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 It should cause a note, not a WARNING, when the system ignores
81 an FTYPE proclamation for a slot accessor.
84 Error reporting on various stream-requiring operations is not
85 very good when the stream argument has the wrong type, because
86 the operation tries to fall through to Gray stream code, and then
87 dies because it's undefined. E.g.
88 (PRINT-UNREADABLE-OBJECT (*STANDARD-OUTPUT* 1)) ..)
89 gives the error message
90 error in SB-KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER:
91 The function SB-IMPL::STREAM-WRITE-STRING is undefined.
92 It would be more useful and correct to signal a TYPE-ERROR:
94 (It wouldn't be terribly difficult to write stubs for all the
95 Gray stream functions that the old CMU CL code expects, with
96 each stub just raising the appropriate TYPE-ERROR.)
99 bogus warnings about undefined functions for magic functions like
100 SB!C::%%DEFUN and SB!C::%DEFCONSTANT when cross-compiling files
101 like src/code/float.lisp
104 The "byte compiling top-level form:" output ought to be condensed.
105 Perhaps any number of such consecutive lines ought to turn into a
106 single "byte compiling top-level forms:" line.
109 Compiling a file containing the erroneous program
113 (DEFSTRUCT (BAR (:INCLUDE FOO))
116 gives only the not-very-useful message
118 (during macroexpansion)
119 Condition PROGRAM-ERROR was signalled.
120 (The specific message which says that the problem was duplicate
121 slot names gets lost.)
124 The handling of IGNORE declarations on lambda list arguments of
125 DEFMETHOD is at least weird, and in fact seems broken and useless.
126 I should fix up another layer of binding, declared IGNORABLE, for
127 typed lambda list arguments.
130 The way that the compiler munges types with arguments together
131 with types with no arguments (in e.g. TYPE-EXPAND) leads to
132 weirdness visible to the user:
133 (DEFTYPE FOO () 'FIXNUM)
135 (TYPEP 11 '(FOO)) => T, which seems weird
136 (TYPEP 11 'FIXNUM) => T
137 (TYPEP 11 '(FIXNUM)) signals an error, as it should
138 The situation is complicated by the presence of Common Lisp types
139 like UNSIGNED-BYTE (which can either be used in list form or alone)
140 so I'm not 100% sure that the behavior above is actually illegal.
141 But I'm 90+% sure, and someday perhaps I'll be motivated to look it up..
144 It would be nice if the
146 (during macroexpansion)
147 said what macroexpansion was at fault, e.g.
149 (during macroexpansion of IN-PACKAGE,
150 during macroexpansion of DEFFOO)
153 The type system doesn't understand the KEYWORD type very well:
154 (SUBTYPEP 'KEYWORD 'SYMBOL) => NIL, NIL
155 It might be possible to fix this by changing the definition of
156 KEYWORD to (AND SYMBOL (SATISFIES KEYWORDP)), but the type system
157 would need to be a bit smarter about AND types, too:
158 (SUBTYPEP '(AND SYMBOL KEYWORD) 'SYMBOL) => NIL, NIL
159 (The type system does know something about AND types already,
160 (SUBTYPEP '(AND INTEGER FLOAT) 'NUMBER) => T, T
161 (SUBTYPEP '(AND INTEGER FIXNUM) 'NUMBER) =>T, T
162 so likely this is a small patch.)
165 Floating point infinities are screwed up. [When I was converting CMU CL
166 to SBCL, I was looking for complexity to delete, and I thought it was safe
167 to just delete support for floating point infinities. It wasn't: they're
168 generated by the floating point hardware even when we remove support
169 for them in software. -- WHN] Support for them should be restored.
172 The ANSI syntax for non-STANDARD method combination types in CLOS is
173 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
174 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
175 If you mess this up, omitting the PROGN qualifier in in DEFMETHOD,
176 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
177 (DEFMETHOD FOO ((X BAR)) (PRINT 'NUMBER))
178 the error mesage is not easy to understand:
179 INVALID-METHOD-ERROR was called outside the dynamic scope
180 of a method combination function (inside the body of
181 DEFINE-METHOD-COMBINATION or a method on the generic
182 function COMPUTE-EFFECTIVE-METHOD).
183 It would be better if it were more informative, a la
184 The method combination type for this method (STANDARD) does
185 not match the method combination type for the generic function
187 Also, after you make the mistake of omitting the PROGN qualifier
188 on a DEFMETHOD, doing a new DEFMETHOD with the correct qualifier
190 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
192 INVALID-METHOD-ERROR was called outside the dynamic scope
193 of a method combination function (inside the body of
194 DEFINE-METHOD-COMBINATION or a method on the generic
195 function COMPUTE-EFFECTIVE-METHOD).
196 This is not very helpful..
199 (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
200 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
201 (Also, when this is fixed, we can enable the code in PROCLAIM which
202 checks for incompatible FTYPE redeclarations.)
205 The ANSI spec says that CONS can be a compound type spec, e.g.
206 (CONS FIXNUM REAL). SBCL doesn't support this.
209 from Paolo Amoroso on the CMU CL mailing list 27 Feb 2000:
210 I use CMU CL 18b under Linux. When COMPILE-FILE is supplied a physical
211 pathname, the type of the corresponding compiled file is X86F:
212 * (compile-file "/home/paolo/lisp/tools/foo")
213 Python version 1.0, VM version Intel x86 on 27 FEB 0 06:00:46 pm.
214 Compiling: /home/paolo/lisp/tools/foo.lisp 27 FEB 0 05:57:42 pm
216 Compiling DEFUN SQUARE:
217 Byte Compiling Top-Level Form:
218 /home/paolo/lisp/tools/foo.x86f written.
219 Compilation finished in 0:00:00.
220 #p"/home/paolo/lisp/tools/foo.x86f"
223 But when the function is called with a logical pathname, the file type
225 * (compile-file "tools:foo")
226 Python version 1.0, VM version Intel x86 on 27 FEB 0 06:01:04 pm.
227 Compiling: /home/paolo/lisp/tools/foo.lisp 27 FEB 0 05:57:42 pm
229 Compiling DEFUN SQUARE:
230 Byte Compiling Top-Level Form:
231 TOOLS:FOO.FASL written.
232 Compilation finished in 0:00:00.
233 #p"/home/paolo/lisp/tools/foo.fasl"
238 from DTC on the CMU CL mailing list 25 Feb 2000:
239 ;;; Compiler fails when this file is compiled.
241 ;;; Problem shows up in delete-block within ir1util.lisp. The assertion
242 ;;; (assert (member (functional-kind lambda) '(:let :mv-let :assignment)))
243 ;;; fails within bind node branch.
245 ;;; Note that if c::*check-consistency* is enabled then an un-reached
246 ;;; entry is also reported.
249 (declare (values nil))
266 (let ((ttt #'(lambda () (go cccc))))
267 (declare (special ttt))
268 (return-from bbbb nil))
271 (return-from bbbb nil))))))
274 (I *think* this is a bug. It certainly seems like strange behavior. But
275 the ANSI spec is scary, dark, and deep..)
276 (FORMAT NIL "~,1G" 1.4) => "1. "
277 (FORMAT NIL "~3,1G" 1.4) => "1. "
280 from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
282 (setf (find-class 'ccc1) (find-class 'ccc))
283 (defmethod zut ((c ccc1)) 123)
284 DTC's recommended workaround from the mailing list 3 Mar 2000:
285 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
288 The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
289 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
290 inside "~<..~>" (without the colon modifier on the closing syntax).
291 However, SBCL doesn't do this:
292 * (FORMAT T "~<munge~wegnum~>" 12)
297 When too many files are opened, OPEN will fail with an
298 uninformative error message
299 error in function OPEN: error opening #P"/tmp/foo.lisp": NIL
300 instead of saying that too many files are open.
303 Right now, when COMPILE-FILE has a read error, it actually pops
304 you into the debugger before giving up on the file. It should
305 instead handle the error, perhaps issuing (and handling)
306 a secondary error "caught ERROR: unrecoverable error during compilation"
307 and then return with FAILURE-P true,
310 from CMU CL mailing list 01 May 2000
312 I realize I can take care of this by doing (proclaim (ignore pcl::.slots1.))
313 but seeing as .slots0. is not-exported, shouldn't it be ignored within the
317 In: DEFMETHOD FOO-BAR-BAZ (RESOURCE-TYPE)
318 (DEFMETHOD FOO-BAR-BAZ
319 ((SELF RESOURCE-TYPE))
320 (SETF (SLOT-VALUE SELF 'NAME) 3))
321 --> BLOCK MACROLET PCL::FAST-LEXICAL-METHOD-FUNCTIONS
322 --> PCL::BIND-FAST-LEXICAL-METHOD-MACROS MACROLET
323 --> PCL::BIND-LEXICAL-METHOD-FUNCTIONS LET PCL::BIND-ARGS LET* PCL::PV-BINDING
324 --> PCL::PV-BINDING1 PCL::PV-ENV LET
326 (LET ((PCL::.SLOTS0. #))
331 Warning: Variable PCL::.SLOTS0. defined but never used.
333 Compilation unit finished.
336 #<Standard-Method FOO-BAR-BAZ (RESOURCE-TYPE) {480918FD}>
339 reported by Sam Steingold on the cmucl-imp mailing list 12 May 2000:
341 Also, there is another bug: `array-displacement' should return an array
342 or nil as first value (as per ANSI CL), while CMUCL declares it as
343 returning an array as first value always.
346 Sometimes (SB-EXT:QUIT) fails with
347 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
348 Process inferior-lisp exited abnormally with code 1
349 I haven't noticed a repeatable case of this yet.
352 some sort of bug in inlining and RETURN-FROM in sbcl-0.6.5: Compiling
355 (BLOCK USED-BY-SOME-Y?
358 (UNLESS (REJECTED? Y)
359 (RETURN-FROM USED-BY-SOME-Y? T)))))
360 (DECLARE (INLINE FROB))
365 error in function SB-KERNEL:ASSERT-ERROR:
366 The assertion (EQ (SB-C::CONTINUATION-KIND SB-C::CONT) :BLOCK-START) failed.
367 This is still present in sbcl-0.6.8.
370 The CMU CL reader code takes liberties in binding the standard read table
371 when reading the names of characters. Tim Moore posted a patch to the
372 CMU CL mailing list Mon, 22 May 2000 21:30:41 -0700.
375 In some cases the compiler believes type declarations on array
376 elements without checking them, e.g.
377 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
380 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
381 (WHEN (CONSP (AREF X 0))
383 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
386 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
387 all cases, e.g. the type assumption *is* checked if the array
388 elements are declared to be of some structure type instead of CONS.
391 The printer doesn't report closures very well. This is true in
395 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
396 It would be nice to make closures have a settable name slot,
397 and make things like DEFSTRUCT and FLET, which create closures,
398 set helpful values into this slot.
401 And as long as we're wishing, it would be awfully nice if INSPECT could
402 also report on closures, telling about the values of the bound variables.
405 as reported by Robert Strandh on the CMU CL mailing list 12 Jun 2000:
407 (defconstant +a-constant+ (make-instance 'a-class))
408 (defconstant +another-constant+ (vector +a-constant+))
410 CMU Common Lisp release x86-linux 2.4.19 8 February 2000 build 456,
413 Send bug reports and questions to your local CMU CL maintainer,
414 or to pvaneynd@debian.org
415 or to cmucl-help@cons.org. (prefered)
416 type (help) for help, (quit) to exit, and (demo) to see the demos
418 Python 1.0, target Intel x86
419 CLOS based on PCL version: September 16 92 PCL (f)
420 * (defclass a-class () ())
421 #<STANDARD-CLASS A-CLASS {48027BD5}>
422 * (compile-file "xx.lisp")
423 Python version 1.0, VM version Intel x86 on 12 JUN 00 08:12:55 am.
425 /home/strandh/Research/Functional/Common-Lisp/CLIM/Development/McCLIM
426 /xx.lisp 12 JUN 00 07:47:14 am
427 Compiling Load Time Value of (PCL::GET-MAKE-INSTANCE-FUNCTION-SYMBOL
429 Byte Compiling Top-Level Form:
430 Error in function C::DUMP-STRUCTURE: Attempt to dump invalid
432 #<A-CLASS {4803A5B5}>
436 The compiler assumes that any time a function of declared FTYPE
437 doesn't signal an error, its arguments were of the declared type.
438 E.g. compiling and loading
439 (DECLAIM (OPTIMIZE (SAFETY 3)))
440 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
441 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE) FACTORIAL)))
443 (COND ((> (FACTORIAL X) 1.0E6)
444 (FORMAT T "too big~%"))
446 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
448 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
451 will cause the INTEGERP case to be selected, giving bogus output a la
453 This violates the "declarations are assertions" principle.
454 According to the ANSI spec, in the section "System Class FUNCTION",
455 this is a case of "lying to the compiler", but the lying is done
456 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
457 not by the unexpectedly general definition of FACTORIAL. In any case,
458 "declarations are assertions" means that lying to the compiler should
459 cause an error to be signalled, and should not cause a bogus
460 result to be returned. Thus, the compiler should not assume
461 that arbitrary functions check their argument types. (It might
462 make sense to add another flag (CHECKED?) to DEFKNOWN to
463 identify functions which *do* check their argument types.)
466 As pointed out by Martin Cracauer on the CMU CL mailing list
467 13 Jun 2000, the :FILE-LENGTH operation for
468 FD-STREAM-MISC-ROUTINE is broken for large files: it says
469 (THE INDEX SIZE) even though SIZE can be larger than INDEX.
472 In SBCL 0.6.5 (and CMU CL 18b) compiling and loading
473 (in-package :cl-user)
474 (declaim (optimize (safety 3)
476 (compilation-speed 2)
479 #+nil (sb-ext:inhibit-warnings 2)))
480 (declaim (ftype (function * (values)) emptyvalues))
481 (defun emptyvalues (&rest rest) (declare (ignore rest)) (values))
483 (defgeneric assertoid ((x t)))
484 (defmethod assertoid ((x t)) "just a placeholder")
486 (declare (type hash-table ht))
492 (assertoid (hash-table-count ht)))))))
493 (unless (typep res 'foo)
495 (common-lisp-user::bad-result-from-assertive-typed-fun
499 (bar (make-hash-table))
501 Error in KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER:
502 the function C::%INSTANCE-TYPEP is undefined.
503 %INSTANCE-TYPEP is always supposed to be IR1-transformed away, but for
504 some reason -- the (VALUES) return value declaration? -- the optimizer is
505 confused and compiles a full call to %INSTANCE-TYPEP (which doesn't exist
506 as a function) instead.
509 The %INSTANCE-TYPEP problem in bug 37 comes up also when compiling
511 (IN-PACKAGE :CL-USER)
513 (DECLARE (OPTIMIZE (SAFETY 3) (SPEED 2) (SPACE 2)))
514 (DECLAIM (FTYPE (FUNCTION (&REST T) (VALUES)) EMPTYVALUES))
515 (DEFUN EMPTYVALUES (&REST REST)
516 (DECLARE (IGNORE REST))
518 (DEFSTRUCT DUMMYSTRUCT X Y)
519 (DEFUN FROB-EMPTYVALUES (X)
520 (LET ((RES (EMPTYVALUES X X X)))
521 (UNLESS (TYPEP RES 'DUMMYSTRUCT)
522 'EXPECTED-RETURN-VALUE))))
523 (ASSERT (EQ (FROB-EMPTYVALUES 11) 'EXPECTED-RETURN-VALUE))
527 DEFMETHOD doesn't check the syntax of &REST argument lists properly,
528 accepting &REST even when it's not followed by an argument name:
529 (DEFMETHOD FOO ((X T) &REST) NIL)
532 On the CMU CL mailing list 26 June 2000, Douglas Crosher wrote
534 Hannu Rummukainen wrote:
536 > There's something weird going on with the compilation of the attached
537 > code. Compiling and loading the file in a fresh lisp, then invoking
539 Thanks for the bug report, nice to have this one fixed. It was a bug
540 in the x86 backend, the < VOP. A fix has been committed to the main
541 source, see the file compiler/x86/float.lisp.
543 Probably the same bug exists in SBCL.
546 TYPEP treats the result of UPGRADED-ARRAY-ELEMENT-TYPE as gospel,
547 so that (TYPEP (MAKE-ARRAY 3) '(VECTOR SOMETHING-NOT-DEFINED-YET))
548 returns (VALUES T T). Probably it should be an error instead,
549 complaining that the type SOMETHING-NOT-DEFINED-YET is not defined.
552 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
553 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
555 (DECLARE (TYPE INDEXOID X))
556 (THE (VALUES INDEXOID)
558 where the implementation of the type check in function FOO
559 includes a full call to %TYPEP. There are also some fundamental problems
560 with the interpretation of VALUES types (inherited from CMU CL, and
561 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
562 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
565 The definitions of SIGCONTEXT-FLOAT-REGISTER and
566 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
567 supported on FreeBSD because the floating point state is not saved,
568 but at least as of FreeBSD 4.0, the floating point state *is* saved,
569 so they could be supported after all. Very likely
570 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
573 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
574 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
575 issue with 'union, 'and, 'or etc. So even though according to the
576 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
577 (and now SBCL) interpret them as legal types.
580 ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
581 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
582 list, and it is probably possible to use substantially the same
583 patches to add it to SBCL.
586 a slew of floating-point-related errors reported by Peter Van Eynde
588 a: (SQRT -9.0) fails, because SB-KERNEL::COMPLEX-SQRT is undefined.
589 Similarly, COMPLEX-ASIN, COMPLEX-ACOS, COMPLEX-ACOSH, and others
591 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
592 should probably be 1.4012985e-45. In SBCL,
593 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
594 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
595 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
596 and LEAST-NEGATIVE-LONG-FLOAT.
597 c: Many expressions generate floating infinity:
602 PVE's regression tests want them to raise errors. SBCL
603 generates the infinities instead, which may or may not be
604 conforming behavior, but then blow it by being unable to
605 output the infinities, since support for infinities is generally
606 broken, and in particular SB-IMPL::OUTPUT-FLOAT-INFINITY is
608 d: (in section12.erg) various forms a la
609 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
610 don't give the right behavior.
613 type safety errors reported by Peter Van Eynde July 25, 2000:
614 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
616 In general lengths of array type specifications aren't
617 checked by COERCE, so it fails when the spec is
618 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
619 b: CONCATENATE has the same problem of not checking the length
620 of specified output array types. MAKE-SEQUENCE and MAP and
621 MERGE also have the same problem.
622 c: (COERCE 'AND 'FUNCTION) returns something related to
623 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
624 d: ELT signals SIMPLE-ERROR if its index argument
625 isn't a valid index for its sequence argument, but should
626 signal TYPE-ERROR instead.
627 e: FILE-LENGTH is supposed to signal a type error when its
628 argument is not a stream associated with a file, but doesn't.
629 f: (FLOAT-RADIX 2/3) should signal an error instead of
631 g: (LOAD "*.lsp") should signal FILE-ERROR.
632 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
633 should signal TYPE-ERROR.
634 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
635 be used for input and output as needed. It should fail with
636 TYPE-ERROR when handed e.g. the results of
637 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
638 the inappropriate positions, but doesn't.
639 j: (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
641 should probably signal an error instead of making a pathname with
643 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
644 not a binary input stream, but instead cheerfully reads from
645 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
648 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
649 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
651 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
652 signal a PROGRAM-ERROR, and doesn't.
653 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
654 and other DEFCLASS forms with duplicate specifications in their
655 slots, should signal a PROGRAM-ERROR, and doesn't.
656 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
657 causes a COMPILER-ERROR.
660 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
661 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
662 PROGRAM-ERROR, but SBCL accepts it instead.
663 b: SYMBOL-MACROLET should refuse to bind something which is
664 declared as a global variable, signalling PROGRAM-ERROR.
665 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
666 it binds is declared SPECIAL inside.
669 LOOP bugs reported by Peter Van Eynde July 25, 2000:
670 a: (LOOP WITH (A B) DO (PRINT 1)) is a syntax error according to
671 the definition of WITH clauses given in the ANSI spec, but
672 compiles and runs happily in SBCL.
673 b: a messy one involving package iteration:
674 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<))))
675 Should be: (("blah" "blah2") ("blah2"))
676 SBCL: (("blah") ("blah2"))
677 * (LET ((X 1)) (LOOP FOR I BY (INCF X) FROM X TO 10 COLLECT I))
678 doesn't work -- SBCL's LOOP says BY isn't allowed in a FOR clause.
681 type system errors reported by Peter Van Eynde July 25, 2000:
682 a: (SUBTYPEP 'BIGNUM 'INTEGER) => NIL, NIL
683 but should be (VALUES T T) instead.
684 b: (SUBTYPEP 'EXTENDED-CHAR 'CHARACTER) => NIL, NIL
685 but should be (VALUES T T) instead.
686 c: (SUBTYPEP '(INTEGER (0) (0)) 'NIL) dies with nested errors.
687 d: In general, the system doesn't like '(INTEGER (0) (0)) -- it
688 blows up at the level of SPECIFIER-TYPE with
689 "Lower bound (0) is greater than upper bound (0)." Probably
690 SPECIFIER-TYPE should return NIL instead.
691 e: (TYPEP 0 '(COMPLEX (EQL 0)) fails with
692 "Component type for Complex is not numeric: (EQL 0)."
693 This might be easy to fix; the type system already knows
694 that (SUBTYPEP '(EQL 0) 'NUMBER) is true.
695 f: The type system doesn't know about the condition system,
696 so that e.g. (TYPEP 'SIMPLE-ERROR 'ERROR)=>NIL.
697 g: The type system isn't all that smart about relationships
698 between hairy types, as shown in the type.erg test results,
699 e.g. (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL.
702 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
704 (DEFGENERIC FOO02 (X))
705 (DEFMETHOD FOO02 ((X NUMBER)) T)
706 (LET ((M (FIND-METHOD (FUNCTION FOO02)
708 (LIST (FIND-CLASS (QUOTE NUMBER))))))
709 (REMOVE-METHOD (FUNCTION FOO02) M)
710 (DEFGENERIC FOO03 (X))
711 (ADD-METHOD (FUNCTION FOO03) M)))
712 should give an error, but SBCL allows it.
713 b: READ should probably return READER-ERROR, not the bare
714 arithmetic error, when input a la "1/0" or "1e1000" causes
718 It has been reported (e.g. by Peter Van Eynde) that there are
719 several metaobject protocol "errors". (In order to fix them, we might
720 need to document exactly what metaobject protocol specification
721 we're following -- the current code is just inherited from PCL.)
724 another error from Peter Van Eynde 5 September 2000:
725 (FORMAT NIL "~F" "FOO") should work, but instead reports an error.
726 PVE submitted a patch to deal with this bug, but it exposes other
727 comparably serious bugs, so I didn't apply it. It looks as though
728 the FORMAT code needs a fair amount of rewriting in order to comply
729 with the various details of the ANSI spec.
732 The implementation of #'+ returns its single argument without
733 type checking, e.g. (+ "illegal") => "illegal".
736 In sbcl-0.6.7, there is no doc string for CL:PUSH, probably
737 because it's defined with the DEFMACRO-MUNDANELY macro and something
738 is wrong with doc string setting in that macro.
741 Attempting to use COMPILE on something defined by DEFMACRO fails:
742 (DEFMACRO FOO (X) (CONS X X))
744 Error in function C::GET-LAMBDA-TO-COMPILE:
745 #<Closure Over Function "DEFUN (SETF MACRO-FUNCTION)" {480E21B1}> was defined in a non-null environment.
748 (SUBTYPEP '(AND ZILCH INTEGER) 'ZILCH)
752 CL:*DEFAULT-PATHNAME-DEFAULTS* doesn't behave as ANSI suggests (reflecting
753 current working directory). And there's no supported way to update
754 or query the current working directory (a la Unix "chdir" and "pwd"),
755 which is functionality that ILISP needs (and currently gets with low-level
759 The debugger LIST-LOCATIONS command doesn't work properly.
762 Compiling and loading
763 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
765 then requesting a BACKTRACE at the debugger prompt gives no information
766 about where in the user program the problem occurred.
769 The compiler is supposed to do type inference well enough that
772 ((SIMPLE-ARRAY SINGLE-FLOAT)
774 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
777 is redundant. However, as reported by Juan Jose Garcia Ripoll for
778 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
779 workaround for the problem for now, but can't be done by the TYPECASE
780 macros themselves, since it's too hard for the macro to detect
781 assignments to the variable within the clause.
782 Note: The compiler *is* smart enough to do the type inference in
783 many cases. This case, derived from a couple of MACROEXPAND-1
784 calls on Ripoll's original test case,
786 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
787 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
788 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
789 (LET ((I 0) (G2554 LENGTH))
790 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
793 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
794 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
795 (GO SB-LOOP::NEXT-LOOP)
796 SB-LOOP::END-LOOP))))))
797 demonstrates the problem; but the problem goes away if the TAGBODY
798 and GO forms are removed (leaving the SETF in ordinary, non-looping
799 code), or if the TAGBODY and GO forms are retained, but the
800 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
803 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
804 I am looking into this problem that showed up on the cmucl-help
805 list. It seems to me that the "implementation specific environment
806 hacking functions" found in pcl/walker.lisp are completely messed
807 up. The good thing is that they appear to be barely used within
808 PCL and the munged environment object is passed to cmucl only
809 in calls to macroexpand-1, which is probably why this case fails.
810 SBCL uses essentially the same code, so if the environment hacking
811 is screwed up, it affects us too.
814 Using the pretty-printer from the command prompt gives funny
815 results, apparently because the pretty-printer doesn't know
816 about user's command input, including the user's carriage return
817 that the user, and therefore the pretty-printer thinks that
818 the new output block should start indented 2 or more characters
819 rightward of the correct location.
822 As reported by Carl Witty on submit@bugs.debian.org 1999-05-08,
824 (in-package "CL-USER")
825 (defun equal-terms (termx termy)
827 ((alpha-equal-bound-term-lists (listx listy)
828 (or (and (null listx) (null listy))
830 (let ((bindings-x (bindings-of-bound-term (car listx)))
831 (bindings-y (bindings-of-bound-term (car listy))))
832 (if (and (null bindings-x) (null bindings-y))
833 (alpha-equal-terms (term-of-bound-term (car listx))
834 (term-of-bound-term (car listy)))
835 (and (= (length bindings-x) (length bindings-y))
837 (enter-binding-pairs (bindings-of-bound-term (car listx))
838 (bindings-of-bound-term (car listy)))
839 (alpha-equal-terms (term-of-bound-term (car listx))
840 (term-of-bound-term (car listy)))
841 (exit-binding-pairs (bindings-of-bound-term (car listx))
842 (bindings-of-bound-term (car listy)))))))
843 (alpha-equal-bound-term-lists (cdr listx) (cdr listy)))))
845 (alpha-equal-terms (termx termy)
846 (if (and (variable-p termx)
848 (equal-bindings (id-of-variable-term termx)
849 (id-of-variable-term termy))
850 (and (equal-operators-p (operator-of-term termx) (operator-of-term termy))
851 (alpha-equal-bound-term-lists (bound-terms-of-term termx)
852 (bound-terms-of-term termy))))))
856 (with-variable-invocation (alpha-equal-terms termx termy))))))
857 causes an assertion failure
858 The assertion (EQ (C::LAMBDA-TAIL-SET C::CALLER)
859 (C::LAMBDA-TAIL-SET (C::LAMBDA-HOME C::CALLEE))) failed.
861 Bob Rogers reports (1999-07-28 on cmucl-imp@cons.org) a smaller test
862 case with the same problem:
863 (defun parse-fssp-alignment ()
864 ;; Given an FSSP alignment file named by the argument . . .
865 (labels ((get-fssp-char ()
869 ;; Stub body, enough to tickle the bug.
870 (list (read-fssp-char)
874 ANSI specifies that the RESULT-TYPE argument of CONCATENATE must be
875 a subtype of SEQUENCE, but CONCATENATE doesn't check this properly:
876 (CONCATENATE 'SIMPLE-ARRAY #(1 2) '(3)) => #(1 2 3)
877 This also leads to funny behavior when derived type specifiers
878 are used, as originally reported by Milan Zamazal for CMU CL (on the
879 Debian bugs mailing list (?) 2000-02-27), then reported by Martin
880 Atzmueller for SBCL (2000-10-01 on sbcl-devel@lists.sourceforge.net):
881 (DEFTYPE FOO () 'SIMPLE-ARRAY)
882 (CONCATENATE 'FOO #(1 2) '(3))
883 => #<ARRAY-TYPE SIMPLE-ARRAY> is a bad type specifier for
885 The derived type specifier FOO should act the same way as the
886 built-in type SIMPLE-ARRAY here, but it doesn't. That problem
887 doesn't seem to exist for sequence types:
888 (DEFTYPE BAR () 'SIMPLE-VECTOR)
889 (CONCATENATE 'BAR #(1 2) '(3)) => #(1 2 3)
892 KNOWN BUGS RELATED TO THE IR1 INTERPRETER
894 (Note: At some point, the pure interpreter (actually a semi-pure
895 interpreter aka "the IR1 interpreter") will probably go away, replaced
897 (DEFUN EVAL (X) (FUNCALL (COMPILE NIL (LAMBDA ..)))))
898 and at that time these bugs should either go away automatically or
899 become more tractable to fix. Until then, they'll probably remain,
900 since some of them aren't considered urgent, and the rest are too hard
901 to fix as long as so many special cases remain. After the IR1
902 interpreter goes away is also the preferred time to start
903 systematically exterminating cases where debugging functionality
904 (backtrace, breakpoint, etc.) breaks down, since getting rid of the
905 IR1 interpreter will reduce the number of special cases we need to
909 The FUNCTION special operator doesn't check properly whether its
910 argument is a function name. E.g. (FUNCTION (X Y)) returns a value
911 instead of failing with an error. (Later attempting to funcall the
912 value does cause an error.)
915 COMPILED-FUNCTION-P bogusly reports T for interpreted functions:
916 * (DEFUN FOO (X) (- 12 X))
918 * (COMPILED-FUNCTION-P #'FOO)
923 (DEFVAR *SUPPRESS-P* T)
924 (EVAL '(UNLESS *SUPPRESS-P*
925 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
926 (FORMAT T "surprise!"))))
927 prints "surprise!". Probably the entire EVAL-WHEN mechanism ought to be
928 rewritten from scratch to conform to the ANSI definition, abandoning
929 the *ALREADY-EVALED-THIS* hack which is used in sbcl-0.6.8.9 (and
930 in the original CMU CL source, too). This should be easier to do --
931 though still nontrivial -- once the various IR1 interpreter special
935 EVAL-WHEN's idea of what's a toplevel form is even more screwed up
936 than the example in IR1-3 would suggest, since COMPILE-FILE and
937 COMPILE both print both "right now!" messages when compiling the
941 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
942 (PRINT "yes! right now!"))
945 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
946 (PRINT "no! right now!"))
948 and while EVAL doesn't print the "right now!" messages, the first
949 FUNCALL on the value returned by EVAL causes both of them to be printed.
952 The system accepts DECLAIM in most places where DECLARE would be
953 accepted, without even issuing a warning. ANSI allows this, but since
954 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
955 meaning is rather different, and it's unlikely that the user
956 has a good reason for doing DECLAIM not at top level, it would be
957 good to issue a STYLE-WARNING when this happens. A possible
958 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
959 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
960 [This is considered an IR1-interpreter-related bug because until
961 EVAL-WHEN is rewritten, which won't happen until after the IR1
962 interpreter is gone, the system's notion of what's a top-level form
963 and what's not will remain too confused to fix this problem.]