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 "compiling top-level form:" output ought to be condensed.
92 Perhaps any number of such consecutive lines ought to turn into a
93 single "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 (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
123 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
124 (Also, when this is fixed, we can enable the code in PROCLAIM which
125 checks for incompatible FTYPE redeclarations.)
128 from DTC on the CMU CL mailing list 25 Feb 2000:
129 ;;; Compiler fails when this file is compiled.
131 ;;; Problem shows up in delete-block within ir1util.lisp. The assertion
132 ;;; (assert (member (functional-kind lambda) '(:let :mv-let :assignment)))
133 ;;; fails within bind node branch.
135 ;;; Note that if c::*check-consistency* is enabled then an un-reached
136 ;;; entry is also reported.
139 (declare (values nil))
156 (let ((ttt #'(lambda () (go cccc))))
157 (declare (special ttt))
158 (return-from bbbb nil))
161 (return-from bbbb nil))))))
164 (I *think* this is a bug. It certainly seems like strange behavior. But
165 the ANSI spec is scary, dark, and deep..)
166 (FORMAT NIL "~,1G" 1.4) => "1. "
167 (FORMAT NIL "~3,1G" 1.4) => "1. "
170 from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
172 (setf (find-class 'ccc1) (find-class 'ccc))
173 (defmethod zut ((c ccc1)) 123)
174 DTC's recommended workaround from the mailing list 3 Mar 2000:
175 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
178 The ANSI spec, in section "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
179 says that an error is signalled if ~W, ~_, ~<...~:>, ~I, or ~:T is used
180 inside "~<..~>" (without the colon modifier on the closing syntax).
181 However, SBCL doesn't do this:
182 * (FORMAT T "~<munge~wegnum~>" 12)
187 Sometimes (SB-EXT:QUIT) fails with
188 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
189 Process inferior-lisp exited abnormally with code 1
190 I haven't noticed a repeatable case of this yet.
193 some sort of bug in inlining and RETURN-FROM in sbcl-0.6.5: Compiling
196 (BLOCK USED-BY-SOME-Y?
199 (UNLESS (REJECTED? Y)
200 (RETURN-FROM USED-BY-SOME-Y? T)))))
201 (DECLARE (INLINE FROB))
206 error in function SB-KERNEL:ASSERT-ERROR:
207 The assertion (EQ (SB-C::CONTINUATION-KIND SB-C::CONT) :BLOCK-START) failed.
208 This is still present in sbcl-0.6.8.
211 In some cases the compiler believes type declarations on array
212 elements without checking them, e.g.
213 (DECLAIM (OPTIMIZE (SAFETY 3) (SPEED 1) (SPACE 1)))
216 (DECLARE (TYPE (SIMPLE-ARRAY CONS 1) X))
217 (WHEN (CONSP (AREF X 0))
219 (BAR (VECTOR (MAKE-FOO :A 11 :B 12)))
222 in SBCL 0.6.5 (and also in CMU CL 18b). This does not happen for
223 all cases, e.g. the type assumption *is* checked if the array
224 elements are declared to be of some structure type instead of CONS.
227 The printer doesn't report closures very well. This is true in
231 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
232 It would be nice to make closures have a settable name slot,
233 and make things like DEFSTRUCT and FLET, which create closures,
234 set helpful values into this slot.
237 And as long as we're wishing, it would be awfully nice if INSPECT could
238 also report on closures, telling about the values of the bound variables.
241 The compiler assumes that any time a function of declared FTYPE
242 doesn't signal an error, its arguments were of the declared type.
243 E.g. compiling and loading
244 (DECLAIM (OPTIMIZE (SAFETY 3)))
245 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
247 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE)) FACTORIAL))
249 (COND ((> (FACTORIAL X) 1.0E6)
250 (FORMAT T "too big~%"))
252 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
254 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
257 will cause the INTEGERP case to be selected, giving bogus output a la
259 This violates the "declarations are assertions" principle.
260 According to the ANSI spec, in the section "System Class FUNCTION",
261 this is a case of "lying to the compiler", but the lying is done
262 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
263 not by the unexpectedly general definition of FACTORIAL. In any case,
264 "declarations are assertions" means that lying to the compiler should
265 cause an error to be signalled, and should not cause a bogus
266 result to be returned. Thus, the compiler should not assume
267 that arbitrary functions check their argument types. (It might
268 make sense to add another flag (CHECKED?) to DEFKNOWN to
269 identify functions which *do* check their argument types.)
270 (Also, verify that the compiler handles declared function
271 return types as assertions.)
274 DEFMETHOD doesn't check the syntax of &REST argument lists properly,
275 accepting &REST even when it's not followed by an argument name:
276 (DEFMETHOD FOO ((X T) &REST) NIL)
279 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
280 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
282 (DECLARE (TYPE INDEXOID X))
283 (THE (VALUES INDEXOID)
285 where the implementation of the type check in function FOO
286 includes a full call to %TYPEP. There are also some fundamental problems
287 with the interpretation of VALUES types (inherited from CMU CL, and
288 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
289 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
292 The definitions of SIGCONTEXT-FLOAT-REGISTER and
293 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
294 supported on FreeBSD because the floating point state is not saved,
295 but at least as of FreeBSD 4.0, the floating point state *is* saved,
296 so they could be supported after all. Very likely
297 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
300 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
301 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
302 issue with 'union, 'and, 'or etc. So even though according to the
303 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
304 (and now SBCL) interpret them as legal types.
307 ANSI specifies DEFINE-SYMBOL-MACRO, but it's not defined in SBCL.
308 CMU CL added it ca. Aug 13, 2000, after some discussion on the mailing
309 list, and it is probably possible to use substantially the same
310 patches to add it to SBCL.
313 a slew of floating-point-related errors reported by Peter Van Eynde
315 a: (fixed in sbcl-0.6.11.25)
316 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
317 should probably be 1.4012985e-45. In SBCL,
318 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
319 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
320 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
321 and LEAST-NEGATIVE-LONG-FLOAT.
322 c: Many expressions generate floating infinity:
327 PVE's regression tests want them to raise errors. SBCL
328 generates the infinities instead, which may or may not be
330 d: (in section12.erg) various forms a la
331 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
332 don't give the right behavior.
335 type safety errors reported by Peter Van Eynde July 25, 2000:
336 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
338 In general lengths of array type specifications aren't
339 checked by COERCE, so it fails when the spec is
340 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
341 b: CONCATENATE has the same problem of not checking the length
342 of specified output array types. MAKE-SEQUENCE and MAP and
343 MERGE also have the same problem.
344 c: (COERCE 'AND 'FUNCTION) returns something related to
345 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
346 f: (FLOAT-RADIX 2/3) should signal an error instead of
348 g: (LOAD "*.lsp") should signal FILE-ERROR.
349 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
350 should signal TYPE-ERROR.
351 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
352 be used for input and output as needed. It should fail with
353 TYPE-ERROR when handed e.g. the results of
354 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
355 the inappropriate positions, but doesn't.
356 j: (PARSE-NAMESTRING (COERCE (LIST #\f #\o #\o (CODE-CHAR 0) #\4 #\8)
358 should probably signal an error instead of making a pathname with
360 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
361 not a binary input stream, but instead cheerfully reads from
362 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
365 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
366 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
368 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
369 signal a PROGRAM-ERROR, and doesn't.
370 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
371 and other DEFCLASS forms with duplicate specifications in their
372 slots, should signal a PROGRAM-ERROR, and doesn't.
373 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
374 causes a COMPILER-ERROR.
377 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
378 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
379 PROGRAM-ERROR, but SBCL accepts it instead.
380 b: SYMBOL-MACROLET should refuse to bind something which is
381 declared as a global variable, signalling PROGRAM-ERROR.
382 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
383 it binds is declared SPECIAL inside.
386 type system errors reported by Peter Van Eynde July 25, 2000:
387 c: (SUBTYPEP '(INTEGER (0) (0)) 'NIL) dies with nested errors.
388 d: In general, the system doesn't like '(INTEGER (0) (0)) -- it
389 blows up at the level of SPECIFIER-TYPE with
390 "Lower bound (0) is greater than upper bound (0)." Probably
391 SPECIFIER-TYPE should return NIL instead.
392 e: (TYPEP 0 '(COMPLEX (EQL 0)) fails with
393 "Component type for Complex is not numeric: (EQL 0)."
394 This might be easy to fix; the type system already knows
395 that (SUBTYPEP '(EQL 0) 'NUMBER) is true.
396 g: The type system isn't all that smart about relationships
397 between hairy types, as shown in the type.erg test results,
398 e.g. (SUBTYPEP 'CONS '(NOT ATOM)) => NIL, NIL.
401 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
403 (DEFGENERIC FOO02 (X))
404 (DEFMETHOD FOO02 ((X NUMBER)) T)
405 (LET ((M (FIND-METHOD (FUNCTION FOO02)
407 (LIST (FIND-CLASS (QUOTE NUMBER))))))
408 (REMOVE-METHOD (FUNCTION FOO02) M)
409 (DEFGENERIC FOO03 (X))
410 (ADD-METHOD (FUNCTION FOO03) M)))
411 should give an error, but SBCL allows it.
412 b: READ should probably return READER-ERROR, not the bare
413 arithmetic error, when input a la "1/0" or "1e1000" causes
417 It has been reported (e.g. by Peter Van Eynde) that there are
418 several metaobject protocol "errors". (In order to fix them, we might
419 need to document exactly what metaobject protocol specification
420 we're following -- the current code is just inherited from PCL.)
423 The implementation of #'+ returns its single argument without
424 type checking, e.g. (+ "illegal") => "illegal".
427 Attempting to use COMPILE on something defined by DEFMACRO fails:
428 (DEFMACRO FOO (X) (CONS X X))
430 Error in function C::GET-LAMBDA-TO-COMPILE:
431 #<Closure Over Function "DEFUN (SETF MACRO-FUNCTION)" {480E21B1}> was defined in a non-null environment.
434 (SUBTYPEP '(AND ZILCH INTEGER) 'ZILCH) => NIL, NIL
435 Note: I looked into fixing this in 0.6.11.15, but gave up. The
436 problem seems to be that there are two relevant type methods for
437 the subtypep operation, HAIRY :COMPLEX-SUBTYPEP-ARG2 and
438 INTERSECTION :COMPLEX-SUBTYPEP-ARG1, and only the first is
439 called. This could be fixed, but type dispatch is messy and
440 confusing enough already, I don't want to complicate it further.
441 Perhaps someday we can make CLOS cross-compiled (instead of compiled
442 after bootstrapping) so that we don't need to have the type system
443 available before CLOS, and then we can rewrite the type methods to
444 CLOS methods, and then expressing the solutions to stuff like this
445 should become much more straightforward. -- WHN 2001-03-14
448 The debugger LIST-LOCATIONS command doesn't work properly.
451 Compiling and loading
452 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
454 then requesting a BACKTRACE at the debugger prompt gives no information
455 about where in the user program the problem occurred.
458 The compiler is supposed to do type inference well enough that
461 ((SIMPLE-ARRAY SINGLE-FLOAT)
463 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
466 is redundant. However, as reported by Juan Jose Garcia Ripoll for
467 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
468 workaround for the problem for now, but can't be done by the TYPECASE
469 macros themselves, since it's too hard for the macro to detect
470 assignments to the variable within the clause.
471 Note: The compiler *is* smart enough to do the type inference in
472 many cases. This case, derived from a couple of MACROEXPAND-1
473 calls on Ripoll's original test case,
475 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
476 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
477 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
478 (LET ((I 0) (G2554 LENGTH))
479 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
482 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
483 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
484 (GO SB-LOOP::NEXT-LOOP)
485 SB-LOOP::END-LOOP))))))
486 demonstrates the problem; but the problem goes away if the TAGBODY
487 and GO forms are removed (leaving the SETF in ordinary, non-looping
488 code), or if the TAGBODY and GO forms are retained, but the
489 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
492 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
493 I am looking into this problem that showed up on the cmucl-help
494 list. It seems to me that the "implementation specific environment
495 hacking functions" found in pcl/walker.lisp are completely messed
496 up. The good thing is that they appear to be barely used within
497 PCL and the munged environment object is passed to cmucl only
498 in calls to macroexpand-1, which is probably why this case fails.
499 SBCL uses essentially the same code, so if the environment hacking
500 is screwed up, it affects us too.
503 Using the pretty-printer from the command prompt gives funny
504 results, apparently because the pretty-printer doesn't know
505 about user's command input, including the user's carriage return
506 that the user, and therefore the pretty-printer thinks that
507 the new output block should start indented 2 or more characters
508 rightward of the correct location.
511 (probably related to bug #70; maybe related to bug #109)
512 As reported by Carl Witty on submit@bugs.debian.org 1999-05-08,
514 (in-package "CL-USER")
515 (defun equal-terms (termx termy)
517 ((alpha-equal-bound-term-lists (listx listy)
518 (or (and (null listx) (null listy))
520 (let ((bindings-x (bindings-of-bound-term (car listx)))
521 (bindings-y (bindings-of-bound-term (car listy))))
522 (if (and (null bindings-x) (null bindings-y))
523 (alpha-equal-terms (term-of-bound-term (car listx))
524 (term-of-bound-term (car listy)))
525 (and (= (length bindings-x) (length bindings-y))
527 (enter-binding-pairs (bindings-of-bound-term (car listx))
528 (bindings-of-bound-term (car listy)))
529 (alpha-equal-terms (term-of-bound-term (car listx))
530 (term-of-bound-term (car listy)))
531 (exit-binding-pairs (bindings-of-bound-term (car listx))
532 (bindings-of-bound-term (car listy)))))))
533 (alpha-equal-bound-term-lists (cdr listx) (cdr listy)))))
535 (alpha-equal-terms (termx termy)
536 (if (and (variable-p termx)
538 (equal-bindings (id-of-variable-term termx)
539 (id-of-variable-term termy))
540 (and (equal-operators-p (operator-of-term termx) (operator-of-term termy))
541 (alpha-equal-bound-term-lists (bound-terms-of-term termx)
542 (bound-terms-of-term termy))))))
546 (with-variable-invocation (alpha-equal-terms termx termy))))))
547 causes an assertion failure
548 The assertion (EQ (C::LAMBDA-TAIL-SET C::CALLER)
549 (C::LAMBDA-TAIL-SET (C::LAMBDA-HOME C::CALLEE))) failed.
551 Bob Rogers reports (1999-07-28 on cmucl-imp@cons.org) a smaller test
552 case with the same problem:
553 (defun parse-fssp-alignment ()
554 ;; Given an FSSP alignment file named by the argument . . .
555 (labels ((get-fssp-char ()
559 ;; Stub body, enough to tickle the bug.
560 (list (read-fssp-char)
564 ANSI specifies that the RESULT-TYPE argument of CONCATENATE must be
565 a subtype of SEQUENCE, but CONCATENATE doesn't check this properly:
566 (CONCATENATE 'SIMPLE-ARRAY #(1 2) '(3)) => #(1 2 3)
567 This also leads to funny behavior when derived type specifiers
568 are used, as originally reported by Milan Zamazal for CMU CL (on the
569 Debian bugs mailing list (?) 2000-02-27), then reported by Martin
570 Atzmueller for SBCL (2000-10-01 on sbcl-devel@lists.sourceforge.net):
571 (DEFTYPE FOO () 'SIMPLE-ARRAY)
572 (CONCATENATE 'FOO #(1 2) '(3))
573 => #<ARRAY-TYPE SIMPLE-ARRAY> is a bad type specifier for
575 The derived type specifier FOO should act the same way as the
576 built-in type SIMPLE-ARRAY here, but it doesn't. That problem
577 doesn't seem to exist for sequence types:
578 (DEFTYPE BAR () 'SIMPLE-VECTOR)
579 (CONCATENATE 'BAR #(1 2) '(3)) => #(1 2 3)
582 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
583 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
584 crashes SBCL. In general tracing anything which is used in the
585 implementation of TRACE is likely to have the same problem.
588 As reported by Daniel Solaz on cmucl-help@cons.org 2000-11-23,
589 SXHASH returns the same value for all non-STRUCTURE-OBJECT instances,
590 notably including all PCL instances. There's a limit to how much
591 SXHASH can do to return unique values for instances, but at least
592 it should probably look at the class name, the way that it does
593 for STRUCTURE-OBJECTs.
596 As reported by Martin Atzmueller on the sbcl-devel list 2000-11-22,
597 > There remains one issue, that is a bug in SBCL:
598 > According to my interpretation of the spec, the ":" and "@" modifiers
599 > should appear _after_ the comma-seperated arguments.
600 > Well, SBCL (and CMUCL for that matter) accept
601 > (ASSERT (STRING= (FORMAT NIL "~:8D" 1) " 1"))
602 > where the correct way (IMHO) should be
603 > (ASSERT (STRING= (FORMAT NIL "~8:D" 1) " 1"))
604 Probably SBCL should stop accepting the "~:8D"-style format arguments,
605 or at least issue a warning.
608 (probably related to bug #65; maybe related to bug #109)
609 The compiler doesn't like &OPTIONAL arguments in LABELS and FLET
611 (DEFUN FIND-BEFORE (ITEM SEQUENCE &KEY (TEST #'EQL))
612 (LABELS ((FIND-ITEM (OBJ SEQ TEST &OPTIONAL (VAL NIL))
613 (LET ((ITEM (FIRST SEQ)))
616 ((FUNCALL TEST OBJ ITEM)
619 (FIND-ITEM OBJ (REST SEQ) TEST (NCONC VAL `(,ITEM))))))))
620 (FIND-ITEM ITEM SEQUENCE TEST)))
621 from David Young's bug report on cmucl-help@cons.org 30 Nov 2000
622 causes sbcl-0.6.9 to fail with
623 error in function SB-KERNEL:ASSERT-ERROR:
624 The assertion (EQ (SB-C::LAMBDA-TAIL-SET SB-C::CALLER)
625 (SB-C::LAMBDA-TAIL-SET
626 (SB-C::LAMBDA-HOME SB-C::CALLEE))) failed.
629 (DECLAIM (OPTIMIZE ..)) doesn't work. E.g. even after
630 (DECLAIM (OPTIMIZE (SPEED 3))), things are still optimized with
631 the previous SPEED policy. This bug will probably get fixed in
632 0.6.9.x in a general cleanup of optimization policy.
635 (DECLAIM (OPTIMIZE ..)) doesn't work properly inside LOCALLY forms.
638 As reported by Martin Atzmueller on sbcl-devel 26 Dec 2000,
639 ANSI says that WITH-OUTPUT-TO-STRING should have a keyword
640 :ELEMENT-TYPE, but in sbcl-0.6.9 this is not defined for
641 WITH-OUTPUT-TO-STRING.
644 ANSI says in one place that type declarations can be abbreviated even
645 when the type name is not a symbol, e.g.
646 (DECLAIM ((VECTOR T) *FOOVECTOR*))
647 SBCL doesn't support this. But ANSI says in another place that this
648 isn't allowed. So it's not clear this is a bug after all. (See the
649 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
653 as pointed out by Dan Barlow on sbcl-devel 2000-07-02:
654 The PICK-TEMPORARY-FILE-NAME utility used by LOAD-FOREIGN uses
655 an easily guessable temporary filename in a way which might open
656 applications using LOAD-FOREIGN to hijacking by malicious users
657 on the same machine. Incantations for doing this safely are
658 floating around the net in various "how to write secure programs
659 despite Unix" documents, and it would be good to (1) fix this in
660 LOAD-FOREIGN, and (2) hunt for any other code which uses temporary
661 files and make it share the same new safe logic.
664 Functions are assigned names based on the context in which they're
665 defined. This is less than ideal for the functions which are
666 used to implement CLOS methods. E.g. the output of
667 (DESCRIBE 'PRINT-OBJECT) lists functions like
668 #<FUNCTION "DEF!STRUCT (TRACE-INFO (:MAKE-LOAD-FORM-FUN SB-KERNEL:JUST-DUMP-IT-NORMALLY) (:PRINT-OBJECT #))" {1020E49}>
670 #<FUNCTION "MACROLET ((FORCE-DELAYED-DEF!METHODS NIL #))" {1242871}>
671 It would be better if these functions' names always identified
672 them as methods, and identified their generic functions and
676 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
677 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
678 the mean of the distribution can be systematically O(0.1%) wrong.
679 Just increasing R-I-E-B is probably not a good solution, since
680 it would decrease efficiency more than is probably necessary. Perhaps
681 using some sort of accept/reject method would be better.
684 (SUBTYPEP '(SATISFIES SOME-UNDEFINED-FUN) NIL)=>NIL,T (should be NIL,NIL)
687 Internally the compiler sometimes evaluates
688 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
689 (I stumbled across this when I added an
690 (assert (not (eq type1 *wild-type*)))
691 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
692 in a type context should probably be translated to T, and so it's
693 probably to ask whether it's equal to the T type and then (using the
694 EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
695 (I haven't tried to investigate this bug enough to guess whether
696 there might be any user-level symptoms.)
699 a latent cross-compilation/bootstrapping bug: The cross-compilation
700 host's CL:CHAR-CODE-LIMIT is used in target code in readtable.lisp
701 and possibly elsewhere. Instead, we should use the target system's
702 CHAR-CODE-LIMIT. This will probably cause problems if we try to
703 bootstrap on a system which uses a different value of CHAR-CODE-LIMIT
707 (subtypep '(or (integer -1 1)
711 (integer -1 1))) => NIL,T
712 An analogous problem with SINGLE-FLOAT and REAL types was fixed in
713 sbcl-0.6.11.22, but some peculiarites of the RATIO type make it
714 awkward to generalize the fix to INTEGER and RATIONAL. It's not
715 clear what's the best fix. (See the "bug in type handling" discussion
716 on cmucl-imp ca. 2001-03-22 and ca. 2001-02-12.)
719 In sbcl-0.6.11.26, (COMPILE 'IN-HOST-COMPILATION-MODE) in
720 src/cold/shared.lisp doesn't correctly translate the
722 (defun in-host-compilation-mode (fn)
723 (let ((*features* (cons :sb-xc-host *features*))
724 ;; the CROSS-FLOAT-INFINITY-KLUDGE, as documented in
725 ;; base-target-features.lisp-expr:
726 (*shebang-features* (set-difference *shebang-features*
727 '(:sb-propagate-float-type
728 :sb-propagate-fun-type))))
729 (with-additional-nickname ("SB-XC" "SB!XC")
731 No error is reported by the compiler, but when the function is executed,
733 TYPE-ERROR in SB-KERNEL::OBJECT-NOT-TYPE-ERROR-HANDLER:
734 (:LINUX :X86 :IEEE-FLOATING-POINT :SB-CONSTRAIN-FLOAT-TYPE :SB-TEST
735 :SB-INTERPRETER :SB-DOC :UNIX ...) is not of type SYMBOL.
738 Inconsistencies between derived and declared VALUES return types for
739 DEFUN aren't checked very well. E.g. the logic which successfully
740 catches problems like
741 (declaim (ftype (function (fixnum) float) foo))
743 (declare (type integer x))
744 (values x)) ; wrong return type, detected, gives warning, good!
746 (declaim (ftype (function (t) (values t t)) bar))
748 (values x)) ; wrong number of return values, no warning, bad!
749 The cause of this is seems to be that (1) the internal function
750 VALUES-TYPES-EQUAL-OR-INTERSECT used to make the check handles its
751 arguments symmetrically, and (2) when the type checking code was
752 written back when when SBCL's code was still CMU CL, the intent
754 (declaim (ftype (function (t) t) bar))
756 (values x x)) ; wrong number of return values; should give warning?
757 not be warned for, because a two-valued return value is considered
758 to be compatible with callers who expects a single value to be
759 returned. That intent is probably not appropriate for modern ANSI
760 Common Lisp, but fixing this might be complicated because of other
761 divergences between auld-style and new-style handling of
762 multiple-VALUES types. (Some issues related to this were discussed
763 on cmucl-imp at some length sometime in 2000.)
766 The facility for dumping a running Lisp image to disk gets confused
767 when run without the PURIFY option, and creates an unnecessarily large
768 core file (apparently representing memory usage up to the previous
769 high-water mark). Moreover, when the file is loaded, it confuses the
770 GC, so that thereafter memory usage can never be reduced below that
774 The TRACE facility can't be used on some kinds of functions.
775 (Basically, the breakpoint facility was incompletely implemented
776 in the X86 port of CMU CL, and hasn't been fixed in SBCL.)
779 In sbcl-0.6.11.41 (and in all earlier SBCL, and in CMU
780 CL), out-of-line structure slot setters are horribly inefficient
781 whenever the type of the slot is declared, because out-of-line
782 structure slot setters are implemented as closures to save space,
783 so the compiler doesn't compile the type test into code, but
784 instead just saves the type in a lexical closure and interprets it
786 A proper solution involves deciding whether it's really worth
787 saving space by implementing structure slot accessors as closures.
788 (If it's not worth it, the problem vanishes automatically. If it
789 is worth it, there are hacks we could use to force type tests to
790 be compiled anyway, and even shared. E.g. we could implement
791 an EQUAL hash table mapping from types to compiled type tests,
792 and save the appropriate compiled type test as part of each lexical
793 closure; or we could make the lexical closures be placeholders
794 which overwrite their old definition as a lexical closure with
795 a new compiled definition the first time that they're called.)
796 As a workaround for the problem, #'(SETF FOO) expressions can
797 be replaced with (EFFICIENT-SETF-FUNCTION FOO), where
798 (defmacro efficient-setf-function (place-function-name)
799 (or #+sbcl (and (sb-impl::info :function :accessor-for place-function-name)
800 ;; a workaround for the problem, encouraging the
801 ;; inline expansion of the structure accessor, so
802 ;; that the compiler can optimize its type test
803 (let ((new-value (gensym "NEW-VALUE-"))
804 (structure-value (gensym "STRUCTURE-VALUE-")))
805 `(lambda (,new-value ,structure-value)
806 (setf (,place-function-name ,structure-value)
808 ;; no problem, can just use the ordinary expansion
809 `(function (setf ,place-function-name))))
812 DESCRIBE interacts poorly with *PRINT-CIRCLE*, e.g. the output from
813 (let ((*print-circle* t)) (describe (make-hash-table)))
815 #<HASH-TABLE :TEST EQL :COUNT 0 {90BBFC5}> is an . (EQL)
817 Its REHASH-SIZE is 1.5. Its REHASH-THRESHOLD is . (1.0)
818 It holds 0 key/value pairs.
819 where the ". (EQL)" and ". (1.0)" substrings are screwups.
820 (This is likely a pretty-printer problem which happens to
821 be exercised by DESCRIBE, not actually a DESCRIBE problem.)
824 There's apparently a bug in CEILING optimization which caused
825 Douglas Crosher to patch the CMU CL version. Martin Atzmueller
826 applied the patches to SBCL and they didn't seem to cause problems
827 (as reported sbcl-devel 2001-05-04). However, since the patches
828 modify nontrivial code which was apparently written incorrectly
829 the first time around, until regression tests are written I'm not
830 comfortable merging the patches in the CVS version of SBCL.
833 As reported by Arthur Lemmens sbcl-devel 2001-05-05, ANSI
834 requires that SYMBOL-MACROLET refuse to rebind special variables,
835 but SBCL doesn't do this. (Also as reported by AL in the same
836 message, SBCL depended on this nonconforming behavior to build
837 itself, because of the way that **CURRENT-SEGMENT** was implemented.
838 As of sbcl-0.6.12.x, this dependence on the nonconforming behavior
839 has been fixed, but the nonconforming behavior remains.)
842 (DESCRIBE 'SB-ALIEN:DEF-ALIEN-TYPE) reports the macro argument list
846 in #<PACKAGE "SB-ALIEN">.
847 Macro-function: #<FUNCTION "DEF!MACRO DEF-ALIEN-TYPE" {19F4A39}>
848 Macro arguments: (#:whole-470 #:environment-471)
849 On Sat, May 26, 2001 09:45:57 AM CDT it was compiled from:
850 /usr/stuff/sbcl/src/code/host-alieneval.lisp
851 Created: Monday, March 12, 2001 07:47:43 AM CST
854 (DESCRIBE 'STREAM-READ-BYTE)
857 (reported by Eric Marsden on cmucl-imp 2001-06-15)
858 (and APD pointed out on sbcl-devel 2001-12-29 that it's the same
862 (TIME (ROOM T)) reports more than 200 Mbytes consed even for
863 a clean, just-started SBCL system. And it seems to be right:
864 (ROOM T) can bring a small computer to its knees for a *long*
865 time trying to GC afterwards. Surely there's some more economical
866 way to implement (ROOM T).
869 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
871 ;;; This file fails to compile.
872 ;;; Maybe this bug is related to bugs #65, #70 in the BUGS file.
873 (in-package :cl-user)
879 ;; Uncomment and it works
882 In SBCL 0.6.12.42, the problem is
883 internal error, failed AVER:
884 "(COMMON-LISP:EQ (SB!C::LAMBDA-TAIL-SET SB!C::CALLER)
885 (SB!C::LAMBDA-TAIL-SET (SB!C::LAMBDA-HOME SB!C::CALLEE)))"
888 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
890 ;;; The compiler is flushing the argument type test, and the default
891 ;;; case in the cond, so that calling with say a fixnum 0 causes a
893 (declaim (optimize (safety 2) (speed 3)))
895 (declare (type (or string stream) x))
896 (cond ((typep x 'string) 'string)
897 ((typep x 'stream) 'stream)
900 The symptom in sbcl-0.6.12.42 on OpenBSD is actually (TST 0)=>STREAM
901 (not the SIGBUS reported in the comment) but that's broken too;
902 type declarations are supposed to be treated as assertions unless
903 SAFETY 0, so we should be getting a TYPE-ERROR.
906 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
908 (in-package :cl-user)
909 ;;; Produces an assertion failures when compiled.
911 (declare (type (or (function (t) t) null) z))
912 (let ((z (or z #'identity)))
913 (declare (type (function (t) t) z))
915 The error in sbcl-0.6.12.42 is
916 internal error, failed AVER:
917 "(COMMON-LISP:NOT (COMMON-LISP:EQ SB!C::CHECK COMMON-LISP:T))"
920 reported by Martin Atzmueller 2001-06-25; taken from CMU CL bugs
921 collection; apparently originally reported by Bruno Haible
922 (in-package :cl-user)
923 ;;; From: Bruno Haible
924 ;;; Subject: scope of SPECIAL declarations
925 ;;; It seems CMUCL has a bug relating to the scope of SPECIAL
926 ;;; declarations. I observe this with "CMU Common Lisp 18a x86-linux
929 (declare (special x))
932 (declare (special x)) y)))
933 ;;; Gives: 0 (this should return 1 according to CLHS)
935 (declare (special x))
938 (declare (special x)) y)))
939 ;;; Gives: 1 (correct).
940 The reported results match what we get from the interpreter
944 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
946 (in-package :cl-user)
947 ;;; From: David Gadbois <gadbois@cyc.com>
949 ;;; Logical pathnames aren't externalizable.
951 (let ((tempfile "/tmp/test.lisp"))
952 (setf (logical-pathname-translations "XXX")
953 '(("XXX:**;*.*" "/tmp/**/*.*")))
954 (with-open-file (out tempfile :direction :output)
955 (write-string "(defvar *path* #P\"XXX:XXX;FOO.LISP\")" out))
956 (compile-file tempfile))
957 The error message in sbcl-0.6.12.42 is
959 ; (while making load form for #<SB-IMPL::LOGICAL-HOST "XXX">)
960 ; A logical host can't be dumped as a constant: #<SB-IMPL::LOGICAL-HOST "XXX">
963 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
965 (in-package :cl-user)
966 ;;; The following invokes a compiler error.
967 (declaim (optimize (speed 2) (debug 3)))
970 (unwind-protect nil)))
974 The error message in sbcl-0.6.12.42 is
975 internal error, failed AVER:
976 "(COMMON-LISP:EQ (SB!C::TN-ENVIRONMENT SB!C:TN) SB!C::TN-ENV)"
979 When the compiler inline expands functions, it may be that different
980 kinds of return values are generated from different code branches.
981 E.g. an inline expansion of POSITION generates integer results
982 from one branch, and NIL results from another. When that inline
983 expansion is used in a context where only one of those results
986 (aref *a1* (position x *a2*)))
987 and the compiler can't prove that the unacceptable branch is
988 never taken, then bogus type mismatch warnings can be generated.
989 If you need to suppress the type mismatch warnings, you can
990 suppress the inline expansion,
992 #+sbcl (declare (notinline position)) ; to suppress bug 117 bogowarnings
993 (aref *a1* (position x *a2*)))
994 or, sometimes, suppress them by declaring the result to be of an
997 (aref *a1* (the integer (position x *a2*))))
999 This is not a new compiler problem in 0.7.0, but the new compiler
1000 transforms for FIND, POSITION, FIND-IF, and POSITION-IF make it
1001 more conspicuous. If you don't need performance from these functions,
1002 and the bogus warnings are a nuisance for you, you can return to
1003 your pre-0.7.0 state of grace with
1004 #+sbcl (declaim (notinline find position find-if position-if)) ; bug 117..
1007 as reported by Eric Marsden on cmucl-imp@cons.org 2001-08-14:
1008 (= (FLOAT 1 DOUBLE-FLOAT-EPSILON)
1009 (+ (FLOAT 1 DOUBLE-FLOAT-EPSILON) DOUBLE-FLOAT-EPSILON)) => T
1010 when of course it should be NIL. (He says it only fails for X86,
1011 not SPARC; dunno about Alpha.)
1013 Also, "the same problem exists for LONG-FLOAT-EPSILON,
1014 DOUBLE-FLOAT-NEGATIVE-EPSILON, LONG-FLOAT-NEGATIVE-EPSILON (though
1015 for the -negative- the + is replaced by a - in the test)."
1017 Raymond Toy comments that this is tricky on the X86 since its FPU
1018 uses 80-bit precision internally.
1021 The compiler incorrectly figures the return type of
1022 (DEFUN FOO (FRAME UP-FRAME)
1029 This problem exists in CMU CL 18c too. When I reported it on
1030 cmucl-imp@cons.org, Raymond Toy replied 23 Aug 2001 with
1031 a partial explanation, but no fix has been found yet.
1034 Even in sbcl-0.pre7.x, which is supposed to be free of the old
1035 non-ANSI behavior of treating the function return type inferred
1036 from the current function definition as a declaration of the
1037 return type from any function of that name, the return type of NIL
1038 is attached to FOO in 120a above, and used to optimize code which
1042 There was some sort of screwup in handling of
1043 (IF (NOT (IGNORE-ERRORS ..))). E.g.
1045 (if (not (ignore-errors
1046 (make-pathname :host "foo" :directory "!bla" :name "bar")))
1048 (error "notunlessnot")))
1049 The (NOT (IGNORE-ERRORS ..)) form evaluates to T, so this should be
1050 printing "ok", but instead it's going to the ERROR. This problem
1051 seems to've been introduced by MNA's HANDLER-CASE patch (sbcl-devel
1052 2001-07-17) and as a workaround (put in sbcl-0.pre7.14.flaky4.12)
1053 I reverted back to the old weird HANDLER-CASE code. However, I
1054 think the problem looks like a compiler bug in handling RETURN-FROM,
1055 so I left the MNA-patched code in HANDLER-CASE (suppressed with
1056 #+NIL) and I'd like to go back to see whether this really is
1057 a compiler bug before I delete this BUGS entry.
1060 The *USE-IMPLEMENTATION-TYPES* hack causes bugs, particularly
1061 (IN-PACKAGE :SB-KERNEL)
1062 (TYPE= (SPECIFIER-TYPE '(VECTOR T))
1063 (SPECIFIER-TYPE '(VECTOR UNDEFTYPE)))
1064 Then because of this, the compiler bogusly optimizes
1065 (TYPEP #(11) '(SIMPLE-ARRAY UNDEF-TYPE 1))
1066 to T. Unfortunately, just setting *USE-IMPLEMENTATION-TYPES* to
1067 NIL around sbcl-0.pre7.14.flaky4.12 didn't work: the compiler complained
1068 about type mismatches (probably harmlessly, another instance of bug 117);
1069 and then cold init died with a segmentation fault.
1072 As of version 0.pre7.14, SBCL's implementation of MACROLET makes
1073 the entire lexical environment at the point of MACROLET available
1074 in the bodies of the macroexpander functions. In particular, it
1075 allows the function bodies (which run at compile time) to try to
1076 access lexical variables (which are only defined at runtime).
1077 It doesn't even issue a warning, which is bad.
1079 The SBCL behavior arguably conforms to the ANSI spec (since the
1080 spec says that the behavior is undefined, ergo anything conforms).
1081 However, it would be better to issue a compile-time error.
1082 Unfortunately I (WHN) don't see any simple way to detect this
1083 condition in order to issue such an error, so for the meantime
1084 SBCL just does this weird broken "conforming" thing.
1086 The ANSI standard says, in the definition of the special operator
1088 The macro-expansion functions defined by MACROLET are defined
1089 in the lexical environment in which the MACROLET form appears.
1090 Declarations and MACROLET and SYMBOL-MACROLET definitions affect
1091 the local macro definitions in a MACROLET, but the consequences
1092 are undefined if the local macro definitions reference any
1093 local variable or function bindings that are visible in that
1094 lexical environment.
1095 Then it seems to contradict itself by giving the example
1097 (macrolet ((fudge (z)
1098 ;The parameters x and flag are not accessible
1099 ; at this point; a reference to flag would be to
1100 ; the global variable of that name.
1101 ` (if flag (* ,z ,z) ,z)))
1102 ;The parameters x and flag are accessible here.
1106 The comment "a reference to flag would be to the global variable
1107 of the same name" sounds like good behavior for the system to have.
1108 but actual specification quoted above says that the actual behavior
1112 (as reported by Gabe Garza on cmucl-help 2001-09-21)
1114 (defun test-pred (x y)
1118 (func (lambda () x)))
1119 (print (eq func func))
1120 (print (test-pred func func))
1121 (delete func (list func))))
1122 Now calling (TEST-CASE) gives output
1125 (#<FUNCTION {500A9EF9}>)
1126 Evidently Python thinks of the lambda as a code transformation so
1127 much that it forgets that it's also an object.
1130 (fixed in 0.pre7.41)
1133 The DEFSTRUCT section of the ANSI spec, in the :CONC-NAME section,
1134 specifies a precedence rule for name collisions between slot accessors of
1135 structure classes related by inheritance. As of 0.7.0, SBCL still
1139 READ-SEQUENCE doesn't work for Gray streams. (reported by Nathan
1140 Froyd sbcl-devel 2001-10-15) As per subsequent discussion on the
1141 list, the Gray streams proposal doesn't mention READ-SEQUENCE and
1142 WRITE-SEQUENCE because it predates them, generalizing it to
1143 cover them is an obvious extension, ACL does it, and there's a
1144 patch for for CMU CL which does it too.
1147 insufficient syntax checking in MACROLET:
1149 (macrolet ((defmacro bar (z) `(+ z z)))
1151 shouldn't compile without error (because of the extra DEFMACRO symbol).
1154 reported by Alexey Dejneka on sbcl-devel 2001-11-03
1156 "Return X if X is a non-negative integer."
1157 (let ((step (lambda (%funcall)
1160 (t (1+ (funcall %funcall (1- n)))))))))
1163 (funcall step (lambda (n)
1164 (funcall (funcall a a) n))))
1166 (funcall step (lambda (n)
1167 (funcall (funcall a a) n)))))
1169 This function returns its argument. But after removing percents it
1170 does not work: "Result of (1- n) is not a function".
1173 As of sbcl-0.pre7.86.flaky7.3, the cross-compiler, and probably
1174 the CL:COMPILE function (which is based on the same %COMPILE
1175 mechanism) get confused by
1177 (labels ((sxhash-number (x)
1179 (fixnum (sxhash x)) ; through DEFTRANSFORM
1180 (integer (sb!bignum:sxhash-bignum x))
1181 (single-float (sxhash x)) ; through DEFTRANSFORM
1182 (double-float (sxhash x)) ; through DEFTRANSFORM
1183 #!+long-float (long-float (error "stub: no LONG-FLOAT"))
1184 (ratio (let ((result 127810327))
1185 (declare (type fixnum result))
1186 (mixf result (sxhash-number (numerator x)))
1187 (mixf result (sxhash-number (denominator x)))
1189 (complex (let ((result 535698211))
1190 (declare (type fixnum result))
1191 (mixf result (sxhash-number (realpart x)))
1192 (mixf result (sxhash-number (imagpart x)))
1194 (sxhash-recurse (x &optional (depthoid +max-hash-depthoid+))
1195 (declare (type index depthoid))
1198 (if (plusp depthoid)
1199 (mix (sxhash-recurse (car x) (1- depthoid))
1200 (sxhash-recurse (cdr x) (1- depthoid)))
1203 (if (typep x 'structure-object)
1205 (sxhash ; through DEFTRANSFORM
1206 (class-name (layout-class (%instance-layout x)))))
1208 (symbol (sxhash x)) ; through DEFTRANSFORM
1209 (number (sxhash-number x))
1212 (simple-string (sxhash x)) ; through DEFTRANSFORM
1213 (string (%sxhash-substring x))
1214 (bit-vector (let ((result 410823708))
1215 (declare (type fixnum result))
1216 (dotimes (i (min depthoid (length x)))
1217 (mixf result (aref x i)))
1219 (t (logxor 191020317 (sxhash (array-rank x))))))
1222 (sxhash (char-code x)))) ; through DEFTRANSFORM
1224 (sxhash-recurse x)))
1225 complaining "function called with two arguments, but wants exactly
1226 one" about SXHASH-RECURSE. (This might not be strictly a new bug,
1227 since IIRC post-fork CMU CL has also had problems with &OPTIONAL
1228 arguments in FLET/LABELS: it might be an old Python bug which is
1229 only exercised by the new arrangement of the SBCL compiler.)
1233 (DEFUN FOO () (CATCH 0 (PRINT 1331)))
1235 #<SB-C:TN '0!1> is not valid as the second argument to VOP:
1236 SB-C:MAKE-CATCH-BLOCK,
1237 since the TN's primitive type SB-VM::POSITIVE-FIXNUM doesn't allow
1238 any of the SCs allowed by the operand restriction:
1239 (SB-VM::DESCRIPTOR-REG)
1240 The (CATCH 0 ...) construct is bad style (because of unportability
1241 of EQ testing of numbers) but it is legal, and shouldn't cause an
1242 internal compiler error. (This error occurs in sbcl-0.6.13 and in
1243 0.pre7.86.flaky7.14.)
1246 Trying to compile something like
1247 (sb!alien:def-alien-routine "breakpoint_remove" sb!c-call:void
1248 (code-obj sb!c-call:unsigned-long)
1249 (pc-offset sb!c-call:int)
1250 (old-inst sb!c-call:unsigned-long))
1251 in SBCL-0.pre7.86.flaky7.22 after warm init fails with an error
1252 cannot use values types here
1253 probably because the SB-C-CALL:VOID type gets translated to (VALUES).
1254 It should be valid to use VOID for a function return type, so perhaps
1255 instead of calling SPECIFIER-TYPE (which excludes all VALUES types
1256 automatically) we should call VALUES-SPECIFIER-TYPE and handle VALUES
1257 types manually, allowing the special case (VALUES) but still excluding
1258 all more-complex VALUES types.
1261 (reported by Alexey Dejneka sbcl-devel 2001-12-07)
1262 (let ((s '((1 2 3))))
1263 (eval (eval ``(vector ,@',@s))))
1265 should return #(1 2 3), instead of this it causes a reader error.
1267 Interior call of BACKQUOTIFY erroneously optimizes ,@': it immediately
1268 splices the temporal representation of ,@S.
1271 Ideally, uninterning a symbol would allow it, and its associated
1272 FDEFINITION and PROCLAIM data, to be reclaimed by the GC. However,
1273 at least as of sbcl-0.7.0, this isn't the case. Information about
1274 FDEFINITIONs and PROCLAIMed properties is stored in globaldb.lisp
1275 essentially in ordinary (non-weak) hash tables keyed by symbols.
1276 Thus, once a system has an entry in this system, it tends to live
1277 forever, even when it is uninterned and all other references to it
1281 (reported by Arnaud Rouanet on cmucl-imp 2001-12-18)
1282 (defmethod foo ((x integer))
1284 (defmethod foo :around ((x integer))
1286 (call-next-method)))
1287 Now (FOO 3) should return 3, but instead it returns 4.
1290 (SB-DEBUG:BACKTRACE) output should start with something
1291 including the name BACKTRACE, not (as in 0.pre7.88)
1292 just "0: (\"hairy arg processor\" ...)". In general
1293 the names in BACKTRACE are all screwed up compared to
1294 the nice useful names in 0.6.13.
1296 Note for those who observe that this is an annoying
1297 bug and doesn't belong in a release: See the "note for the
1300 Note for the ambitious: This is an important bug and I'd
1301 really like to fix it and spent many hours on it. The
1302 obvious ways to fix it are hard, because the underlying
1303 infrastructure seems to be rather broken.
1304 * There are two mostly-separate systems for storing names,
1305 the in-the-function-object system used by e.g.
1306 CL:FUNCTION-LAMBDA-EXPRESSION and the
1307 in-the-DEBUG-FUN-object system used by e.g. BACKTRACE.
1308 The code as of sbcl-0.pre7.94 is smart enough to set
1309 up the first value, but not the second (because I naively
1310 assumed that one mechanism is enough, and didn't proof
1311 read the entire system to see whether there might be
1312 another mechanism?! argh...)
1313 * The systems are not quite separate, but instead weirdly and
1314 fragilely coupled by the FUN-DEBUG-FUN algorithm.
1315 * If you try to refactor this dain bramage away, reducing
1316 things to a single system -- I tried to add a
1317 %SIMPLE-FUN-DEBUG-FUN slot, planning eventually to get
1318 rid of the old %SIMPLE-FUN-NAME slot in favor of indirection
1319 through the new slot -- you get torpedoed by the fragility
1320 of the SIMPLE-FUN primitive object. Just adding the
1321 new slot, without making any other changes in the system,
1322 is enough to make the system fail with what look like
1323 memory corruption problems in warm init.
1324 But please do fix some or all of the problem, I'm tired
1325 of messing with it. -- WHN 2001-12-22
1328 a cross-compiler bug in sbcl-0.pre7.107
1330 $ cat > /tmp/bug139.lisp << EOF
1331 (in-package "SB!KERNEL")
1332 (defun f-c-l (name parent-types)
1333 (let* ((cpl (mapcar (lambda (x)
1334 (condition-class-cpl x))
1337 (concatenate 'simple-vector
1338 (layout-inherits cond-layout))))
1339 (if (not (mismatch (layout-inherits olayout) new-inherits))
1343 $ sbcl --core output/after-xc.core
1345 * (target-compile-file "/tmp/bug139.lisp")
1347 internal error, failed AVER:
1348 "(COMMON-LISP:MEMBER SB!C::FUN (SB!C::COMPONENT-LAMBDAS SB!C:COMPONENT))"
1350 It seems as though this xc bug is likely to correspond to a bug in the
1351 ordinary compiler, but I haven't yet found a test case which causes
1352 this problem in the ordinary compiler.
1354 related weirdness: Using #'(LAMBDA (X) ...) instead of (LAMBDA (X) ...)
1355 makes the assertion failure go away.
1358 In sbcl-0.pre7.107, (DIRECTORY "*.*") is broken, as reported by
1359 Nathan Froyd sbcl-devel 2001-12-28.
1361 Christophe Rhodes suggested (sbcl-devel 2001-12-30) converting
1362 the MERGED-PATHNAME expression in DEFUN DIRECTORY to
1363 (merged-pathname (merge-pathnames pathname
1364 *default-pathname-defaults*))
1365 This looks right, and fixes this bug, but it interacts with the NODES
1366 logic in %ENUMERATE-PATHNAMES to create a new bug, so that
1367 (DIRECTORY "../**/*.*") no longer shows files in the current working
1368 directory. Probably %ENUMERATE-PATHNAMES (or related logic like
1369 %ENUMERATE-MATCHES) needs to be patched as well.
1371 Note: The MERGED-PATHNAME change changes behavior incompatibly,
1372 making e.g. (DIRECTORY "*") no longer equivalent to (DIRECTORY "*.*"),
1373 so deserves a NEWS entry. E.g.
1374 * minor incompatible change (part of a bug fix by Christophe Rhodes
1375 to DIRECTORY behavior): DIRECTORY no longer implicitly promotes
1376 NIL slots of its pathname argument to :WILD, and in particular
1377 asking for the contents of a directory, which you used to be able
1378 to do without explicit wildcards, e.g. (DIRECTORY "/tmp/"),
1379 now needs explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
1382 KNOWN BUGS RELATED TO THE IR1 INTERPRETER
1384 (Now that the IR1 interpreter has gone away, these should be
1385 relatively straightforward to fix.)
1388 The system accepts DECLAIM in most places where DECLARE would be
1389 accepted, without even issuing a warning. ANSI allows this, but since
1390 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
1391 meaning is rather different, and it's unlikely that the user
1392 has a good reason for doing DECLAIM not at top level, it would be
1393 good to issue a STYLE-WARNING when this happens. A possible
1394 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
1395 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
1396 [This is considered an IR1-interpreter-related bug because until
1397 EVAL-WHEN is rewritten, which won't happen until after the IR1
1398 interpreter is gone, the system's notion of what's a top-level form
1399 and what's not will remain too confused to fix this problem.]
1402 (another wishlist thing..) Reimplement DEFMACRO to be basically
1403 like DEFMACRO-MUNDANELY, just using EVAL-WHEN.