3 Bugs can be reported on the help mailing list
4 sbcl-help@lists.sourceforge.net
5 or on the development mailing list
6 sbcl-devel@lists.sourceforge.net
8 Please include enough information in a bug report that someone reading
9 it can reproduce the problem, i.e. don't write
10 Subject: apparent bug in PRINT-OBJECT (or *PRINT-LENGTH*?)
11 PRINT-OBJECT doesn't seem to work with *PRINT-LENGTH*. Is this a bug?
13 Subject: apparent bug in PRINT-OBJECT (or *PRINT-LENGTH*?)
14 In sbcl-1.2.3 running under OpenBSD 4.5 on my Alpha box, when
15 I compile and load the file
16 (DEFSTRUCT (FOO (:PRINT-OBJECT (LAMBDA (X Y)
17 (LET ((*PRINT-LENGTH* 4))
20 then at the command line type
22 the program loops endlessly instead of printing the object.
27 There is also some information on bugs in the manual page and
28 in the TODO file. Eventually more such information may move here.
30 The gaps in the number sequence belong to old bug descriptions which
31 have gone away (typically because they were fixed, but sometimes for
32 other reasons, e.g. because they were moved elsewhere).
35 KNOWN BUGS OF NO SPECIAL CLASS:
38 DEFSTRUCT almost certainly should overwrite the old LAYOUT information
39 instead of just punting when a contradictory structure definition
40 is loaded. As it is, if you redefine DEFSTRUCTs in a way which
41 changes their layout, you probably have to rebuild your entire
42 program, even if you know or guess enough about the internals of
43 SBCL to wager that this (undefined in ANSI) operation would be safe.
46 ANSI specifies that a type mismatch in a structure slot
47 initialization value should not cause a warning.
49 This one might not be fixed for a while because while we're big
50 believers in ANSI compatibility and all, (1) there's no obvious
51 simple way to do it (short of disabling all warnings for type
52 mismatches everywhere), and (2) there's a good portable
53 workaround. ANSI justifies this specification by saying
54 The restriction against issuing a warning for type mismatches
55 between a slot-initform and the corresponding slot's :TYPE
56 option is necessary because a slot-initform must be specified
57 in order to specify slot options; in some cases, no suitable
59 In SBCL, as in CMU CL (or, for that matter, any compiler which
60 really understands Common Lisp types) a suitable default does
61 exist, in all cases, because the compiler understands the concept
62 of functions which never return (i.e. has return type NIL, e.g.
63 ERROR). Thus, as a portable workaround, you can use a call to
64 some known-never-to-return function as the default. E.g.
66 (BAR (ERROR "missing :BAR argument")
67 :TYPE SOME-TYPE-TOO-HAIRY-TO-CONSTRUCT-AN-INSTANCE-OF))
69 (DECLAIM (FTYPE () NIL) MISSING-ARG)
70 (DEFUN REQUIRED-ARG () ; workaround for SBCL non-ANSI slot init typing
71 (ERROR "missing required argument"))
73 (BAR (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
74 (BLETCH (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
75 (N-REFS-SO-FAR 0 :TYPE (INTEGER 0)))
76 Such code will compile without complaint and work correctly either
77 on SBCL or on a completely compliant Common Lisp system.
80 bogus warnings about undefined functions for magic functions like
81 SB!C::%%DEFUN and SB!C::%DEFCONSTANT when cross-compiling files
82 like src/code/float.lisp. Fixing this will probably require
83 straightening out enough bootstrap consistency issues that
84 the cross-compiler can run with *TYPE-SYSTEM-INITIALIZED*.
85 Instead, the cross-compiler runs in a slightly flaky state
86 which is sane enough to compile SBCL itself, but which is
87 also unstable in several ways, including its inability
88 to really grok function declarations.
91 The "byte compiling top-level form:" output ought to be condensed.
92 Perhaps any number of such consecutive lines ought to turn into a
93 single "byte compiling top-level forms:" line.
96 The handling of IGNORE declarations on lambda list arguments of
97 DEFMETHOD is at least weird, and in fact seems broken and useless.
98 I should fix up another layer of binding, declared IGNORABLE, for
99 typed lambda list arguments.
102 The way that the compiler munges types with arguments together
103 with types with no arguments (in e.g. TYPE-EXPAND) leads to
104 weirdness visible to the user:
105 (DEFTYPE FOO () 'FIXNUM)
107 (TYPEP 11 '(FOO)) => T, which seems weird
108 (TYPEP 11 'FIXNUM) => T
109 (TYPEP 11 '(FIXNUM)) signals an error, as it should
110 The situation is complicated by the presence of Common Lisp types
111 like UNSIGNED-BYTE (which can either be used in list form or alone)
112 so I'm not 100% sure that the behavior above is actually illegal.
113 But I'm 90+% sure, and someday perhaps I'll be motivated to look it up..
116 It would be nice if the
118 (during macroexpansion)
119 said what macroexpansion was at fault, e.g.
121 (during macroexpansion of IN-PACKAGE,
122 during macroexpansion of DEFFOO)
125 The type system doesn't understand the KEYWORD type very well:
126 (SUBTYPEP 'KEYWORD 'SYMBOL) => NIL, NIL
127 It might be possible to fix this by changing the definition of
128 KEYWORD to (AND SYMBOL (SATISFIES KEYWORDP)), but the type system
129 would need to be a bit smarter about AND types, too:
130 (SUBTYPEP '(AND SYMBOL KEYWORD) 'SYMBOL) => NIL, NIL
131 (The type system does know something about AND types already,
132 (SUBTYPEP '(AND INTEGER FLOAT) 'NUMBER) => T, T
133 (SUBTYPEP '(AND INTEGER FIXNUM) 'NUMBER) =>T, T
134 so likely this is a small patch.)
137 Floating point infinities are screwed up. [When I was converting CMU CL
138 to SBCL, I was looking for complexity to delete, and I thought it was safe
139 to just delete support for floating point infinities. It wasn't: they're
140 generated by the floating point hardware even when we remove support
141 for them in software. -- WHN] Support for them should be restored.
144 The ANSI syntax for non-STANDARD method combination types in CLOS is
145 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
146 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
147 If you mess this up, omitting the PROGN qualifier in in DEFMETHOD,
148 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
149 (DEFMETHOD FOO ((X BAR)) (PRINT 'NUMBER))
150 the error mesage is not easy to understand:
151 INVALID-METHOD-ERROR was called outside the dynamic scope
152 of a method combination function (inside the body of
153 DEFINE-METHOD-COMBINATION or a method on the generic
154 function COMPUTE-EFFECTIVE-METHOD).
155 It would be better if it were more informative, a la
156 The method combination type for this method (STANDARD) does
157 not match the method combination type for the generic function
159 Also, after you make the mistake of omitting the PROGN qualifier
160 on a DEFMETHOD, doing a new DEFMETHOD with the correct qualifier
162 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
164 INVALID-METHOD-ERROR was called outside the dynamic scope
165 of a method combination function (inside the body of
166 DEFINE-METHOD-COMBINATION or a method on the generic
167 function COMPUTE-EFFECTIVE-METHOD).
168 This is not very helpful..
171 (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
172 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
173 (Also, when this is fixed, we can enable the code in PROCLAIM which
174 checks for incompatible FTYPE redeclarations.)
177 The ANSI spec says that CONS can be a compound type spec, e.g.
178 (CONS FIXNUM REAL). SBCL doesn't support this.
181 from DTC on the CMU CL mailing list 25 Feb 2000:
182 ;;; Compiler fails when this file is compiled.
184 ;;; Problem shows up in delete-block within ir1util.lisp. The assertion
185 ;;; (assert (member (functional-kind lambda) '(:let :mv-let :assignment)))
186 ;;; fails within bind node branch.
188 ;;; Note that if c::*check-consistency* is enabled then an un-reached
189 ;;; entry is also reported.
192 (declare (values nil))
209 (let ((ttt #'(lambda () (go cccc))))
210 (declare (special ttt))
211 (return-from bbbb nil))
214 (return-from bbbb nil))))))
217 (I *think* this is a bug. It certainly seems like strange behavior. But
218 the ANSI spec is scary, dark, and deep..)
219 (FORMAT NIL "~,1G" 1.4) => "1. "
220 (FORMAT NIL "~3,1G" 1.4) => "1. "
223 from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
225 (setf (find-class 'ccc1) (find-class 'ccc))
226 (defmethod zut ((c ccc1)) 123)
227 DTC's recommended workaround from the mailing list 3 Mar 2000:
228 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
231 The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
232 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
233 inside "~<..~>" (without the colon modifier on the closing syntax).
234 However, SBCL doesn't do this:
235 * (FORMAT T "~<munge~wegnum~>" 12)
240 When too many files are opened, OPEN will fail with an
241 uninformative error message
242 error in function OPEN: error opening #P"/tmp/foo.lisp": NIL
243 instead of saying that too many files are open.
246 Right now, when COMPILE-FILE has a read error, it actually pops
247 you into the debugger before giving up on the file. It should
248 instead handle the error, perhaps issuing (and handling)
249 a secondary error "caught ERROR: unrecoverable error during compilation"
250 and then return with FAILURE-P true,
253 from CMU CL mailing list 01 May 2000
255 I realize I can take care of this by doing (proclaim (ignore pcl::.slots1.))
256 but seeing as .slots0. is not-exported, shouldn't it be ignored within the
260 In: DEFMETHOD FOO-BAR-BAZ (RESOURCE-TYPE)
261 (DEFMETHOD FOO-BAR-BAZ
262 ((SELF RESOURCE-TYPE))
263 (SETF (SLOT-VALUE SELF 'NAME) 3))
264 --> BLOCK MACROLET PCL::FAST-LEXICAL-METHOD-FUNCTIONS
265 --> PCL::BIND-FAST-LEXICAL-METHOD-MACROS MACROLET
266 --> PCL::BIND-LEXICAL-METHOD-FUNCTIONS LET PCL::BIND-ARGS LET* PCL::PV-BINDING
267 --> PCL::PV-BINDING1 PCL::PV-ENV LET
269 (LET ((PCL::.SLOTS0. #))
274 Warning: Variable PCL::.SLOTS0. defined but never used.
276 Compilation unit finished.
279 #<Standard-Method FOO-BAR-BAZ (RESOURCE-TYPE) {480918FD}>
282 reported by Sam Steingold on the cmucl-imp mailing list 12 May 2000:
284 Also, there is another bug: `array-displacement' should return an array
285 or nil as first value (as per ANSI CL), while CMUCL declares it as
286 returning an array as first value always.
289 Sometimes (SB-EXT:QUIT) fails with
290 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
291 Process inferior-lisp exited abnormally with code 1
292 I haven't noticed a repeatable case of this yet.
295 some sort of bug in inlining and RETURN-FROM in sbcl-0.6.5: Compiling
298 (BLOCK USED-BY-SOME-Y?
301 (UNLESS (REJECTED? Y)
302 (RETURN-FROM USED-BY-SOME-Y? T)))))
303 (DECLARE (INLINE FROB))
308 error in function SB-KERNEL:ASSERT-ERROR:
309 The assertion (EQ (SB-C::CONTINUATION-KIND SB-C::CONT) :BLOCK-START) failed.
310 This is still present in sbcl-0.6.8.
313 The CMU CL reader code takes liberties in binding the standard read table
314 when reading the names of characters. Tim Moore posted a patch to the
315 CMU CL mailing list Mon, 22 May 2000 21:30:41 -0700.
318 In some cases the compiler believes type declarations on array
319 elements without checking them, e.g.
320 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
323 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
324 (WHEN (CONSP (AREF X 0))
326 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
329 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
330 all cases, e.g. the type assumption *is* checked if the array
331 elements are declared to be of some structure type instead of CONS.
334 The printer doesn't report closures very well. This is true in
338 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
339 It would be nice to make closures have a settable name slot,
340 and make things like DEFSTRUCT and FLET, which create closures,
341 set helpful values into this slot.
344 And as long as we're wishing, it would be awfully nice if INSPECT could
345 also report on closures, telling about the values of the bound variables.
348 as reported by Robert Strandh on the CMU CL mailing list 12 Jun 2000:
350 (defconstant +a-constant+ (make-instance 'a-class))
351 (defconstant +another-constant+ (vector +a-constant+))
353 CMU Common Lisp release x86-linux 2.4.19 8 February 2000 build 456,
356 Send bug reports and questions to your local CMU CL maintainer,
357 or to pvaneynd@debian.org
358 or to cmucl-help@cons.org. (prefered)
359 type (help) for help, (quit) to exit, and (demo) to see the demos
361 Python 1.0, target Intel x86
362 CLOS based on PCL version: September 16 92 PCL (f)
363 * (defclass a-class () ())
364 #<STANDARD-CLASS A-CLASS {48027BD5}>
365 * (compile-file "xx.lisp")
366 Python version 1.0, VM version Intel x86 on 12 JUN 00 08:12:55 am.
368 /home/strandh/Research/Functional/Common-Lisp/CLIM/Development/McCLIM
369 /xx.lisp 12 JUN 00 07:47:14 am
370 Compiling Load Time Value of (PCL::GET-MAKE-INSTANCE-FUNCTION-SYMBOL
372 Byte Compiling Top-Level Form:
373 Error in function C::DUMP-STRUCTURE: Attempt to dump invalid
375 #<A-CLASS {4803A5B5}>
379 The compiler assumes that any time a function of declared FTYPE
380 doesn't signal an error, its arguments were of the declared type.
381 E.g. compiling and loading
382 (DECLAIM (OPTIMIZE (SAFETY 3)))
383 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
384 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE) FACTORIAL)))
386 (COND ((> (FACTORIAL X) 1.0E6)
387 (FORMAT T "too big~%"))
389 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
391 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
394 will cause the INTEGERP case to be selected, giving bogus output a la
396 This violates the "declarations are assertions" principle.
397 According to the ANSI spec, in the section "System Class FUNCTION",
398 this is a case of "lying to the compiler", but the lying is done
399 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
400 not by the unexpectedly general definition of FACTORIAL. In any case,
401 "declarations are assertions" means that lying to the compiler should
402 cause an error to be signalled, and should not cause a bogus
403 result to be returned. Thus, the compiler should not assume
404 that arbitrary functions check their argument types. (It might
405 make sense to add another flag (CHECKED?) to DEFKNOWN to
406 identify functions which *do* check their argument types.)
409 As pointed out by Martin Cracauer on the CMU CL mailing list
410 13 Jun 2000, the :FILE-LENGTH operation for
411 FD-STREAM-MISC-ROUTINE is broken for large files: it says
412 (THE INDEX SIZE) even though SIZE can be larger than INDEX.
415 In SBCL 0.6.5 (and CMU CL 18b) compiling and loading
416 (in-package :cl-user)
417 (declaim (optimize (safety 3)
419 (compilation-speed 2)
422 #+nil (sb-ext:inhibit-warnings 2)))
423 (declaim (ftype (function * (values)) emptyvalues))
424 (defun emptyvalues (&rest rest) (declare (ignore rest)) (values))
426 (defgeneric assertoid ((x t)))
427 (defmethod assertoid ((x t)) "just a placeholder")
429 (declare (type hash-table ht))
435 (assertoid (hash-table-count ht)))))))
436 (unless (typep res 'foo)
438 (common-lisp-user::bad-result-from-assertive-typed-fun
442 (bar (make-hash-table))
444 Error in KERNEL::UNDEFINED-SYMBOL-ERROR-HANDLER:
445 the function C::%INSTANCE-TYPEP is undefined.
446 %INSTANCE-TYPEP is always supposed to be IR1-transformed away, but for
447 some reason -- the (VALUES) return value declaration? -- the optimizer is
448 confused and compiles a full call to %INSTANCE-TYPEP (which doesn't exist
449 as a function) instead.
452 The %INSTANCE-TYPEP problem in bug 37 comes up also when compiling
454 (IN-PACKAGE :CL-USER)
456 (DECLARE (OPTIMIZE (SAFETY 3) (SPEED 2) (SPACE 2)))
457 (DECLAIM (FTYPE (FUNCTION (&REST T) (VALUES)) EMPTYVALUES))
458 (DEFUN EMPTYVALUES (&REST REST)
459 (DECLARE (IGNORE REST))
461 (DEFSTRUCT DUMMYSTRUCT X Y)
462 (DEFUN FROB-EMPTYVALUES (X)
463 (LET ((RES (EMPTYVALUES X X X)))
464 (UNLESS (TYPEP RES 'DUMMYSTRUCT)
465 'EXPECTED-RETURN-VALUE))))
466 (ASSERT (EQ (FROB-EMPTYVALUES 11) 'EXPECTED-RETURN-VALUE))
470 DEFMETHOD doesn't check the syntax of &REST argument lists properly,
471 accepting &REST even when it's not followed by an argument name:
472 (DEFMETHOD FOO ((X T) &REST) NIL)
475 On the CMU CL mailing list 26 June 2000, Douglas Crosher wrote
477 Hannu Rummukainen wrote:
479 > There's something weird going on with the compilation of the attached
480 > code. Compiling and loading the file in a fresh lisp, then invoking
482 Thanks for the bug report, nice to have this one fixed. It was a bug
483 in the x86 backend, the < VOP. A fix has been committed to the main
484 source, see the file compiler/x86/float.lisp.
486 Probably the same bug exists in SBCL.
489 TYPEP treats the result of UPGRADED-ARRAY-ELEMENT-TYPE as gospel,
490 so that (TYPEP (MAKE-ARRAY 3) '(VECTOR SOMETHING-NOT-DEFINED-YET))
491 returns (VALUES T T). Probably it should be an error instead,
492 complaining that the type SOMETHING-NOT-DEFINED-YET is not defined.
495 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
496 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
498 (DECLARE (TYPE INDEXOID X))
499 (THE (VALUES INDEXOID)
501 where the implementation of the type check in function FOO
502 includes a full call to %TYPEP. There are also some fundamental problems
503 with the interpretation of VALUES types (inherited from CMU CL, and
504 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
505 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
508 The definitions of SIGCONTEXT-FLOAT-REGISTER and
509 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
510 supported on FreeBSD because the floating point state is not saved,
511 but at least as of FreeBSD 4.0, the floating point state *is* saved,
512 so they could be supported after all. Very likely
513 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
516 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
517 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
518 issue with 'union, 'and, 'or etc. So even though according to the
519 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
520 (and now SBCL) interpret them as legal types.
523 ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
524 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
525 list, and it is probably possible to use substantially the same
526 patches to add it to SBCL.
529 a slew of floating-point-related errors reported by Peter Van Eynde
531 a: (SQRT -9.0) fails, because SB-KERNEL::COMPLEX-SQRT is undefined.
532 Similarly, COMPLEX-ASIN, COMPLEX-ACOS, COMPLEX-ACOSH, and others
534 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
535 should probably be 1.4012985e-45. In SBCL,
536 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
537 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
538 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
539 and LEAST-NEGATIVE-LONG-FLOAT.
540 c: Many expressions generate floating infinity:
545 PVE's regression tests want them to raise errors. SBCL
546 generates the infinities instead, which may or may not be
547 conforming behavior, but then blow it by being unable to
548 output the infinities, since support for infinities is generally
549 broken, and in particular SB-IMPL::OUTPUT-FLOAT-INFINITY is
551 d: (in section12.erg) various forms a la
552 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
553 don't give the right behavior.
556 type safety errors reported by Peter Van Eynde July 25, 2000:
557 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
559 In general lengths of array type specifications aren't
560 checked by COERCE, so it fails when the spec is
561 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
562 b: CONCATENATE has the same problem of not checking the length
563 of specified output array types. MAKE-SEQUENCE and MAP and
564 MERGE also have the same problem.
565 c: (COERCE 'AND 'FUNCTION) returns something related to
566 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
567 d: ELT signals SIMPLE-ERROR if its index argument
568 isn't a valid index for its sequence argument, but should
569 signal TYPE-ERROR instead.
570 e: FILE-LENGTH is supposed to signal a type error when its
571 argument is not a stream associated with a file, but doesn't.
572 f: (FLOAT-RADIX 2/3) should signal an error instead of
574 g: (LOAD "*.lsp") should signal FILE-ERROR.
575 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
576 should signal TYPE-ERROR.
577 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
578 be used for input and output as needed. It should fail with
579 TYPE-ERROR when handed e.g. the results of
580 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
581 the inappropriate positions, but doesn't.
582 j: (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
584 should probably signal an error instead of making a pathname with
586 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
587 not a binary input stream, but instead cheerfully reads from
588 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
591 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
592 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
594 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
595 signal a PROGRAM-ERROR, and doesn't.
596 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
597 and other DEFCLASS forms with duplicate specifications in their
598 slots, should signal a PROGRAM-ERROR, and doesn't.
599 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
600 causes a COMPILER-ERROR.
603 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
604 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
605 PROGRAM-ERROR, but SBCL accepts it instead.
606 b: SYMBOL-MACROLET should refuse to bind something which is
607 declared as a global variable, signalling PROGRAM-ERROR.
608 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
609 it binds is declared SPECIAL inside.
612 LOOP bugs reported by Peter Van Eynde July 25, 2000:
613 a: (LOOP WITH (A B) DO (PRINT 1)) is a syntax error according to
614 the definition of WITH clauses given in the ANSI spec, but
615 compiles and runs happily in SBCL.
616 b: a messy one involving package iteration:
617 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<))))
618 Should be: (("blah" "blah2") ("blah2"))
619 SBCL: (("blah") ("blah2"))
620 * (LET ((X 1)) (LOOP FOR I BY (INCF X) FROM X TO 10 COLLECT I))
621 doesn't work -- SBCL's LOOP says BY isn't allowed in a FOR clause.
624 type system errors reported by Peter Van Eynde July 25, 2000:
625 a: (SUBTYPEP 'BIGNUM 'INTEGER) => NIL, NIL
626 but should be (VALUES T T) instead.
627 b: (SUBTYPEP 'EXTENDED-CHAR 'CHARACTER) => NIL, NIL
628 but should be (VALUES T T) instead.
629 c: (SUBTYPEP '(INTEGER (0) (0)) 'NIL) dies with nested errors.
630 d: In general, the system doesn't like '(INTEGER (0) (0)) -- it
631 blows up at the level of SPECIFIER-TYPE with
632 "Lower bound (0) is greater than upper bound (0)." Probably
633 SPECIFIER-TYPE should return NIL instead.
634 e: (TYPEP 0 '(COMPLEX (EQL 0)) fails with
635 "Component type for Complex is not numeric: (EQL 0)."
636 This might be easy to fix; the type system already knows
637 that (SUBTYPEP '(EQL 0) 'NUMBER) is true.
638 f: The type system doesn't know about the condition system,
639 so that e.g. (TYPEP 'SIMPLE-ERROR 'ERROR)=>NIL.
640 g: The type system isn't all that smart about relationships
641 between hairy types, as shown in the type.erg test results,
642 e.g. (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL.
645 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
647 (DEFGENERIC FOO02 (X))
648 (DEFMETHOD FOO02 ((X NUMBER)) T)
649 (LET ((M (FIND-METHOD (FUNCTION FOO02)
651 (LIST (FIND-CLASS (QUOTE NUMBER))))))
652 (REMOVE-METHOD (FUNCTION FOO02) M)
653 (DEFGENERIC FOO03 (X))
654 (ADD-METHOD (FUNCTION FOO03) M)))
655 should give an error, but SBCL allows it.
656 b: READ should probably return READER-ERROR, not the bare
657 arithmetic error, when input a la "1/0" or "1e1000" causes
661 It has been reported (e.g. by Peter Van Eynde) that there are
662 several metaobject protocol "errors". (In order to fix them, we might
663 need to document exactly what metaobject protocol specification
664 we're following -- the current code is just inherited from PCL.)
667 another error from Peter Van Eynde 5 September 2000:
668 (FORMAT NIL "~F" "FOO") should work, but instead reports an error.
669 PVE submitted a patch to deal with this bug, but it exposes other
670 comparably serious bugs, so I didn't apply it. It looks as though
671 the FORMAT code needs a fair amount of rewriting in order to comply
672 with the various details of the ANSI spec.
675 The implementation of #'+ returns its single argument without
676 type checking, e.g. (+ "illegal") => "illegal".
679 In sbcl-0.6.7, there is no doc string for CL:PUSH, probably
680 because it's defined with the DEFMACRO-MUNDANELY macro and something
681 is wrong with doc string setting in that macro.
684 Attempting to use COMPILE on something defined by DEFMACRO fails:
685 (DEFMACRO FOO (X) (CONS X X))
687 Error in function C::GET-LAMBDA-TO-COMPILE:
688 #<Closure Over Function "DEFUN (SETF MACRO-FUNCTION)" {480E21B1}> was defined in a non-null environment.
691 (SUBTYPEP '(AND ZILCH INTEGER) 'ZILCH)
695 CL:*DEFAULT-PATHNAME-DEFAULTS* doesn't behave as ANSI suggests (reflecting
696 current working directory). And there's no supported way to update
697 or query the current working directory (a la Unix "chdir" and "pwd"),
698 which is functionality that ILISP needs (and currently gets with low-level
702 The debugger LIST-LOCATIONS command doesn't work properly.
705 Compiling and loading
706 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
708 then requesting a BACKTRACE at the debugger prompt gives no information
709 about where in the user program the problem occurred.
712 The compiler is supposed to do type inference well enough that
715 ((SIMPLE-ARRAY SINGLE-FLOAT)
717 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
720 is redundant. However, as reported by Juan Jose Garcia Ripoll for
721 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
722 workaround for the problem for now, but can't be done by the TYPECASE
723 macros themselves, since it's too hard for the macro to detect
724 assignments to the variable within the clause.
725 Note: The compiler *is* smart enough to do the type inference in
726 many cases. This case, derived from a couple of MACROEXPAND-1
727 calls on Ripoll's original test case,
729 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
730 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
731 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
732 (LET ((I 0) (G2554 LENGTH))
733 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
736 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
737 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
738 (GO SB-LOOP::NEXT-LOOP)
739 SB-LOOP::END-LOOP))))))
740 demonstrates the problem; but the problem goes away if the TAGBODY
741 and GO forms are removed (leaving the SETF in ordinary, non-looping
742 code), or if the TAGBODY and GO forms are retained, but the
743 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
746 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
747 I am looking into this problem that showed up on the cmucl-help
748 list. It seems to me that the "implementation specific environment
749 hacking functions" found in pcl/walker.lisp are completely messed
750 up. The good thing is that they appear to be barely used within
751 PCL and the munged environment object is passed to cmucl only
752 in calls to macroexpand-1, which is probably why this case fails.
753 SBCL uses essentially the same code, so if the environment hacking
754 is screwed up, it affects us too.
757 Using the pretty-printer from the command prompt gives funny
758 results, apparently because the pretty-printer doesn't know
759 about user's command input, including the user's carriage return
760 that the user, and therefore the pretty-printer thinks that
761 the new output block should start indented 2 or more characters
762 rightward of the correct location.
765 As reported by Carl Witty on submit@bugs.debian.org 1999-05-08,
767 (in-package "CL-USER")
768 (defun equal-terms (termx termy)
770 ((alpha-equal-bound-term-lists (listx listy)
771 (or (and (null listx) (null listy))
773 (let ((bindings-x (bindings-of-bound-term (car listx)))
774 (bindings-y (bindings-of-bound-term (car listy))))
775 (if (and (null bindings-x) (null bindings-y))
776 (alpha-equal-terms (term-of-bound-term (car listx))
777 (term-of-bound-term (car listy)))
778 (and (= (length bindings-x) (length bindings-y))
780 (enter-binding-pairs (bindings-of-bound-term (car listx))
781 (bindings-of-bound-term (car listy)))
782 (alpha-equal-terms (term-of-bound-term (car listx))
783 (term-of-bound-term (car listy)))
784 (exit-binding-pairs (bindings-of-bound-term (car listx))
785 (bindings-of-bound-term (car listy)))))))
786 (alpha-equal-bound-term-lists (cdr listx) (cdr listy)))))
788 (alpha-equal-terms (termx termy)
789 (if (and (variable-p termx)
791 (equal-bindings (id-of-variable-term termx)
792 (id-of-variable-term termy))
793 (and (equal-operators-p (operator-of-term termx) (operator-of-term termy))
794 (alpha-equal-bound-term-lists (bound-terms-of-term termx)
795 (bound-terms-of-term termy))))))
799 (with-variable-invocation (alpha-equal-terms termx termy))))))
800 causes an assertion failure
801 The assertion (EQ (C::LAMBDA-TAIL-SET C::CALLER)
802 (C::LAMBDA-TAIL-SET (C::LAMBDA-HOME C::CALLEE))) failed.
804 Bob Rogers reports (1999-07-28 on cmucl-imp@cons.org) a smaller test
805 case with the same problem:
806 (defun parse-fssp-alignment ()
807 ;; Given an FSSP alignment file named by the argument . . .
808 (labels ((get-fssp-char ()
812 ;; Stub body, enough to tickle the bug.
813 (list (read-fssp-char)
817 ANSI specifies that the RESULT-TYPE argument of CONCATENATE must be
818 a subtype of SEQUENCE, but CONCATENATE doesn't check this properly:
819 (CONCATENATE 'SIMPLE-ARRAY #(1 2) '(3)) => #(1 2 3)
820 This also leads to funny behavior when derived type specifiers
821 are used, as originally reported by Milan Zamazal for CMU CL (on the
822 Debian bugs mailing list (?) 2000-02-27), then reported by Martin
823 Atzmueller for SBCL (2000-10-01 on sbcl-devel@lists.sourceforge.net):
824 (DEFTYPE FOO () 'SIMPLE-ARRAY)
825 (CONCATENATE 'FOO #(1 2) '(3))
826 => #<ARRAY-TYPE SIMPLE-ARRAY> is a bad type specifier for
828 The derived type specifier FOO should act the same way as the
829 built-in type SIMPLE-ARRAY here, but it doesn't. That problem
830 doesn't seem to exist for sequence types:
831 (DEFTYPE BAR () 'SIMPLE-VECTOR)
832 (CONCATENATE 'BAR #(1 2) '(3)) => #(1 2 3)
835 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
836 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
837 crashes SBCL. In general tracing anything which is used in the
838 implementation of TRACE is likely to have the same problem.
841 As reported by Daniel Solaz on cmucl-help@cons.org 2000-11-23,
842 SXHASH returns the same value for all non-STRUCTURE-OBJECT instances,
843 notably including all PCL instances. There's a limit to how much
844 SXHASH can do to return unique values for instances, but at least
845 it should probably look at the class name, the way that it does
846 for STRUCTURE-OBJECTs.
849 As reported by Martin Atzmueller on the sbcl-devel list 2000-11-22,
850 > There remains one issue, that is a bug in SBCL:
851 > According to my interpretation of the spec, the ":" and "@" modifiers
852 > should appear _after_ the comma-seperated arguments.
853 > Well, SBCL (and CMUCL for that matter) accept
854 > (ASSERT (STRING= (FORMAT NIL "~:8D" 1) " 1"))
855 > where the correct way (IMHO) should be
856 > (ASSERT (STRING= (FORMAT NIL "~8:D" 1) " 1"))
857 Probably SBCL should stop accepting the "~:8D"-style format arguments,
858 or at least issue a warning.
861 The compiler doesn't like &OPTIONAL arguments in LABELS and FLET
863 (DEFUN FIND-BEFORE (ITEM SEQUENCE &KEY (TEST #'EQL))
864 (LABELS ((FIND-ITEM (OBJ SEQ TEST &OPTIONAL (VAL NIL))
865 (LET ((ITEM (FIRST SEQ)))
868 ((FUNCALL TEST OBJ ITEM)
871 (FIND-ITEM OBJ (REST SEQ) TEST (NCONC VAL `(,ITEM))))))))
872 (FIND-ITEM ITEM SEQUENCE TEST)))
873 from David Young's bug report on cmucl-help@cons.org 30 Nov 2000
874 causes sbcl-0.6.9 to fail with
875 error in function SB-KERNEL:ASSERT-ERROR:
876 The assertion (EQ (SB-C::LAMBDA-TAIL-SET SB-C::CALLER)
877 (SB-C::LAMBDA-TAIL-SET
878 (SB-C::LAMBDA-HOME SB-C::CALLEE))) failed.
881 (DECLAIM (OPTIMIZE ..)) doesn't work. E.g. even after
882 (DECLAIM (OPTIMIZE (SPEED 3))), things are still optimized with
883 the previous SPEED policy. This bug will probably get fixed in
884 0.6.9.x in a general cleanup of optimization policy.
887 KNOWN BUGS RELATED TO THE IR1 INTERPRETER
889 (Note: At some point, the pure interpreter (actually a semi-pure
890 interpreter aka "the IR1 interpreter") will probably go away, replaced
892 (DEFUN EVAL (X) (FUNCALL (COMPILE NIL (LAMBDA ..)))))
893 and at that time these bugs should either go away automatically or
894 become more tractable to fix. Until then, they'll probably remain,
895 since some of them aren't considered urgent, and the rest are too hard
896 to fix as long as so many special cases remain. After the IR1
897 interpreter goes away is also the preferred time to start
898 systematically exterminating cases where debugging functionality
899 (backtrace, breakpoint, etc.) breaks down, since getting rid of the
900 IR1 interpreter will reduce the number of special cases we need to
904 The FUNCTION special operator doesn't check properly whether its
905 argument is a function name. E.g. (FUNCTION (X Y)) returns a value
906 instead of failing with an error. (Later attempting to funcall the
907 value does cause an error.)
910 COMPILED-FUNCTION-P bogusly reports T for interpreted functions:
911 * (DEFUN FOO (X) (- 12 X))
913 * (COMPILED-FUNCTION-P #'FOO)
918 (DEFVAR *SUPPRESS-P* T)
919 (EVAL '(UNLESS *SUPPRESS-P*
920 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
921 (FORMAT T "surprise!"))))
922 prints "surprise!". Probably the entire EVAL-WHEN mechanism ought to be
923 rewritten from scratch to conform to the ANSI definition, abandoning
924 the *ALREADY-EVALED-THIS* hack which is used in sbcl-0.6.8.9 (and
925 in the original CMU CL source, too). This should be easier to do --
926 though still nontrivial -- once the various IR1 interpreter special
930 EVAL-WHEN's idea of what's a toplevel form is even more screwed up
931 than the example in IR1-3 would suggest, since COMPILE-FILE and
932 COMPILE both print both "right now!" messages when compiling the
936 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
937 (PRINT "yes! right now!"))
940 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
941 (PRINT "no! right now!"))
943 and while EVAL doesn't print the "right now!" messages, the first
944 FUNCALL on the value returned by EVAL causes both of them to be printed.
947 The system accepts DECLAIM in most places where DECLARE would be
948 accepted, without even issuing a warning. ANSI allows this, but since
949 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
950 meaning is rather different, and it's unlikely that the user
951 has a good reason for doing DECLAIM not at top level, it would be
952 good to issue a STYLE-WARNING when this happens. A possible
953 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
954 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
955 [This is considered an IR1-interpreter-related bug because until
956 EVAL-WHEN is rewritten, which won't happen until after the IR1
957 interpreter is gone, the system's notion of what's a top-level form
958 and what's not will remain too confused to fix this problem.]