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).
36 DEFSTRUCT almost certainly should overwrite the old LAYOUT information
37 instead of just punting when a contradictory structure definition
38 is loaded. As it is, if you redefine DEFSTRUCTs in a way which
39 changes their layout, you probably have to rebuild your entire
40 program, even if you know or guess enough about the internals of
41 SBCL to wager that this (undefined in ANSI) operation would be safe.
43 3: "type checking of structure slots"
45 ANSI specifies that a type mismatch in a structure slot
46 initialization value should not cause a warning.
48 This one might not be fixed for a while because while we're big
49 believers in ANSI compatibility and all, (1) there's no obvious
50 simple way to do it (short of disabling all warnings for type
51 mismatches everywhere), and (2) there's a good portable
52 workaround, and (3) by their own reasoning, it looks as though
53 ANSI may have gotten it wrong. ANSI justifies this specification
55 The restriction against issuing a warning for type mismatches
56 between a slot-initform and the corresponding slot's :TYPE
57 option is necessary because a slot-initform must be specified
58 in order to specify slot options; in some cases, no suitable
60 However, in SBCL (as in CMU CL or, for that matter, any compiler
61 which really understands Common Lisp types) a suitable default
62 does exist, in all cases, because the compiler understands the
63 concept of functions which never return (i.e. has return type NIL).
64 Thus, as a portable workaround, you can use a call to some
65 known-never-to-return function as the default. E.g.
67 (BAR (ERROR "missing :BAR argument")
68 :TYPE SOME-TYPE-TOO-HAIRY-TO-CONSTRUCT-AN-INSTANCE-OF))
70 (DECLAIM (FTYPE (FUNCTION () NIL) MISSING-ARG))
71 (DEFUN REQUIRED-ARG () ; workaround for SBCL non-ANSI slot init typing
72 (ERROR "missing required argument"))
74 (BAR (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
75 (BLETCH (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
76 (N-REFS-SO-FAR 0 :TYPE (INTEGER 0)))
77 Such code should compile without complaint and work correctly either
78 on SBCL or on any other completely compliant Common Lisp system.
81 And as long as we're wishing, it would be awfully nice if INSPECT could
82 also report on closures, telling about the values of the bound variables.
84 Currently INSPECT and DESCRIBE do show the values, but showing the
85 names of the bindings would be even nicer.
88 The compiler assumes that any time a function of declared FTYPE
89 doesn't signal an error, its arguments were of the declared type.
90 E.g. compiling and loading
91 (DECLAIM (OPTIMIZE (SAFETY 3)))
92 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
94 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE)) FACTORIAL))
96 (COND ((> (FACTORIAL X) 1.0E6)
97 (FORMAT T "too big~%"))
99 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
101 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
104 will cause the INTEGERP case to be selected, giving bogus output a la
106 This violates the "declarations are assertions" principle.
107 According to the ANSI spec, in the section "System Class FUNCTION",
108 this is a case of "lying to the compiler", but the lying is done
109 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
110 not by the unexpectedly general definition of FACTORIAL. In any case,
111 "declarations are assertions" means that lying to the compiler should
112 cause an error to be signalled, and should not cause a bogus
113 result to be returned. Thus, the compiler should not assume
114 that arbitrary functions check their argument types. (It might
115 make sense to add another flag (CHECKED?) to DEFKNOWN to
116 identify functions which *do* check their argument types.)
117 (Also, verify that the compiler handles declared function
118 return types as assertions.)
121 The definitions of SIGCONTEXT-FLOAT-REGISTER and
122 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
123 supported on FreeBSD because the floating point state is not saved,
124 but at least as of FreeBSD 4.0, the floating point state *is* saved,
125 so they could be supported after all. Very likely
126 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
129 Compiling and loading
130 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
132 then requesting a BACKTRACE at the debugger prompt gives no information
133 about where in the user program the problem occurred.
135 (this is apparently mostly fixed on the SPARC, PPC, and x86 architectures:
136 while giving the backtrace the non-x86 systems complains about "unknown
137 source location: using block start", but apart from that the
138 backtrace seems reasonable. On x86 this is masked by bug 353. See
139 tests/debug.impure.lisp for a test case)
142 Using the pretty-printer from the command prompt gives funny
143 results, apparently because the pretty-printer doesn't know
144 about user's command input, including the user's carriage return
145 that the user, and therefore the pretty-printer thinks that
146 the new output block should start indented 2 or more characters
147 rightward of the correct location.
150 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
151 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
152 crashes SBCL. In general tracing anything which is used in the
153 implementation of TRACE is likely to have the same problem.
156 ANSI says in one place that type declarations can be abbreviated even
157 when the type name is not a symbol, e.g.
158 (DECLAIM ((VECTOR T) *FOOVECTOR*))
159 SBCL doesn't support this. But ANSI says in another place that this
160 isn't allowed. So it's not clear this is a bug after all. (See the
161 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
164 (Actually this has changed changed since, and types as above are
165 now supported. This may be a bug.)
168 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
169 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
170 the mean of the distribution can be systematically O(0.1%) wrong.
171 Just increasing R-I-E-B is probably not a good solution, since
172 it would decrease efficiency more than is probably necessary. Perhaps
173 using some sort of accept/reject method would be better.
176 Internally the compiler sometimes evaluates
177 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
178 (I stumbled across this when I added an
179 (assert (not (eq type1 *wild-type*)))
180 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
181 in a type context should probably be translated to T, and so it's
182 probably wrong to ask whether it's equal to the T type and then (using
183 the EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
184 (I haven't tried to investigate this bug enough to guess whether
185 there might be any user-level symptoms.)
187 In fact, the type system is likely to depend on this inequality not
188 holding... * is not equivalent to T in many cases, such as
189 (VECTOR *) /= (VECTOR T).
192 In sbcl-0.6.11.41 (and in all earlier SBCL, and in CMU
193 CL), out-of-line structure slot setters are horribly inefficient
194 whenever the type of the slot is declared, because out-of-line
195 structure slot setters are implemented as closures to save space,
196 so the compiler doesn't compile the type test into code, but
197 instead just saves the type in a lexical closure and interprets it
199 To exercise the problem, compile and load
200 (cl:in-package :cl-user)
202 (bar (error "missing") :type bar))
205 (loop (setf (foo-bar *foo*) x)))
207 (defvar *bar* (make-bar))
208 (defvar *foo* (make-foo :bar *bar*))
209 (defvar *setf-foo-bar* #'(setf foo-bar))
211 (loop (funcall *setf-foo-bar* x *foo*)))
212 then run (WASTREL1 *BAR*) or (WASTREL2 *BAR*), hit Ctrl-C, and
213 use BACKTRACE, to see it's spending all essentially all its time
214 in %TYPEP and VALUES-SPECIFIER-TYPE and so forth.
215 One possible solution would be simply to give up on
216 representing structure slot accessors as functions, and represent
217 them as macroexpansions instead. This can be inconvenient for users,
218 but it's not clear that it's worse than trying to help by expanding
219 into a horribly inefficient implementation.
220 As a workaround for the problem, #'(SETF FOO) expressions
221 can be replaced with (EFFICIENT-SETF-FUNCTION FOO), where
222 (defmacro efficient-setf-function (place-function-name)
223 (or #+sbcl (and (sb-int:info :function :accessor-for place-function-name)
224 ;; a workaround for the problem, encouraging the
225 ;; inline expansion of the structure accessor, so
226 ;; that the compiler can optimize its type test
227 (let ((new-value (gensym "NEW-VALUE-"))
228 (structure-value (gensym "STRUCTURE-VALUE-")))
229 `(lambda (,new-value ,structure-value)
230 (setf (,place-function-name ,structure-value)
232 ;; no problem, can just use the ordinary expansion
233 `(function (setf ,place-function-name))))
236 There's apparently a bug in CEILING optimization which caused
237 Douglas Crosher to patch the CMU CL version. Martin Atzmueller
238 applied the patches to SBCL and they didn't seem to cause problems
239 (as reported sbcl-devel 2001-05-04). However, since the patches
240 modify nontrivial code which was apparently written incorrectly
241 the first time around, until regression tests are written I'm not
242 comfortable merging the patches in the CVS version of SBCL.
247 a) ROOM works by walking over the heap linearly, instead of
248 following the object graph. Hence, it report garbage objects that
249 are unreachable. (Maybe this is a feature and not a bug?)
251 b) ROOM uses MAP-ALLOCATED-OBJECTS to walk the heap, which doesn't
252 check all pointers as well as it should, and can hence become
253 confused, leading to aver failures. As of 1.0.13.21 these (the
254 SAP= aver in particular) should be mostly under control, but push
255 ROOM hard enough and it still might croak.
258 When the compiler inline expands functions, it may be that different
259 kinds of return values are generated from different code branches.
260 E.g. an inline expansion of POSITION generates integer results
261 from one branch, and NIL results from another. When that inline
262 expansion is used in a context where only one of those results
265 (aref *a1* (position x *a2*)))
266 and the compiler can't prove that the unacceptable branch is
267 never taken, then bogus type mismatch warnings can be generated.
268 If you need to suppress the type mismatch warnings, you can
269 suppress the inline expansion,
271 #+sbcl (declare (notinline position)) ; to suppress bug 117 bogowarnings
272 (aref *a1* (position x *a2*)))
273 or, sometimes, suppress them by declaring the result to be of an
276 (aref *a1* (the integer (position x *a2*))))
278 This is not a new compiler problem in 0.7.0, but the new compiler
279 transforms for FIND, POSITION, FIND-IF, and POSITION-IF make it
280 more conspicuous. If you don't need performance from these functions,
281 and the bogus warnings are a nuisance for you, you can return to
282 your pre-0.7.0 state of grace with
283 #+sbcl (declaim (notinline find position find-if position-if)) ; bug 117..
288 As of version 0.pre7.14, SBCL's implementation of MACROLET makes
289 the entire lexical environment at the point of MACROLET available
290 in the bodies of the macroexpander functions. In particular, it
291 allows the function bodies (which run at compile time) to try to
292 access lexical variables (which are only defined at runtime).
293 It doesn't even issue a warning, which is bad.
295 The SBCL behavior arguably conforms to the ANSI spec (since the
296 spec says that the behavior is undefined, ergo anything conforms).
297 However, it would be better to issue a compile-time error.
298 Unfortunately I (WHN) don't see any simple way to detect this
299 condition in order to issue such an error, so for the meantime
300 SBCL just does this weird broken "conforming" thing.
302 The ANSI standard says, in the definition of the special operator
304 The macro-expansion functions defined by MACROLET are defined
305 in the lexical environment in which the MACROLET form appears.
306 Declarations and MACROLET and SYMBOL-MACROLET definitions affect
307 the local macro definitions in a MACROLET, but the consequences
308 are undefined if the local macro definitions reference any
309 local variable or function bindings that are visible in that
311 Then it seems to contradict itself by giving the example
313 (macrolet ((fudge (z)
314 ;The parameters x and flag are not accessible
315 ; at this point; a reference to flag would be to
316 ; the global variable of that name.
317 ` (if flag (* ,z ,z) ,z)))
318 ;The parameters x and flag are accessible here.
322 The comment "a reference to flag would be to the global variable
323 of the same name" sounds like good behavior for the system to have.
324 but actual specification quoted above says that the actual behavior
327 (Since 0.7.8.23 macroexpanders are defined in a restricted version
328 of the lexical environment, containing no lexical variables and
329 functions, which seems to conform to ANSI and CLtL2, but signalling
330 a STYLE-WARNING for references to variables similar to locals might
334 Ideally, uninterning a symbol would allow it, and its associated
335 FDEFINITION and PROCLAIM data, to be reclaimed by the GC. However,
336 at least as of sbcl-0.7.0, this isn't the case. Information about
337 FDEFINITIONs and PROCLAIMed properties is stored in globaldb.lisp
338 essentially in ordinary (non-weak) hash tables keyed by symbols.
339 Thus, once a system has an entry in this system, it tends to live
340 forever, even when it is uninterned and all other references to it
344 (reported by Jesse Bouwman 2001-10-24 through the unfortunately
345 prominent SourceForge web/db bug tracking system, which is
346 unfortunately not a reliable way to get a timely response from
347 the SBCL maintainers)
348 In the course of trying to build a test case for an
349 application error, I encountered this behavior:
350 If you start up sbcl, and then lay on CTRL-C for a
351 minute or two, the lisp process will eventually say:
352 %PRIMITIVE HALT called; the party is over.
353 and throw you into the monitor. If I start up lisp,
354 attach to the process with strace, and then do the same
355 (abusive) thing, I get instead:
356 access failure in heap page not marked as write-protected
357 and the monitor again. I don't know enough to have the
358 faintest idea of what is going on here.
359 This is with sbcl 6.12, uname -a reports:
360 Linux prep 2.2.19 #4 SMP Tue Apr 24 13:59:52 CDT 2001 i686 unknown
361 I (WHN) have verified that the same thing occurs on sbcl-0.pre7.141
362 under OpenBSD 2.9 on my X86 laptop. Do be patient when you try it:
363 it took more than two minutes (but less than five) for me.
367 ANSI allows types `(COMPLEX ,FOO) to use very hairy values for
368 FOO, e.g. (COMPLEX (AND REAL (SATISFIES ODDP))). The old CMU CL
369 COMPLEX implementation didn't deal with this, and hasn't been
370 upgraded to do so. (This doesn't seem to be a high priority
371 conformance problem, since seems hard to construct useful code
374 [ partially fixed by CSR in 0.8.17.17 because of a PFD ansi-tests
375 report that (COMPLEX RATIO) was failing; still failing on types of
376 the form (AND NUMBER (SATISFIES REALP) (SATISFIES ZEROP)). ]
378 b. (fixed in 0.8.3.43)
381 Floating point errors are reported poorly. E.g. on x86 OpenBSD
384 debugger invoked on condition of type SB-KERNEL:FLOATING-POINT-EXCEPTION:
385 An arithmetic error SB-KERNEL:FLOATING-POINT-EXCEPTION was signalled.
386 No traps are enabled? How can this be?
387 It should be possible to be much more specific (overflow, division
388 by zero, etc.) and of course the "How can this be?" should be fixable.
390 See also bugs #45.c and #183
393 (reported by Robert E. Brown 2002-04-16)
394 When a function is called with too few arguments, causing the
395 debugger to be entered, the uninitialized slots in the bad call frame
396 seem to cause GCish problems, being interpreted as tagged data even
397 though they're not. In particular, executing ROOM in the
398 debugger at that point causes AVER failures:
401 * (lisp-implementation-version)
407 failed AVER: "(SAP= CURRENT END)"
408 (Christophe Rhodes reports that this doesn't occur on the SPARC, which
409 isn't too surprising since there are many differences in stack
410 implementation and GC conservatism between the X86 and other ports.)
412 (Can't reproduce on x86 linux as of 1.0.20.23 - MGL)
414 This is probably the same bug as 216
417 The compiler sometimes tries to constant-fold expressions before
418 it checks to see whether they can be reached. This can lead to
419 bogus warnings about errors in the constant folding, e.g. in code
422 (WRITE-STRING (> X 0) "+" "0"))
423 compiled in a context where the compiler can prove that X is NIL,
424 and the compiler complains that (> X 0) causes a type error because
425 NIL isn't a valid argument to #'>. Until sbcl-0.7.4.10 or so this
426 caused a full WARNING, which made the bug really annoying because then
427 COMPILE and COMPILE-FILE returned FAILURE-P=T for perfectly legal
428 code. Since then the warning has been downgraded to STYLE-WARNING,
429 so it's still a bug but at least it's a little less annoying.
431 183: "IEEE floating point issues"
432 Even where floating point handling is being dealt with relatively
433 well (as of sbcl-0.7.5, on sparc/sunos and alpha; see bug #146), the
434 accrued-exceptions and current-exceptions part of the fp control
435 word don't seem to bear much relation to reality. E.g. on
439 debugger invoked on condition of type DIVISION-BY-ZERO:
440 arithmetic error DIVISION-BY-ZERO signalled
441 0] (sb-vm::get-floating-point-modes)
443 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
444 :ROUNDING-MODE :NEAREST
445 :CURRENT-EXCEPTIONS NIL
446 :ACCRUED-EXCEPTIONS (:INEXACT)
449 * (sb-vm::get-floating-point-modes)
450 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
451 :ROUNDING-MODE :NEAREST
452 :CURRENT-EXCEPTIONS (:INEXACT)
453 :ACCRUED-EXCEPTIONS (:INEXACT)
456 188: "compiler performance fiasco involving type inference and UNION-TYPE"
460 (declare (optimize (safety 3)))
461 (declare (optimize (compilation-speed 2)))
462 (declare (optimize (speed 1) (debug 1) (space 1)))
464 (declare (type (integer 0) start))
465 (print (incf start 22))
466 (print (incf start 26))
467 (print (incf start 28)))
469 (declare (type (integer 0) start))
470 (print (incf start 22))
471 (print (incf start 26)))
473 (declare (type (integer 0) start))
474 (print (incf start 22))
475 (print (incf start 26))))))
477 [ Update: 1.0.14.36 improved this quite a bit (20-25%) by
478 eliminating useless work from PROPAGATE-FROM-SETS -- but as alluded
479 below, maybe we should be smarter about when to decide a derived
480 type is "good enough". ]
482 This example could be solved with clever enough constraint
483 propagation or with SSA, but consider
488 The careful type of X is {2k} :-(. Is it really important to be
489 able to work with unions of many intervals?
491 191: "Miscellaneous PCL deficiencies"
492 (reported by Alexey Dejneka sbcl-devel 2002-08-04)
493 a. DEFCLASS does not inform the compiler about generated
494 functions. Compiling a file with
498 (WITH-SLOTS (A-CLASS-X) A
500 results in a STYLE-WARNING:
502 SB-SLOT-ACCESSOR-NAME::|COMMON-LISP-USER A-CLASS-X slot READER|
504 APD's fix for this was checked in to sbcl-0.7.6.20, but Pierre
505 Mai points out that the declamation of functions is in fact
506 incorrect in some cases (most notably for structure
507 classes). This means that at present erroneous attempts to use
508 WITH-SLOTS and the like on classes with metaclass STRUCTURE-CLASS
509 won't get the corresponding STYLE-WARNING.
511 [much later, in 2006-08] in fact it's no longer erroneous to use
512 WITH-SLOTS on structure-classes. However, including :METACLASS
513 STRUCTURE-CLASS in the class definition gives a whole bunch of
514 function redefinition warnings, so we're still not good to close
517 c. (fixed in 0.8.4.23)
519 201: "Incautious type inference from compound types"
520 a. (reported by APD sbcl-devel 2002-09-17)
522 (LET ((Y (CAR (THE (CONS INTEGER *) X))))
524 (FORMAT NIL "~S IS ~S, Y = ~S"
531 (FOO ' (1 . 2)) => "NIL IS INTEGER, Y = 1"
535 (declare (type (array * (4 4)) x))
537 (setq x (make-array '(4 4)))
538 (adjust-array y '(3 5))
539 (= (array-dimension y 0) (eval `(array-dimension ,y 0)))))
541 * (foo (make-array '(4 4) :adjustable t))
544 205: "environment issues in cross compiler"
545 (These bugs have no impact on user code, but should be fixed or
547 a. Macroexpanders introduced with MACROLET are defined in the null
549 b. The body of (EVAL-WHEN (:COMPILE-TOPLEVEL) ...) is evaluated in
550 the null lexical environment.
551 c. The cross-compiler cannot inline functions defined in a non-null
554 207: "poorly distributed SXHASH results for compound data"
555 SBCL's SXHASH could probably try a little harder. ANSI: "the
556 intent is that an implementation should make a good-faith
557 effort to produce hash-codes that are well distributed
558 within the range of non-negative fixnums". But
559 (let ((hits (make-hash-table)))
562 (let* ((ij (cons i j))
563 (newlist (push ij (gethash (sxhash ij) hits))))
565 (format t "~&collision: ~S~%" newlist))))))
566 reports lots of collisions in sbcl-0.7.8. A stronger MIX function
567 would be an obvious way of fix. Maybe it would be acceptably efficient
568 to redo MIX using a lookup into a 256-entry s-box containing
569 29-bit pseudorandom numbers?
571 211: "keywords processing"
572 a. :ALLOW-OTHER-KEYS T should allow a function to receive an odd
573 number of keyword arguments.
575 212: "Sequence functions and circular arguments"
576 COERCE, MERGE and CONCATENATE go into an infinite loop when given
577 circular arguments; it would be good for the user if they could be
578 given an error instead (ANSI 17.1.1 allows this behaviour on the part
579 of the implementation, as conforming code cannot give non-proper
580 sequences to these functions. MAP also has this problem (and
581 solution), though arguably the convenience of being able to do
582 (MAP 'LIST '+ FOO '#1=(1 . #1#))
583 might be classed as more important (though signalling an error when
584 all of the arguments are circular is probably desireable).
586 213: "Sequence functions and type checking"
587 b. MAP, when given a type argument that is SUBTYPEP LIST, does not
588 check that it will return a sequence of the given type. Fixing
589 it along the same lines as the others (cf. work done around
590 sbcl-0.7.8.45) is possible, but doing so efficiently didn't look
591 entirely straightforward.
592 c. All of these functions will silently accept a type of the form
594 whether or not the return value is of this type. This is
595 probably permitted by ANSI (see "Exceptional Situations" under
596 ANSI MAKE-SEQUENCE), but the DERIVE-TYPE mechanism does not
597 know about this escape clause, so code of the form
598 (INTEGERP (CAR (MAKE-SEQUENCE '(CONS INTEGER *) 2)))
599 can erroneously return T.
601 215: ":TEST-NOT handling by functions"
603 We should verify that our handling of :TEST-NOT and :TEST is consistent
604 for all functions that accept them: that is, signal an error if both
607 Similarly, a compile-time full warning for calls with both would be good.
609 We might also consider a compile-time style warning for :TEST-NOT.
611 216: "debugger confused by frames with invalid number of arguments"
612 In sbcl-0.7.8.51, executing e.g. (VECTOR-PUSH-EXTEND T), BACKTRACE, Q
613 leaves the system confused, enough so that (QUIT) no longer works.
614 It's as though the process of working with the uninitialized slot in
615 the bad VECTOR-PUSH-EXTEND frame causes GC problems, though that may
616 not be the actual problem. (CMU CL 18c doesn't have problems with this.)
618 (Can't reproduce on x86 linux as of 1.0.20.22 - MGL)
620 This is probably the same bug as 162
622 235: "type system and inline expansion"
624 (declaim (ftype (function (cons) number) acc))
625 (declaim (inline acc))
627 (the number (car c)))
630 (values (locally (declare (optimize (safety 0)))
632 (locally (declare (optimize (safety 3)))
635 (foo '(nil) '(t)) => NIL, T.
637 As of 0.9.15.41 this seems to be due to ACC being inlined only once
638 inside FOO, which results in the second call reusing the FUNCTIONAL
639 resulting from the first -- which doesn't check the type.
641 237: "Environment arguments to type functions"
642 a. Functions SUBTYPEP, TYPEP, UPGRADED-ARRAY-ELEMENT-TYPE, and
643 UPGRADED-COMPLEX-PART-TYPE now have an optional environment
644 argument, but they ignore it completely. This is almost
645 certainly not correct.
646 b. Also, the compiler's optimizers for TYPEP have not been informed
647 about the new argument; consequently, they will not transform
648 calls of the form (TYPEP 1 'INTEGER NIL), even though this is
649 just as optimizeable as (TYPEP 1 'INTEGER).
651 238: "REPL compiler overenthusiasm for CLOS code"
653 * (defclass foo () ())
654 * (defmethod bar ((x foo) (foo foo)) (call-next-method))
655 causes approximately 100 lines of code deletion notes. Some
656 discussion on this issue happened under the title 'Three "interesting"
657 bugs in PCL', resulting in a fix for this oververbosity from the
658 compiler proper; however, the problem persists in the interactor
659 because the notion of original source is not preserved: for the
660 compiler, the original source of the above expression is (DEFMETHOD
661 BAR ((X FOO) (FOO FOO)) (CALL-NEXT-METHOD)), while by the time the
662 compiler gets its hands on the code needing compilation from the REPL,
663 it has been macroexpanded several times.
665 A symptom of the same underlying problem, reported by Tony Martinez:
667 (with-input-from-string (*query-io* " no")
669 (simple-type-error () 'error))
671 ; (SB-KERNEL:FLOAT-WAIT)
673 ; note: deleting unreachable code
674 ; compilation unit finished
677 242: "WRITE-SEQUENCE suboptimality"
678 (observed from clx performance)
679 In sbcl-0.7.13, WRITE-SEQUENCE of a sequence of type
680 (SIMPLE-ARRAY (UNSIGNED-BYTE 8) (*)) on a stream with element-type
681 (UNSIGNED-BYTE 8) will write to the stream one byte at a time,
682 rather than writing the sequence in one go, leading to severe
683 performance degradation.
684 As of sbcl-0.9.0.36, this is solved for fd-streams, so is less of a
685 problem in practice. (Fully fixing this would require adding a
686 ansi-stream-n-bout slot and associated methods to write a byte
687 sequence to ansi-stream, similar to the existing ansi-stream-sout
690 243: "STYLE-WARNING overenthusiasm for unused variables"
691 (observed from clx compilation)
692 In sbcl-0.7.14, in the presence of the macros
693 (DEFMACRO FOO (X) `(BAR ,X))
694 (DEFMACRO BAR (X) (DECLARE (IGNORABLE X)) 'NIL)
695 somewhat surprising style warnings are emitted for
696 (COMPILE NIL '(LAMBDA (Y) (FOO Y))):
698 ; (LAMBDA (Y) (FOO Y))
700 ; caught STYLE-WARNING:
701 ; The variable Y is defined but never used.
703 245: bugs in disassembler
704 b. On X86 operand size prefix is not recognized.
707 (defun foo (&key (a :x))
711 does not cause a warning. (BTW: old SBCL issued a warning, but for a
712 function, which was never called!)
715 Compiler does not emit warnings for
717 a. (lambda () (svref (make-array 8 :adjustable t) 1))
719 b. (fixed at some point before 1.0.4.10)
722 (declare (optimize (debug 0)))
723 (declare (type vector x))
724 (list (fill-pointer x)
728 Complex array type does not have corresponding type specifier.
730 This is a problem because the compiler emits optimization notes when
731 you use a non-simple array, and without a type specifier for hairy
732 array types, there's no good way to tell it you're doing it
733 intentionally so that it should shut up and just compile the code.
735 Another problem is confusing error message "asserted type ARRAY
736 conflicts with derived type (VALUES SIMPLE-VECTOR &OPTIONAL)" during
737 compiling (LAMBDA (V) (VALUES (SVREF V 0) (VECTOR-POP V))).
739 The last problem is that when type assertions are converted to type
740 checks, types are represented with type specifiers, so we could lose
741 complex attribute. (Now this is probably not important, because
742 currently checks for complex arrays seem to be performed by
746 (compile nil '(lambda () (aref (make-array 0) 0))) compiles without
747 warning. Analogous cases with the index and length being equal and
748 greater than 0 are warned for; the problem here seems to be that the
749 type required for an array reference of this type is (INTEGER 0 (0))
750 which is canonicalized to NIL.
755 (t1 (specifier-type s)))
756 (eval `(defstruct ,s))
757 (type= t1 (specifier-type s)))
762 b. The same for CSUBTYPEP.
764 262: "yet another bug in inline expansion of local functions"
765 During inline expansion of a local function Python can try to
766 reference optimized away objects (functions, variables, CTRANs from
767 tags and blocks), which later may lead to problems. Some of the
768 cases are worked around by forbidding expansion in such cases, but
769 the better way would be to reimplement inline expansion by copying
773 David Lichteblau provided (sbcl-devel 2003-06-01) a patch to fix
774 behaviour of streams with element-type (SIGNED-BYTE 8). The patch
775 looks reasonable, if not obviously correct; however, it caused the
776 PPC/Linux port to segfault during warm-init while loading
777 src/pcl/std-class.fasl. A workaround patch was made, but it would
778 be nice to understand why the first patch caused problems, and to
779 fix the cause if possible.
781 268: "wrong free declaration scope"
782 The following code must signal type error:
784 (locally (declare (optimize (safety 3)))
785 (flet ((foo (x &optional (y (car x)))
786 (declare (optimize (safety 0)))
788 (funcall (eval #'foo) 1)))
791 In the following function constraint propagator optimizes nothing:
794 (declare (integer x))
795 (declare (optimize speed))
803 Compilation of the following two forms causes "X is unbound" error:
805 (symbol-macrolet ((x pi))
806 (macrolet ((foo (y) (+ x y)))
807 (declaim (inline bar))
813 (See (COERCE (CDR X) 'FUNCTION) in IR1-CONVERT-INLINE-LAMBDA.)
816 CLHS says that type declaration of a symbol macro should not affect
817 its expansion, but in SBCL it does. (If you like magic and want to
818 fix it, don't forget to change all uses of MACROEXPAND to
822 The following code (taken from CLOCC) takes a lot of time to compile:
825 (declare (type (integer 0 #.large-constant) n))
828 (fixed in 0.8.2.51, but a test case would be good)
830 279: type propagation error -- correctly inferred type goes astray?
831 In sbcl-0.8.3 and sbcl-0.8.1.47, the warning
832 The binding of ABS-FOO is a (VALUES (INTEGER 0 0)
833 &OPTIONAL), not a (INTEGER 1 536870911)
834 is emitted when compiling this file:
835 (declaim (ftype (function ((integer 0 #.most-positive-fixnum))
836 (integer #.most-negative-fixnum 0))
841 (let* (;; Uncomment this for a type mismatch warning indicating
842 ;; that the type of (FOO X) is correctly understood.
843 #+nil (fs-foo (float-sign (foo x)))
844 ;; Uncomment this for a type mismatch warning
845 ;; indicating that the type of (ABS (FOO X)) is
846 ;; correctly understood.
847 #+nil (fs-abs-foo (float-sign (abs (foo x))))
848 ;; something wrong with this one though
849 (abs-foo (abs (foo x))))
850 (declare (type (integer 1 100) abs-foo))
855 283: Thread safety: libc functions
856 There are places that we call unsafe-for-threading libc functions
857 that we should find alternatives for, or put locks around. Known or
858 strongly suspected problems, as of 1.0.3.13: please update this
859 bug instead of creating new ones
861 284: Thread safety: special variables
862 There are lots of special variables in SBCL, and I feel sure that at
863 least some of them are indicative of potentially thread-unsafe
864 parts of the system. See doc/internals/notes/threading-specials
866 286: "recursive known functions"
867 Self-call recognition conflicts with known function
868 recognition. Currently cross compiler and target COMPILE do not
869 recognize recursion, and in target compiler it can be disabled. We
870 can always disable it for known functions with RECURSIVE attribute,
871 but there remains a possibility of a function with a
872 (tail)-recursive simplification pass and transforms/VOPs for base
875 288: fundamental cross-compilation issues (from old UGLINESS file)
876 Using host floating point numbers to represent target floating point
877 numbers, or host characters to represent target characters, is
878 theoretically shaky. (The characters are OK as long as the characters
879 are in the ANSI-guaranteed character set, though, so they aren't a
880 real problem as long as the sources don't need anything but that;
881 the floats are a real problem.)
883 289: "type checking and source-transforms"
885 (block nil (let () (funcall #'+ (eval 'nil) (eval '1) (return :good))))
888 Our policy is to check argument types at the moment of a call. It
889 disagrees with ANSI, which says that type assertions are put
890 immediately onto argument expressions, but is easier to implement in
891 IR1 and is more compatible to type inference, inline expansion,
892 etc. IR1-transforms automatically keep this policy, but source
893 transforms for associative functions (such as +), being applied
894 during IR1-convertion, do not. It may be tolerable for direct calls
895 (+ x y z), but for (FUNCALL #'+ x y z) it is non-conformant.
897 b. Another aspect of this problem is efficiency. [x y + z +]
898 requires less registers than [x y z + +]. This transformation is
899 currently performed with source transforms, but it would be good to
900 also perform it in IR1 optimization phase.
902 290: Alpha floating point and denormalized traps
903 In SBCL 0.8.3.6x on the alpha, we work around what appears to be a
904 hardware or kernel deficiency: the status of the enable/disable
905 denormalized-float traps bit seems to be ambiguous; by the time we
906 get to os_restore_fp_control after a trap, denormalized traps seem
907 to be enabled. Since we don't want a trap every time someone uses a
908 denormalized float, in general, we mask out that bit when we restore
909 the control word; however, this clobbers any change the user might
913 LOOP with non-constant arithmetic step clauses suffers from overzealous
914 type constraint: code of the form
915 (loop for d of-type double-float from 0d0 to 10d0 by x collect d)
916 compiles to a type restriction on X of (AND DOUBLE-FLOAT (REAL
917 (0))). However, an integral value of X should be legal, because
918 successive adds of integers to double-floats produces double-floats,
919 so none of the type restrictions in the code is violated.
921 300: (reported by Peter Graves) Function PEEK-CHAR checks PEEK-TYPE
922 argument type only after having read a character. This is caused
923 with EXPLICIT-CHECK attribute in DEFKNOWN. The similar problem
924 exists with =, /=, <, >, <=, >=. They were fixed, but it is probably
925 less error prone to have EXPLICIT-CHECK be a local declaration,
926 being put into the definition, instead of an attribute being kept in
927 a separate file; maybe also put it into SB-EXT?
929 301: ARRAY-SIMPLE-=-TYPE-METHOD breaks on corner cases which can arise
930 in NOTE-ASSUMED-TYPES
931 In sbcl-0.8.7.32, compiling the file
933 (declare (type integer x))
934 (declare (type (vector (or hash-table bit)) y))
937 (declare (type integer x))
938 (declare (type (simple-array base (2)) y))
941 failed AVER: "(NOT (AND (NOT EQUALP) CERTAINP))"
943 303: "nonlinear LVARs" (aka MISC.293)
945 (multiple-value-call #'list
947 (multiple-value-prog1
948 (eval '(values :a :b :c))
954 (throw 'bar (values 3 4)))))))))))
956 (BUU 1) returns garbage.
958 The problem is that both EVALs sequentially write to the same LVAR.
960 306: "Imprecise unions of array types"
962 a. fixed in SBCL 0.9.15.48
967 ,@(loop for x across sb-vm:*specialized-array-element-type-properties*
968 collect `(array ,(sb-vm:saetp-specifier x)))))
969 => NIL, T (when it should be T, T)
971 309: "Dubious values for implementation limits"
972 (reported by Bruno Haible sbcl-devel "Incorrect value of
973 multiple-values-limit" 2004-04-19)
974 (values-list (make-list 1000000)), on x86/linux, signals a stack
975 exhaustion condition, despite MULTIPLE-VALUES-LIMIT being
976 significantly larger than 1000000. There are probably similar
977 dubious values for CALL-ARGUMENTS-LIMIT (see cmucl-help/cmucl-imp
978 around the same time regarding a call to LIST on sparc with 1000
979 arguments) and other implementation limit constants.
981 314: "LOOP :INITIALLY clauses and scope of initializers"
982 reported by Bruno Haible sbcl-devel "various SBCL bugs" from CLISP
983 test suite, originally by Thomas F. Burdick.
984 ;; <http://www.lisp.org/HyperSpec/Body/sec_6-1-7-2.html>
985 ;; According to the HyperSpec 6.1.2.1.4, in for-as-equals-then, var is
986 ;; initialized to the result of evaluating form1. 6.1.7.2 says that
987 ;; initially clauses are evaluated in the loop prologue, which precedes all
988 ;; loop code except for the initial settings provided by with, for, or as.
989 (loop :for x = 0 :then (1+ x)
990 :for y = (1+ x) :then (ash y 1)
991 :for z :across #(1 3 9 27 81 243)
993 :initially (assert (zerop x)) :initially (assert (= 2 w))
994 :until (>= w 100) :collect w)
995 Expected: (2 6 15 38)
998 318: "stack overflow in compiler warning with redefined class"
999 reported by Bruno Haible sbcl-devel "various SBCL bugs" from CLISP
1002 (setf (find-class 'foo) nil)
1003 (defstruct foo slot-1)
1004 This used to give a stack overflow from within the printer, which has
1005 been fixed as of 0.8.16.11. Current result:
1007 ; can't compile TYPEP of anonymous or undefined class:
1008 ; #<SB-KERNEL:STRUCTURE-CLASSOID FOO>
1010 debugger invoked on a TYPE-ERROR in thread 19973:
1011 The value NIL is not of type FUNCTION.
1013 CSR notes: it's not really clear what it should give: is (SETF FIND-CLASS)
1014 meant to be enough to delete structure classes from the system?
1016 319: "backquote with comma inside array"
1017 reported by Bruno Haible sbcl-devel "various SBCL bugs" from CLISP
1019 (read-from-string "`#1A(1 2 ,(+ 2 2) 4)")
1021 #(1 2 ((SB-IMPL::|,|) + 2 2) 4)
1022 which probably isn't intentional.
1024 324: "STREAMs and :ELEMENT-TYPE with large bytesize"
1025 In theory, (open foo :element-type '(unsigned-byte <x>)) should work
1026 for all positive integral <x>. At present, it only works for <x> up
1027 to about 1024 (and similarly for signed-byte), so
1028 (open "/dev/zero" :element-type '(unsigned-byte 1025))
1029 gives an error in sbcl-0.8.10.
1031 325: "CLOSE :ABORT T on superseding streams"
1032 Closing a stream opened with :IF-EXISTS :SUPERSEDE with :ABORT T leaves no
1033 file on disk, even if one existed before opening.
1035 The illegality of this is not crystal clear, as the ANSI dictionary
1036 entry for CLOSE says that when :ABORT is T superseded files are not
1037 superseded (ie. the original should be restored), whereas the OPEN
1038 entry says about :IF-EXISTS :SUPERSEDE "If possible, the
1039 implementation should not destroy the old file until the new stream
1040 is closed." -- implying that even though undesirable, early deletion
1041 is legal. Restoring the original would none the less be the polite
1044 326: "*PRINT-CIRCLE* crosstalk between streams"
1045 In sbcl-0.8.10.48 it's possible for *PRINT-CIRCLE* references to be
1046 mixed between streams when output operations are intermingled closely
1047 enough (as by doing output on S2 from within (PRINT-OBJECT X S1) in the
1048 test case below), so that e.g. the references #2# appears on a stream
1049 with no preceding #2= on that stream to define it (because the #2= was
1050 sent to another stream).
1051 (cl:in-package :cl-user)
1052 (defstruct foo index)
1053 (defparameter *foo* (make-foo :index 4))
1055 (defparameter *bar* (make-bar))
1056 (defparameter *tangle* (list *foo* *bar* *foo*))
1057 (defmethod print-object ((foo foo) stream)
1058 (let ((index (foo-index foo)))
1059 (format *trace-output*
1060 "~&-$- emitting FOO ~D, ambient *BAR*=~S~%"
1062 (format stream "[FOO ~D]" index))
1064 (let ((tsos (make-string-output-stream))
1065 (ssos (make-string-output-stream)))
1066 (let ((*print-circle* t)
1067 (*trace-output* tsos)
1068 (*standard-output* ssos))
1069 (prin1 *tangle* *standard-output*))
1070 (let ((string (get-output-stream-string ssos)))
1071 (unless (string= string "(#1=[FOO 4] #S(BAR) #1#)")
1072 ;; In sbcl-0.8.10.48 STRING was "(#1=[FOO 4] #2# #1#)".:-(
1073 (error "oops: ~S" string)))))
1074 It might be straightforward to fix this by turning the
1075 *CIRCULARITY-HASH-TABLE* and *CIRCULARITY-COUNTER* variables into
1076 per-stream slots, but (1) it would probably be sort of messy faking
1077 up the special variable binding semantics using UNWIND-PROTECT and
1078 (2) it might be sort of a pain to test that no other bugs had been
1081 328: "Profiling generic functions", transplanted from #241
1082 (from tonyms on #lisp IRC 2003-02-25)
1083 In sbcl-0.7.12.55, typing
1084 (defclass foo () ((bar :accessor foo-bar)))
1087 (defclass foo () ((bar :accessor foo-bar)))
1088 gives the error message
1089 "#:FOO-BAR already names an ordinary function or a macro."
1091 Problem: when a generic function is profiled, it appears as an ordinary
1092 function to PCL. (Remembering the uninterned accessor is OK, as the
1093 redefinition must be able to remove old accessors from their generic
1096 329: "Sequential class redefinition"
1097 reported by Bruno Haible:
1098 (defclass reactor () ((max-temp :initform 10000000)))
1099 (defvar *r1* (make-instance 'reactor))
1100 (defvar *r2* (make-instance 'reactor))
1101 (slot-value *r1* 'max-temp)
1102 (slot-value *r2* 'max-temp)
1103 (defclass reactor () ((uptime :initform 0)))
1104 (slot-value *r1* 'uptime)
1105 (defclass reactor () ((uptime :initform 0) (max-temp :initform 10000)))
1106 (slot-value *r1* 'max-temp) ; => 10000
1107 (slot-value *r2* 'max-temp) ; => 10000000 oops...
1110 The method effective when the wrapper is obsoleted can be saved
1111 in the wrapper, and then to update the instance just run through
1112 all the old wrappers in order from oldest to newest.
1114 336: "slot-definitions must retain the generic functions of accessors"
1115 reported by Tony Martinez:
1116 (defclass foo () ((bar :reader foo-bar)))
1117 (defun foo-bar (x) x)
1118 (defclass foo () ((bar :reader get-bar))) ; => error, should work
1120 Note: just punting the accessor removal if the fdefinition
1121 is not a generic function is not enough:
1123 (defclass foo () ((bar :reader foo-bar)))
1124 (defvar *reader* #'foo-bar)
1125 (defun foo-bar (x) x)
1126 (defclass foo () ((bar :initform 'ok :reader get-bar)))
1127 (funcall *reader* (make-instance 'foo)) ; should be an error, since
1128 ; the method must be removed
1129 ; by the class redefinition
1131 Fixing this should also fix a subset of #328 -- update the
1132 description with a new test-case then.
1134 339: "DEFINE-METHOD-COMBINATION bugs"
1135 (reported by Bruno Haible via the clisp test suite)
1137 a. Syntax checking laxity (should produce errors):
1138 i. (define-method-combination foo :documentation :operator)
1139 ii. (define-method-combination foo :documentation nil)
1140 iii. (define-method-combination foo nil)
1141 iv. (define-method-combination foo nil nil
1142 (:arguments order &aux &key))
1143 v. (define-method-combination foo nil nil (:arguments &whole))
1144 vi. (define-method-combination foo nil nil (:generic-function))
1145 vii. (define-method-combination foo nil nil (:generic-function bar baz))
1146 viii. (define-method-combination foo nil nil (:generic-function (bar)))
1147 ix. (define-method-combination foo nil ((3)))
1148 x. (define-method-combination foo nil ((a)))
1150 b. define-method-combination arguments lambda list badness
1151 i. &aux args are currently unsupported;
1152 ii. default values of &optional and &key arguments are ignored;
1153 iii. supplied-p variables for &optional and &key arguments are not
1156 c. (fixed in sbcl-0.9.15.15)
1158 344: more (?) ROOM T problems (possibly part of bug 108)
1159 In sbcl-0.8.12.51, and off and on leading up to it, the
1160 SB!VM:MEMORY-USAGE operations in ROOM T caused
1161 unhandled condition (of type SB-INT:BUG):
1162 failed AVER: "(SAP= CURRENT END)"
1163 Several clever people have taken a shot at this without fixing
1164 it; this time around (before sbcl-0.8.13 release) I (WHN) just
1165 commented out the SB!VM:MEMORY-USAGE calls until someone figures
1166 out how to make them work reliably with the rest of the GC.
1168 (Note: there's at least one dubious thing in room.lisp: see the
1169 comment in VALID-OBJ)
1171 346: alpha backtrace
1172 In sbcl-0.8.13, all backtraces from errors caused by internal errors
1173 on the alpha seem to have a "bogus stack frame".
1175 349: PPRINT-INDENT rounding implementation decisions
1176 At present, pprint-indent (and indeed the whole pretty printer)
1177 more-or-less assumes that it's using a monospace font. That's
1178 probably not too silly an assumption, but one piece of information
1179 the current implementation loses is from requests to indent by a
1180 non-integral amount. As of sbcl-0.8.15.9, the system silently
1181 truncates the indentation to an integer at the point of request, but
1182 maybe the non-integral value should be propagated through the
1183 pprinter and only truncated at output? (So that indenting by 1/2
1184 then 3/2 would indent by two spaces, not one?)
1186 352: forward-referenced-class trouble
1187 reported by Bruno Haible on sbcl-devel
1189 (setf (class-name (find-class 'a)) 'b)
1193 Expected: an instance of c, with a slot named x
1194 Got: debugger invoked on a SIMPLE-ERROR in thread 78906:
1195 While computing the class precedence list of the class named C.
1196 The class named B is a forward referenced class.
1197 The class named B is a direct superclass of the class named C.
1199 [ Is this actually a bug? DEFCLASS only replaces an existing class
1200 when the class name is the proper name of that class, and in the
1201 above code the class found by (FIND-CLASS 'A) does not have a
1202 proper name. CSR, 2006-08-07 ]
1204 353: debugger suboptimalities on x86
1205 On x86 backtraces for undefined functions start with a bogus stack
1206 frame, and backtraces for throws to unknown catch tags with a "no
1207 debug information" frame. These are both due to CODE-COMPONENT-FROM-BITS
1208 (used on non-x86 platforms) being a more complete solution then what
1211 On x86/linux large portions of tests/debug.impure.lisp have been commented
1212 out as failures. The probable culprit for these problems is in x86-call-context
1213 (things work fine on x86/freebsd).
1215 More generally, the debugger internals suffer from excessive x86/non-x86
1216 conditionalization and OAOOMization: refactoring the common parts would
1219 354: XEPs in backtraces
1220 Under default compilation policy
1224 Has the XEP for TEST in the backtrace, not the TEST frame itself.
1225 (sparc and x86 at least)
1227 Since SBCL 0.8.20.1 this is hidden unless *SHOW-ENTRY-POINT-DETAILS*
1228 is true (instead there appear two TEST frames at least on ppc). The
1229 underlying cause seems to be that SB-C::TAIL-ANNOTATE will not merge
1230 the tail-call for the XEP, since Python has by that time proved that
1231 the function can never return; same happens if the function holds an
1232 unconditional call to ERROR.
1235 (reported by Bruno Haible)
1236 After the "layout depth conflict" error, the CLOS is left in a state where
1237 it's not possible to define new standard-class subclasses any more.
1239 (defclass prioritized-dispatcher ()
1240 ((dependents :type list :initform nil)))
1241 (defmethod sb-pcl:validate-superclass ((c1 sb-pcl:funcallable-standard-class)
1242 (c2 (eql (find-class 'prioritized-dispatcher))))
1244 (defclass prioritized-generic-function (prioritized-dispatcher standard-generic-function)
1246 (:metaclass sb-pcl:funcallable-standard-class))
1247 ;; ERROR, Quit the debugger with ABORT
1248 (defclass typechecking-reader-class (standard-class)
1250 Expected: #<STANDARD-CLASS TYPECHECKING-READER-CLASS>
1251 Got: ERROR "The assertion SB-PCL::WRAPPERS failed."
1253 [ This test case does not cause the error any more. However,
1254 similar problems can be observed with
1256 (defclass foo (standard-class) ()
1257 (:metaclass sb-mop:funcallable-standard-class))
1258 (sb-mop:finalize-inheritance (find-class 'foo))
1260 (defclass bar (standard-class) ())
1261 (make-instance 'bar)
1264 357: defstruct inheritance of initforms
1265 (reported by Bruno Haible)
1266 When defstruct and defclass (with :metaclass structure-class) are mixed,
1267 1. some slot initforms are ignored by the DEFSTRUCT generated constructor
1269 2. all slot initforms are ignored by MAKE-INSTANCE. (This can be arguably
1270 OK for initforms that were given in a DEFSTRUCT form, but for those
1271 given in a DEFCLASS form, I think it qualifies as a bug.)
1273 (defstruct structure02a
1277 (defclass structure02b (structure02a)
1278 ((slot4 :initform -44)
1281 (slot7 :initform (floor (* pi pi)))
1282 (slot8 :initform 88))
1283 (:metaclass structure-class))
1284 (defstruct (structure02c (:include structure02b (slot8 -88)))
1287 (slot11 (floor (exp 3))))
1289 (let ((a (make-structure02c)))
1290 (list (structure02c-slot4 a)
1291 (structure02c-slot5 a)
1292 (structure02c-slot6 a)
1293 (structure02c-slot7 a)))
1294 Expected: (-44 nil t 9)
1295 Got: (SB-PCL::..SLOT-UNBOUND.. SB-PCL::..SLOT-UNBOUND..
1296 SB-PCL::..SLOT-UNBOUND.. SB-PCL::..SLOT-UNBOUND..)
1298 (let ((b (make-instance 'structure02c)))
1299 (list (structure02c-slot2 b)
1300 (structure02c-slot3 b)
1301 (structure02c-slot4 b)
1302 (structure02c-slot6 b)
1303 (structure02c-slot7 b)
1304 (structure02c-slot8 b)
1305 (structure02c-slot10 b)
1306 (structure02c-slot11 b)))
1307 Expected: (t 3 -44 t 9 -88 t 20)
1308 Got: (0 0 0 0 0 0 0 0)
1310 359: wrong default value for ensure-generic-function's :generic-function-class argument
1311 (reported by Bruno Haible)
1312 ANSI CL is silent on this, but the MOP's specification of ENSURE-GENERIC-FUNCTION says:
1313 "The remaining arguments are the complete set of keyword arguments
1314 received by ENSURE-GENERIC-FUNCTION."
1315 and the spec of ENSURE-GENERIC-FUNCTION-USING-CLASS:
1316 ":GENERIC-FUNCTION-CLASS - a class metaobject or a class name. If it is not
1317 supplied, it defaults to the class named STANDARD-GENERIC-FUNCTION."
1318 This is not the case in SBCL. Test case:
1319 (defclass my-generic-function (standard-generic-function)
1321 (:metaclass sb-pcl:funcallable-standard-class))
1322 (setf (fdefinition 'foo1)
1323 (make-instance 'my-generic-function :name 'foo1))
1324 (ensure-generic-function 'foo1
1325 :generic-function-class (find-class 'standard-generic-function))
1327 ; => #<SB-MOP:FUNCALLABLE-STANDARD-CLASS STANDARD-GENERIC-FUNCTION>
1328 (setf (fdefinition 'foo2)
1329 (make-instance 'my-generic-function :name 'foo2))
1330 (ensure-generic-function 'foo2)
1332 Expected: #<SB-MOP:FUNCALLABLE-STANDARD-CLASS STANDARD-GENERIC-FUNCTION>
1333 Got: #<SB-MOP:FUNCALLABLE-STANDARD-CLASS MY-GENERIC-FUNCTION>
1335 362: missing error when a slot-definition is created without a name
1336 (reported by Bruno Haible)
1337 The MOP says about slot-definition initialization:
1338 "The :NAME argument is a slot name. An ERROR is SIGNALled if this argument
1339 is not a symbol which can be used as a variable name. An ERROR is SIGNALled
1340 if this argument is not supplied."
1342 (make-instance (find-class 'sb-pcl:standard-direct-slot-definition))
1344 Got: #<SB-MOP:STANDARD-DIRECT-SLOT-DEFINITION NIL>
1346 363: missing error when a slot-definition is created with a wrong documentation object
1347 (reported by Bruno Haible)
1348 The MOP says about slot-definition initialization:
1349 "The :DOCUMENTATION argument is a STRING or NIL. An ERROR is SIGNALled
1350 if it is not. This argument default to NIL during initialization."
1352 (make-instance (find-class 'sb-pcl:standard-direct-slot-definition)
1354 :documentation 'not-a-string)
1356 Got: #<SB-MOP:STANDARD-DIRECT-SLOT-DEFINITION FOO>
1358 370: reader misbehaviour on large-exponent floats
1359 (read-from-string "1.0s1000000000000000000000000000000000000000")
1360 causes the reader to attempt to create a very large bignum (which it
1361 will then attempt to coerce to a rational). While this isn't
1362 completely wrong, it is probably not ideal -- checking the floating
1363 point control word state and then returning the relevant float
1364 (most-positive-short-float or short-float-infinity) or signalling an
1365 error immediately would seem to make more sense.
1367 372: floating-point overflow not signalled on ppc/darwin
1368 The following assertions in float.pure.lisp fail on ppc/darwin
1369 (Mac OS X version 10.3.7):
1370 (assert (raises-error? (scale-float 1.0 most-positive-fixnum)
1371 floating-point-overflow))
1372 (assert (raises-error? (scale-float 1.0d0 (1+ most-positive-fixnum))
1373 floating-point-overflow)))
1374 as the SCALE-FLOAT just returns
1375 #.SB-EXT:SINGLE/DOUBLE-FLOAT-POSITIVE-INFINITY. These tests have been
1376 disabled on Darwin for now.
1378 377: Memory fault error reporting
1379 On those architectures where :C-STACK-IS-CONTROL-STACK is in
1380 *FEATURES*, we handle SIG_MEMORY_FAULT (SEGV or BUS) on an altstack,
1381 so we cannot handle the signal directly (as in interrupt_handle_now())
1382 in the case when the signal comes from some external agent (the user
1383 using kill(1), or a fault in some foreign code, for instance). As
1384 of sbcl-0.8.20.20, this is fixed by calling
1385 arrange_return_to_lisp_function() to a new error-signalling
1386 function, but as a result the error reporting is poor: we cannot
1387 even tell the user at which address the fault occurred. We should
1388 arrange such that arguments can be passed to the function called from
1389 arrange_return_to_lisp_function(), but this looked hard to do in
1390 general without suffering from memory leaks.
1392 379: TRACE :ENCAPSULATE NIL broken on ppc/darwin
1393 See commented-out test-case in debug.impure.lisp.
1395 382: externalization unexpectedly changes array simplicity
1396 COMPILE-FILE and LOAD
1398 (let ((x #.(make-array 4 :fill-pointer 0)))
1399 (values (eval `(typep ',x 'simple-array))
1400 (typep x 'simple-array))))
1401 then (FOO) => T, NIL.
1403 Similar problems exist with SIMPLE-ARRAY-P, ARRAY-HEADER accessors
1404 and all array dimension functions.
1406 383: ASH'ing non-constant zeros
1409 (declare (type (integer -2 14) b))
1410 (declare (ignorable b))
1411 (ash (imagpart b) 57))
1412 on PPC (and other platforms, presumably) gives an error during the
1413 emission of FASH-ASH-LEFT/FIXNUM=>FIXNUM as the assembler attempts to
1414 stuff a too-large constant into the immediate field of a PPC
1415 instruction. Either the VOP should be fixed or the compiler should be
1416 taught how to transform this case away, paying particular attention
1417 to side-effects that might occur in the arguments to ASH.
1419 384: Compiler runaway on very large character types
1421 (compile nil '(lambda (x)
1422 (declare (type (member #\a 1) x))
1423 (the (member 1 nil) x)))
1425 The types apparently normalize into a very large type, and the compiler
1426 gets lost in REMOVE-DUPLICATES. Perhaps the latter should use
1427 a better algorithm (one based on hash tables, say) on very long lists
1428 when :TEST has its default value?
1432 (compile nil '(lambda (x) (the (not (eql #\a)) x)))
1434 (partially fixed in 0.9.3.1, but a better representation for these
1438 (format nil "~4,1F" 0.001) => "0.00" (should be " 0.0");
1439 (format nil "~4,1@F" 0.001) => "+.00" (should be "+0.0").
1440 (format nil "~E" 0.01) => "10.e-3" (should be "1.e-2");
1441 (format nil "~G" 0.01) => "10.e-3" (should be "1.e-2");
1443 386: SunOS/x86 stack exhaustion handling broken
1444 According to <http://alfa.s145.xrea.com/sbcl/solaris-x86.html>, the
1445 stack exhaustion checking (implemented with a write-protected guard
1446 page) does not work on SunOS/x86.
1449 (found by Dmitry Bogomolov)
1451 (defclass foo () ((x :type (unsigned-byte 8))))
1452 (defclass bar () ((x :type symbol)))
1453 (defclass baz (foo bar) ())
1457 SB-PCL::SPECIALIZER-APPLICABLE-USING-TYPE-P cannot handle the second argument
1460 [ Can't trigger this any more, as of 2006-08-07 ]
1463 (reported several times on sbcl-devel, by Rick Taube, Brian Rowe and
1466 ROUND-NUMERIC-BOUND assumes that float types always have a FORMAT
1467 specifying whether they're SINGLE or DOUBLE. This is true for types
1468 computed by the type system itself, but the compiler type derivation
1469 short-circuits this and constructs non-canonical types. A temporary
1470 fix was made to ROUND-NUMERIC-BOUND for the sbcl-0.9.6 release, but
1471 the right fix is to remove the abstraction violation in the
1472 compiler's type deriver.
1474 393: Wrong error from methodless generic function
1475 (DEFGENERIC FOO (X))
1477 gives NO-APPLICABLE-METHOD rather than an argument count error.
1479 395: Unicode and streams
1480 One of the remaining problems in SBCL's Unicode support is the lack
1481 of generality in certain streams.
1482 a. FILL-POINTER-STREAMs: SBCL refuses to write (e.g. using FORMAT)
1483 to streams made from strings that aren't character strings with
1485 (let ((v (make-array 5 :fill-pointer 0 :element-type 'standard-char)))
1488 should return a non-simple base string containing "foo" but
1491 (reported on sbcl-help by "tichy")
1493 396: block-compilation bug
1497 (when (funcall (eval #'(lambda (x) (eql x 2))) y)
1499 (incf x (incf y z))))))
1503 (bar 1) => 11, should be 4.
1506 The more interrupts arrive the less accurate SLEEP's timing gets.
1507 (time (sb-thread:terminate-thread
1508 (prog1 (sb-thread:make-thread (lambda ()
1515 398: GC-unsafe SB-ALIEN string deporting
1516 Translating a Lisp string to an alien string by taking a SAP to it
1517 as done by the :DEPORT-GEN methods for C-STRING and UTF8-STRING
1518 is not safe, since the Lisp string can move. For example the
1519 following code will fail quickly on both cheneygc and pre-0.9.8.19
1522 (setf (bytes-consed-between-gcs) 4096)
1523 (define-alien-routine "strcmp" int (s1 c-string) (s2 c-string))
1526 (let ((string "hello, world"))
1527 (assert (zerop (strcmp string string)))))
1529 (This will appear to work on post-0.9.8.19 GENCGC, since
1530 the GC no longer zeroes memory immediately after releasing
1531 it after a minor GC. Either enabling the READ_PROTECT_FREE_PAGES
1532 #define in gencgc.c or modifying the example so that a major
1533 GC will occasionally be triggered would unmask the bug.)
1535 On cheneygc the only solution would seem to be allocating some alien
1536 memory, copying the data over, and arranging that it's freed once we
1537 return. For GENCGC we could instead try to arrange that the string
1538 from which the SAP is taken is always pinned.
1540 For some more details see comments for (define-alien-type-method
1541 (c-string :deport-gen) ...) in host-c-call.lisp.
1543 403: FORMAT/PPRINT-LOGICAL-BLOCK of CONDITIONs ignoring *PRINT-CIRCLE*
1546 (make-condition 'simple-error
1547 :format-control "ow... ~S"
1548 :format-arguments '(#1=(#1#))))
1549 (setf *print-circle* t *print-level* 4)
1550 (format nil "~@<~A~:@>" *c*)
1553 where I (WHN) believe the correct result is "ow... #1=(#1#)",
1554 like the result from (PRINC-TO-STRING *C*). The question of
1555 what the correct result is is complicated by the hairy text in
1556 the Hyperspec "22.3.5.2 Tilde Less-Than-Sign: Logical Block",
1557 Other than the difference in its argument, ~@<...~:> is
1558 exactly the same as ~<...~:> except that circularity detection
1559 is not applied if ~@<...~:> is encountered at top level in a
1561 But because the odd behavior happens even without the at-sign,
1562 (format nil "~<~A~:@>" (list *c*)) ; => "ow... (((#)))"
1563 and because something seemingly similar can happen even in
1564 PPRINT-LOGICAL-BLOCK invoked directly without FORMAT,
1565 (pprint-logical-block (*standard-output* '(some nonempty list))
1566 (format *standard-output* "~A" '#1=(#1#)))
1567 (which prints "(((#)))" to *STANDARD-OUTPUT*), I don't think
1568 that the 22.3.5.2 trickiness is fundamental to the problem.
1570 My guess is that the problem is related to the logic around the MODE
1571 argument to CHECK-FOR-CIRCULARITY, but I haven't reverse-engineered
1572 enough of the intended meaning of the different MODE values to be
1575 404: nonstandard DWIMness in LOOP with unportably-ordered clauses
1576 In sbcl-0.9.13, the code
1577 (loop with stack = (make-array 2 :fill-pointer 2 :initial-element t)
1578 for length = (length stack)
1579 while (plusp length)
1580 for element = (vector-pop stack)
1582 compiles without error or warning and returns (T T). Unfortunately,
1583 it is inconsistent with the ANSI definition of the LOOP macro,
1584 because it mixes up VARIABLE-CLAUSEs with MAIN-CLAUSEs. Furthermore,
1585 SBCL's interpretation of the intended meaning is only one possible,
1586 unportable interpretation of the noncompliant code; in CLISP 2.33.2,
1587 the code compiles with a warning
1588 LOOP: FOR clauses should occur before the loop's main body
1589 and then fails at runtime with
1590 VECTOR-POP: #() has length zero
1591 perhaps because CLISP has shuffled the clauses into an
1592 ANSI-compliant order before proceeding.
1594 406: functional has external references -- failed aver
1595 Given the following food in a single file
1596 (eval-when (:compile-toplevel :load-toplevel :execute)
1599 (foo #.(make-foo3)))
1600 as of 0.9.18.11 the file compiler breaks on it:
1601 failed AVER: "(NOT (FUNCTIONAL-HAS-EXTERNAL-REFERENCES-P CLAMBDA))"
1602 Defining the missing MAKE-LOAD-FORM method makes the error go away.
1604 407: misoptimization of loop, COERCE 'FLOAT, and HANDLER-CASE for bignums
1605 (reported by Ariel Badichi on sbcl-devel 2007-01-09)
1606 407a: In sbcl-1.0.1 on Linux x86,
1608 (loop for n from (expt 2 1024) do
1610 (coerce n 'single-float)
1611 (simple-type-error ()
1612 (format t "Got here.~%")
1613 (return-from foo)))))
1615 causes an infinite loop, where handling the error would be expected.
1616 407b: In sbcl-1.0.1 on Linux x86,
1618 (loop for n from (expt 2 1024) do
1620 (format t "~E~%" (coerce n 'single-float))
1621 (simple-type-error ()
1622 (format t "Got here.~%")
1623 (return-from bar)))))
1624 fails to compile, with
1625 Too large to be represented as a SINGLE-FLOAT: ...
1627 0: ((LABELS SB-BIGNUM::CHECK-EXPONENT) ...)
1628 1: ((LABELS SB-BIGNUM::FLOAT-FROM-BITS) ...)
1629 2: (SB-KERNEL:%SINGLE-FLOAT ...)
1630 3: (SB-C::BOUND-FUNC ...)
1631 4: (SB-C::%SINGLE-FLOAT-DERIVE-TYPE-AUX ...)
1633 These are now fixed, but (COERCE HUGE 'SINGLE-FLOAT) still signals a
1634 type-error at runtime. The question is, should it instead signal a
1635 floating-point overflow, or return an infinity?
1637 408: SUBTYPEP confusion re. OR of SATISFIES of not-yet-defined predicate
1638 As reported by Levente M\'{e}sz\'{a}ros sbcl-devel 2006-02-20,
1639 (aver (equal (multiple-value-list
1640 (subtypep '(or (satisfies x) string)
1641 '(or (satisfies x) integer)))
1643 fails. Also, beneath that failure lurks another failure,
1644 (aver (equal (multiple-value-list
1646 '(or (satisfies x) integer)))
1648 Having looked at this for an hour or so in sbcl-1.0.2, and
1649 specifically having looked at the output from
1652 (y '(or (satisfies x) integer)))
1653 (trace sb-kernel::union-complex-subtypep-arg2
1654 sb-kernel::invoke-complex-subtypep-arg1-method
1655 sb-kernel::type-union
1656 sb-kernel::type-intersection
1659 my (WHN) impression is that the problem is that the semantics of TYPE=
1660 are wrong for what the UNION-COMPLEX-SUBTYPEP-ARG2 code is trying
1661 to use it for. The comments on the definition of TYPE= probably
1662 date back to CMU CL and seem to define it as a confusing thing:
1663 its primary value is something like "certainly equal," and its
1664 secondary value is something like "certain about that certainty."
1665 I'm left uncertain how to fix UNION-COMPLEX-SUBTYPEP-ARG2 without
1666 reducing its generality by removing the TYPE= cleverness. Possibly
1667 the tempting TYPE/= relative defined next to it might be a
1668 suitable replacement for the purpose. Probably, though, it would
1669 be best to start by reverse engineering exactly what TYPE= and
1670 TYPE/= do, and writing an explanation which is so clear that one
1671 can see immediately what it's supposed to mean in odd cases like
1672 (TYPE= '(SATISFIES X) 'INTEGER) when X isn't defined yet.
1674 409: MORE TYPE SYSTEM PROBLEMS
1675 Found while investigating an optimization failure for extended
1676 sequences. The extended sequence type implementation was altered to
1677 work around the problem, but the fundamental problem remains, to wit:
1678 (sb-kernel:type= (sb-kernel:specifier-type '(or float ratio))
1679 (sb-kernel:specifier-type 'single-float))
1680 returns NIL, NIL on sbcl-1.0.3.
1681 (probably related to bug #408)
1683 410: read circularities and type declarations
1684 Consider the definition
1685 (defstruct foo (a 0 :type (not symbol)))
1687 (setf *print-circle* t) ; just in case
1688 (read-from-string "#1=#s(foo :a #1#)")
1689 This gives a type error (#:G1 is not a (NOT SYMBOL)) because of the
1690 implementation of read circularity, using a symbol as a marker for
1691 the previously-referenced object.
1693 416: backtrace confusion
1704 gives the correct error, but the backtrace shows
1705 1: (SB-KERNEL:FDEFINITION-OBJECT 13 NIL)
1706 as the second frame.
1708 418: SUBSEQ on lists doesn't support bignum indexes
1710 LIST-SUBSEQ* now has all the works necessary to support bignum indexes,
1711 but it needs to be verified that changing the DEFKNOWN doesn't kill
1712 performance elsewhere.
1714 Other generic sequence functions have this problem as well.
1716 419: stack-allocated indirect closure variables are not popped
1719 (multiple-value-call #'list
1720 (eval '(values 1 2 3))
1722 (declare (sb-int:truly-dynamic-extent x))
1727 (declare (dynamic-extent #'mget #'mset))
1728 ((lambda (f g) (eval `(progn ,f ,g (values 4 5 6)))) #'mget #'mset)))))
1730 (ASSERT (EQUAL (BUG419 42) '(1 2 3 4 5 6))) => failure
1732 Note: as of SBCL 1.0.16.29 this bug no longer affects user code, as
1733 SB-INT:TRULY-DYNAMIC-EXTENT needs to be used instead of
1734 DYNAMIC-EXTENT for this to happen. Proper fix for this bug requires
1735 (Nikodemus thinks) storing the relevant LAMBDA-VARs in a
1736 :DYNAMIC-EXTENT cleanup, and teaching stack analysis how to deal
1739 421: READ-CHAR-NO-HANG misbehaviour on Windows Console:
1741 It seems that on Windows READ-CHAR-NO-HANG hangs if the user
1742 has pressed a key, but not yet enter (ie. SYSREAD-MAY-BLOCK-P
1743 seems to lie if the OS is buffering input for us on Console.)
1745 reported by Elliot Slaughter on sbcl-devel 2008/1/10.
1747 422: out-of-extent return not checked in safe code
1749 (declaim (optimize safety))
1750 (funcall (catch 't (block nil (throw 't (lambda () (return))))))
1752 behaves ...erratically. Reported by Kevin Reid on sbcl-devel
1753 2007-07-06. (We don't _have_ to check things like this, but we
1754 generally try to check returns in safe code, so we should here too.)
1756 424: toplevel closures and *CHECK-CONSISTENCY*
1758 The following breaks under COMPILE-FILE if *CHECK-CONSISTENCY* is true.
1760 (let ((exported-symbols-alist
1761 (loop for symbol being the external-symbols of :cl
1762 collect (cons symbol
1763 (concatenate 'string
1765 (string-downcase symbol))))))
1766 (defun hyperdoc-lookup (symbol)
1767 (cdr (assoc symbol exported-symbols-alist))))
1769 (Test-case adapted from CL-PPCRE.)
1771 428: TIMER SCHEDULE-STRESS and PARALLEL-UNSCHEDULE in
1772 timer.impure.lisp fails
1774 Failure modes vary. Core problem seems to be (?) recursive entry to
1779 Compiling a file with this contents makes the compiler loop in
1782 (declaim (inline storage))
1784 (the (simple-array flt (*)) (unknown x)))
1786 (defun test1 (lumps &key cg)
1787 (let ((nodes (map 'list (lambda (lump) (storage lump))
1789 (setf (aref nodes 0) 2)
1790 (assert (every #'~= (apply #'concatenate 'list nodes) '(2 3 6 9)))))
1792 431: alien strucure redefinition doesn't work as expected