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, and (3) by their own reasoning, it looks as though
54 ANSI may have gotten it wrong. ANSI justifies this specification
56 The restriction against issuing a warning for type mismatches
57 between a slot-initform and the corresponding slot's :TYPE
58 option is necessary because a slot-initform must be specified
59 in order to specify slot options; in some cases, no suitable
61 However, in SBCL (as in CMU CL or, for that matter, any compiler
62 which really understands Common Lisp types) a suitable default
63 does exist, in all cases, because the compiler understands the
64 concept of functions which never return (i.e. has return type NIL).
65 Thus, as a portable workaround, you can use a call to some
66 known-never-to-return function as the default. E.g.
68 (BAR (ERROR "missing :BAR argument")
69 :TYPE SOME-TYPE-TOO-HAIRY-TO-CONSTRUCT-AN-INSTANCE-OF))
71 (DECLAIM (FTYPE (FUNCTION () NIL) MISSING-ARG))
72 (DEFUN REQUIRED-ARG () ; workaround for SBCL non-ANSI slot init typing
73 (ERROR "missing required argument"))
75 (BAR (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
76 (BLETCH (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
77 (N-REFS-SO-FAR 0 :TYPE (INTEGER 0)))
78 Such code should compile without complaint and work correctly either
79 on SBCL or on any other completely compliant Common Lisp system.
82 bogus warnings about undefined functions for magic functions like
83 SB!C::%%DEFUN and SB!C::%DEFCONSTANT when cross-compiling files
84 like src/code/float.lisp. Fixing this will probably require
85 straightening out enough bootstrap consistency issues that
86 the cross-compiler can run with *TYPE-SYSTEM-INITIALIZED*.
87 Instead, the cross-compiler runs in a slightly flaky state
88 which is sane enough to compile SBCL itself, but which is
89 also unstable in several ways, including its inability
90 to really grok function declarations.
92 As of sbcl-0.7.5, sbcl's cross-compiler does run with
93 *TYPE-SYSTEM-INITIALIZED*; however, this bug remains.
96 The "compiling top-level form:" output ought to be condensed.
97 Perhaps any number of such consecutive lines ought to turn into a
98 single "compiling top-level forms:" line.
101 The way that the compiler munges types with arguments together
102 with types with no arguments (in e.g. TYPE-EXPAND) leads to
103 weirdness visible to the user:
104 (DEFTYPE FOO () 'FIXNUM)
106 (TYPEP 11 '(FOO)) => T, which seems weird
107 (TYPEP 11 'FIXNUM) => T
108 (TYPEP 11 '(FIXNUM)) signals an error, as it should
109 The situation is complicated by the presence of Common Lisp types
110 like UNSIGNED-BYTE (which can either be used in list form or alone)
111 so I'm not 100% sure that the behavior above is actually illegal.
112 But I'm 90+% sure, and the following related behavior,
114 treating the bare symbol AND as equivalent to '(AND), is specifically
115 forbidden (by the ANSI specification of the AND type).
118 It would be nice if the
120 (during macroexpansion)
121 said what macroexpansion was at fault, e.g.
123 (during macroexpansion of IN-PACKAGE,
124 during macroexpansion of DEFFOO)
127 (SUBTYPEP '(FUNCTION (T BOOLEAN) NIL)
128 '(FUNCTION (FIXNUM FIXNUM) NIL)) => T, T
129 (Also, when this is fixed, we can enable the code in PROCLAIM which
130 checks for incompatible FTYPE redeclarations.)
133 (I *think* this is a bug. It certainly seems like strange behavior. But
134 the ANSI spec is scary, dark, and deep.. -- WHN)
135 (FORMAT NIL "~,1G" 1.4) => "1. "
136 (FORMAT NIL "~3,1G" 1.4) => "1. "
139 from Marco Antoniotti on cmucl-imp mailing list 1 Mar 2000:
141 (setf (find-class 'ccc1) (find-class 'ccc))
142 (defmethod zut ((c ccc1)) 123)
143 In sbcl-0.7.1.13, this gives an error,
144 There is no class named CCC1.
145 DTC's recommended workaround from the mailing list 3 Mar 2000:
146 (setf (pcl::find-class 'ccc1) (pcl::find-class 'ccc))
149 Sometimes (SB-EXT:QUIT) fails with
150 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
151 Process inferior-lisp exited abnormally with code 1
152 I haven't noticed a repeatable case of this yet.
155 The printer doesn't report closures very well. This is true in
159 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
160 It would be nice to make closures have a settable name slot,
161 and make things like DEFSTRUCT and FLET, which create closures,
162 set helpful values into this slot.
165 And as long as we're wishing, it would be awfully nice if INSPECT could
166 also report on closures, telling about the values of the bound variables.
169 The compiler assumes that any time a function of declared FTYPE
170 doesn't signal an error, its arguments were of the declared type.
171 E.g. compiling and loading
172 (DECLAIM (OPTIMIZE (SAFETY 3)))
173 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
175 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE)) FACTORIAL))
177 (COND ((> (FACTORIAL X) 1.0E6)
178 (FORMAT T "too big~%"))
180 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
182 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
185 will cause the INTEGERP case to be selected, giving bogus output a la
187 This violates the "declarations are assertions" principle.
188 According to the ANSI spec, in the section "System Class FUNCTION",
189 this is a case of "lying to the compiler", but the lying is done
190 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
191 not by the unexpectedly general definition of FACTORIAL. In any case,
192 "declarations are assertions" means that lying to the compiler should
193 cause an error to be signalled, and should not cause a bogus
194 result to be returned. Thus, the compiler should not assume
195 that arbitrary functions check their argument types. (It might
196 make sense to add another flag (CHECKED?) to DEFKNOWN to
197 identify functions which *do* check their argument types.)
198 (Also, verify that the compiler handles declared function
199 return types as assertions.)
202 TYPEP of VALUES types is sometimes implemented very inefficiently, e.g. in
203 (DEFTYPE INDEXOID () '(INTEGER 0 1000))
205 (DECLARE (TYPE INDEXOID X))
206 (THE (VALUES INDEXOID)
208 where the implementation of the type check in function FOO
209 includes a full call to %TYPEP. There are also some fundamental problems
210 with the interpretation of VALUES types (inherited from CMU CL, and
211 from the ANSI CL standard) as discussed on the cmucl-imp@cons.org
212 mailing list, e.g. in Robert Maclachlan's post of 21 Jun 2000.
215 The definitions of SIGCONTEXT-FLOAT-REGISTER and
216 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
217 supported on FreeBSD because the floating point state is not saved,
218 but at least as of FreeBSD 4.0, the floating point state *is* saved,
219 so they could be supported after all. Very likely
220 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
223 (as discussed by Douglas Crosher on the cmucl-imp mailing list ca.
224 Aug. 10, 2000): CMUCL currently interprets 'member as '(member); same
225 issue with 'union, 'and, 'or etc. So even though according to the
226 ANSI spec, bare 'MEMBER, 'AND, and 'OR are not legal types, CMUCL
227 (and now SBCL) interpret them as legal types.
230 a slew of floating-point-related errors reported by Peter Van Eynde
232 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT is bogus, and
233 should probably be 1.4012985e-45. In SBCL,
234 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
235 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
236 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
237 and LEAST-NEGATIVE-LONG-FLOAT.
238 c: Many expressions generate floating infinity on x86/Linux:
243 PVE's regression tests want them to raise errors. sbcl-0.7.0.5
244 on x86/Linux generates the infinities instead. That might or
245 might not be conforming behavior, but it's also inconsistent,
246 which is almost certainly wrong. (Inconsistency: (/ 1 0.0)
247 should give the same result as (/ 1.0 0.0), but instead (/ 1 0.0)
248 generates SINGLE-FLOAT-POSITIVE-INFINITY and (/ 1.0 0.0)
250 d: (in section12.erg) various forms a la
251 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
252 don't give the right behavior.
255 type safety errors reported by Peter Van Eynde July 25, 2000:
256 a: (COERCE (QUOTE (A B C)) (QUOTE (VECTOR * 4)))
258 In general lengths of array type specifications aren't
259 checked by COERCE, so it fails when the spec is
260 (VECTOR 4), (STRING 2), (SIMPLE-BIT-VECTOR 3), or whatever.
261 b: CONCATENATE has the same problem of not checking the length
262 of specified output array types. MAKE-SEQUENCE and MAP and
263 MERGE also have the same problem.
264 c: (COERCE 'AND 'FUNCTION) returns something related to
265 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
266 h: (MAKE-CONCATENATED-STREAM (MAKE-STRING-OUTPUT-STREAM))
267 should signal TYPE-ERROR.
268 i: MAKE-TWO-WAY-STREAM doesn't check that its arguments can
269 be used for input and output as needed. It should fail with
270 TYPE-ERROR when handed e.g. the results of
271 MAKE-STRING-INPUT-STREAM or MAKE-STRING-OUTPUT-STREAM in
272 the inappropriate positions, but doesn't.
273 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
274 not a binary input stream, but instead cheerfully reads from
275 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
278 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
279 a: (DEFCLASS FOO () (A B A)) should signal a PROGRAM-ERROR, and
281 b: (DEFCLASS FOO () (A B A) (:DEFAULT-INITARGS X A X B)) should
282 signal a PROGRAM-ERROR, and doesn't.
283 c: (DEFCLASS FOO07 NIL ((A :ALLOCATION :CLASS :ALLOCATION :CLASS))),
284 and other DEFCLASS forms with duplicate specifications in their
285 slots, should signal a PROGRAM-ERROR, and doesn't.
286 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
287 causes a COMPILER-ERROR.
290 SYMBOL-MACROLET bugs reported by Peter Van Eynde July 25, 2000:
291 a: (SYMBOL-MACROLET ((T TRUE)) ..) should probably signal
292 PROGRAM-ERROR, but SBCL accepts it instead.
293 b: SYMBOL-MACROLET should refuse to bind something which is
294 declared as a global variable, signalling PROGRAM-ERROR.
295 c: SYMBOL-MACROLET should signal PROGRAM-ERROR if something
296 it binds is declared SPECIAL inside.
299 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
301 (DEFGENERIC FOO02 (X))
302 (DEFMETHOD FOO02 ((X NUMBER)) T)
303 (LET ((M (FIND-METHOD (FUNCTION FOO02)
305 (LIST (FIND-CLASS (QUOTE NUMBER))))))
306 (REMOVE-METHOD (FUNCTION FOO02) M)
307 (DEFGENERIC FOO03 (X))
308 (ADD-METHOD (FUNCTION FOO03) M)))
309 should give an error, but SBCL allows it.
310 b: READ should probably return READER-ERROR, not the bare
311 arithmetic error, when input a la "1/0" or "1e1000" causes
315 It has been reported (e.g. by Peter Van Eynde) that there are
316 several metaobject protocol "errors". (In order to fix them, we might
317 need to document exactly what metaobject protocol specification
318 we're following -- the current code is just inherited from PCL.)
321 The implementation of #'+ returns its single argument without
322 type checking, e.g. (+ "illegal") => "illegal".
325 The debugger LIST-LOCATIONS command doesn't work properly.
328 Compiling and loading
329 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
331 then requesting a BACKTRACE at the debugger prompt gives no information
332 about where in the user program the problem occurred.
335 The compiler is supposed to do type inference well enough that
338 ((SIMPLE-ARRAY SINGLE-FLOAT)
340 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
343 is redundant. However, as reported by Juan Jose Garcia Ripoll for
344 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
345 workaround for the problem for now, but can't be done by the TYPECASE
346 macros themselves, since it's too hard for the macro to detect
347 assignments to the variable within the clause.
348 Note: The compiler *is* smart enough to do the type inference in
349 many cases. This case, derived from a couple of MACROEXPAND-1
350 calls on Ripoll's original test case,
352 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
353 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
354 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
355 (LET ((I 0) (G2554 LENGTH))
356 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
359 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
360 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
361 (GO SB-LOOP::NEXT-LOOP)
362 SB-LOOP::END-LOOP))))))
363 demonstrates the problem; but the problem goes away if the TAGBODY
364 and GO forms are removed (leaving the SETF in ordinary, non-looping
365 code), or if the TAGBODY and GO forms are retained, but the
366 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
369 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
370 I am looking into this problem that showed up on the cmucl-help
371 list. It seems to me that the "implementation specific environment
372 hacking functions" found in pcl/walker.lisp are completely messed
373 up. The good thing is that they appear to be barely used within
374 PCL and the munged environment object is passed to cmucl only
375 in calls to macroexpand-1, which is probably why this case fails.
376 SBCL uses essentially the same code, so if the environment hacking
377 is screwed up, it affects us too.
380 Using the pretty-printer from the command prompt gives funny
381 results, apparently because the pretty-printer doesn't know
382 about user's command input, including the user's carriage return
383 that the user, and therefore the pretty-printer thinks that
384 the new output block should start indented 2 or more characters
385 rightward of the correct location.
388 (probably related to bug #70; maybe related to bug #109)
389 As reported by Carl Witty on submit@bugs.debian.org 1999-05-08,
391 (in-package "CL-USER")
392 (defun equal-terms (termx termy)
394 ((alpha-equal-bound-term-lists (listx listy)
395 (or (and (null listx) (null listy))
397 (let ((bindings-x (bindings-of-bound-term (car listx)))
398 (bindings-y (bindings-of-bound-term (car listy))))
399 (if (and (null bindings-x) (null bindings-y))
400 (alpha-equal-terms (term-of-bound-term (car listx))
401 (term-of-bound-term (car listy)))
402 (and (= (length bindings-x) (length bindings-y))
404 (enter-binding-pairs (bindings-of-bound-term (car listx))
405 (bindings-of-bound-term (car listy)))
406 (alpha-equal-terms (term-of-bound-term (car listx))
407 (term-of-bound-term (car listy)))
408 (exit-binding-pairs (bindings-of-bound-term (car listx))
409 (bindings-of-bound-term (car listy)))))))
410 (alpha-equal-bound-term-lists (cdr listx) (cdr listy)))))
412 (alpha-equal-terms (termx termy)
413 (if (and (variable-p termx)
415 (equal-bindings (id-of-variable-term termx)
416 (id-of-variable-term termy))
417 (and (equal-operators-p (operator-of-term termx) (operator-of-term termy))
418 (alpha-equal-bound-term-lists (bound-terms-of-term termx)
419 (bound-terms-of-term termy))))))
423 (with-variable-invocation (alpha-equal-terms termx termy))))))
424 causes an assertion failure
425 The assertion (EQ (C::LAMBDA-TAIL-SET C::CALLER)
426 (C::LAMBDA-TAIL-SET (C::LAMBDA-HOME C::CALLEE))) failed.
428 Bob Rogers reports (1999-07-28 on cmucl-imp@cons.org) a smaller test
429 case with the same problem:
430 (defun parse-fssp-alignment ()
431 ;; Given an FSSP alignment file named by the argument . . .
432 (labels ((get-fssp-char ()
436 ;; Stub body, enough to tickle the bug.
437 (list (read-fssp-char)
441 ANSI specifies that the RESULT-TYPE argument of CONCATENATE must be
442 a subtype of SEQUENCE, but CONCATENATE doesn't check this properly:
443 (CONCATENATE 'SIMPLE-ARRAY #(1 2) '(3)) => #(1 2 3)
444 This also leads to funny behavior when derived type specifiers
445 are used, as originally reported by Milan Zamazal for CMU CL (on the
446 Debian bugs mailing list (?) 2000-02-27), then reported by Martin
447 Atzmueller for SBCL (2000-10-01 on sbcl-devel@lists.sourceforge.net):
448 (DEFTYPE FOO () 'SIMPLE-ARRAY)
449 (CONCATENATE 'FOO #(1 2) '(3))
450 => #<ARRAY-TYPE SIMPLE-ARRAY> is a bad type specifier for
452 The derived type specifier FOO should act the same way as the
453 built-in type SIMPLE-ARRAY here, but it doesn't. That problem
454 doesn't seem to exist for sequence types:
455 (DEFTYPE BAR () 'SIMPLE-VECTOR)
456 (CONCATENATE 'BAR #(1 2) '(3)) => #(1 2 3)
457 See also bug #46a./b., and discussion and patch sbcl-devel and
461 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
462 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
463 crashes SBCL. In general tracing anything which is used in the
464 implementation of TRACE is likely to have the same problem.
467 (probably related to bug #65; maybe related to bug #109)
468 The compiler doesn't like &OPTIONAL arguments in LABELS and FLET
470 (DEFUN FIND-BEFORE (ITEM SEQUENCE &KEY (TEST #'EQL))
471 (LABELS ((FIND-ITEM (OBJ SEQ TEST &OPTIONAL (VAL NIL))
472 (LET ((ITEM (FIRST SEQ)))
475 ((FUNCALL TEST OBJ ITEM)
478 (FIND-ITEM OBJ (REST SEQ) TEST (NCONC VAL `(,ITEM))))))))
479 (FIND-ITEM ITEM SEQUENCE TEST)))
480 from David Young's bug report on cmucl-help@cons.org 30 Nov 2000
481 causes sbcl-0.6.9 to fail with
482 error in function SB-KERNEL:ASSERT-ERROR:
483 The assertion (EQ (SB-C::LAMBDA-TAIL-SET SB-C::CALLER)
484 (SB-C::LAMBDA-TAIL-SET
485 (SB-C::LAMBDA-HOME SB-C::CALLEE))) failed.
488 (DECLAIM (OPTIMIZE ..)) doesn't work properly inside LOCALLY forms.
491 As reported by Martin Atzmueller on sbcl-devel 26 Dec 2000,
492 ANSI says that WITH-OUTPUT-TO-STRING should have a keyword
493 :ELEMENT-TYPE, but in sbcl-0.6.9 this is not defined for
494 WITH-OUTPUT-TO-STRING.
497 ANSI says in one place that type declarations can be abbreviated even
498 when the type name is not a symbol, e.g.
499 (DECLAIM ((VECTOR T) *FOOVECTOR*))
500 SBCL doesn't support this. But ANSI says in another place that this
501 isn't allowed. So it's not clear this is a bug after all. (See the
502 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
506 as pointed out by Dan Barlow on sbcl-devel 2000-07-02:
507 The PICK-TEMPORARY-FILE-NAME utility used by LOAD-FOREIGN uses
508 an easily guessable temporary filename in a way which might open
509 applications using LOAD-FOREIGN to hijacking by malicious users
510 on the same machine. Incantations for doing this safely are
511 floating around the net in various "how to write secure programs
512 despite Unix" documents, and it would be good to (1) fix this in
513 LOAD-FOREIGN, and (2) hunt for any other code which uses temporary
514 files and make it share the same new safe logic.
517 Functions are assigned names based on the context in which they're
518 defined. This is less than ideal for the functions which are
519 used to implement CLOS methods. E.g. the output of
520 (DESCRIBE 'PRINT-OBJECT) lists functions like
521 #<FUNCTION "DEF!STRUCT (TRACE-INFO (:MAKE-LOAD-FORM-FUN SB-KERNEL:JUST-DUMP-IT-NORMALLY) (:PRINT-OBJECT #))" {1020E49}>
523 #<FUNCTION "MACROLET ((FORCE-DELAYED-DEF!METHODS NIL #))" {1242871}>
524 It would be better if these functions' names always identified
525 them as methods, and identified their generic functions and
529 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
530 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
531 the mean of the distribution can be systematically O(0.1%) wrong.
532 Just increasing R-I-E-B is probably not a good solution, since
533 it would decrease efficiency more than is probably necessary. Perhaps
534 using some sort of accept/reject method would be better.
537 Internally the compiler sometimes evaluates
538 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
539 (I stumbled across this when I added an
540 (assert (not (eq type1 *wild-type*)))
541 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
542 in a type context should probably be translated to T, and so it's
543 probably wrong to ask whether it's equal to the T type and then (using
544 the EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
545 (I haven't tried to investigate this bug enough to guess whether
546 there might be any user-level symptoms.)
549 Inconsistencies between derived and declared VALUES return types for
550 DEFUN aren't checked very well. E.g. the logic which successfully
551 catches problems like
552 (declaim (ftype (function (fixnum) float) foo))
554 (declare (type integer x))
555 (values x)) ; wrong return type, detected, gives warning, good!
557 (declaim (ftype (function (t) (values t t)) bar))
559 (values x)) ; wrong number of return values, no warning, bad!
560 The cause of this is seems to be that (1) the internal function
561 VALUES-TYPES-EQUAL-OR-INTERSECT used to make the check handles its
562 arguments symmetrically, and (2) when the type checking code was
563 written back when when SBCL's code was still CMU CL, the intent
565 (declaim (ftype (function (t) t) bar))
567 (values x x)) ; wrong number of return values; should give warning?
568 not be warned for, because a two-valued return value is considered
569 to be compatible with callers who expects a single value to be
570 returned. That intent is probably not appropriate for modern ANSI
571 Common Lisp, but fixing this might be complicated because of other
572 divergences between auld-style and new-style handling of
573 multiple-VALUES types. (Some issues related to this were discussed
574 on cmucl-imp at some length sometime in 2000.)
577 The facility for dumping a running Lisp image to disk gets confused
578 when run without the PURIFY option, and creates an unnecessarily large
579 core file (apparently representing memory usage up to the previous
580 high-water mark). Moreover, when the file is loaded, it confuses the
581 GC, so that thereafter memory usage can never be reduced below that
585 In sbcl-0.6.11.41 (and in all earlier SBCL, and in CMU
586 CL), out-of-line structure slot setters are horribly inefficient
587 whenever the type of the slot is declared, because out-of-line
588 structure slot setters are implemented as closures to save space,
589 so the compiler doesn't compile the type test into code, but
590 instead just saves the type in a lexical closure and interprets it
592 A proper solution involves deciding whether it's really worth
593 saving space by implementing structure slot accessors as closures.
594 (If it's not worth it, the problem vanishes automatically. If it
595 is worth it, there are hacks we could use to force type tests to
596 be compiled anyway, and even shared. E.g. we could implement
597 an EQUAL hash table mapping from types to compiled type tests,
598 and save the appropriate compiled type test as part of each lexical
599 closure; or we could make the lexical closures be placeholders
600 which overwrite their old definition as a lexical closure with
601 a new compiled definition the first time that they're called.)
602 As a workaround for the problem, #'(SETF FOO) expressions can
603 be replaced with (EFFICIENT-SETF-FUNCTION FOO), where
604 (defmacro efficient-setf-function (place-function-name)
605 (or #+sbcl (and (sb-impl::info :function :accessor-for place-function-name)
606 ;; a workaround for the problem, encouraging the
607 ;; inline expansion of the structure accessor, so
608 ;; that the compiler can optimize its type test
609 (let ((new-value (gensym "NEW-VALUE-"))
610 (structure-value (gensym "STRUCTURE-VALUE-")))
611 `(lambda (,new-value ,structure-value)
612 (setf (,place-function-name ,structure-value)
614 ;; no problem, can just use the ordinary expansion
615 `(function (setf ,place-function-name))))
618 There's apparently a bug in CEILING optimization which caused
619 Douglas Crosher to patch the CMU CL version. Martin Atzmueller
620 applied the patches to SBCL and they didn't seem to cause problems
621 (as reported sbcl-devel 2001-05-04). However, since the patches
622 modify nontrivial code which was apparently written incorrectly
623 the first time around, until regression tests are written I'm not
624 comfortable merging the patches in the CVS version of SBCL.
627 As reported by Arthur Lemmens sbcl-devel 2001-05-05, ANSI
628 requires that SYMBOL-MACROLET refuse to rebind special variables,
629 but SBCL doesn't do this. (Also as reported by AL in the same
630 message, SBCL depended on this nonconforming behavior to build
631 itself, because of the way that **CURRENT-SEGMENT** was implemented.
632 As of sbcl-0.7.3.x, this dependence on the nonconforming behavior
633 has been fixed, but the nonconforming behavior remains.)
636 (DESCRIBE 'SB-ALIEN:DEF-ALIEN-TYPE) reports the macro argument list
640 in #<PACKAGE "SB-ALIEN">.
641 Macro-function: #<FUNCTION "DEF!MACRO DEF-ALIEN-TYPE" {19F4A39}>
642 Macro arguments: (#:whole-470 #:environment-471)
643 On Sat, May 26, 2001 09:45:57 AM CDT it was compiled from:
644 /usr/stuff/sbcl/src/code/host-alieneval.lisp
645 Created: Monday, March 12, 2001 07:47:43 AM CST
648 (TIME (ROOM T)) reports more than 200 Mbytes consed even for
649 a clean, just-started SBCL system. And it seems to be right:
650 (ROOM T) can bring a small computer to its knees for a *long*
651 time trying to GC afterwards. Surely there's some more economical
652 way to implement (ROOM T).
655 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
657 ;;; This file fails to compile.
658 ;;; Maybe this bug is related to bugs #65, #70 in the BUGS file.
659 (in-package :cl-user)
665 ;; Uncomment and it works
668 In SBCL 0.6.12.42, the problem is
669 internal error, failed AVER:
670 "(COMMON-LISP:EQ (SB!C::LAMBDA-TAIL-SET SB!C::CALLER)
671 (SB!C::LAMBDA-TAIL-SET (SB!C::LAMBDA-HOME SB!C::CALLEE)))"
674 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
676 ;;; The compiler is flushing the argument type test, and the default
677 ;;; case in the cond, so that calling with say a fixnum 0 causes a
679 (declaim (optimize (safety 2) (speed 3)))
681 (declare (type (or string stream) x))
682 (cond ((typep x 'string) 'string)
683 ((typep x 'stream) 'stream)
686 The symptom in sbcl-0.6.12.42 on OpenBSD is actually (TST 0)=>STREAM
687 (not the SIGBUS reported in the comment) but that's broken too;
688 type declarations are supposed to be treated as assertions unless
689 SAFETY 0, so we should be getting a TYPE-ERROR.
692 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
694 (in-package :cl-user)
695 ;;; From: David Gadbois <gadbois@cyc.com>
697 ;;; Logical pathnames aren't externalizable.
699 (let ((tempfile "/tmp/test.lisp"))
700 (setf (logical-pathname-translations "XXX")
701 '(("XXX:**;*.*" "/tmp/**/*.*")))
702 (with-open-file (out tempfile :direction :output)
703 (write-string "(defvar *path* #P\"XXX:XXX;FOO.LISP\")" out))
704 (compile-file tempfile))
705 The error message in sbcl-0.6.12.42 is
707 ; (while making load form for #<SB-IMPL::LOGICAL-HOST "XXX">)
708 ; A logical host can't be dumped as a constant: #<SB-IMPL::LOGICAL-HOST "XXX">
711 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
713 (in-package :cl-user)
714 ;;; The following invokes a compiler error.
715 (declaim (optimize (speed 2) (debug 3)))
718 (unwind-protect nil)))
722 The error message in sbcl-0.6.12.42 is
723 internal error, failed AVER:
724 "(COMMON-LISP:EQ (SB!C::TN-ENVIRONMENT SB!C:TN) SB!C::TN-ENV)"
727 When the compiler inline expands functions, it may be that different
728 kinds of return values are generated from different code branches.
729 E.g. an inline expansion of POSITION generates integer results
730 from one branch, and NIL results from another. When that inline
731 expansion is used in a context where only one of those results
734 (aref *a1* (position x *a2*)))
735 and the compiler can't prove that the unacceptable branch is
736 never taken, then bogus type mismatch warnings can be generated.
737 If you need to suppress the type mismatch warnings, you can
738 suppress the inline expansion,
740 #+sbcl (declare (notinline position)) ; to suppress bug 117 bogowarnings
741 (aref *a1* (position x *a2*)))
742 or, sometimes, suppress them by declaring the result to be of an
745 (aref *a1* (the integer (position x *a2*))))
747 This is not a new compiler problem in 0.7.0, but the new compiler
748 transforms for FIND, POSITION, FIND-IF, and POSITION-IF make it
749 more conspicuous. If you don't need performance from these functions,
750 and the bogus warnings are a nuisance for you, you can return to
751 your pre-0.7.0 state of grace with
752 #+sbcl (declaim (notinline find position find-if position-if)) ; bug 117..
755 as reported by Eric Marsden on cmucl-imp@cons.org 2001-08-14:
756 (= (FLOAT 1 DOUBLE-FLOAT-EPSILON)
757 (+ (FLOAT 1 DOUBLE-FLOAT-EPSILON) DOUBLE-FLOAT-EPSILON)) => T
758 when of course it should be NIL. (He says it only fails for X86,
759 not SPARC; dunno about Alpha.)
761 Also, "the same problem exists for LONG-FLOAT-EPSILON,
762 DOUBLE-FLOAT-NEGATIVE-EPSILON, LONG-FLOAT-NEGATIVE-EPSILON (though
763 for the -negative- the + is replaced by a - in the test)."
765 Raymond Toy comments that this is tricky on the X86 since its FPU
766 uses 80-bit precision internally.
769 The compiler incorrectly figures the return type of
770 (DEFUN FOO (FRAME UP-FRAME)
777 This problem exists in CMU CL 18c too. When I reported it on
778 cmucl-imp@cons.org, Raymond Toy replied 23 Aug 2001 with
779 a partial explanation, but no fix has been found yet.
782 Even in sbcl-0.pre7.x, which is supposed to be free of the old
783 non-ANSI behavior of treating the function return type inferred
784 from the current function definition as a declaration of the
785 return type from any function of that name, the return type of NIL
786 is attached to FOO in 120a above, and used to optimize code which
790 There was some sort of screwup in handling of
791 (IF (NOT (IGNORE-ERRORS ..))). E.g.
793 (if (not (ignore-errors
794 (make-pathname :host "foo" :directory "!bla" :name "bar")))
796 (error "notunlessnot")))
797 The (NOT (IGNORE-ERRORS ..)) form evaluates to T, so this should be
798 printing "ok", but instead it's going to the ERROR. This problem
799 seems to've been introduced by MNA's HANDLER-CASE patch (sbcl-devel
800 2001-07-17) and as a workaround (put in sbcl-0.pre7.14.flaky4.12)
801 I reverted back to the old weird HANDLER-CASE code. However, I
802 think the problem looks like a compiler bug in handling RETURN-FROM,
803 so I left the MNA-patched code in HANDLER-CASE (suppressed with
804 #+NIL) and I'd like to go back to see whether this really is
805 a compiler bug before I delete this BUGS entry.
808 The *USE-IMPLEMENTATION-TYPES* hack causes bugs, particularly
809 (IN-PACKAGE :SB-KERNEL)
810 (TYPE= (SPECIFIER-TYPE '(VECTOR T))
811 (SPECIFIER-TYPE '(VECTOR UNDEFTYPE)))
812 Then because of this, the compiler bogusly optimizes
813 (TYPEP #(11) '(SIMPLE-ARRAY UNDEF-TYPE 1))
814 to T. Unfortunately, just setting *USE-IMPLEMENTATION-TYPES* to
815 NIL around sbcl-0.pre7.14.flaky4.12 didn't work: the compiler complained
816 about type mismatches (probably harmlessly, another instance of bug 117);
817 and then cold init died with a segmentation fault.
820 As of version 0.pre7.14, SBCL's implementation of MACROLET makes
821 the entire lexical environment at the point of MACROLET available
822 in the bodies of the macroexpander functions. In particular, it
823 allows the function bodies (which run at compile time) to try to
824 access lexical variables (which are only defined at runtime).
825 It doesn't even issue a warning, which is bad.
827 The SBCL behavior arguably conforms to the ANSI spec (since the
828 spec says that the behavior is undefined, ergo anything conforms).
829 However, it would be better to issue a compile-time error.
830 Unfortunately I (WHN) don't see any simple way to detect this
831 condition in order to issue such an error, so for the meantime
832 SBCL just does this weird broken "conforming" thing.
834 The ANSI standard says, in the definition of the special operator
836 The macro-expansion functions defined by MACROLET are defined
837 in the lexical environment in which the MACROLET form appears.
838 Declarations and MACROLET and SYMBOL-MACROLET definitions affect
839 the local macro definitions in a MACROLET, but the consequences
840 are undefined if the local macro definitions reference any
841 local variable or function bindings that are visible in that
843 Then it seems to contradict itself by giving the example
845 (macrolet ((fudge (z)
846 ;The parameters x and flag are not accessible
847 ; at this point; a reference to flag would be to
848 ; the global variable of that name.
849 ` (if flag (* ,z ,z) ,z)))
850 ;The parameters x and flag are accessible here.
854 The comment "a reference to flag would be to the global variable
855 of the same name" sounds like good behavior for the system to have.
856 but actual specification quoted above says that the actual behavior
860 (as reported by Gabe Garza on cmucl-help 2001-09-21)
862 (defun test-pred (x y)
866 (func (lambda () x)))
867 (print (eq func func))
868 (print (test-pred func func))
869 (delete func (list func))))
870 Now calling (TEST-CASE) gives output
873 (#<FUNCTION {500A9EF9}>)
874 Evidently Python thinks of the lambda as a code transformation so
875 much that it forgets that it's also an object.
878 The DEFSTRUCT section of the ANSI spec, in the :CONC-NAME section,
879 specifies a precedence rule for name collisions between slot accessors of
880 structure classes related by inheritance. As of 0.7.0, SBCL still
884 insufficient syntax checking in MACROLET:
886 (macrolet ((defmacro bar (z) `(+ z z)))
888 shouldn't compile without error (because of the extra DEFMACRO symbol).
891 As of sbcl-0.pre7.86.flaky7.3, the cross-compiler, and probably
892 the CL:COMPILE function (which is based on the same %COMPILE
893 mechanism) get confused by
895 (labels ((sxhash-number (x)
897 (fixnum (sxhash x)) ; through DEFTRANSFORM
898 (integer (sb!bignum:sxhash-bignum x))
899 (single-float (sxhash x)) ; through DEFTRANSFORM
900 (double-float (sxhash x)) ; through DEFTRANSFORM
901 #!+long-float (long-float (error "stub: no LONG-FLOAT"))
902 (ratio (let ((result 127810327))
903 (declare (type fixnum result))
904 (mixf result (sxhash-number (numerator x)))
905 (mixf result (sxhash-number (denominator x)))
907 (complex (let ((result 535698211))
908 (declare (type fixnum result))
909 (mixf result (sxhash-number (realpart x)))
910 (mixf result (sxhash-number (imagpart x)))
912 (sxhash-recurse (x &optional (depthoid +max-hash-depthoid+))
913 (declare (type index depthoid))
917 (mix (sxhash-recurse (car x) (1- depthoid))
918 (sxhash-recurse (cdr x) (1- depthoid)))
921 (if (typep x 'structure-object)
923 (sxhash ; through DEFTRANSFORM
924 (class-name (layout-class (%instance-layout x)))))
926 (symbol (sxhash x)) ; through DEFTRANSFORM
927 (number (sxhash-number x))
930 (simple-string (sxhash x)) ; through DEFTRANSFORM
931 (string (%sxhash-substring x))
932 (bit-vector (let ((result 410823708))
933 (declare (type fixnum result))
934 (dotimes (i (min depthoid (length x)))
935 (mixf result (aref x i)))
937 (t (logxor 191020317 (sxhash (array-rank x))))))
940 (sxhash (char-code x)))) ; through DEFTRANSFORM
943 complaining "function called with two arguments, but wants exactly
944 one" about SXHASH-RECURSE. (This might not be strictly a new bug,
945 since IIRC post-fork CMU CL has also had problems with &OPTIONAL
946 arguments in FLET/LABELS: it might be an old Python bug which is
947 only exercised by the new arrangement of the SBCL compiler.)
950 Ideally, uninterning a symbol would allow it, and its associated
951 FDEFINITION and PROCLAIM data, to be reclaimed by the GC. However,
952 at least as of sbcl-0.7.0, this isn't the case. Information about
953 FDEFINITIONs and PROCLAIMed properties is stored in globaldb.lisp
954 essentially in ordinary (non-weak) hash tables keyed by symbols.
955 Thus, once a system has an entry in this system, it tends to live
956 forever, even when it is uninterned and all other references to it
960 (reported by Arnaud Rouanet on cmucl-imp 2001-12-18)
961 (defmethod foo ((x integer))
963 (defmethod foo :around ((x integer))
966 Now (FOO 3) should return 3, but instead it returns 4.
969 (reported by Alexey Dejneka sbcl-devel 2002-01-03)
971 SUBTYPEP does not work well with redefined classes:
980 * (defclass b (a) ())
993 This bug was fixed in sbcl-0.7.4.1 by invalidating the PCL wrapper
994 class upon redefinition. Unfortunately, doing so causes bug #176 to
995 appear. Pending further investication, one or other of these bugs
996 might be present at any given time.
999 Pretty-printing nested backquotes doesn't work right, as
1000 reported by Alexey Dejneka sbcl-devel 2002-01-13:
1002 ``(FOO SB-IMPL::BACKQ-COMMA-AT S)
1003 * (lisp-implementation-version)
1007 (as reported by Lynn Quam on cmucl-imp ca. 2002-01-16)
1008 %NATURALIZE-C-STRING conses a lot, like 16 bytes per byte
1009 of the naturalized string. We could probably port the patches
1010 from the cmucl-imp mailing list.
1013 (reported by Jesse Bouwman 2001-10-24 through the unfortunately
1014 prominent SourceForge web/db bug tracking system, which is
1015 unfortunately not a reliable way to get a timely response from
1016 the SBCL maintainers)
1017 In the course of trying to build a test case for an
1018 application error, I encountered this behavior:
1019 If you start up sbcl, and then lay on CTRL-C for a
1020 minute or two, the lisp process will eventually say:
1021 %PRIMITIVE HALT called; the party is over.
1022 and throw you into the monitor. If I start up lisp,
1023 attach to the process with strace, and then do the same
1024 (abusive) thing, I get instead:
1025 access failure in heap page not marked as write-protected
1026 and the monitor again. I don't know enough to have the
1027 faintest idea of what is going on here.
1028 This is with sbcl 6.12, uname -a reports:
1029 Linux prep 2.2.19 #4 SMP Tue Apr 24 13:59:52 CDT 2001 i686 unknown
1030 I (WHN) have verified that the same thing occurs on sbcl-0.pre7.141
1031 under OpenBSD 2.9 on my X86 laptop. Do be patient when you try it:
1032 it took more than two minutes (but less than five) for me.
1035 (This was once known as IR1-4, but it lived on even after the
1036 IR1 interpreter went to the big bit bucket in the sky.)
1037 The system accepts DECLAIM in most places where DECLARE would be
1038 accepted, without even issuing a warning. ANSI allows this, but since
1039 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
1040 meaning is rather different, and it's unlikely that the user
1041 has a good reason for doing DECLAIM not at top level, it would be
1042 good to issue a STYLE-WARNING when this happens. A possible
1043 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
1044 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
1045 [This is considered an IR1-interpreter-related bug because until
1046 EVAL-WHEN is rewritten, which won't happen until after the IR1
1047 interpreter is gone, the system's notion of what's a top-level form
1048 and what's not will remain too confused to fix this problem.]
1051 ANSI allows types `(COMPLEX ,FOO) to use very hairy values for
1052 FOO, e.g. (COMPLEX (AND REAL (SATISFIES ODDP))). The old CMU CL
1053 COMPLEX implementation didn't deal with this, and hasn't been
1054 upgraded to do so. (This doesn't seem to be a high priority
1055 conformance problem, since seems hard to construct useful code
1059 Floating point errors are reported poorly. E.g. on x86 OpenBSD
1062 debugger invoked on condition of type SB-KERNEL:FLOATING-POINT-EXCEPTION:
1063 An arithmetic error SB-KERNEL:FLOATING-POINT-EXCEPTION was signalled.
1064 No traps are enabled? How can this be?
1065 It should be possible to be much more specific (overflow, division
1066 by zero, etc.) and of course the "How can this be?" should be fixable.
1068 See also bugs #45.c and #183
1071 In sbcl-0.7.1.3 on x86, COMPILE-FILE on the file
1072 (in-package :cl-user)
1075 (defstruct foo bar bletch)
1077 (labels ((kidify1 (kid)
1082 (m+ (frobnicate kid)
1085 (declare (inline kid-frob))
1088 (the simple-vector (foo-bar perd)))))
1090 debugger invoked on condition of type TYPE-ERROR:
1091 The value NIL is not of type SB-C::NODE.
1092 The location of this failure has moved around as various related
1093 issues were cleaned up. As of sbcl-0.7.1.9, it occurs in
1094 NODE-BLOCK called by LAMBDA-COMPONENT called by IR2-CONVERT-CLOSURE.
1097 (essentially the same problem as a CMU CL bug reported by Martin
1098 Cracauer on cmucl-imp 2002-02-19)
1099 There is a hole in structure slot type checking. Compiling and LOADing
1100 (declaim (optimize safety))
1102 (bla 0 :type fixnum))
1104 (let ((foo (make-foo)))
1105 (setf (foo-bla foo) '(1 . 1))
1106 (format t "Is ~a of type ~a a cons? => ~a~%"
1108 (type-of (foo-bla foo))
1109 (consp (foo-bla foo)))))
1111 should signal an error, but in sbcl-0.7.1.21 instead gives the output
1112 Is (1 . 1) of type CONS a cons? => NIL
1113 without signalling an error.
1116 Functions SUBTYPEP, TYPEP, UPGRADED-ARRAY-ELEMENT-TYPE, and
1117 UPGRADED-COMPLEX-PART-TYPE should have an optional environment argument.
1118 (reported by Alexey Dejneka sbcl-devel 2002-04-12)
1121 (reported by Robert E. Brown 2002-04-16)
1122 When a function is called with too few arguments, causing the
1123 debugger to be entered, the uninitialized slots in the bad call frame
1124 seem to cause GCish problems, being interpreted as tagged data even
1125 though they're not. In particular, executing ROOM in the
1126 debugger at that point causes AVER failures:
1129 * (lisp-implementation-version)
1135 failed AVER: "(SAP= CURRENT END)"
1136 (Christophe Rhodes reports that this doesn't occur on the SPARC, which
1137 isn't too surprising since there are many differences in stack
1138 implementation and GC conservatism between the X86 and other ports.)
1141 Array types with element-types of some unknown type are falsely being
1142 assumed to be of type (ARRAY T) by the compiler in some cases. The
1143 following code demonstrates the problem:
1146 (declare (type (vector bar) x))
1148 (deftype bar () 'single-float)
1149 (foo (make-array 3 :element-type 'bar))
1150 -> TYPE-ERROR "The value #(0.0 0.0 0.0) is not of type (VECTOR BAR)."
1151 (typep (make-array 3 :element-type 'bar) '(vector bar))
1154 The easy solution is to make the functions which depend on knowing
1155 the upgraded-array-element-type (in compiler/array-tran and
1156 compiler/generic/vm-tran as of sbcl-0.7.3.x) be slightly smarter about
1157 unknown types; an alternative is to have the
1158 specialized-element-type slot in the ARRAY-TYPE structure be
1159 *WILD-TYPE* for UNKNOWN-TYPE element types.
1163 (in-package :cl-user)
1165 (defmethod permanentize ((uustk uustk))
1166 (flet ((frob (hash-table test-for-deletion)
1168 (obj-entry.stale? (oe)
1169 (destructuring-bind (key . datum) oe
1170 (declare (type simple-vector key))
1171 (deny0 (void? datum))
1172 (some #'stale? key))))
1173 (declare (inline frob obj-entry.stale?))
1174 (frob (uustk.args-hash->obj-alist uustk)
1176 (frob (uustk.hash->memoized-objs-list uustk)
1179 in sbcl-0.7.3.11 causes an assertion failure,
1182 (AND (NULL (BLOCK-SUCC B))
1183 (NOT (BLOCK-DELETE-P B))
1184 (NOT (EQ B (COMPONENT-HEAD #)))))"
1187 In sbcl-0.7.3.11, compiling the (illegal) code
1188 (in-package :cl-user)
1189 (defmethod prove ((uustk uustk))
1190 (zap ((frob () nil))
1192 gives the (not terribly clear) error message
1194 ; (during macroexpansion of (DEFMETHOD PROVE ...))
1195 ; can't get template for (FROB NIL NIL)
1196 The problem seems to be that the code walker used by the DEFMETHOD
1197 macro is unhappy with the illegal syntax in the method body, and
1198 is giving an unclear error message.
1201 (reported by Dan Barlow on sbcl-devel 2002-05-10)
1202 In sbcl-0.7.3.12, doing
1203 (defstruct foo bar baz)
1204 (compile nil (lambda (x) (or x (foo-baz x))))
1206 debugger invoked on condition of type SB-INT:BUG:
1207 full call to SB-KERNEL:%INSTANCE-REF
1208 This is probably a bug in SBCL itself. [...]
1209 Since this is a reasonable user error, it shouldn't be reported as
1213 (reported by Pierre Mai while investigating bug 47):
1214 (DEFCLASS FOO () ((A :SILLY T)))
1215 signals a SIMPLE-ERROR, not a PROGRAM-ERROR.
1218 sbcl's treatment of at least macro lambda lists is too permissive;
1219 e.g., in sbcl-0.7.3.7:
1220 (defmacro foo (&rest rest bar) `(,bar ,rest))
1221 (macroexpand '(foo quux zot)) -> (QUUX (QUUX ZOT))
1222 whereas section 3.4.4 of the CLHS doesn't allow required parameters
1223 to come after the rest argument.
1226 The compiler sometimes tries to constant-fold expressions before
1227 it checks to see whether they can be reached. This can lead to
1228 bogus warnings about errors in the constant folding, e.g. in code
1231 (WRITE-STRING (> X 0) "+" "0"))
1232 compiled in a context where the compiler can prove that X is NIL,
1233 and the compiler complains that (> X 0) causes a type error because
1234 NIL isn't a valid argument to #'>. Until sbcl-0.7.4.10 or so this
1235 caused a full WARNING, which made the bug really annoying because then
1236 COMPILE and COMPILE-FILE returned FAILURE-P=T for perfectly legal
1237 code. Since then the warning has been downgraded to STYLE-WARNING,
1238 so it's still a bug but at least it's a little less annoying.
1241 The error message from attempting to use a #\Return format
1243 (format nil "~^M") ; replace "^M" with a literal #\Return
1244 debugger invoked on condition of type SB-FORMAT::FORMAT-ERROR:
1245 error in format: unknown format directive
1248 is not terribly helpful; this is more noticeable than parallel cases
1249 with e.g. #\Backspace because of the differing newline conventions
1250 on various operating systems. (reported by Harald Hanche-Olsen on
1251 cmucl-help 2002-05-31)
1254 reported by Alexey Dejneka 08 Jun 2002 in sbcl-devel:
1255 Playing with McCLIM, I've received an error "Unbound variable WRAPPER
1256 in SB-PCL::CHECK-WRAPPER-VALIDITY".
1257 (defun check-wrapper-validity (instance)
1258 (let* ((owrapper (wrapper-of instance)))
1259 (if (not (invalid-wrapper-p owrapper))
1261 (let* ((state (wrapper-state wrapper)) ; !!!
1263 I've tried to replace it with OWRAPPER, but now OBSOLETE-INSTANCE-TRAP
1264 breaks with "NIL is not of type SB-KERNEL:LAYOUT".
1266 partial fix: The undefined variable WRAPPER resulted from an error
1267 in recent refactoring, as can be seen by comparing to the code in e.g.
1268 sbcl-0.7.2. Replacing WRAPPER with OWRAPPER (done by WHN in sbcl-0.7.4.22)
1269 should bring the code back to its behavior as of sbcl-0.7.2, but
1270 that still leaves the OBSOLETE-INSTANCE-TRAP bug. An example of
1271 input which triggers that bug is
1273 (let ((lastname (intern (format nil "C~D" (1- i))))
1274 (name (intern (format nil "C~D" i))))
1275 (eval `(defclass ,name
1276 (,@(if (= i 0) nil (list lastname)))
1278 (eval `(defmethod initialize-instance :after ((x ,name) &rest any)
1279 (declare (ignore any))))))
1281 (defclass c0 (b) ())
1282 (make-instance 'c19)
1286 178: "AVER failure compiling confused THEs in FUNCALL"
1287 In sbcl-0.7.4.24, compiling
1289 (funcall (the function (the standard-object x))))
1292 "(AND (EQ (IR2-CONTINUATION-PRIMITIVE-TYPE 2CONT) FUNCTION-PTYPE) (EQ CHECK T))"
1293 This variant compiles OK, though:
1294 (defun bug178alternative (x)
1295 (funcall (the nil x)))
1297 181: "bad type specifier drops compiler into debugger"
1299 (in-package :cl-user)
1301 (declare (type 0 x))
1304 bad thing to be a type specifier: 0
1305 which seems fine, but also enters the debugger (instead of having
1306 the compiler handle the error, convert it into a COMPILER-ERROR, and
1307 continue compiling) which seems wrong.
1309 183: "IEEE floating point issues"
1310 Even where floating point handling is being dealt with relatively
1311 well (as of sbcl-0.7.5, on sparc/sunos and alpha; see bug #146), the
1312 accrued-exceptions and current-exceptions part of the fp control
1313 word don't seem to bear much relation to reality. E.g. on
1317 debugger invoked on condition of type DIVISION-BY-ZERO:
1318 arithmetic error DIVISION-BY-ZERO signalled
1319 0] (sb-vm::get-floating-point-modes)
1321 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
1322 :ROUNDING-MODE :NEAREST
1323 :CURRENT-EXCEPTIONS NIL
1324 :ACCRUED-EXCEPTIONS (:INEXACT)
1327 * (sb-vm::get-floating-point-modes)
1328 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
1329 :ROUNDING-MODE :NEAREST
1330 :CURRENT-EXCEPTIONS (:INEXACT)
1331 :ACCRUED-EXCEPTIONS (:INEXACT)
1334 185: "top-level forms at the REPL"
1335 * (locally (defstruct foo (a 0 :type fixnum)))
1338 ; (in macroexpansion of (SB-KERNEL::%DELAYED-GET-COMPILER-LAYOUT BAR))
1339 however, compiling and loading the same expression in a file works
1342 187: "type inference confusion around DEFTRANSFORM time"
1343 (reported even more verbosely on sbcl-devel 2002-06-28 as "strange
1344 bug in DEFTRANSFORM")
1345 After the file below is compiled and loaded in sbcl-0.7.5, executing
1346 (TCX (MAKE-ARRAY 4 :FILL-POINTER 2) 0)
1347 at the REPL returns an adjustable vector, which is wrong. Presumably
1348 somehow the DERIVE-TYPE information for the output values of %WAD is
1349 being mispropagated as a type constraint on the input values of %WAD,
1350 and so causing the type test to be optimized away. It's unclear how
1351 hand-expanding the DEFTRANSFORM would change this, but it suggests
1352 the DEFTRANSFORM machinery (or at least the way DEFTRANSFORMs are
1353 invoked at a particular phase) is involved.
1354 (cl:in-package :sb-c)
1355 (eval-when (:compile-toplevel)
1356 ;;; standin for %DATA-VECTOR-AND-INDEX
1357 (defknown %dvai (array index)
1359 (foldable flushable))
1360 (deftransform %dvai ((array index)
1364 (let* ((atype (continuation-type array))
1365 (eltype (array-type-specialized-element-type atype)))
1366 (when (eq eltype *wild-type*)
1367 (give-up-ir1-transform
1368 "specialized array element type not known at compile-time"))
1369 (when (not (array-type-complexp atype))
1370 (give-up-ir1-transform "SIMPLE array!"))
1371 `(if (array-header-p array)
1372 (%wad array index nil)
1373 (values array index))))
1374 ;;; standin for %WITH-ARRAY-DATA
1375 (defknown %wad (array index (or index null))
1376 (values (simple-array * (*)) index index index)
1377 (foldable flushable))
1378 ;;; (Commenting out this optimizer causes the bug to go away.)
1379 (defoptimizer (%wad derive-type) ((array start end))
1380 (let ((atype (continuation-type array)))
1381 (when (array-type-p atype)
1382 (values-specifier-type
1383 `(values (simple-array ,(type-specifier
1384 (array-type-specialized-element-type atype))
1386 index index index)))))
1388 (defun %wad (array start end)
1389 (format t "~&in %WAD~%")
1390 (%with-array-data array start end))
1391 (cl:in-package :cl-user)
1393 (declare (type (vector t) v))
1394 (declare (notinline sb-kernel::%with-array-data))
1395 ;; (Hand-expending DEFTRANSFORM %DVAI here also causes the bug to
1399 188: "compiler performance fiasco involving type inference and UNION-TYPE"
1400 (In sbcl-0.7.6.10, DEFTRANSFORM CONCATENATE was commented out until this
1401 bug could be fixed properly, so you won't see the bug unless you restore
1402 the DEFTRANSFORM by hand.) In sbcl-0.7.5.11 on a 700 MHz Pentium III,
1406 (declare (optimize (safety 3)))
1407 (declare (optimize (compilation-speed 2)))
1408 (declare (optimize (speed 1) (debug 1) (space 1)))
1409 (let ((fn "if-this-file-exists-the-universe-is-strange"))
1410 (load fn :if-does-not-exist nil)
1411 (load (concatenate 'string fn ".lisp") :if-does-not-exist nil)
1412 (load (concatenate 'string fn ".fasl") :if-does-not-exist nil)
1413 (load (concatenate 'string fn ".misc-garbage")
1414 :if-does-not-exist nil)))))
1416 134.552 seconds of real time
1417 133.35156 seconds of user run time
1418 0.03125 seconds of system run time
1419 [Run times include 2.787 seconds GC run time.]
1421 246883368 bytes consed.
1422 BACKTRACE from Ctrl-C in the compilation shows that the compiler is
1423 thinking about type relationships involving types like
1425 (OR (INTEGER 576 576)
1436 190: "PPC/Linux pipe? buffer? bug"
1437 In sbcl-0.7.6, the run-program.test.sh test script sometimes hangs
1438 on the PPC/Linux platform, waiting for a zombie env process. This
1439 is a classic symptom of buffer filling and deadlock, but it seems
1440 only sporadically reproducible.
1442 191: "Miscellaneous PCL deficiencies"
1443 (reported by Alexey Dejenka sbcl-devel 2002-08-04)
1444 a. DEFCLASS does not inform the compiler about generated
1445 functions. Compiling a file with
1446 (DEFCLASS A-CLASS ()
1448 (DEFUN A-CLASS-X (A)
1449 (WITH-SLOTS (A-CLASS-X) A
1451 results in a STYLE-WARNING:
1453 SB-SLOT-ACCESSOR-NAME::|COMMON-LISP-USER A-CLASS-X slot READER|
1454 b. DEFGENERIC does not check lambda list syntax; from the REPL:
1455 * (defgeneric gf ("a" #p"b"))
1457 #<STANDARD-GENERIC-FUNCTION GF (0)>
1459 c. the examples in CLHS 7.6.5.1 (regarding generic function lambda
1460 lists and &KEY arguments) do not signal errors when they should.
1463 DEFUNCT CATEGORIES OF BUGS
1465 These labels were used for bugs related to the old IR1 interpreter.
1466 The # values reached 6 before the category was closed down.