3 Bugs can be reported on the help mailing list
4 sbcl-help@lists.sourceforge.net
5 or on the development mailing list
6 sbcl-devel@lists.sourceforge.net
8 Please include enough information in a bug report that someone reading
9 it can reproduce the problem, i.e. don't write
10 Subject: apparent bug in PRINT-OBJECT (or *PRINT-LENGTH*?)
11 PRINT-OBJECT doesn't seem to work with *PRINT-LENGTH*. Is this a bug?
13 Subject: apparent bug in PRINT-OBJECT (or *PRINT-LENGTH*?)
14 In sbcl-1.2.3 running under OpenBSD 4.5 on my Alpha box, when
15 I compile and load the file
16 (DEFSTRUCT (FOO (:PRINT-OBJECT (LAMBDA (X Y)
17 (LET ((*PRINT-LENGTH* 4))
20 then at the command line type
22 the program loops endlessly instead of printing the object.
27 There is also some information on bugs in the manual page and
28 in the TODO file. Eventually more such information may move here.
30 The gaps in the number sequence belong to old bug descriptions which
31 have gone away (typically because they were fixed, but sometimes for
32 other reasons, e.g. because they were moved elsewhere).
35 KNOWN BUGS OF NO SPECIAL CLASS:
38 DEFSTRUCT almost certainly should overwrite the old LAYOUT information
39 instead of just punting when a contradictory structure definition
40 is loaded. As it is, if you redefine DEFSTRUCTs in a way which
41 changes their layout, you probably have to rebuild your entire
42 program, even if you know or guess enough about the internals of
43 SBCL to wager that this (undefined in ANSI) operation would be safe.
46 ANSI specifies that a type mismatch in a structure slot
47 initialization value should not cause a warning.
49 This one might not be fixed for a while because while we're big
50 believers in ANSI compatibility and all, (1) there's no obvious
51 simple way to do it (short of disabling all warnings for type
52 mismatches everywhere), and (2) there's a good portable
53 workaround. ANSI justifies this specification by saying
54 The restriction against issuing a warning for type mismatches
55 between a slot-initform and the corresponding slot's :TYPE
56 option is necessary because a slot-initform must be specified
57 in order to specify slot options; in some cases, no suitable
59 In SBCL, as in CMU CL (or, for that matter, any compiler which
60 really understands Common Lisp types) a suitable default does
61 exist, in all cases, because the compiler understands the concept
62 of functions which never return (i.e. has return type NIL, e.g.
63 ERROR). Thus, as a portable workaround, you can use a call to
64 some known-never-to-return function as the default. E.g.
66 (BAR (ERROR "missing :BAR argument")
67 :TYPE SOME-TYPE-TOO-HAIRY-TO-CONSTRUCT-AN-INSTANCE-OF))
69 (DECLAIM (FTYPE () NIL) MISSING-ARG)
70 (DEFUN REQUIRED-ARG () ; workaround for SBCL non-ANSI slot init typing
71 (ERROR "missing required argument"))
73 (BAR (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
74 (BLETCH (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
75 (N-REFS-SO-FAR 0 :TYPE (INTEGER 0)))
76 Such code will compile without complaint and work correctly either
77 on SBCL or on a completely compliant Common Lisp system.
80 bogus warnings about undefined functions for magic functions like
81 SB!C::%%DEFUN and SB!C::%DEFCONSTANT when cross-compiling files
82 like src/code/float.lisp. Fixing this will probably require
83 straightening out enough bootstrap consistency issues that
84 the cross-compiler can run with *TYPE-SYSTEM-INITIALIZED*.
85 Instead, the cross-compiler runs in a slightly flaky state
86 which is sane enough to compile SBCL itself, but which is
87 also unstable in several ways, including its inability
88 to really grok function declarations.
91 The "byte compiling top-level form:" output ought to be condensed.
92 Perhaps any number of such consecutive lines ought to turn into a
93 single "byte compiling top-level forms:" line.
96 The way that the compiler munges types with arguments together
97 with types with no arguments (in e.g. TYPE-EXPAND) leads to
98 weirdness visible to the user:
99 (DEFTYPE FOO () 'FIXNUM)
101 (TYPEP 11 '(FOO)) => T, which seems weird
102 (TYPEP 11 'FIXNUM) => T
103 (TYPEP 11 '(FIXNUM)) signals an error, as it should
104 The situation is complicated by the presence of Common Lisp types
105 like UNSIGNED-BYTE (which can either be used in list form or alone)
106 so I'm not 100% sure that the behavior above is actually illegal.
107 But I'm 90+% sure, and the following related behavior,
109 treating the bare symbol AND as equivalent to '(AND), is specifically
110 forbidden (by the ANSI specification of the AND type).
113 It would be nice if the
115 (during macroexpansion)
116 said what macroexpansion was at fault, e.g.
118 (during macroexpansion of IN-PACKAGE,
119 during macroexpansion of DEFFOO)
122 Floating point infinities are screwed up. [When I was converting CMU CL
123 to SBCL, I was looking for complexity to delete, and I thought it was safe
124 to just delete support for floating point infinities. It wasn't: they're
125 generated by the floating point hardware even when we remove support
126 for them in software. Also we claim the :IEEE-FLOATING-POINT feature,
127 and I think that means we should support infinities.-- WHN] Support
128 for them should be restored.
131 The ANSI syntax for non-STANDARD method combination types in CLOS is
132 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
133 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
134 If you mess this up, omitting the PROGN qualifier in in DEFMETHOD,
135 (DEFGENERIC FOO (X) (:METHOD-COMBINATION PROGN))
136 (DEFMETHOD FOO ((X BAR)) (PRINT 'NUMBER))
137 the error mesage is not easy to understand:
138 INVALID-METHOD-ERROR was called outside the dynamic scope
139 of a method combination function (inside the body of
140 DEFINE-METHOD-COMBINATION or a method on the generic
141 function COMPUTE-EFFECTIVE-METHOD).
142 It would be better if it were more informative, a la
143 The method combination type for this method (STANDARD) does
144 not match the method combination type for the generic function
146 Also, after you make the mistake of omitting the PROGN qualifier
147 on a DEFMETHOD, doing a new DEFMETHOD with the correct qualifier
149 (DEFMETHOD FOO PROGN ((X BAR)) (PRINT 'NUMBER))
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 This is not very helpful..
158 (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
159 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
160 (Also, when this is fixed, we can enable the code in PROCLAIM which
161 checks for incompatible FTYPE redeclarations.)
164 from DTC on the CMU CL mailing list 25 Feb 2000:
165 ;;; Compiler fails when this file is compiled.
167 ;;; Problem shows up in delete-block within ir1util.lisp. The assertion
168 ;;; (assert (member (functional-kind lambda) '(:let :mv-let :assignment)))
169 ;;; fails within bind node branch.
171 ;;; Note that if c::*check-consistency* is enabled then an un-reached
172 ;;; entry is also reported.
175 (declare (values nil))
192 (let ((ttt #'(lambda () (go cccc))))
193 (declare (special ttt))
194 (return-from bbbb nil))
197 (return-from bbbb nil))))))
200 (I *think* this is a bug. It certainly seems like strange behavior. But
201 the ANSI spec is scary, dark, and deep..)
202 (FORMAT NIL "~,1G" 1.4) => "1. "
203 (FORMAT NIL "~3,1G" 1.4) => "1. "
206 from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
208 (setf (find-class 'ccc1) (find-class 'ccc))
209 (defmethod zut ((c ccc1)) 123)
210 DTC's recommended workaround from the mailing list 3 Mar 2000:
211 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
214 The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
215 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
216 inside "~<..~>" (without the colon modifier on the closing syntax).
217 However, SBCL doesn't do this:
218 * (FORMAT T "~<munge~wegnum~>" 12)
223 When too many files are opened, OPEN will fail with an
224 uninformative error message
225 error in function OPEN: error opening #P"/tmp/foo.lisp": NIL
226 instead of saying that too many files are open.
229 Right now, when COMPILE-FILE has a read error, it actually pops
230 you into the debugger before giving up on the file. It should
231 instead handle the error, perhaps issuing (and handling)
232 a secondary error "caught ERROR: unrecoverable error during compilation"
233 and then return with FAILURE-P true,
236 reported by Sam Steingold on the cmucl-imp mailing list 12 May 2000:
238 Also, there is another bug: `array-displacement' should return an array
239 or nil as first value (as per ANSI CL), while CMUCL declares it as
240 returning an array as first value always.
243 Sometimes (SB-EXT:QUIT) fails with
244 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
245 Process inferior-lisp exited abnormally with code 1
246 I haven't noticed a repeatable case of this yet.
249 some sort of bug in inlining and RETURN-FROM in sbcl-0.6.5: Compiling
252 (BLOCK USED-BY-SOME-Y?
255 (UNLESS (REJECTED? Y)
256 (RETURN-FROM USED-BY-SOME-Y? T)))))
257 (DECLARE (INLINE FROB))
262 error in function SB-KERNEL:ASSERT-ERROR:
263 The assertion (EQ (SB-C::CONTINUATION-KIND SB-C::CONT) :BLOCK-START) failed.
264 This is still present in sbcl-0.6.8.
267 In some cases the compiler believes type declarations on array
268 elements without checking them, e.g.
269 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
272 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
273 (WHEN (CONSP (AREF X 0))
275 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
278 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
279 all cases, e.g. the type assumption *is* checked if the array
280 elements are declared to be of some structure type instead of CONS.
283 The printer doesn't report closures very well. This is true in
287 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
288 It would be nice to make closures have a settable name slot,
289 and make things like DEFSTRUCT and FLET, which create closures,
290 set helpful values into this slot.
293 And as long as we're wishing, it would be awfully nice if INSPECT could
294 also report on closures, telling about the values of the bound variables.
297 The compiler assumes that any time a function of declared FTYPE
298 doesn't signal an error, its arguments were of the declared type.
299 E.g. compiling and loading
300 (DECLAIM (OPTIMIZE (SAFETY 3)))
301 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
302 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE) FACTORIAL)))
304 (COND ((> (FACTORIAL X) 1.0E6)
305 (FORMAT T "too big~%"))
307 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
309 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
312 will cause the INTEGERP case to be selected, giving bogus output a la
314 This violates the "declarations are assertions" principle.
315 According to the ANSI spec, in the section "System Class FUNCTION",
316 this is a case of "lying to the compiler", but the lying is done
317 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
318 not by the unexpectedly general definition of FACTORIAL. In any case,
319 "declarations are assertions" means that lying to the compiler should
320 cause an error to be signalled, and should not cause a bogus
321 result to be returned. Thus, the compiler should not assume
322 that arbitrary functions check their argument types. (It might
323 make sense to add another flag (CHECKED?) to DEFKNOWN to
324 identify functions which *do* check their argument types.)
327 DEFMETHOD doesn't check the syntax of &REST argument lists properly,
328 accepting &REST even when it's not followed by an argument name:
329 (DEFMETHOD FOO ((X T) &REST) NIL)
332 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
333 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
335 (DECLARE (TYPE INDEXOID X))
336 (THE (VALUES INDEXOID)
338 where the implementation of the type check in function FOO
339 includes a full call to %TYPEP. There are also some fundamental problems
340 with the interpretation of VALUES types (inherited from CMU CL, and
341 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
342 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
345 The definitions of SIGCONTEXT-FLOAT-REGISTER and
346 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
347 supported on FreeBSD because the floating point state is not saved,
348 but at least as of FreeBSD 4.0, the floating point state *is* saved,
349 so they could be supported after all. Very likely
350 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
353 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
354 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
355 issue with 'union, 'and, 'or etc. So even though according to the
356 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
357 (and now SBCL) interpret them as legal types.
360 ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
361 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
362 list, and it is probably possible to use substantially the same
363 patches to add it to SBCL.
366 a slew of floating-point-related errors reported by Peter Van Eynde
368 a: (SQRT -9.0) fails, because SB-KERNEL::COMPLEX-SQRT is undefined.
369 Similarly, COMPLEX-ASIN, COMPLEX-ACOS, COMPLEX-ACOSH, and others
371 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
372 should probably be 1.4012985e-45. In SBCL,
373 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
374 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
375 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
376 and LEAST-NEGATIVE-LONG-FLOAT.
377 c: Many expressions generate floating infinity:
382 PVE's regression tests want them to raise errors. SBCL
383 generates the infinities instead, which may or may not be
384 conforming behavior, but then blow it by being unable to
385 output the infinities, since support for infinities is generally
386 broken, and in particular SB-IMPL::OUTPUT-FLOAT-INFINITY is
388 d: (in section12.erg) various forms a la
389 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
390 don't give the right behavior.
393 type safety errors reported by Peter Van Eynde July 25, 2000:
394 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
396 In general lengths of array type specifications aren't
397 checked by COERCE, so it fails when the spec is
398 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
399 b: CONCATENATE has the same problem of not checking the length
400 of specified output array types. MAKE-SEQUENCE and MAP and
401 MERGE also have the same problem.
402 c: (COERCE 'AND 'FUNCTION) returns something related to
403 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
404 d: ELT signals SIMPLE-ERROR if its index argument
405 isn't a valid index for its sequence argument, but should
406 signal TYPE-ERROR instead.
407 e: FILE-LENGTH is supposed to signal a type error when its
408 argument is not a stream associated with a file, but doesn't.
409 f: (FLOAT-RADIX 2/3) should signal an error instead of
411 g: (LOAD "*.lsp") should signal FILE-ERROR.
412 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
413 should signal TYPE-ERROR.
414 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
415 be used for input and output as needed. It should fail with
416 TYPE-ERROR when handed e.g. the results of
417 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
418 the inappropriate positions, but doesn't.
419 j: (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
421 should probably signal an error instead of making a pathname with
423 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
424 not a binary input stream, but instead cheerfully reads from
425 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
428 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
429 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
431 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
432 signal a PROGRAM-ERROR, and doesn't.
433 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
434 and other DEFCLASS forms with duplicate specifications in their
435 slots, should signal a PROGRAM-ERROR, and doesn't.
436 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
437 causes a COMPILER-ERROR.
440 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
441 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
442 PROGRAM-ERROR, but SBCL accepts it instead.
443 b: SYMBOL-MACROLET should refuse to bind something which is
444 declared as a global variable, signalling PROGRAM-ERROR.
445 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
446 it binds is declared SPECIAL inside.
449 LOOP bugs reported by Peter Van Eynde July 25, 2000:
450 a: (LOOP WITH (A B) DO (PRINT 1)) is a syntax error according to
451 the definition of WITH clauses given in the ANSI spec, but
452 compiles and runs happily in SBCL.
453 b: a messy one involving package iteration:
454 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<))))
455 Should be: (("blah" "blah2") ("blah2"))
456 SBCL: (("blah") ("blah2"))
457 * (LET ((X 1)) (LOOP FOR I BY (INCF X) FROM X TO 10 COLLECT I))
458 doesn't work -- SBCL's LOOP says BY isn't allowed in a FOR clause.
461 type system errors reported by Peter Van Eynde July 25, 2000:
462 a: (SUBTYPEP 'BIGNUM 'INTEGER) => NIL, NIL
463 but should be (VALUES T T) instead.
464 b: (SUBTYPEP 'EXTENDED-CHAR 'CHARACTER) => NIL, NIL
465 but should be (VALUES T T) instead.
466 c: (SUBTYPEP '(INTEGER (0) (0)) 'NIL) dies with nested errors.
467 d: In general, the system doesn't like '(INTEGER (0) (0)) -- it
468 blows up at the level of SPECIFIER-TYPE with
469 "Lower bound (0) is greater than upper bound (0)." Probably
470 SPECIFIER-TYPE should return NIL instead.
471 e: (TYPEP 0 '(COMPLEX (EQL 0)) fails with
472 "Component type for Complex is not numeric: (EQL 0)."
473 This might be easy to fix; the type system already knows
474 that (SUBTYPEP '(EQL 0) 'NUMBER) is true.
475 f: The type system doesn't know about the condition system,
476 so that e.g. (TYPEP 'SIMPLE-ERROR 'ERROR)=>NIL.
477 g: The type system isn't all that smart about relationships
478 between hairy types, as shown in the type.erg test results,
479 e.g. (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL.
482 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
484 (DEFGENERIC FOO02 (X))
485 (DEFMETHOD FOO02 ((X NUMBER)) T)
486 (LET ((M (FIND-METHOD (FUNCTION FOO02)
488 (LIST (FIND-CLASS (QUOTE NUMBER))))))
489 (REMOVE-METHOD (FUNCTION FOO02) M)
490 (DEFGENERIC FOO03 (X))
491 (ADD-METHOD (FUNCTION FOO03) M)))
492 should give an error, but SBCL allows it.
493 b: READ should probably return READER-ERROR, not the bare
494 arithmetic error, when input a la "1/0" or "1e1000" causes
498 It has been reported (e.g. by Peter Van Eynde) that there are
499 several metaobject protocol "errors". (In order to fix them, we might
500 need to document exactly what metaobject protocol specification
501 we're following -- the current code is just inherited from PCL.)
504 another error from Peter Van Eynde 5 September 2000:
505 (FORMAT NIL "~F" "FOO") should work, but instead reports an error.
506 PVE submitted a patch to deal with this bug, but it exposes other
507 comparably serious bugs, so I didn't apply it. It looks as though
508 the FORMAT code needs a fair amount of rewriting in order to comply
509 with the various details of the ANSI spec.
512 The implementation of #'+ returns its single argument without
513 type checking, e.g. (+ "illegal") => "illegal".
516 In sbcl-0.6.7, there is no doc string for CL:PUSH, probably
517 because it's defined with the DEFMACRO-MUNDANELY macro and something
518 is wrong with doc string setting in that macro.
521 Attempting to use COMPILE on something defined by DEFMACRO fails:
522 (DEFMACRO FOO (X) (CONS X X))
524 Error in function C::GET-LAMBDA-TO-COMPILE:
525 #<Closure Over Function "DEFUN (SETF MACRO-FUNCTION)" {480E21B1}> was defined in a non-null environment.
528 (SUBTYPEP '(AND ZILCH INTEGER) 'ZILCH) => NIL, NIL
529 Note: I looked into fixing this in 0.6.11.15, but gave up. The
530 problem seems to be that there are two relevant type methods for
531 the subtypep operation, HAIRY :COMPLEX-SUBTYPEP-ARG2 and
532 INTERSECTION :COMPLEX-SUBTYPEP-ARG1, and only the first is
533 called. This could be fixed, but type dispatch is messy and
534 confusing enough already, I don't want to complicate it further.
535 Perhaps someday we can make CLOS cross-compiled (instead of compiled
536 after bootstrapping) so that we don't need to have the type system
537 available before CLOS, and then we can rewrite the type methods to
538 CLOS methods, and then expressing the solutions to stuff like this
539 should become much more straightforward. -- WHN 2001-03-14
542 CL:*DEFAULT-PATHNAME-DEFAULTS* doesn't behave as ANSI suggests (reflecting
543 current working directory). And there's no supported way to update
544 or query the current working directory (a la Unix "chdir" and "pwd"),
545 which is functionality that ILISP needs (and currently gets with low-level
549 The debugger LIST-LOCATIONS command doesn't work properly.
552 Compiling and loading
553 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
555 then requesting a BACKTRACE at the debugger prompt gives no information
556 about where in the user program the problem occurred.
559 The compiler is supposed to do type inference well enough that
562 ((SIMPLE-ARRAY SINGLE-FLOAT)
564 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
567 is redundant. However, as reported by Juan Jose Garcia Ripoll for
568 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
569 workaround for the problem for now, but can't be done by the TYPECASE
570 macros themselves, since it's too hard for the macro to detect
571 assignments to the variable within the clause.
572 Note: The compiler *is* smart enough to do the type inference in
573 many cases. This case, derived from a couple of MACROEXPAND-1
574 calls on Ripoll's original test case,
576 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
577 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
578 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
579 (LET ((I 0) (G2554 LENGTH))
580 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
583 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
584 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
585 (GO SB-LOOP::NEXT-LOOP)
586 SB-LOOP::END-LOOP))))))
587 demonstrates the problem; but the problem goes away if the TAGBODY
588 and GO forms are removed (leaving the SETF in ordinary, non-looping
589 code), or if the TAGBODY and GO forms are retained, but the
590 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
593 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
594 I am looking into this problem that showed up on the cmucl-help
595 list. It seems to me that the "implementation specific environment
596 hacking functions" found in pcl/walker.lisp are completely messed
597 up. The good thing is that they appear to be barely used within
598 PCL and the munged environment object is passed to cmucl only
599 in calls to macroexpand-1, which is probably why this case fails.
600 SBCL uses essentially the same code, so if the environment hacking
601 is screwed up, it affects us too.
604 Using the pretty-printer from the command prompt gives funny
605 results, apparently because the pretty-printer doesn't know
606 about user's command input, including the user's carriage return
607 that the user, and therefore the pretty-printer thinks that
608 the new output block should start indented 2 or more characters
609 rightward of the correct location.
612 (probably related to bug #70)
613 As reported by Carl Witty on submit@bugs.debian.org 1999-05-08,
615 (in-package "CL-USER")
616 (defun equal-terms (termx termy)
618 ((alpha-equal-bound-term-lists (listx listy)
619 (or (and (null listx) (null listy))
621 (let ((bindings-x (bindings-of-bound-term (car listx)))
622 (bindings-y (bindings-of-bound-term (car listy))))
623 (if (and (null bindings-x) (null bindings-y))
624 (alpha-equal-terms (term-of-bound-term (car listx))
625 (term-of-bound-term (car listy)))
626 (and (= (length bindings-x) (length bindings-y))
628 (enter-binding-pairs (bindings-of-bound-term (car listx))
629 (bindings-of-bound-term (car listy)))
630 (alpha-equal-terms (term-of-bound-term (car listx))
631 (term-of-bound-term (car listy)))
632 (exit-binding-pairs (bindings-of-bound-term (car listx))
633 (bindings-of-bound-term (car listy)))))))
634 (alpha-equal-bound-term-lists (cdr listx) (cdr listy)))))
636 (alpha-equal-terms (termx termy)
637 (if (and (variable-p termx)
639 (equal-bindings (id-of-variable-term termx)
640 (id-of-variable-term termy))
641 (and (equal-operators-p (operator-of-term termx) (operator-of-term termy))
642 (alpha-equal-bound-term-lists (bound-terms-of-term termx)
643 (bound-terms-of-term termy))))))
647 (with-variable-invocation (alpha-equal-terms termx termy))))))
648 causes an assertion failure
649 The assertion (EQ (C::LAMBDA-TAIL-SET C::CALLER)
650 (C::LAMBDA-TAIL-SET (C::LAMBDA-HOME C::CALLEE))) failed.
652 Bob Rogers reports (1999-07-28 on cmucl-imp@cons.org) a smaller test
653 case with the same problem:
654 (defun parse-fssp-alignment ()
655 ;; Given an FSSP alignment file named by the argument . . .
656 (labels ((get-fssp-char ()
660 ;; Stub body, enough to tickle the bug.
661 (list (read-fssp-char)
665 ANSI specifies that the RESULT-TYPE argument of CONCATENATE must be
666 a subtype of SEQUENCE, but CONCATENATE doesn't check this properly:
667 (CONCATENATE 'SIMPLE-ARRAY #(1 2) '(3)) => #(1 2 3)
668 This also leads to funny behavior when derived type specifiers
669 are used, as originally reported by Milan Zamazal for CMU CL (on the
670 Debian bugs mailing list (?) 2000-02-27), then reported by Martin
671 Atzmueller for SBCL (2000-10-01 on sbcl-devel@lists.sourceforge.net):
672 (DEFTYPE FOO () 'SIMPLE-ARRAY)
673 (CONCATENATE 'FOO #(1 2) '(3))
674 => #<ARRAY-TYPE SIMPLE-ARRAY> is a bad type specifier for
676 The derived type specifier FOO should act the same way as the
677 built-in type SIMPLE-ARRAY here, but it doesn't. That problem
678 doesn't seem to exist for sequence types:
679 (DEFTYPE BAR () 'SIMPLE-VECTOR)
680 (CONCATENATE 'BAR #(1 2) '(3)) => #(1 2 3)
683 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
684 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
685 crashes SBCL. In general tracing anything which is used in the
686 implementation of TRACE is likely to have the same problem.
689 As reported by Daniel Solaz on cmucl-help@cons.org 2000-11-23,
690 SXHASH returns the same value for all non-STRUCTURE-OBJECT instances,
691 notably including all PCL instances. There's a limit to how much
692 SXHASH can do to return unique values for instances, but at least
693 it should probably look at the class name, the way that it does
694 for STRUCTURE-OBJECTs.
697 As reported by Martin Atzmueller on the sbcl-devel list 2000-11-22,
698 > There remains one issue, that is a bug in SBCL:
699 > According to my interpretation of the spec, the ":" and "@" modifiers
700 > should appear _after_ the comma-seperated arguments.
701 > Well, SBCL (and CMUCL for that matter) accept
702 > (ASSERT (STRING= (FORMAT NIL "~:8D" 1) " 1"))
703 > where the correct way (IMHO) should be
704 > (ASSERT (STRING= (FORMAT NIL "~8:D" 1) " 1"))
705 Probably SBCL should stop accepting the "~:8D"-style format arguments,
706 or at least issue a warning.
709 (probably related to bug #65)
710 The compiler doesn't like &OPTIONAL arguments in LABELS and FLET
712 (DEFUN FIND-BEFORE (ITEM SEQUENCE &KEY (TEST #'EQL))
713 (LABELS ((FIND-ITEM (OBJ SEQ TEST &OPTIONAL (VAL NIL))
714 (LET ((ITEM (FIRST SEQ)))
717 ((FUNCALL TEST OBJ ITEM)
720 (FIND-ITEM OBJ (REST SEQ) TEST (NCONC VAL `(,ITEM))))))))
721 (FIND-ITEM ITEM SEQUENCE TEST)))
722 from David Young's bug report on cmucl-help@cons.org 30 Nov 2000
723 causes sbcl-0.6.9 to fail with
724 error in function SB-KERNEL:ASSERT-ERROR:
725 The assertion (EQ (SB-C::LAMBDA-TAIL-SET SB-C::CALLER)
726 (SB-C::LAMBDA-TAIL-SET
727 (SB-C::LAMBDA-HOME SB-C::CALLEE))) failed.
730 (DECLAIM (OPTIMIZE ..)) doesn't work. E.g. even after
731 (DECLAIM (OPTIMIZE (SPEED 3))), things are still optimized with
732 the previous SPEED policy. This bug will probably get fixed in
733 0.6.9.x in a general cleanup of optimization policy.
736 (DECLAIM (OPTIMIZE ..)) doesn't work properly inside LOCALLY forms.
739 As noted in the ANSI specification for COERCE, (COERCE 3 'COMPLEX)
740 gives a result which isn't COMPLEX. The result type optimizer
741 for COERCE doesn't know this, perhaps because it was written before
742 ANSI threw this curveball: the optimizer thinks that COERCE always
743 returns a result of the specified type. Thus while the interpreted
745 (DEFUN TRICKY (X) (TYPEP (COERCE X 'COMPLEX) 'COMPLEX))
746 returns the correct result,
748 the compiled function
754 As reported by Martin Atzmueller on sbcl-devel 26 Dec 2000,
755 ANSI says that WITH-OUTPUT-TO-STRING should have a keyword
756 :ELEMENT-TYPE, but in sbcl-0.6.9 this is not defined for
757 WITH-OUTPUT-TO-STRING.
760 ANSI says in one place that type declarations can be abbreviated even
761 when the type name is not a symbol, e.g.
762 (DECLAIM ((VECTOR T) *FOOVECTOR*))
763 SBCL doesn't support this. But ANSI says in another place that this
764 isn't allowed. So it's not clear this is a bug after all. (See the
765 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
769 as pointed out by Dan Barlow on sbcl-devel 2000-07-02:
770 The PICK-TEMPORARY-FILE-NAME utility used by LOAD-FOREIGN uses
771 an easily guessable temporary filename in a way which might open
772 applications using LOAD-FOREIGN to hijacking by malicious users
773 on the same machine. Incantations for doing this safely are
774 floating around the net in various "how to write secure programs
775 despite Unix" documents, and it would be good to (1) fix this in
776 LOAD-FOREIGN, and (2) hunt for any other code which uses temporary
777 files and make it share the same new safe logic.
780 (fixed early Feb 2001 by MNA)
783 As reported by wbuss@TELDA.NET (Wolfhard Buss) on cmucl-help
786 (loop with (a . b) of-type float = '(0.0 . 1.0)
787 and (c . d) of-type float = '(2.0 . 3.0)
788 return (list a b c d))
789 should evaluate to (0.0 1.0 2.0 3.0). cmucl-18c disagrees and
790 invokes the debugger: "B is not of type list".
791 SBCL does the same thing.
794 Functions are assigned names based on the context in which they're
795 defined. This is less than ideal for the functions which are
796 used to implement CLOS methods. E.g. the output of
797 (DESCRIBE 'PRINT-OBJECT) lists functions like
798 #<FUNCTION "DEF!STRUCT (TRACE-INFO (:MAKE-LOAD-FORM-FUN SB-KERNEL:JUST-DUMP-IT-NORMALLY) (:PRINT-OBJECT #))" {1020E49}>
800 #<FUNCTION "MACROLET ((FORCE-DELAYED-DEF!METHODS NIL #))" {1242871}>
801 It would be better if these functions' names always identified
802 them as methods, and identified their generic functions and
806 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
807 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
808 the mean of the distribution can be systematically O(0.1%) wrong.
809 Just increasing R-I-E-B is probably not a good solution, since
810 it would decrease efficiency more than is probably necessary. Perhaps
811 using some sort of accept/reject method would be better.
814 (SUBTYPEP '(SATISFIES SOME-UNDEFINED-FUN) NIL)=>NIL,T (should be NIL,NIL)
817 Internally the compiler sometimes evaluates
818 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
819 (I stumbled across this when I added an
820 (assert (not (eq type1 *wild-type*)))
821 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
822 in a type context should probably be translated to T, and so it's
823 probably to ask whether it's equal to the T type and then (using the
824 EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
825 (I haven't tried to investigate this bug enough to guess whether
826 there might be any user-level symptoms.)
829 Despite what the manual says, (DECLAIM (SPEED 0)) doesn't cause
830 things to be byte compiled. This seems to be true in cmucl-2.4.19,
831 too: (COMPILE-FILE .. :BYTE-COMPILE T) causes byte-compilation,
832 but ordinary COMPILE-FILE of a file containing (DECLAIM (SPEED 0))
836 The type system doesn't understand the the intersection of the types
837 KEYWORD and (OR KEYWORD NULL) is KEYWORD, perhaps because KEYWORD
838 is itself an intersection type and that causes technical problems
839 with the simplification. Thus, the optimizer can't make some useful
840 valid type inferences.
843 a latent cross-compilation/bootstrapping bug: The cross-compilation
844 host's CL:CHAR-CODE-LIMIT is used in target code in readtable.lisp
845 and possibly elsewhere. Instead, we should use the target system's
846 CHAR-CODE-LIMIT. This will probably cause problems if we try to
847 bootstrap on a system which uses a different value of CHAR-CODE-LIMIT
851 KNOWN BUGS RELATED TO THE IR1 INTERPRETER
853 (Note: At some point, the pure interpreter (actually a semi-pure
854 interpreter aka "the IR1 interpreter") will probably go away, replaced
856 (DEFUN EVAL (X) (FUNCALL (COMPILE NIL (LAMBDA ..)))))
857 and at that time these bugs should either go away automatically or
858 become more tractable to fix. Until then, they'll probably remain,
859 since some of them aren't considered urgent, and the rest are too hard
860 to fix as long as so many special cases remain. After the IR1
861 interpreter goes away is also the preferred time to start
862 systematically exterminating cases where debugging functionality
863 (backtrace, breakpoint, etc.) breaks down, since getting rid of the
864 IR1 interpreter will reduce the number of special cases we need to
868 The FUNCTION special operator doesn't check properly whether its
869 argument is a function name. E.g. (FUNCTION (X Y)) returns a value
870 instead of failing with an error. (Later attempting to funcall the
871 value does cause an error.)
874 COMPILED-FUNCTION-P bogusly reports T for interpreted functions:
875 * (DEFUN FOO (X) (- 12 X))
877 * (COMPILED-FUNCTION-P #'FOO)
882 (DEFVAR *SUPPRESS-P* T)
883 (EVAL '(UNLESS *SUPPRESS-P*
884 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
885 (FORMAT T "surprise!"))))
886 prints "surprise!". Probably the entire EVAL-WHEN mechanism ought to be
887 rewritten from scratch to conform to the ANSI definition, abandoning
888 the *ALREADY-EVALED-THIS* hack which is used in sbcl-0.6.8.9 (and
889 in the original CMU CL source, too). This should be easier to do --
890 though still nontrivial -- once the various IR1 interpreter special
894 EVAL-WHEN's idea of what's a toplevel form is even more screwed up
895 than the example in IR1-3 would suggest, since COMPILE-FILE and
896 COMPILE both print both "right now!" messages when compiling the
900 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
901 (PRINT "yes! right now!"))
904 (EVAL-WHEN (:COMPILE-TOPLEVEL :LOAD-TOPLEVEL :EXECUTE)
905 (PRINT "no! right now!"))
907 and while EVAL doesn't print the "right now!" messages, the first
908 FUNCALL on the value returned by EVAL causes both of them to be printed.
911 The system accepts DECLAIM in most places where DECLARE would be
912 accepted, without even issuing a warning. ANSI allows this, but since
913 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
914 meaning is rather different, and it's unlikely that the user
915 has a good reason for doing DECLAIM not at top level, it would be
916 good to issue a STYLE-WARNING when this happens. A possible
917 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
918 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
919 [This is considered an IR1-interpreter-related bug because until
920 EVAL-WHEN is rewritten, which won't happen until after the IR1
921 interpreter is gone, the system's notion of what's a top-level form
922 and what's not will remain too confused to fix this problem.]