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 bugs which have been
34 KNOWN BUGS OF NO SPECIAL CLASS:
37 DEFSTRUCT should almost certainly overwrite the old LAYOUT information
38 instead of just punting when a contradictory structure definition
42 It should cause a STYLE-WARNING, not a full WARNING, when a structure
43 slot default value does not match the declared structure slot type.
44 (The current behavior is consistent with SBCL's behavior elsewhere,
45 and would not be a problem, except that the other behavior is
46 specifically required by the ANSI spec.)
49 It should cause a note, not a WARNING, when the system ignores
50 an FTYPE proclamation for a slot accessor.
53 Error reporting on various stream-requiring operations is not
54 very good when the stream argument has the wrong type, because
55 the operation tries to fall through to Gray stream code, and then
56 dies because it's undefined. E.g.
57 (PRINT-UNREADABLE-OBJECT (*STANDARD-OUTPUT* 1))
58 gives the error message
59 error in SB-KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER:
60 The function SB-IMPL::STREAM-WRITE-STRING is undefined.
61 It would be more useful and correct to signal a TYPE-ERROR:
63 (It wouldn't be terribly difficult to write stubs for all the
64 Gray stream functions that the old CMU CL code expects, with
65 each stub just raising the appropriate TYPE-ERROR.)
68 bogus warnings about undefined functions for magic functions like
69 SB!C::%%DEFUN and SB!C::%DEFCONSTANT when cross-compiling files
70 like src/code/float.lisp
73 The "byte compiling top-level form:" output ought to be condensed.
74 Perhaps any number of such consecutive lines ought to turn into a
75 single "byte compiling top-level forms:" line.
78 Compiling a file containing the erroneous program
82 (DEFSTRUCT (BAR (:INCLUDE FOO))
85 gives only the not-very-useful message
87 (during macroexpansion)
88 Condition PROGRAM-ERROR was signalled.
89 (The specific message which says that the problem was duplicate
90 slot names gets lost.)
93 The handling of IGNORE declarations on lambda list arguments of
94 DEFMETHOD is at least weird, and in fact seems broken and useless.
95 I should fix up another layer of binding, declared IGNORABLE, for
96 typed lambda list arguments.
99 The way that the compiler munges types with arguments together
100 with types with no arguments (in e.g. TYPE-EXPAND) leads to
101 weirdness visible to the user:
102 (DEFTYPE FOO () 'FIXNUM)
104 (TYPEP 11 '(FOO)) => T, which seems weird
105 (TYPEP 11 'FIXNUM) => T
106 (TYPEP 11 '(FIXNUM)) signals an error, as it should
107 The situation is complicated by the presence of Common Lisp types
108 like UNSIGNED-BYTE (which can either be used in list form or alone)
109 so I'm not 100% sure that the behavior above is actually illegal.
110 But I'm 90+% sure, and someday perhaps I'll be motivated to look it up..
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 type system doesn't understand the KEYWORD type very well:
123 (SUBTYPEP 'KEYWORD 'SYMBOL) => NIL, NIL
124 It might be possible to fix this by changing the definition of
125 KEYWORD to (AND SYMBOL (SATISFIES KEYWORDP)), but the type system
126 would need to be a bit smarter about AND types, too:
127 (SUBTYPEP '(AND SYMBOL KEYWORD) 'SYMBOL) => NIL, NIL
128 (The type system does know something about AND types already,
129 (SUBTYPEP '(AND INTEGER FLOAT) 'NUMBER) => T, T
130 (SUBTYPEP '(AND INTEGER FIXNUM) 'NUMBER) =>T, T
131 so likely this is a small patch.)
134 Floating point infinities are screwed up. [When I was converting CMU CL
135 to SBCL, I was looking for complexity to delete, and I thought it was safe
136 to just delete support for floating point infinities. It wasn't: they're
137 generated by the floating point hardware even when we remove support
138 for them in software. -- WHN] Support for them should be restored.
141 The ANSI syntax for non-STANDARD method combination types in CLOS is
142 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
143 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
144 If you mess this up, omitting the PROGN qualifier in in DEFMETHOD,
145 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
146 (DEFMETHOD FOO ((X BAR)) (PRINT 'NUMBER))
147 the error mesage is not easy to understand:
148 INVALID-METHOD-ERROR was called outside the dynamic scope
149 of a method combination function (inside the body of
150 DEFINE-METHOD-COMBINATION or a method on the generic
151 function COMPUTE-EFFECTIVE-METHOD).
152 It would be better if it were more informative, a la
153 The method combination type for this method (STANDARD) does
154 not match the method combination type for the generic function
156 Also, after you make the mistake of omitting the PROGN qualifier
157 on a DEFMETHOD, doing a new DEFMETHOD with the correct qualifier
159 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
161 INVALID-METHOD-ERROR was called outside the dynamic scope
162 of a method combination function (inside the body of
163 DEFINE-METHOD-COMBINATION or a method on the generic
164 function COMPUTE-EFFECTIVE-METHOD).
165 This is not very helpful..
168 (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
169 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
170 (Also, when this is fixed, we can enable the code in PROCLAIM which
171 checks for incompatible FTYPE redeclarations.)
174 The ANSI spec says that CONS can be a compound type spec, e.g.
175 (CONS FIXNUM REAL). SBCL doesn't support this.
178 from Paolo Amoroso on the CMU CL mailing list 27 Feb 2000:
179 I use CMU CL 18b under Linux. When COMPILE-FILE is supplied a physical
180 pathname, the type of the corresponding compiled file is X86F:
181 * (compile-file "/home/paolo/lisp/tools/foo")
182 Python version 1.0, VM version Intel x86 on 27 FEB 0 06:00:46 pm.
183 Compiling: /home/paolo/lisp/tools/foo.lisp 27 FEB 0 05:57:42 pm
185 Compiling DEFUN SQUARE:
186 Byte Compiling Top-Level Form:
187 /home/paolo/lisp/tools/foo.x86f written.
188 Compilation finished in 0:00:00.
189 #p"/home/paolo/lisp/tools/foo.x86f"
192 But when the function is called with a logical pathname, the file type
194 * (compile-file "tools:foo")
195 Python version 1.0, VM version Intel x86 on 27 FEB 0 06:01:04 pm.
196 Compiling: /home/paolo/lisp/tools/foo.lisp 27 FEB 0 05:57:42 pm
198 Compiling DEFUN SQUARE:
199 Byte Compiling Top-Level Form:
200 TOOLS:FOO.FASL written.
201 Compilation finished in 0:00:00.
202 #p"/home/paolo/lisp/tools/foo.fasl"
207 from DTC on the CMU CL mailing list 25 Feb 2000:
208 ;;; Compiler fails when this file is compiled.
210 ;;; Problem shows up in delete-block within ir1util.lisp. The assertion
211 ;;; (assert (member (functional-kind lambda) '(:let :mv-let :assignment)))
212 ;;; fails within bind node branch.
214 ;;; Note that if c::*check-consistency* is enabled then an un-reached
215 ;;; entry is also reported.
218 (declare (values nil))
235 (let ((ttt #'(lambda () (go cccc))))
236 (declare (special ttt))
237 (return-from bbbb nil))
240 (return-from bbbb nil))))))
243 (I *think* this is a bug. It certainly seems like strange behavior. But
244 the ANSI spec is scary, dark, and deep..)
245 (FORMAT NIL "~,1G" 1.4) => "1. "
246 (FORMAT NIL "~3,1G" 1.4) => "1. "
249 from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
251 (setf (find-class 'ccc1) (find-class 'ccc))
252 (defmethod zut ((c ccc1)) 123)
253 DTC's recommended workaround from the mailing list 3 Mar 2000:
254 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
257 The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
258 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
259 inside "~<..~>" (without the colon modifier on the closing syntax).
260 However, SBCL doesn't do this:
261 * (FORMAT T "~<munge~wegnum~>" 12)
266 When too many files are opened, OPEN will fail with an
267 uninformative error message
268 error in function OPEN: error opening #P"/tmp/foo.lisp": NIL
269 instead of saying that too many files are open.
272 Right now, when COMPILE-FILE has a read error, it actually pops
273 you into the debugger before giving up on the file. It should
274 instead handle the error, perhaps issuing (and handling)
275 a secondary error "caught ERROR: unrecoverable error during compilation"
276 and then return with FAILURE-P true,
279 from CMU CL mailing list 01 May 2000
281 I realize I can take care of this by doing (proclaim (ignore pcl::.slots1.))
282 but seeing as .slots0. is not-exported, shouldn't it be ignored within the
286 In: DEFMETHOD FOO-BAR-BAZ (RESOURCE-TYPE)
287 (DEFMETHOD FOO-BAR-BAZ
288 ((SELF RESOURCE-TYPE))
289 (SETF (SLOT-VALUE SELF 'NAME) 3))
290 --> BLOCK MACROLET PCL::FAST-LEXICAL-METHOD-FUNCTIONS
291 --> PCL::BIND-FAST-LEXICAL-METHOD-MACROS MACROLET
292 --> PCL::BIND-LEXICAL-METHOD-FUNCTIONS LET PCL::BIND-ARGS LET* PCL::PV-BINDING
293 --> PCL::PV-BINDING1 PCL::PV-ENV LET
295 (LET ((PCL::.SLOTS0. #))
300 Warning: Variable PCL::.SLOTS0. defined but never used.
302 Compilation unit finished.
305 #<Standard-Method FOO-BAR-BAZ (RESOURCE-TYPE) {480918FD}>
308 reported by Sam Steingold on the cmucl-imp mailing list 12 May 2000:
310 Also, there is another bug: `array-displacement' should return an array
311 or nil as first value (as per ANSI CL), while CMUCL declares it as
312 returning an array as first value always.
315 Sometimes (SB-EXT:QUIT) fails with
316 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
317 Process inferior-lisp exited abnormally with code 1
318 I haven't noticed a repeatable case of this yet.
321 The system accepts DECLAIM in most places where DECLARE would be
322 accepted, without even issuing a warning. ANSI allows this, but since
323 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
324 meaning is rather different, and it's unlikely that the user
325 has a good reason for doing DECLAIM not at top level, it would be
326 good to issue a STYLE-WARNING when this happens. A possible
327 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
328 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
331 some sort of bug in inlining and RETURN-FROM in sbcl-0.6.5: Compiling
334 (BLOCK USED-BY-SOME-Y?
337 (UNLESS (REJECTED? Y)
338 (RETURN-FROM USED-BY-SOME-Y? T)))))
339 (DECLARE (INLINE FROB))
344 error in function SB-KERNEL:ASSERT-ERROR:
345 The assertion (EQ (SB-C::CONTINUATION-KIND SB-C::CONT) :BLOCK-START) failed.
346 This is still present in sbcl-0.6.8.
349 The CMU CL reader code takes liberties in binding the standard read table
350 when reading the names of characters. Tim Moore posted a patch to the
351 CMU CL mailing list Mon, 22 May 2000 21:30:41 -0700.
354 In some cases the compiler believes type declarations on array
355 elements without checking them, e.g.
356 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
359 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
360 (WHEN (CONSP (AREF X 0))
362 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
365 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
366 all cases, e.g. the type assumption *is* checked if the array
367 elements are declared to be of some structure type instead of CONS.
370 The printer doesn't report closures very well. This is true in
374 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
375 It would be nice to make closures have a settable name slot,
376 and make things like DEFSTRUCT and FLET, which create closures,
377 set helpful values into this slot.
380 And as long as we're wishing, it would be awfully nice if INSPECT could
381 also report on closures, telling about the values of the bound variables.
384 as reported by Robert Strandh on the CMU CL mailing list 12 Jun 2000:
386 (defconstant +a-constant+ (make-instance 'a-class))
387 (defconstant +another-constant+ (vector +a-constant+))
389 CMU Common Lisp release x86-linux 2.4.19 8 February 2000 build 456,
392 Send bug reports and questions to your local CMU CL maintainer,
393 or to pvaneynd@debian.org
394 or to cmucl-help@cons.org. (prefered)
395 type (help) for help, (quit) to exit, and (demo) to see the demos
397 Python 1.0, target Intel x86
398 CLOS based on PCL version: September 16 92 PCL (f)
399 * (defclass a-class () ())
400 #<STANDARD-CLASS A-CLASS {48027BD5}>
401 * (compile-file "xx.lisp")
402 Python version 1.0, VM version Intel x86 on 12 JUN 00 08:12:55 am.
404 /home/strandh/Research/Functional/Common-Lisp/CLIM/Development/McCLIM
405 /xx.lisp 12 JUN 00 07:47:14 am
406 Compiling Load Time Value of (PCL::GET-MAKE-INSTANCE-FUNCTION-SYMBOL
408 Byte Compiling Top-Level Form:
409 Error in function C::DUMP-STRUCTURE: Attempt to dump invalid
411 #<A-CLASS {4803A5B5}>
415 The compiler assumes that any time a function of declared FTYPE
416 doesn't signal an error, its arguments were of the declared type.
417 E.g. compiling and loading
418 (DECLAIM (OPTIMIZE (SAFETY 3)))
419 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
420 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE) FACTORIAL)))
422 (COND ((> (FACTORIAL X) 1.0E6)
423 (FORMAT T "too big~%"))
425 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
427 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
430 will cause the INTEGERP case to be selected, giving bogus output a la
432 This violates the "declarations are assertions" principle.
433 According to the ANSI spec, in the section "System Class FUNCTION",
434 this is a case of "lying to the compiler", but the lying is done
435 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
436 not by the unexpectedly general definition of FACTORIAL. In any case,
437 "declarations are assertions" means that lying to the compiler should
438 cause an error to be signalled, and should not cause a bogus
439 result to be returned. Thus, the compiler should not assume
440 that arbitrary functions check their argument types. (It might
441 make sense to add another flag (CHECKED?) to DEFKNOWN to
442 identify functions which *do* check their argument types.)
445 As pointed out by Martin Cracauer on the CMU CL mailing list
446 13 Jun 2000, the :FILE-LENGTH operation for
447 FD-STREAM-MISC-ROUTINE is broken for large files: it says
448 (THE INDEX SIZE) even though SIZE can be larger than INDEX.
451 In SBCL 0.6.5 (and CMU CL 18b) compiling and loading
452 (in-package :cl-user)
453 (declaim (optimize (safety 3)
455 (compilation-speed 2)
458 #+nil (sb-ext:inhibit-warnings 2)))
459 (declaim (ftype (function * (values)) emptyvalues))
460 (defun emptyvalues (&rest rest) (declare (ignore rest)) (values))
462 (defgeneric assertoid ((x t)))
463 (defmethod assertoid ((x t)) "just a placeholder")
465 (declare (type hash-table ht))
471 (assertoid (hash-table-count ht)))))))
472 (unless (typep res 'foo)
474 (common-lisp-user::bad-result-from-assertive-typed-fun
478 (bar (make-hash-table))
480 Error in KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER:
481 the function C::%INSTANCE-TYPEP is undefined.
482 %INSTANCE-TYPEP is always supposed to be IR1-transformed away, but for
483 some reason -- the (VALUES) return value declaration? -- the optimizer is
484 confused and compiles a full call to %INSTANCE-TYPEP (which doesn't exist
485 as a function) instead.
488 The %INSTANCE-TYPEP problem in bug 37 comes up also when compiling
490 (IN-PACKAGE :CL-USER)
492 (DECLARE (OPTIMIZE (SAFETY 3) (SPEED 2) (SPACE 2)))
493 (DECLAIM (FTYPE (FUNCTION (&REST T) (VALUES)) EMPTYVALUES))
494 (DEFUN EMPTYVALUES (&REST REST)
495 (DECLARE (IGNORE REST))
497 (DEFSTRUCT DUMMYSTRUCT X Y)
498 (DEFUN FROB-EMPTYVALUES (X)
499 (LET ((RES (EMPTYVALUES X X X)))
500 (UNLESS (TYPEP RES 'DUMMYSTRUCT)
501 'EXPECTED-RETURN-VALUE))))
502 (ASSERT (EQ (FROB-EMPTYVALUES 11) 'EXPECTED-RETURN-VALUE))
506 DEFMETHOD doesn't check the syntax of &REST argument lists properly,
507 accepting &REST even when it's not followed by an argument name:
508 (DEFMETHOD FOO ((X T) &REST) NIL)
511 On the CMU CL mailing list 26 June 2000, Douglas Crosher wrote
513 Hannu Rummukainen wrote:
515 > There's something weird going on with the compilation of the attached
516 > code. Compiling and loading the file in a fresh lisp, then invoking
518 Thanks for the bug report, nice to have this one fixed. It was a bug
519 in the x86 backend, the < VOP. A fix has been committed to the main
520 source, see the file compiler/x86/float.lisp.
522 Probably the same bug exists in SBCL.
525 TYPEP treats the result of UPGRADED-ARRAY-ELEMENT-TYPE as gospel,
526 so that (TYPEP (MAKE-ARRAY 3) '(VECTOR SOMETHING-NOT-DEFINED-YET))
527 returns (VALUES T T). Probably it should be an error instead,
528 complaining that the type SOMETHING-NOT-DEFINED-YET is not defined.
531 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
532 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
534 (DECLARE (TYPE INDEXOID X))
535 (THE (VALUES INDEXOID)
537 where the implementation of the type check in function FOO
538 includes a full call to %TYPEP. There are also some fundamental problems
539 with the interpretation of VALUES types (inherited from CMU CL, and
540 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
541 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
544 The definitions of SIGCONTEXT-FLOAT-REGISTER and
545 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
546 supported on FreeBSD because the floating point state is not saved,
547 but at least as of FreeBSD 4.0, the floating point state *is* saved,
548 so they could be supported after all. Very likely
549 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
552 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
553 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
554 issue with 'union, 'and, 'or etc. So even though according to the
555 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
556 (and now SBCL) interpret them as legal types.
559 ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
560 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
561 list, and it is probably possible to use substantially the same
562 patches to add it to SBCL.
565 a slew of floating-point-related errors reported by Peter Van Eynde
567 a: (SQRT -9.0) fails, because SB-KERNEL::COMPLEX-SQRT is undefined.
568 Similarly, COMPLEX-ASIN, COMPLEX-ACOS, COMPLEX-ACOSH, and others
570 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
571 should probably be 1.4012985e-45. In SBCL,
572 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
573 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
574 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
575 and LEAST-NEGATIVE-LONG-FLOAT.
576 c: Many expressions generate floating infinity:
581 PVE's regression tests want them to raise errors. SBCL
582 generates the infinities instead, which may or may not be
583 conforming behavior, but then blow it by being unable to
584 output the infinities, since support for infinities is generally
585 broken, and in particular SB-IMPL::OUTPUT-FLOAT-INFINITY is
587 d: (in section12.erg) various forms a la
588 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
589 don't give the right behavior.
592 type safety errors reported by Peter Van Eynde July 25, 2000:
593 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
595 In general lengths of array type specifications aren't
596 checked by COERCE, so it fails when the spec is
597 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
598 b: CONCATENATE has the same problem of not checking the length
599 of specified output array types. MAKE-SEQUENCE and MAP and
600 MERGE also have the same problem.
601 c: (COERCE 'AND 'FUNCTION) returns something related to
602 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
603 d: ELT signals SIMPLE-ERROR if its index argument
604 isn't a valid index for its sequence argument, but should
605 signal TYPE-ERROR instead.
606 e: FILE-LENGTH is supposed to signal a type error when its
607 argument is not a stream associated with a file, but doesn't.
608 f: (FLOAT-RADIX 2/3) should signal an error instead of
610 g: (LOAD "*.lsp") should signal FILE-ERROR.
611 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
612 should signal TYPE-ERROR.
613 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
614 be used for input and output as needed. It should fail with
615 TYPE-ERROR when handed e.g. the results of
616 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
617 the inappropriate positions, but doesn't.
618 j: (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
620 should probably signal an error instead of making a pathname with
622 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
623 not a binary input stream, but instead cheerfully reads from
624 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
627 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
628 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
630 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
631 signal a PROGRAM-ERROR, and doesn't.
632 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
633 and other DEFCLASS forms with duplicate specifications in their
634 slots, should signal a PROGRAM-ERROR, and doesn't.
635 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
636 causes a COMPILER-ERROR.
639 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
640 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
641 PROGRAM-ERROR, but SBCL accepts it instead.
642 b: SYMBOL-MACROLET should refuse to bind something which is
643 declared as a global variable, signalling PROGRAM-ERROR.
644 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
645 it binds is declared SPECIAL inside.
648 LOOP bugs reported by Peter Van Eynde July 25, 2000:
649 a: (LOOP WITH (A B) DO (PRINT 1)) is a syntax error according to
650 the definition of WITH clauses given in the ANSI spec, but
651 compiles and runs happily in SBCL.
652 b: a messy one involving package iteration:
653 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<))))
654 Should be: (("blah" "blah2") ("blah2"))
655 SBCL: (("blah") ("blah2"))
656 * (LET ((X 1)) (LOOP FOR I BY (INCF X) FROM X TO 10 COLLECT I))
657 doesn't work -- SBCL's LOOP says BY isn't allowed in a FOR clause.
660 type system errors reported by Peter Van Eynde July 25, 2000:
661 a: (SUBTYPEP 'BIGNUM 'INTEGER) => NIL, NIL
662 but should be (VALUES T T) instead.
663 b: (SUBTYPEP 'EXTENDED-CHAR 'CHARACTER) => NIL, NIL
664 but should be (VALUES T T) instead.
665 c: (SUBTYPEP '(INTEGER (0) (0)) 'NIL) dies with nested errors.
666 d: In general, the system doesn't like '(INTEGER (0) (0)) -- it
667 blows up at the level of SPECIFIER-TYPE with
668 "Lower bound (0) is greater than upper bound (0)." Probably
669 SPECIFIER-TYPE should return NIL instead.
670 e: (TYPEP 0 '(COMPLEX (EQL 0)) fails with
671 "Component type for Complex is not numeric: (EQL 0)."
672 This might be easy to fix; the type system already knows
673 that (SUBTYPEP '(EQL 0) 'NUMBER) is true.
674 f: The type system doesn't know about the condition system,
675 so that e.g. (TYPEP 'SIMPLE-ERROR 'ERROR)=>NIL.
676 g: The type system isn't all that smart about relationships
677 between hairy types, as shown in the type.erg test results,
678 e.g. (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL.
681 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
683 (DEFGENERIC FOO02 (X))
684 (DEFMETHOD FOO02 ((X NUMBER)) T)
685 (LET ((M (FIND-METHOD (FUNCTION FOO02)
687 (LIST (FIND-CLASS (QUOTE NUMBER))))))
688 (REMOVE-METHOD (FUNCTION FOO02) M)
689 (DEFGENERIC FOO03 (X))
690 (ADD-METHOD (FUNCTION FOO03) M)))
691 should give an error, but SBCL allows it.
692 b: READ should probably return READER-ERROR, not the bare
693 arithmetic error, when input a la "1/0" or "1e1000" causes
697 It has been reported (e.g. by Peter Van Eynde) that there are
698 several metaobject protocol "errors". (In order to fix them, we might
699 need to document exactly what metaobject protocol specification
700 we're following -- the current code is just inherited from PCL.)
703 another error from Peter Van Eynde 5 September 2000:
704 (FORMAT NIL "~F" "FOO") should work, but instead reports an error.
705 PVE submitted a patch to deal with this bug, but it exposes other
706 comparably serious bugs, so I didn't apply it. It looks as though
707 the FORMAT code needs a fair amount of rewriting in order to comply
708 with the various details of the ANSI spec.
711 The implementation of #'+ returns its single argument without
712 type checking, e.g. (+ "illegal") => "illegal".
715 In sbcl-0.6.7, there is no doc string for CL:PUSH, probably
716 because it's defined with the DEFMACRO-MUNDANELY macro and something
717 is wrong with doc string setting in that macro.
720 Attempting to use COMPILE on something defined by DEFMACRO fails:
721 (DEFMACRO FOO (X) (CONS X X))
723 Error in function C::GET-LAMBDA-TO-COMPILE:
724 #<Closure Over Function "DEFUN (SETF MACRO-FUNCTION)" {480E21B1}> was defined in a non-null environment.
727 (SUBTYPEP '(AND ZILCH INTEGER) 'ZILCH)
731 CL:*DEFAULT-PATHNAME-DEFAULTS* doesn't behave as ANSI suggests (reflecting
732 current working directory). And there's no supported way to update
733 or query the current working directory (a la Unix "chdir" and "pwd"),
734 which is functionality that ILISP needs (and currently gets with low-level
738 The debugger LIST-LOCATIONS command doesn't work properly.
741 Compiling and loading
742 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
744 then requesting a BACKTRACE at the debugger prompt gives no information
745 about where in the user program the problem occurred.
748 The compiler is supposed to do type inference well enough that
751 ((SIMPLE-ARRAY SINGLE-FLOAT)
753 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
756 is redundant. However, as reported by Juan Jose Garcia Ripoll for
757 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
758 workaround for the problem for now, but can't be done by the TYPECASE
759 macros themselves, since it's too hard for the macro to detect
760 assignments to the variable within the clause.
761 Note: The compiler *is* smart enough to do the type inference in
762 many cases. This case, derived from a couple of MACROEXPAND-1
763 calls on Ripoll's original test case,
765 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
766 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
767 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
768 (LET ((I 0) (G2554 LENGTH))
769 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
772 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
773 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
774 (GO SB-LOOP::NEXT-LOOP)
775 SB-LOOP::END-LOOP))))))
776 demonstrates the problem; but the problem goes away if the TAGBODY
777 and GO forms are removed (leaving the SETF in ordinary, non-looping
778 code), or if the TAGBODY and GO forms are retained, but the
779 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
782 KNOWN BUGS RELATED TO THE IR1 INTERPRETER
784 (Note: At some point, the pure interpreter (actually a semi-pure
785 interpreter aka "the IR1 interpreter") will probably go away, replaced
787 (DEFUN EVAL (X) (FUNCALL (COMPILE NIL (LAMBDA ..)))))
788 and at that time these bugs should either go away automatically or
789 become more tractable to fix. Until then, they'll probably remain,
790 since some of them aren't considered urgent, and the rest are too hard
791 to fix as long as so many special cases remain. After the IR1
792 interpreter goes away is also the preferred time to start
793 systematically exterminating cases where debugging functionality
794 (backtrace, breakpoint, etc.) breaks down, since getting rid of the
795 IR1 interpreter will reduce the number of special cases we need to
799 The FUNCTION special operator doesn't check properly whether its
800 argument is a function name. E.g. (FUNCTION (X Y)) returns a value
801 instead of failing with an error. (Later attempting to funcall the
802 value does cause an error.)
805 COMPILED-FUNCTION-P bogusly reports T for interpreted functions:
806 * (DEFUN FOO (X) (- 12 X))
808 * (COMPILED-FUNCTION-P #'FOO)
813 (DEFVAR *SUPPRESS-P* T)
814 (EVAL '(UNLESS *SUPPRESS-P*
815 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
816 (FORMAT T "surprise!"))))
817 prints "surprise!". Probably the entire EVAL-WHEN mechanism ought to be
818 rewritten from scratch to conform to the ANSI definition, abandoning
819 the *ALREADY-EVALED-THIS* hack which is used in sbcl-0.6.8.9 (and
820 in the original CMU CL source, too). This should be easier to do --
821 though still nontrivial -- once the various IR1 interpreter special
825 EVAL-WHEN's idea of what's a toplevel form is even more screwed up
826 than the example in IR1-3 would suggest, since COMPILE-FILE and
827 COMPILE both print both "right now!" messages when compiling the
831 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
832 (PRINT "yes! right now!"))
835 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
836 (PRINT "no! right now!"))
838 and while EVAL doesn't print the "right now!" messages, the first
839 FUNCALL on the value returned by EVAL causes both of them to be printed.