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 It should cause a STYLE-WARNING, not a full WARNING, when a structure
47 slot default value does not match the declared structure slot type.
48 (The current behavior is consistent with SBCL's behavior elsewhere,
49 and would not be a problem, except that the other behavior is
50 specifically required by the ANSI spec.)
53 It should cause a note, not a WARNING, when the system ignores
54 an FTYPE proclamation for a slot accessor.
57 Error reporting on various stream-requiring operations is not
58 very good when the stream argument has the wrong type, because
59 the operation tries to fall through to Gray stream code, and then
60 dies because it's undefined. E.g.
61 (PRINT-UNREADABLE-OBJECT (*STANDARD-OUTPUT* 1)) ..)
62 gives the error message
63 error in SB-KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER:
64 The function SB-IMPL::STREAM-WRITE-STRING is undefined.
65 It would be more useful and correct to signal a TYPE-ERROR:
67 (It wouldn't be terribly difficult to write stubs for all the
68 Gray stream functions that the old CMU CL code expects, with
69 each stub just raising the appropriate TYPE-ERROR.)
72 bogus warnings about undefined functions for magic functions like
73 SB!C::%%DEFUN and SB!C::%DEFCONSTANT when cross-compiling files
74 like src/code/float.lisp
77 The "byte compiling top-level form:" output ought to be condensed.
78 Perhaps any number of such consecutive lines ought to turn into a
79 single "byte compiling top-level forms:" line.
82 Compiling a file containing the erroneous program
86 (DEFSTRUCT (BAR (:INCLUDE FOO))
89 gives only the not-very-useful message
91 (during macroexpansion)
92 Condition PROGRAM-ERROR was signalled.
93 (The specific message which says that the problem was duplicate
94 slot names gets lost.)
97 The handling of IGNORE declarations on lambda list arguments of
98 DEFMETHOD is at least weird, and in fact seems broken and useless.
99 I should fix up another layer of binding, declared IGNORABLE, for
100 typed lambda list arguments.
103 The way that the compiler munges types with arguments together
104 with types with no arguments (in e.g. TYPE-EXPAND) leads to
105 weirdness visible to the user:
106 (DEFTYPE FOO () 'FIXNUM)
108 (TYPEP 11 '(FOO)) => T, which seems weird
109 (TYPEP 11 'FIXNUM) => T
110 (TYPEP 11 '(FIXNUM)) signals an error, as it should
111 The situation is complicated by the presence of Common Lisp types
112 like UNSIGNED-BYTE (which can either be used in list form or alone)
113 so I'm not 100% sure that the behavior above is actually illegal.
114 But I'm 90+% sure, and someday perhaps I'll be motivated to look it up..
117 It would be nice if the
119 (during macroexpansion)
120 said what macroexpansion was at fault, e.g.
122 (during macroexpansion of IN-PACKAGE,
123 during macroexpansion of DEFFOO)
126 The type system doesn't understand the KEYWORD type very well:
127 (SUBTYPEP 'KEYWORD 'SYMBOL) => NIL, NIL
128 It might be possible to fix this by changing the definition of
129 KEYWORD to (AND SYMBOL (SATISFIES KEYWORDP)), but the type system
130 would need to be a bit smarter about AND types, too:
131 (SUBTYPEP '(AND SYMBOL KEYWORD) 'SYMBOL) => NIL, NIL
132 (The type system does know something about AND types already,
133 (SUBTYPEP '(AND INTEGER FLOAT) 'NUMBER) => T, T
134 (SUBTYPEP '(AND INTEGER FIXNUM) 'NUMBER) =>T, T
135 so likely this is a small patch.)
138 Floating point infinities are screwed up. [When I was converting CMU CL
139 to SBCL, I was looking for complexity to delete, and I thought it was safe
140 to just delete support for floating point infinities. It wasn't: they're
141 generated by the floating point hardware even when we remove support
142 for them in software. -- WHN] Support for them should be restored.
145 The ANSI syntax for non-STANDARD method combination types in CLOS is
146 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
147 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
148 If you mess this up, omitting the PROGN qualifier in in DEFMETHOD,
149 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
150 (DEFMETHOD FOO ((X BAR)) (PRINT 'NUMBER))
151 the error mesage is not easy to understand:
152 INVALID-METHOD-ERROR was called outside the dynamic scope
153 of a method combination function (inside the body of
154 DEFINE-METHOD-COMBINATION or a method on the generic
155 function COMPUTE-EFFECTIVE-METHOD).
156 It would be better if it were more informative, a la
157 The method combination type for this method (STANDARD) does
158 not match the method combination type for the generic function
160 Also, after you make the mistake of omitting the PROGN qualifier
161 on a DEFMETHOD, doing a new DEFMETHOD with the correct qualifier
163 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
165 INVALID-METHOD-ERROR was called outside the dynamic scope
166 of a method combination function (inside the body of
167 DEFINE-METHOD-COMBINATION or a method on the generic
168 function COMPUTE-EFFECTIVE-METHOD).
169 This is not very helpful..
172 (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
173 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
174 (Also, when this is fixed, we can enable the code in PROCLAIM which
175 checks for incompatible FTYPE redeclarations.)
178 The ANSI spec says that CONS can be a compound type spec, e.g.
179 (CONS FIXNUM REAL). SBCL doesn't support this.
182 from Paolo Amoroso on the CMU CL mailing list 27 Feb 2000:
183 I use CMU CL 18b under Linux. When COMPILE-FILE is supplied a physical
184 pathname, the type of the corresponding compiled file is X86F:
185 * (compile-file "/home/paolo/lisp/tools/foo")
186 Python version 1.0, VM version Intel x86 on 27 FEB 0 06:00:46 pm.
187 Compiling: /home/paolo/lisp/tools/foo.lisp 27 FEB 0 05:57:42 pm
189 Compiling DEFUN SQUARE:
190 Byte Compiling Top-Level Form:
191 /home/paolo/lisp/tools/foo.x86f written.
192 Compilation finished in 0:00:00.
193 #p"/home/paolo/lisp/tools/foo.x86f"
196 But when the function is called with a logical pathname, the file type
198 * (compile-file "tools:foo")
199 Python version 1.0, VM version Intel x86 on 27 FEB 0 06:01:04 pm.
200 Compiling: /home/paolo/lisp/tools/foo.lisp 27 FEB 0 05:57:42 pm
202 Compiling DEFUN SQUARE:
203 Byte Compiling Top-Level Form:
204 TOOLS:FOO.FASL written.
205 Compilation finished in 0:00:00.
206 #p"/home/paolo/lisp/tools/foo.fasl"
211 from DTC on the CMU CL mailing list 25 Feb 2000:
212 ;;; Compiler fails when this file is compiled.
214 ;;; Problem shows up in delete-block within ir1util.lisp. The assertion
215 ;;; (assert (member (functional-kind lambda) '(:let :mv-let :assignment)))
216 ;;; fails within bind node branch.
218 ;;; Note that if c::*check-consistency* is enabled then an un-reached
219 ;;; entry is also reported.
222 (declare (values nil))
239 (let ((ttt #'(lambda () (go cccc))))
240 (declare (special ttt))
241 (return-from bbbb nil))
244 (return-from bbbb nil))))))
247 (I *think* this is a bug. It certainly seems like strange behavior. But
248 the ANSI spec is scary, dark, and deep..)
249 (FORMAT NIL "~,1G" 1.4) => "1. "
250 (FORMAT NIL "~3,1G" 1.4) => "1. "
253 from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
255 (setf (find-class 'ccc1) (find-class 'ccc))
256 (defmethod zut ((c ccc1)) 123)
257 DTC's recommended workaround from the mailing list 3 Mar 2000:
258 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
261 The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
262 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
263 inside "~<..~>" (without the colon modifier on the closing syntax).
264 However, SBCL doesn't do this:
265 * (FORMAT T "~<munge~wegnum~>" 12)
270 When too many files are opened, OPEN will fail with an
271 uninformative error message
272 error in function OPEN: error opening #P"/tmp/foo.lisp": NIL
273 instead of saying that too many files are open.
276 Right now, when COMPILE-FILE has a read error, it actually pops
277 you into the debugger before giving up on the file. It should
278 instead handle the error, perhaps issuing (and handling)
279 a secondary error "caught ERROR: unrecoverable error during compilation"
280 and then return with FAILURE-P true,
283 from CMU CL mailing list 01 May 2000
285 I realize I can take care of this by doing (proclaim (ignore pcl::.slots1.))
286 but seeing as .slots0. is not-exported, shouldn't it be ignored within the
290 In: DEFMETHOD FOO-BAR-BAZ (RESOURCE-TYPE)
291 (DEFMETHOD FOO-BAR-BAZ
292 ((SELF RESOURCE-TYPE))
293 (SETF (SLOT-VALUE SELF 'NAME) 3))
294 --> BLOCK MACROLET PCL::FAST-LEXICAL-METHOD-FUNCTIONS
295 --> PCL::BIND-FAST-LEXICAL-METHOD-MACROS MACROLET
296 --> PCL::BIND-LEXICAL-METHOD-FUNCTIONS LET PCL::BIND-ARGS LET* PCL::PV-BINDING
297 --> PCL::PV-BINDING1 PCL::PV-ENV LET
299 (LET ((PCL::.SLOTS0. #))
304 Warning: Variable PCL::.SLOTS0. defined but never used.
306 Compilation unit finished.
309 #<Standard-Method FOO-BAR-BAZ (RESOURCE-TYPE) {480918FD}>
312 reported by Sam Steingold on the cmucl-imp mailing list 12 May 2000:
314 Also, there is another bug: `array-displacement' should return an array
315 or nil as first value (as per ANSI CL), while CMUCL declares it as
316 returning an array as first value always.
319 Sometimes (SB-EXT:QUIT) fails with
320 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
321 Process inferior-lisp exited abnormally with code 1
322 I haven't noticed a repeatable case of this yet.
325 some sort of bug in inlining and RETURN-FROM in sbcl-0.6.5: Compiling
328 (BLOCK USED-BY-SOME-Y?
331 (UNLESS (REJECTED? Y)
332 (RETURN-FROM USED-BY-SOME-Y? T)))))
333 (DECLARE (INLINE FROB))
338 error in function SB-KERNEL:ASSERT-ERROR:
339 The assertion (EQ (SB-C::CONTINUATION-KIND SB-C::CONT) :BLOCK-START) failed.
340 This is still present in sbcl-0.6.8.
343 The CMU CL reader code takes liberties in binding the standard read table
344 when reading the names of characters. Tim Moore posted a patch to the
345 CMU CL mailing list Mon, 22 May 2000 21:30:41 -0700.
348 In some cases the compiler believes type declarations on array
349 elements without checking them, e.g.
350 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
353 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
354 (WHEN (CONSP (AREF X 0))
356 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
359 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
360 all cases, e.g. the type assumption *is* checked if the array
361 elements are declared to be of some structure type instead of CONS.
364 The printer doesn't report closures very well. This is true in
368 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
369 It would be nice to make closures have a settable name slot,
370 and make things like DEFSTRUCT and FLET, which create closures,
371 set helpful values into this slot.
374 And as long as we're wishing, it would be awfully nice if INSPECT could
375 also report on closures, telling about the values of the bound variables.
378 as reported by Robert Strandh on the CMU CL mailing list 12 Jun 2000:
380 (defconstant +a-constant+ (make-instance 'a-class))
381 (defconstant +another-constant+ (vector +a-constant+))
383 CMU Common Lisp release x86-linux 2.4.19 8 February 2000 build 456,
386 Send bug reports and questions to your local CMU CL maintainer,
387 or to pvaneynd@debian.org
388 or to cmucl-help@cons.org. (prefered)
389 type (help) for help, (quit) to exit, and (demo) to see the demos
391 Python 1.0, target Intel x86
392 CLOS based on PCL version: September 16 92 PCL (f)
393 * (defclass a-class () ())
394 #<STANDARD-CLASS A-CLASS {48027BD5}>
395 * (compile-file "xx.lisp")
396 Python version 1.0, VM version Intel x86 on 12 JUN 00 08:12:55 am.
398 /home/strandh/Research/Functional/Common-Lisp/CLIM/Development/McCLIM
399 /xx.lisp 12 JUN 00 07:47:14 am
400 Compiling Load Time Value of (PCL::GET-MAKE-INSTANCE-FUNCTION-SYMBOL
402 Byte Compiling Top-Level Form:
403 Error in function C::DUMP-STRUCTURE: Attempt to dump invalid
405 #<A-CLASS {4803A5B5}>
409 The compiler assumes that any time a function of declared FTYPE
410 doesn't signal an error, its arguments were of the declared type.
411 E.g. compiling and loading
412 (DECLAIM (OPTIMIZE (SAFETY 3)))
413 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
414 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE) FACTORIAL)))
416 (COND ((> (FACTORIAL X) 1.0E6)
417 (FORMAT T "too big~%"))
419 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
421 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
424 will cause the INTEGERP case to be selected, giving bogus output a la
426 This violates the "declarations are assertions" principle.
427 According to the ANSI spec, in the section "System Class FUNCTION",
428 this is a case of "lying to the compiler", but the lying is done
429 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
430 not by the unexpectedly general definition of FACTORIAL. In any case,
431 "declarations are assertions" means that lying to the compiler should
432 cause an error to be signalled, and should not cause a bogus
433 result to be returned. Thus, the compiler should not assume
434 that arbitrary functions check their argument types. (It might
435 make sense to add another flag (CHECKED?) to DEFKNOWN to
436 identify functions which *do* check their argument types.)
439 As pointed out by Martin Cracauer on the CMU CL mailing list
440 13 Jun 2000, the :FILE-LENGTH operation for
441 FD-STREAM-MISC-ROUTINE is broken for large files: it says
442 (THE INDEX SIZE) even though SIZE can be larger than INDEX.
445 In SBCL 0.6.5 (and CMU CL 18b) compiling and loading
446 (in-package :cl-user)
447 (declaim (optimize (safety 3)
449 (compilation-speed 2)
452 #+nil (sb-ext:inhibit-warnings 2)))
453 (declaim (ftype (function * (values)) emptyvalues))
454 (defun emptyvalues (&rest rest) (declare (ignore rest)) (values))
456 (defgeneric assertoid ((x t)))
457 (defmethod assertoid ((x t)) "just a placeholder")
459 (declare (type hash-table ht))
465 (assertoid (hash-table-count ht)))))))
466 (unless (typep res 'foo)
468 (common-lisp-user::bad-result-from-assertive-typed-fun
472 (bar (make-hash-table))
474 Error in KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER:
475 the function C::%INSTANCE-TYPEP is undefined.
476 %INSTANCE-TYPEP is always supposed to be IR1-transformed away, but for
477 some reason -- the (VALUES) return value declaration? -- the optimizer is
478 confused and compiles a full call to %INSTANCE-TYPEP (which doesn't exist
479 as a function) instead.
482 The %INSTANCE-TYPEP problem in bug 37 comes up also when compiling
484 (IN-PACKAGE :CL-USER)
486 (DECLARE (OPTIMIZE (SAFETY 3) (SPEED 2) (SPACE 2)))
487 (DECLAIM (FTYPE (FUNCTION (&REST T) (VALUES)) EMPTYVALUES))
488 (DEFUN EMPTYVALUES (&REST REST)
489 (DECLARE (IGNORE REST))
491 (DEFSTRUCT DUMMYSTRUCT X Y)
492 (DEFUN FROB-EMPTYVALUES (X)
493 (LET ((RES (EMPTYVALUES X X X)))
494 (UNLESS (TYPEP RES 'DUMMYSTRUCT)
495 'EXPECTED-RETURN-VALUE))))
496 (ASSERT (EQ (FROB-EMPTYVALUES 11) 'EXPECTED-RETURN-VALUE))
500 DEFMETHOD doesn't check the syntax of &REST argument lists properly,
501 accepting &REST even when it's not followed by an argument name:
502 (DEFMETHOD FOO ((X T) &REST) NIL)
505 On the CMU CL mailing list 26 June 2000, Douglas Crosher wrote
507 Hannu Rummukainen wrote:
509 > There's something weird going on with the compilation of the attached
510 > code. Compiling and loading the file in a fresh lisp, then invoking
512 Thanks for the bug report, nice to have this one fixed. It was a bug
513 in the x86 backend, the < VOP. A fix has been committed to the main
514 source, see the file compiler/x86/float.lisp.
516 Probably the same bug exists in SBCL.
519 TYPEP treats the result of UPGRADED-ARRAY-ELEMENT-TYPE as gospel,
520 so that (TYPEP (MAKE-ARRAY 3) '(VECTOR SOMETHING-NOT-DEFINED-YET))
521 returns (VALUES T T). Probably it should be an error instead,
522 complaining that the type SOMETHING-NOT-DEFINED-YET is not defined.
525 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
526 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
528 (DECLARE (TYPE INDEXOID X))
529 (THE (VALUES INDEXOID)
531 where the implementation of the type check in function FOO
532 includes a full call to %TYPEP. There are also some fundamental problems
533 with the interpretation of VALUES types (inherited from CMU CL, and
534 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
535 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
538 The definitions of SIGCONTEXT-FLOAT-REGISTER and
539 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
540 supported on FreeBSD because the floating point state is not saved,
541 but at least as of FreeBSD 4.0, the floating point state *is* saved,
542 so they could be supported after all. Very likely
543 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
546 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
547 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
548 issue with 'union, 'and, 'or etc. So even though according to the
549 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
550 (and now SBCL) interpret them as legal types.
553 ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
554 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
555 list, and it is probably possible to use substantially the same
556 patches to add it to SBCL.
559 a slew of floating-point-related errors reported by Peter Van Eynde
561 a: (SQRT -9.0) fails, because SB-KERNEL::COMPLEX-SQRT is undefined.
562 Similarly, COMPLEX-ASIN, COMPLEX-ACOS, COMPLEX-ACOSH, and others
564 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
565 should probably be 1.4012985e-45. In SBCL,
566 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
567 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
568 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
569 and LEAST-NEGATIVE-LONG-FLOAT.
570 c: Many expressions generate floating infinity:
575 PVE's regression tests want them to raise errors. SBCL
576 generates the infinities instead, which may or may not be
577 conforming behavior, but then blow it by being unable to
578 output the infinities, since support for infinities is generally
579 broken, and in particular SB-IMPL::OUTPUT-FLOAT-INFINITY is
581 d: (in section12.erg) various forms a la
582 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
583 don't give the right behavior.
586 type safety errors reported by Peter Van Eynde July 25, 2000:
587 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
589 In general lengths of array type specifications aren't
590 checked by COERCE, so it fails when the spec is
591 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
592 b: CONCATENATE has the same problem of not checking the length
593 of specified output array types. MAKE-SEQUENCE and MAP and
594 MERGE also have the same problem.
595 c: (COERCE 'AND 'FUNCTION) returns something related to
596 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
597 d: ELT signals SIMPLE-ERROR if its index argument
598 isn't a valid index for its sequence argument, but should
599 signal TYPE-ERROR instead.
600 e: FILE-LENGTH is supposed to signal a type error when its
601 argument is not a stream associated with a file, but doesn't.
602 f: (FLOAT-RADIX 2/3) should signal an error instead of
604 g: (LOAD "*.lsp") should signal FILE-ERROR.
605 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
606 should signal TYPE-ERROR.
607 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
608 be used for input and output as needed. It should fail with
609 TYPE-ERROR when handed e.g. the results of
610 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
611 the inappropriate positions, but doesn't.
612 j: (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
614 should probably signal an error instead of making a pathname with
616 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
617 not a binary input stream, but instead cheerfully reads from
618 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
621 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
622 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
624 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
625 signal a PROGRAM-ERROR, and doesn't.
626 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
627 and other DEFCLASS forms with duplicate specifications in their
628 slots, should signal a PROGRAM-ERROR, and doesn't.
629 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
630 causes a COMPILER-ERROR.
633 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
634 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
635 PROGRAM-ERROR, but SBCL accepts it instead.
636 b: SYMBOL-MACROLET should refuse to bind something which is
637 declared as a global variable, signalling PROGRAM-ERROR.
638 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
639 it binds is declared SPECIAL inside.
642 LOOP bugs reported by Peter Van Eynde July 25, 2000:
643 a: (LOOP WITH (A B) DO (PRINT 1)) is a syntax error according to
644 the definition of WITH clauses given in the ANSI spec, but
645 compiles and runs happily in SBCL.
646 b: a messy one involving package iteration:
647 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<))))
648 Should be: (("blah" "blah2") ("blah2"))
649 SBCL: (("blah") ("blah2"))
650 * (LET ((X 1)) (LOOP FOR I BY (INCF X) FROM X TO 10 COLLECT I))
651 doesn't work -- SBCL's LOOP says BY isn't allowed in a FOR clause.
654 type system errors reported by Peter Van Eynde July 25, 2000:
655 a: (SUBTYPEP 'BIGNUM 'INTEGER) => NIL, NIL
656 but should be (VALUES T T) instead.
657 b: (SUBTYPEP 'EXTENDED-CHAR 'CHARACTER) => NIL, NIL
658 but should be (VALUES T T) instead.
659 c: (SUBTYPEP '(INTEGER (0) (0)) 'NIL) dies with nested errors.
660 d: In general, the system doesn't like '(INTEGER (0) (0)) -- it
661 blows up at the level of SPECIFIER-TYPE with
662 "Lower bound (0) is greater than upper bound (0)." Probably
663 SPECIFIER-TYPE should return NIL instead.
664 e: (TYPEP 0 '(COMPLEX (EQL 0)) fails with
665 "Component type for Complex is not numeric: (EQL 0)."
666 This might be easy to fix; the type system already knows
667 that (SUBTYPEP '(EQL 0) 'NUMBER) is true.
668 f: The type system doesn't know about the condition system,
669 so that e.g. (TYPEP 'SIMPLE-ERROR 'ERROR)=>NIL.
670 g: The type system isn't all that smart about relationships
671 between hairy types, as shown in the type.erg test results,
672 e.g. (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL.
675 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
677 (DEFGENERIC FOO02 (X))
678 (DEFMETHOD FOO02 ((X NUMBER)) T)
679 (LET ((M (FIND-METHOD (FUNCTION FOO02)
681 (LIST (FIND-CLASS (QUOTE NUMBER))))))
682 (REMOVE-METHOD (FUNCTION FOO02) M)
683 (DEFGENERIC FOO03 (X))
684 (ADD-METHOD (FUNCTION FOO03) M)))
685 should give an error, but SBCL allows it.
686 b: READ should probably return READER-ERROR, not the bare
687 arithmetic error, when input a la "1/0" or "1e1000" causes
691 It has been reported (e.g. by Peter Van Eynde) that there are
692 several metaobject protocol "errors". (In order to fix them, we might
693 need to document exactly what metaobject protocol specification
694 we're following -- the current code is just inherited from PCL.)
697 another error from Peter Van Eynde 5 September 2000:
698 (FORMAT NIL "~F" "FOO") should work, but instead reports an error.
699 PVE submitted a patch to deal with this bug, but it exposes other
700 comparably serious bugs, so I didn't apply it. It looks as though
701 the FORMAT code needs a fair amount of rewriting in order to comply
702 with the various details of the ANSI spec.
705 The implementation of #'+ returns its single argument without
706 type checking, e.g. (+ "illegal") => "illegal".
709 In sbcl-0.6.7, there is no doc string for CL:PUSH, probably
710 because it's defined with the DEFMACRO-MUNDANELY macro and something
711 is wrong with doc string setting in that macro.
714 Attempting to use COMPILE on something defined by DEFMACRO fails:
715 (DEFMACRO FOO (X) (CONS X X))
717 Error in function C::GET-LAMBDA-TO-COMPILE:
718 #<Closure Over Function "DEFUN (SETF MACRO-FUNCTION)" {480E21B1}> was defined in a non-null environment.
721 (SUBTYPEP '(AND ZILCH INTEGER) 'ZILCH)
725 CL:*DEFAULT-PATHNAME-DEFAULTS* doesn't behave as ANSI suggests (reflecting
726 current working directory). And there's no supported way to update
727 or query the current working directory (a la Unix "chdir" and "pwd"),
728 which is functionality that ILISP needs (and currently gets with low-level
732 The debugger LIST-LOCATIONS command doesn't work properly.
735 Compiling and loading
736 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
738 then requesting a BACKTRACE at the debugger prompt gives no information
739 about where in the user program the problem occurred.
742 The compiler is supposed to do type inference well enough that
745 ((SIMPLE-ARRAY SINGLE-FLOAT)
747 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
750 is redundant. However, as reported by Juan Jose Garcia Ripoll for
751 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
752 workaround for the problem for now, but can't be done by the TYPECASE
753 macros themselves, since it's too hard for the macro to detect
754 assignments to the variable within the clause.
755 Note: The compiler *is* smart enough to do the type inference in
756 many cases. This case, derived from a couple of MACROEXPAND-1
757 calls on Ripoll's original test case,
759 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
760 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
761 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
762 (LET ((I 0) (G2554 LENGTH))
763 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
766 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
767 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
768 (GO SB-LOOP::NEXT-LOOP)
769 SB-LOOP::END-LOOP))))))
770 demonstrates the problem; but the problem goes away if the TAGBODY
771 and GO forms are removed (leaving the SETF in ordinary, non-looping
772 code), or if the TAGBODY and GO forms are retained, but the
773 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
776 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
777 I am looking into this problem that showed up on the cmucl-help
778 list. It seems to me that the "implementation specific environment
779 hacking functions" found in pcl/walker.lisp are completely messed
780 up. The good thing is that they appear to be barely used within
781 PCL and the munged environment object is passed to cmucl only
782 in calls to macroexpand-1, which is probably why this case fails.
783 SBCL uses essentially the same code, so if the environment hacking
784 is screwed up, it affects us too.
787 Using the pretty-printer from the command prompt gives funny
788 results, apparently because the pretty-printer doesn't know
789 about user's command input, including the user's carriage return
790 that the user, and therefore the pretty-printer thinks that
791 the new output block should start indented 2 or more characters
792 rightward of the correct location.
795 As reported by Carl Witty on submit@bugs.debian.org 1999-05-08,
797 (in-package "CL-USER")
798 (defun equal-terms (termx termy)
800 ((alpha-equal-bound-term-lists (listx listy)
801 (or (and (null listx) (null listy))
803 (let ((bindings-x (bindings-of-bound-term (car listx)))
804 (bindings-y (bindings-of-bound-term (car listy))))
805 (if (and (null bindings-x) (null bindings-y))
806 (alpha-equal-terms (term-of-bound-term (car listx))
807 (term-of-bound-term (car listy)))
808 (and (= (length bindings-x) (length bindings-y))
810 (enter-binding-pairs (bindings-of-bound-term (car listx))
811 (bindings-of-bound-term (car listy)))
812 (alpha-equal-terms (term-of-bound-term (car listx))
813 (term-of-bound-term (car listy)))
814 (exit-binding-pairs (bindings-of-bound-term (car listx))
815 (bindings-of-bound-term (car listy)))))))
816 (alpha-equal-bound-term-lists (cdr listx) (cdr listy)))))
818 (alpha-equal-terms (termx termy)
819 (if (and (variable-p termx)
821 (equal-bindings (id-of-variable-term termx)
822 (id-of-variable-term termy))
823 (and (equal-operators-p (operator-of-term termx) (operator-of-term termy))
824 (alpha-equal-bound-term-lists (bound-terms-of-term termx)
825 (bound-terms-of-term termy))))))
829 (with-variable-invocation (alpha-equal-terms termx termy))))))
830 causes an assertion failure
831 The assertion (EQ (C::LAMBDA-TAIL-SET C::CALLER)
832 (C::LAMBDA-TAIL-SET (C::LAMBDA-HOME C::CALLEE))) failed.
834 Bob Rogers reports (1999-07-28 on cmucl-imp@cons.org) a smaller test
835 case with the same problem:
836 (defun parse-fssp-alignment ()
837 ;; Given an FSSP alignment file named by the argument . . .
838 (labels ((get-fssp-char ()
842 ;; Stub body, enough to tickle the bug.
843 (list (read-fssp-char)
847 ANSI specifies that the RESULT-TYPE argument of CONCATENATE must be
848 a subtype of SEQUENCE, but CONCATENATE doesn't check this properly:
849 (CONCATENATE 'SIMPLE-ARRAY #(1 2) '(3)) => #(1 2 3)
850 This also leads to funny behavior when derived type specifiers
851 are used, as originally reported by Milan Zamazal for CMU CL (on the
852 Debian bugs mailing list (?) 2000-02-27), then reported by Martin
853 Atzmueller for SBCL (2000-10-01 on sbcl-devel@lists.sourceforge.net):
854 (DEFTYPE FOO () 'SIMPLE-ARRAY)
855 (CONCATENATE 'FOO #(1 2) '(3))
856 => #<ARRAY-TYPE SIMPLE-ARRAY> is a bad type specifier for
858 The derived type specifier FOO should act the same way as the
859 built-in type SIMPLE-ARRAY here, but it doesn't. That problem
860 doesn't seem to exist for sequence types:
861 (DEFTYPE BAR () 'SIMPLE-VECTOR)
862 (CONCATENATE 'BAR #(1 2) '(3)) => #(1 2 3)
865 KNOWN BUGS RELATED TO THE IR1 INTERPRETER
867 (Note: At some point, the pure interpreter (actually a semi-pure
868 interpreter aka "the IR1 interpreter") will probably go away, replaced
870 (DEFUN EVAL (X) (FUNCALL (COMPILE NIL (LAMBDA ..)))))
871 and at that time these bugs should either go away automatically or
872 become more tractable to fix. Until then, they'll probably remain,
873 since some of them aren't considered urgent, and the rest are too hard
874 to fix as long as so many special cases remain. After the IR1
875 interpreter goes away is also the preferred time to start
876 systematically exterminating cases where debugging functionality
877 (backtrace, breakpoint, etc.) breaks down, since getting rid of the
878 IR1 interpreter will reduce the number of special cases we need to
882 The FUNCTION special operator doesn't check properly whether its
883 argument is a function name. E.g. (FUNCTION (X Y)) returns a value
884 instead of failing with an error. (Later attempting to funcall the
885 value does cause an error.)
888 COMPILED-FUNCTION-P bogusly reports T for interpreted functions:
889 * (DEFUN FOO (X) (- 12 X))
891 * (COMPILED-FUNCTION-P #'FOO)
896 (DEFVAR *SUPPRESS-P* T)
897 (EVAL '(UNLESS *SUPPRESS-P*
898 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
899 (FORMAT T "surprise!"))))
900 prints "surprise!". Probably the entire EVAL-WHEN mechanism ought to be
901 rewritten from scratch to conform to the ANSI definition, abandoning
902 the *ALREADY-EVALED-THIS* hack which is used in sbcl-0.6.8.9 (and
903 in the original CMU CL source, too). This should be easier to do --
904 though still nontrivial -- once the various IR1 interpreter special
908 EVAL-WHEN's idea of what's a toplevel form is even more screwed up
909 than the example in IR1-3 would suggest, since COMPILE-FILE and
910 COMPILE both print both "right now!" messages when compiling the
914 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
915 (PRINT "yes! right now!"))
918 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
919 (PRINT "no! right now!"))
921 and while EVAL doesn't print the "right now!" messages, the first
922 FUNCALL on the value returned by EVAL causes both of them to be printed.
925 The system accepts DECLAIM in most places where DECLARE would be
926 accepted, without even issuing a warning. ANSI allows this, but since
927 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
928 meaning is rather different, and it's unlikely that the user
929 has a good reason for doing DECLAIM not at top level, it would be
930 good to issue a STYLE-WARNING when this happens. A possible
931 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
932 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
933 [This is considered an IR1-interpreter-related bug because until
934 EVAL-WHEN is rewritten, which won't happen until after the IR1
935 interpreter is gone, the system's notion of what's a top-level form
936 and what's not will remain too confused to fix this problem.]