA-TAIL-SET
715 (SB-C::LAMBDA-HOME SB-C::CALLEE))) failed.
718 (DECLAIM (OPTIMIZE ..)) doesn't work. E.g. even after
719 (DECLAIM (OPTIMIZE (SPEED 3))), things are still optimized with
720 the previous SPEED policy. This bug will probably get fixed in
721 0.6.9.x in a general cleanup of optimization policy.
724 (DECLAIM (OPTIMIZE ..)) doesn't work properly inside LOCALLY forms.
727 As noted in the ANSI specification for COERCE, (COERCE 3 'COMPLEX)
728 gives a result which isn't COMPLEX. The result type optimizer
729 for COERCE doesn't know this, perhaps because it was written before
730 ANSI threw this curveball: the optimizer thinks that COERCE always
731 returns a result of the specified type. Thus while the interpreted
733 (DEFUN TRICKY (X) (TYPEP (COERCE X 'COMPLEX) 'COMPLEX))
734 returns the correct result,
736 the compiled function
742 As reported by Martin Atzmueller on sbcl-devel 26 Dec 2000,
743 ANSI says that WITH-OUTPUT-TO-STRING should have a keyword
744 :ELEMENT-TYPE, but in sbcl-0.6.9 this is not defined for
745 WITH-OUTPUT-TO-STRING.
748 ANSI says in one place that type declarations can be abbreviated even
749 when the type name is not a symbol, e.g.
750 (DECLAIM ((VECTOR T) *FOOVECTOR*))
751 SBCL doesn't support this. But ANSI says in another place that this
752 isn't allowed. So it's not clear this is a bug after all. (See the
753 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
757 as pointed out by Dan Barlow on sbcl-devel 2000-07-02:
758 The PICK-TEMPORARY-FILE-NAME utility used by LOAD-FOREIGN uses
759 an easily guessable temporary filename in a way which might open
760 applications using LOAD-FOREIGN to hijacking by malicious users
761 on the same machine. Incantations for doing this safely are
762 floating around the net in various "how to write secure programs
763 despite Unix" documents, and it would be good to (1) fix this in
764 LOAD-FOREIGN, and (2) hunt for any other code which uses temporary
765 files and make it share the same new safe logic.
768 (fixed early Feb 2001 by MNA)
771 As reported by wbuss@TELDA.NET (Wolfhard Buss) on cmucl-help
774 (loop with (a . b) of-type float = '(0.0 . 1.0)
775 and (c . d) of-type float = '(2.0 . 3.0)
776 return (list a b c d))
777 should evaluate to (0.0 1.0 2.0 3.0). cmucl-18c disagrees and
778 invokes the debugger: "B is not of type list".
779 SBCL does the same thing.
782 Functions are assigned names based on the context in which they're
783 defined. This is less than ideal for the functions which are
784 used to implement CLOS methods. E.g. the output of
785 (DESCRIBE 'PRINT-OBJECT) lists functions like
786 #<FUNCTION "DEF!STRUCT (TRACE-INFO (:MAKE-LOAD-FORM-FUN SB-KERNEL:JUST-DUMP-IT-NORMALLY) (:PRINT-OBJECT #))" {1020E49}>
788 #<FUNCTION "MACROLET ((FORCE-DELAYED-DEF!METHODS NIL #))" {1242871}>
789 It would be better if these functions' names always identified
790 them as methods, and identified their generic functions and
794 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
795 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
796 the mean of the distribution can be systematically O(0.1%) wrong.
797 Just increasing R-I-E-B is probably not a good solution, since
798 it would decrease efficiency more than is probably necessary. Perhaps
799 using some sort of accept/reject method would be better.
802 (SUBTYPEP '(SATISFIES SOME-UNDEFINED-FUN) NIL)=>NIL,T (should be NIL,NIL)
805 Internally the compiler sometimes evaluates
806 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
807 (I stumbled across this when I added an
808 (assert (not (eq type1 *wild-type*)))
809 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
810 in a type context should probably be translated to T, and so it's
811 probably to ask whether it's equal to the T type and then (using the
812 EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
813 (I haven't tried to investigate this bug enough to guess whether
814 there might be any user-level symptoms.)
817 a latent cross-compilation/bootstrapping bug: The cross-compilation
818 host's CL:CHAR-CODE-LIMIT is used in target code in readtable.lisp
819 and possibly elsewhere. Instead, we should use the target system's
820 CHAR-CODE-LIMIT. This will probably cause problems if we try to
821 bootstrap on a system which uses a different value of CHAR-CODE-LIMIT
825 (subtypep '(or (integer -1 1)
829 (integer -1 1))) => NIL,T
830 An analogous problem with SINGLE-FLOAT and REAL types was fixed in
831 sbcl-0.6.11.22, but some peculiarites of the RATIO type make it
832 awkward to generalize the fix to INTEGER and RATIONAL. It's not
833 clear what's the best fix. (See the "bug in type handling" discussion
834 on cmucl-imp ca. 2001-03-22 and ca. 2001-02-12.)
837 In sbcl-0.6.11.26, (COMPILE 'IN-HOST-COMPILATION-MODE) in
838 src/cold/shared.lisp doesn't correctly translate the
840 (defun in-host-compilation-mode (fn)
841 (let ((*features* (cons :sb-xc-host *features*))
842 ;; the CROSS-FLOAT-INFINITY-KLUDGE, as documented in
843 ;; base-target-features.lisp-expr:
844 (*shebang-features* (set-difference *shebang-features*
845 '(:sb-propagate-float-type
846 :sb-propagate-fun-type))))
847 (with-additional-nickname ("SB-XC" "SB!XC")
849 No error is reported by the compiler, but when the function is executed,
851 TYPE-ERROR in SB-KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER:
852 (:LINUX :X86 :IEEE-FLOATING-POINT :SB-CONSTRAIN-FLOAT-TYPE :SB-TEST
853 :SB-INTERPRETER :SB-DOC :UNIX ...) is not of type SYMBOL.
856 Inconsistencies between derived and declared VALUES return types for
857 DEFUN aren't checked very well. E.g. the logic which successfully
858 catches problems like
859 (declaim (ftype (function (fixnum) float) foo))
861 (declare (type integer x))
862 (values x)) ; wrong return type, detected, gives warning, good!
864 (declaim (ftype (function (t) (values t t)) bar))
866 (values x)) ; wrong number of return values, no warning, bad!
867 The cause of this is seems to be that (1) the internal function
868 VALUES-TYPES-EQUAL-OR-INTERSECT used to make the check handles its
869 arguments symmetrically, and (2) when the type checking code was
870 written back when when SBCL's code was still CMU CL, the intent
872 (declaim (ftype (function (t) t) bar))
874 (values x x)) ; wrong number of return values; should give warning?
875 not be warned for, because a two-valued return value is considered
876 to be compatible with callers who expects a single value to be
877 returned. That intent is probably not appropriate for modern ANSI
878 Common Lisp, but fixing this might be complicated because of other
879 divergences between auld-style and new-style handling of
880 multiple-VALUES types. (Some issues related to this were discussed
881 on cmucl-imp at some length sometime in 2000.)
884 The facility for dumping a running Lisp image to disk gets confused
885 when run without the PURIFY option, and creates an unnecessarily large
886 core file (apparently representing memory usage up to the previous
887 high-water mark). Moreover, when the file is loaded, it confuses the
888 GC, so that thereafter memory usage can never be reduced below that
892 The TRACE facility can't be used on some kinds of functions.
893 Basically, the breakpoint facility wasn incompletely implemented
894 in the X86 port of CMU CL, and we haven't fixed it in SBCL.
896 KNOWN BUGS RELATED TO THE IR1 INTERPRETER
898 (Note: At some point, the pure interpreter (actually a semi-pure
899 interpreter aka "the IR1 interpreter") will probably go away, replaced
901 (DEFUN EVAL (X) (FUNCALL (COMPILE NIL (LAMBDA ..)))))
902 and at that time these bugs should either go away automatically or
903 become more tractable to fix. Until then, they'll probably remain,
904 since some of them aren't considered urgent, and the rest are too hard
905 to fix as long as so many special cases remain. After the IR1
906 interpreter goes away is also the preferred time to start
907 systematically exterminating cases where debugging functionality
908 (backtrace, breakpoint, etc.) breaks down, since getting rid of the
909 IR1 interpreter will reduce the number of special cases we need to
913 The FUNCTION special operator doesn't check properly whether its
914 argument is a function name. E.g. (FUNCTION (X Y)) returns a value
915 instead of failing with an error. (Later attempting to funcall the
916 value does cause an error.)
919 COMPILED-FUNCTION-P bogusly reports T for interpreted functions:
920 * (DEFUN FOO (X) (- 12 X))
922 * (COMPILED-FUNCTION-P #'FOO)
927 (DEFVAR *SUPPRESS-P* T)
928 (EVAL '(UNLESS *SUPPRESS-P*
929 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
930 (FORMAT T "surprise!"))))
931 prints "surprise!". Probably the entire EVAL-WHEN mechanism ought to be
932 rewritten from scratch to conform to the ANSI definition, abandoning
933 the *ALREADY-EVALED-THIS* hack which is used in sbcl-0.6.8.9 (and
934 in the original CMU CL source, too). This should be easier to do --
935 though still nontrivial -- once the various IR1 interpreter special
939 EVAL-WHEN's idea of what's a toplevel form is even more screwed up
940 than the example in IR1-3 would suggest, since COMPILE-FILE and
941 COMPILE both print both "right now!" messages when compiling the
945 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
946 (PRINT "yes! right now!"))
949 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
950 (PRINT "no! right now!"))
952 and while EVAL doesn't print the "right now!" messages, the first
953 FUNCALL on the value returned by EVAL causes both of them to be printed.
956 The system accepts DECLAIM in most places where DECLARE would be
957 accepted, without even issuing a warning. ANSI allows this, but since
958 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
959 meaning is rather different, and it's unlikely that the user
960 has a good reason for doing DECLAIM not at top level, it would be
961 good to issue a STYLE-WARNING when this happens. A possible
962 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
963 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
964 [This is considered an IR1-interpreter-related bug because until
965 EVAL-WHEN is rewritten, which won't happen until after the IR1
966 interpreter is gone, the system's notion of what's a top-level form
967 and what's not will remain too confused to fix this problem.]