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 c: (COERCE 'AND 'FUNCTION) returns something related to
257 (MACRO-FUNCTION 'AND), but ANSI says it should raise an error.
258 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
259 not a binary input stream, but instead cheerfully reads from
260 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
263 DEFCLASS bugs reported by Peter Van Eynde July 25, 2000:
264 d: (DEFGENERIC IF (X)) should signal a PROGRAM-ERROR, but instead
265 causes a COMPILER-ERROR.
268 miscellaneous errors reported by Peter Van Eynde July 25, 2000:
270 (DEFGENERIC FOO02 (X))
271 (DEFMETHOD FOO02 ((X NUMBER)) T)
272 (LET ((M (FIND-METHOD (FUNCTION FOO02)
274 (LIST (FIND-CLASS (QUOTE NUMBER))))))
275 (REMOVE-METHOD (FUNCTION FOO02) M)
276 (DEFGENERIC FOO03 (X))
277 (ADD-METHOD (FUNCTION FOO03) M)))
278 should give an error, but SBCL allows it.
281 It has been reported (e.g. by Peter Van Eynde) that there are
282 several metaobject protocol "errors". (In order to fix them, we might
283 need to document exactly what metaobject protocol specification
284 we're following -- the current code is just inherited from PCL.)
287 The debugger LIST-LOCATIONS command doesn't work properly.
290 Compiling and loading
291 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
293 then requesting a BACKTRACE at the debugger prompt gives no information
294 about where in the user program the problem occurred.
297 The compiler is supposed to do type inference well enough that
300 ((SIMPLE-ARRAY SINGLE-FLOAT)
302 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
305 is redundant. However, as reported by Juan Jose Garcia Ripoll for
306 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
307 workaround for the problem for now, but can't be done by the TYPECASE
308 macros themselves, since it's too hard for the macro to detect
309 assignments to the variable within the clause.
310 Note: The compiler *is* smart enough to do the type inference in
311 many cases. This case, derived from a couple of MACROEXPAND-1
312 calls on Ripoll's original test case,
314 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
315 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
316 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
317 (LET ((I 0) (G2554 LENGTH))
318 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
321 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
322 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
323 (GO SB-LOOP::NEXT-LOOP)
324 SB-LOOP::END-LOOP))))))
325 demonstrates the problem; but the problem goes away if the TAGBODY
326 and GO forms are removed (leaving the SETF in ordinary, non-looping
327 code), or if the TAGBODY and GO forms are retained, but the
328 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
331 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
332 I am looking into this problem that showed up on the cmucl-help
333 list. It seems to me that the "implementation specific environment
334 hacking functions" found in pcl/walker.lisp are completely messed
335 up. The good thing is that they appear to be barely used within
336 PCL and the munged environment object is passed to cmucl only
337 in calls to macroexpand-1, which is probably why this case fails.
338 SBCL uses essentially the same code, so if the environment hacking
339 is screwed up, it affects us too.
342 Using the pretty-printer from the command prompt gives funny
343 results, apparently because the pretty-printer doesn't know
344 about user's command input, including the user's carriage return
345 that the user, and therefore the pretty-printer thinks that
346 the new output block should start indented 2 or more characters
347 rightward of the correct location.
350 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
351 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
352 crashes SBCL. In general tracing anything which is used in the
353 implementation of TRACE is likely to have the same problem.
356 As reported by Martin Atzmueller on sbcl-devel 26 Dec 2000,
357 ANSI says that WITH-OUTPUT-TO-STRING should have a keyword
358 :ELEMENT-TYPE, but in sbcl-0.6.9 this is not defined for
359 WITH-OUTPUT-TO-STRING.
362 ANSI says in one place that type declarations can be abbreviated even
363 when the type name is not a symbol, e.g.
364 (DECLAIM ((VECTOR T) *FOOVECTOR*))
365 SBCL doesn't support this. But ANSI says in another place that this
366 isn't allowed. So it's not clear this is a bug after all. (See the
367 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
371 as pointed out by Dan Barlow on sbcl-devel 2000-07-02:
372 The PICK-TEMPORARY-FILE-NAME utility used by LOAD-FOREIGN uses
373 an easily guessable temporary filename in a way which might open
374 applications using LOAD-FOREIGN to hijacking by malicious users
375 on the same machine. Incantations for doing this safely are
376 floating around the net in various "how to write secure programs
377 despite Unix" documents, and it would be good to (1) fix this in
378 LOAD-FOREIGN, and (2) hunt for any other code which uses temporary
379 files and make it share the same new safe logic.
381 (partially alleviated in sbcl-0.7.9.32 by a fix by Matthew Danish to
382 make the temporary filename less easily guessable)
385 Functions are assigned names based on the context in which they're
386 defined. This is less than ideal for the functions which are
387 used to implement CLOS methods. E.g. the output of
388 (DESCRIBE 'PRINT-OBJECT) lists functions like
389 #<FUNCTION "DEF!STRUCT (TRACE-INFO (:MAKE-LOAD-FORM-FUN SB-KERNEL:JUST-DUMP-IT-NORMALLY) (:PRINT-OBJECT #))" {1020E49}>
391 #<FUNCTION "MACROLET ((FORCE-DELAYED-DEF!METHODS NIL #))" {1242871}>
392 It would be better if these functions' names always identified
393 them as methods, and identified their generic functions and
397 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
398 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
399 the mean of the distribution can be systematically O(0.1%) wrong.
400 Just increasing R-I-E-B is probably not a good solution, since
401 it would decrease efficiency more than is probably necessary. Perhaps
402 using some sort of accept/reject method would be better.
405 Internally the compiler sometimes evaluates
406 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
407 (I stumbled across this when I added an
408 (assert (not (eq type1 *wild-type*)))
409 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
410 in a type context should probably be translated to T, and so it's
411 probably wrong to ask whether it's equal to the T type and then (using
412 the EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
413 (I haven't tried to investigate this bug enough to guess whether
414 there might be any user-level symptoms.)
416 In fact, the type system is likely to depend on this inequality not
417 holding... * is not equivalent to T in many cases, such as
418 (VECTOR *) /= (VECTOR T).
421 Inconsistencies between derived and declared VALUES return types for
422 DEFUN aren't checked very well. E.g. the logic which successfully
423 catches problems like
424 (declaim (ftype (function (fixnum) float) foo))
426 (declare (type integer x))
427 (values x)) ; wrong return type, detected, gives warning, good!
429 (declaim (ftype (function (t) (values t t)) bar))
431 (values x)) ; wrong number of return values, no warning, bad!
432 The cause of this is seems to be that (1) the internal function
433 VALUES-TYPES-EQUAL-OR-INTERSECT used to make the check handles its
434 arguments symmetrically, and (2) when the type checking code was
435 written back when when SBCL's code was still CMU CL, the intent
437 (declaim (ftype (function (t) t) bar))
439 (values x x)) ; wrong number of return values; should give warning?
440 not be warned for, because a two-valued return value is considered
441 to be compatible with callers who expects a single value to be
442 returned. That intent is probably not appropriate for modern ANSI
443 Common Lisp, but fixing this might be complicated because of other
444 divergences between auld-style and new-style handling of
445 multiple-VALUES types. (Some issues related to this were discussed
446 on cmucl-imp at some length sometime in 2000.)
449 The facility for dumping a running Lisp image to disk gets confused
450 when run without the PURIFY option, and creates an unnecessarily large
451 core file (apparently representing memory usage up to the previous
452 high-water mark). Moreover, when the file is loaded, it confuses the
453 GC, so that thereafter memory usage can never be reduced below that
457 In sbcl-0.6.11.41 (and in all earlier SBCL, and in CMU
458 CL), out-of-line structure slot setters are horribly inefficient
459 whenever the type of the slot is declared, because out-of-line
460 structure slot setters are implemented as closures to save space,
461 so the compiler doesn't compile the type test into code, but
462 instead just saves the type in a lexical closure and interprets it
464 A proper solution involves deciding whether it's really worth
465 saving space by implementing structure slot accessors as closures.
466 (If it's not worth it, the problem vanishes automatically. If it
467 is worth it, there are hacks we could use to force type tests to
468 be compiled anyway, and even shared. E.g. we could implement
469 an EQUAL hash table mapping from types to compiled type tests,
470 and save the appropriate compiled type test as part of each lexical
471 closure; or we could make the lexical closures be placeholders
472 which overwrite their old definition as a lexical closure with
473 a new compiled definition the first time that they're called.)
474 As a workaround for the problem, #'(SETF FOO) expressions can
475 be replaced with (EFFICIENT-SETF-FUNCTION FOO), where
476 (defmacro efficient-setf-function (place-function-name)
477 (or #+sbcl (and (sb-impl::info :function :accessor-for place-function-name)
478 ;; a workaround for the problem, encouraging the
479 ;; inline expansion of the structure accessor, so
480 ;; that the compiler can optimize its type test
481 (let ((new-value (gensym "NEW-VALUE-"))
482 (structure-value (gensym "STRUCTURE-VALUE-")))
483 `(lambda (,new-value ,structure-value)
484 (setf (,place-function-name ,structure-value)
486 ;; no problem, can just use the ordinary expansion
487 `(function (setf ,place-function-name))))
490 There's apparently a bug in CEILING optimization which caused
491 Douglas Crosher to patch the CMU CL version. Martin Atzmueller
492 applied the patches to SBCL and they didn't seem to cause problems
493 (as reported sbcl-devel 2001-05-04). However, since the patches
494 modify nontrivial code which was apparently written incorrectly
495 the first time around, until regression tests are written I'm not
496 comfortable merging the patches in the CVS version of SBCL.
499 (TIME (ROOM T)) reports more than 200 Mbytes consed even for
500 a clean, just-started SBCL system. And it seems to be right:
501 (ROOM T) can bring a small computer to its knees for a *long*
502 time trying to GC afterwards. Surely there's some more economical
503 way to implement (ROOM T).
506 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
508 (in-package :cl-user)
509 ;;; The following invokes a compiler error.
510 (declaim (optimize (speed 2) (debug 3)))
513 (unwind-protect nil)))
517 The error message in sbcl-0.6.12.42 is
518 internal error, failed AVER:
519 "(COMMON-LISP:EQ (SB!C::TN-ENVIRONMENT SB!C:TN) SB!C::TN-ENV)"
521 This examples better illustrates the problem:
524 (declare (optimize (speed 2) (debug 3)))
533 (X is allocated in the physical environment of M1; X is :WRITE in
534 the call of LET [convert-to-global]; IF makes sure that a block
535 exists in M1 before this call.)
537 Because X is :DEBUG-ENVIRONMENT, it is :LIVE by default in all
538 blocks in the environment, particularly it is :LIVE in the start of
539 M1 (where it is not yet :WRITE) [setup-environment-tn-conflicts].
541 Then :LIVE is propagated backwards, i.e. into the caller of M1
542 where X does not exist [lifetime-flow-analysis].
544 (CATCH NIL) causes all TNs to be saved; Python fails on saving
545 non-existent variable; if it is replaced with (FOO), the problem
546 appears when debugging TST: LIST-LOCALS says
548 debugger invoked on condition of type SB-DI:UNKNOWN-DEBUG-VAR:
550 #<SB-DI::COMPILED-DEBUG-VAR X 0
551 {905FF7D}> is not in #<SB-DI::COMPILED-DEBUG-FUNCTION TST>.
553 (in those old versions, in which debugger worked :-().
556 When the compiler inline expands functions, it may be that different
557 kinds of return values are generated from different code branches.
558 E.g. an inline expansion of POSITION generates integer results
559 from one branch, and NIL results from another. When that inline
560 expansion is used in a context where only one of those results
563 (aref *a1* (position x *a2*)))
564 and the compiler can't prove that the unacceptable branch is
565 never taken, then bogus type mismatch warnings can be generated.
566 If you need to suppress the type mismatch warnings, you can
567 suppress the inline expansion,
569 #+sbcl (declare (notinline position)) ; to suppress bug 117 bogowarnings
570 (aref *a1* (position x *a2*)))
571 or, sometimes, suppress them by declaring the result to be of an
574 (aref *a1* (the integer (position x *a2*))))
576 This is not a new compiler problem in 0.7.0, but the new compiler
577 transforms for FIND, POSITION, FIND-IF, and POSITION-IF make it
578 more conspicuous. If you don't need performance from these functions,
579 and the bogus warnings are a nuisance for you, you can return to
580 your pre-0.7.0 state of grace with
581 #+sbcl (declaim (notinline find position find-if position-if)) ; bug 117..
584 as reported by Eric Marsden on cmucl-imp@cons.org 2001-08-14:
585 (= (FLOAT 1 DOUBLE-FLOAT-EPSILON)
586 (+ (FLOAT 1 DOUBLE-FLOAT-EPSILON) DOUBLE-FLOAT-EPSILON)) => T
587 when of course it should be NIL. (He says it only fails for X86,
588 not SPARC; dunno about Alpha.)
590 Also, "the same problem exists for LONG-FLOAT-EPSILON,
591 DOUBLE-FLOAT-NEGATIVE-EPSILON, LONG-FLOAT-NEGATIVE-EPSILON (though
592 for the -negative- the + is replaced by a - in the test)."
594 Raymond Toy comments that this is tricky on the X86 since its FPU
595 uses 80-bit precision internally.
598 Even in sbcl-0.pre7.x, which is supposed to be free of the old
599 non-ANSI behavior of treating the function return type inferred
600 from the current function definition as a declaration of the
601 return type from any function of that name, the return type of NIL
602 is attached to FOO in 120a above, and used to optimize code which
606 As of version 0.pre7.14, SBCL's implementation of MACROLET makes
607 the entire lexical environment at the point of MACROLET available
608 in the bodies of the macroexpander functions. In particular, it
609 allows the function bodies (which run at compile time) to try to
610 access lexical variables (which are only defined at runtime).
611 It doesn't even issue a warning, which is bad.
613 The SBCL behavior arguably conforms to the ANSI spec (since the
614 spec says that the behavior is undefined, ergo anything conforms).
615 However, it would be better to issue a compile-time error.
616 Unfortunately I (WHN) don't see any simple way to detect this
617 condition in order to issue such an error, so for the meantime
618 SBCL just does this weird broken "conforming" thing.
620 The ANSI standard says, in the definition of the special operator
622 The macro-expansion functions defined by MACROLET are defined
623 in the lexical environment in which the MACROLET form appears.
624 Declarations and MACROLET and SYMBOL-MACROLET definitions affect
625 the local macro definitions in a MACROLET, but the consequences
626 are undefined if the local macro definitions reference any
627 local variable or function bindings that are visible in that
629 Then it seems to contradict itself by giving the example
631 (macrolet ((fudge (z)
632 ;The parameters x and flag are not accessible
633 ; at this point; a reference to flag would be to
634 ; the global variable of that name.
635 ` (if flag (* ,z ,z) ,z)))
636 ;The parameters x and flag are accessible here.
640 The comment "a reference to flag would be to the global variable
641 of the same name" sounds like good behavior for the system to have.
642 but actual specification quoted above says that the actual behavior
645 (Since 0.7.8.23 macroexpanders are defined in a restricted version
646 of the lexical environment, containing no lexical variables and
647 functions, which seems to conform to ANSI and CLtL2, but signalling
648 a STYLE-WARNING for references to variables similar to locals might
652 (as reported by Gabe Garza on cmucl-help 2001-09-21)
654 (defun test-pred (x y)
658 (func (lambda () x)))
659 (print (eq func func))
660 (print (test-pred func func))
661 (delete func (list func))))
662 Now calling (TEST-CASE) gives output
665 (#<FUNCTION {500A9EF9}>)
666 Evidently Python thinks of the lambda as a code transformation so
667 much that it forgets that it's also an object.
670 The DEFSTRUCT section of the ANSI spec, in the :CONC-NAME section,
671 specifies a precedence rule for name collisions between slot accessors of
672 structure classes related by inheritance. As of 0.7.0, SBCL still
676 Ideally, uninterning a symbol would allow it, and its associated
677 FDEFINITION and PROCLAIM data, to be reclaimed by the GC. However,
678 at least as of sbcl-0.7.0, this isn't the case. Information about
679 FDEFINITIONs and PROCLAIMed properties is stored in globaldb.lisp
680 essentially in ordinary (non-weak) hash tables keyed by symbols.
681 Thus, once a system has an entry in this system, it tends to live
682 forever, even when it is uninterned and all other references to it
686 Pretty-printing nested backquotes doesn't work right, as
687 reported by Alexey Dejneka sbcl-devel 2002-01-13:
689 ``(FOO SB-IMPL::BACKQ-COMMA-AT S)
690 * (lisp-implementation-version)
694 (reported by Jesse Bouwman 2001-10-24 through the unfortunately
695 prominent SourceForge web/db bug tracking system, which is
696 unfortunately not a reliable way to get a timely response from
697 the SBCL maintainers)
698 In the course of trying to build a test case for an
699 application error, I encountered this behavior:
700 If you start up sbcl, and then lay on CTRL-C for a
701 minute or two, the lisp process will eventually say:
702 %PRIMITIVE HALT called; the party is over.
703 and throw you into the monitor. If I start up lisp,
704 attach to the process with strace, and then do the same
705 (abusive) thing, I get instead:
706 access failure in heap page not marked as write-protected
707 and the monitor again. I don't know enough to have the
708 faintest idea of what is going on here.
709 This is with sbcl 6.12, uname -a reports:
710 Linux prep 2.2.19 #4 SMP Tue Apr 24 13:59:52 CDT 2001 i686 unknown
711 I (WHN) have verified that the same thing occurs on sbcl-0.pre7.141
712 under OpenBSD 2.9 on my X86 laptop. Do be patient when you try it:
713 it took more than two minutes (but less than five) for me.
716 (This was once known as IR1-4, but it lived on even after the
717 IR1 interpreter went to the big bit bucket in the sky.)
718 The system accepts DECLAIM in most places where DECLARE would be
719 accepted, without even issuing a warning. ANSI allows this, but since
720 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
721 meaning is rather different, and it's unlikely that the user
722 has a good reason for doing DECLAIM not at top level, it would be
723 good to issue a STYLE-WARNING when this happens. A possible
724 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
725 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
726 [This is considered an IR1-interpreter-related bug because until
727 EVAL-WHEN is rewritten, which won't happen until after the IR1
728 interpreter is gone, the system's notion of what's a top-level form
729 and what's not will remain too confused to fix this problem.]
732 ANSI allows types `(COMPLEX ,FOO) to use very hairy values for
733 FOO, e.g. (COMPLEX (AND REAL (SATISFIES ODDP))). The old CMU CL
734 COMPLEX implementation didn't deal with this, and hasn't been
735 upgraded to do so. (This doesn't seem to be a high priority
736 conformance problem, since seems hard to construct useful code
740 Floating point errors are reported poorly. E.g. on x86 OpenBSD
743 debugger invoked on condition of type SB-KERNEL:FLOATING-POINT-EXCEPTION:
744 An arithmetic error SB-KERNEL:FLOATING-POINT-EXCEPTION was signalled.
745 No traps are enabled? How can this be?
746 It should be possible to be much more specific (overflow, division
747 by zero, etc.) and of course the "How can this be?" should be fixable.
749 See also bugs #45.c and #183
752 In sbcl-0.7.1.3 on x86, COMPILE-FILE on the file
753 (in-package :cl-user)
756 (defstruct foo bar bletch)
758 (labels ((kidify1 (kid)
766 (declare (inline kid-frob))
769 (the simple-vector (foo-bar perd)))))
771 debugger invoked on condition of type TYPE-ERROR:
772 The value NIL is not of type SB-C::NODE.
773 The location of this failure has moved around as various related
774 issues were cleaned up. As of sbcl-0.7.1.9, it occurs in
775 NODE-BLOCK called by LAMBDA-COMPONENT called by IR2-CONVERT-CLOSURE.
777 (Python LET-converts KIDIFY1 into KID-FROB, then tries to inline
778 expand KID-FROB into %ZEEP. Having partially done it, it sees a call
779 of KIDIFY1, which already does not exist. So it gives up on
780 expansion, leaving garbage consisting of infinished blocks of the
781 partially converted function.)
784 Functions SUBTYPEP, TYPEP, UPGRADED-ARRAY-ELEMENT-TYPE, and
785 UPGRADED-COMPLEX-PART-TYPE should have an optional environment argument.
786 (reported by Alexey Dejneka sbcl-devel 2002-04-12)
789 (reported by Robert E. Brown 2002-04-16)
790 When a function is called with too few arguments, causing the
791 debugger to be entered, the uninitialized slots in the bad call frame
792 seem to cause GCish problems, being interpreted as tagged data even
793 though they're not. In particular, executing ROOM in the
794 debugger at that point causes AVER failures:
797 * (lisp-implementation-version)
803 failed AVER: "(SAP= CURRENT END)"
804 (Christophe Rhodes reports that this doesn't occur on the SPARC, which
805 isn't too surprising since there are many differences in stack
806 implementation and GC conservatism between the X86 and other ports.)
809 In sbcl-0.7.3.11, compiling the (illegal) code
810 (in-package :cl-user)
811 (defmethod prove ((uustk uustk))
814 gives the (not terribly clear) error message
816 ; (during macroexpansion of (DEFMETHOD PROVE ...))
817 ; can't get template for (FROB NIL NIL)
818 The problem seems to be that the code walker used by the DEFMETHOD
819 macro is unhappy with the illegal syntax in the method body, and
820 is giving an unclear error message.
823 The compiler sometimes tries to constant-fold expressions before
824 it checks to see whether they can be reached. This can lead to
825 bogus warnings about errors in the constant folding, e.g. in code
828 (WRITE-STRING (> X 0) "+" "0"))
829 compiled in a context where the compiler can prove that X is NIL,
830 and the compiler complains that (> X 0) causes a type error because
831 NIL isn't a valid argument to #'>. Until sbcl-0.7.4.10 or so this
832 caused a full WARNING, which made the bug really annoying because then
833 COMPILE and COMPILE-FILE returned FAILURE-P=T for perfectly legal
834 code. Since then the warning has been downgraded to STYLE-WARNING,
835 so it's still a bug but at least it's a little less annoying.
837 178: "AVER failure compiling confused THEs in FUNCALL"
838 In sbcl-0.7.4.24, compiling
840 (funcall (the function (the standard-object x))))
843 "(AND (EQ (IR2-CONTINUATION-PRIMITIVE-TYPE 2CONT) FUNCTION-PTYPE) (EQ CHECK T))"
844 This variant compiles OK, though:
845 (defun bug178alternative (x)
846 (funcall (the nil x)))
848 (since 0.7.8.9 it does not signal an error; see also bug 199)
850 183: "IEEE floating point issues"
851 Even where floating point handling is being dealt with relatively
852 well (as of sbcl-0.7.5, on sparc/sunos and alpha; see bug #146), the
853 accrued-exceptions and current-exceptions part of the fp control
854 word don't seem to bear much relation to reality. E.g. on
858 debugger invoked on condition of type DIVISION-BY-ZERO:
859 arithmetic error DIVISION-BY-ZERO signalled
860 0] (sb-vm::get-floating-point-modes)
862 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
863 :ROUNDING-MODE :NEAREST
864 :CURRENT-EXCEPTIONS NIL
865 :ACCRUED-EXCEPTIONS (:INEXACT)
868 * (sb-vm::get-floating-point-modes)
869 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
870 :ROUNDING-MODE :NEAREST
871 :CURRENT-EXCEPTIONS (:INEXACT)
872 :ACCRUED-EXCEPTIONS (:INEXACT)
875 187: "type inference confusion around DEFTRANSFORM time"
876 (reported even more verbosely on sbcl-devel 2002-06-28 as "strange
877 bug in DEFTRANSFORM")
878 After the file below is compiled and loaded in sbcl-0.7.5, executing
879 (TCX (MAKE-ARRAY 4 :FILL-POINTER 2) 0)
880 at the REPL returns an adjustable vector, which is wrong. Presumably
881 somehow the DERIVE-TYPE information for the output values of %WAD is
882 being mispropagated as a type constraint on the input values of %WAD,
883 and so causing the type test to be optimized away. It's unclear how
884 hand-expanding the DEFTRANSFORM would change this, but it suggests
885 the DEFTRANSFORM machinery (or at least the way DEFTRANSFORMs are
886 invoked at a particular phase) is involved.
887 (cl:in-package :sb-c)
888 (eval-when (:compile-toplevel)
889 ;;; standin for %DATA-VECTOR-AND-INDEX
890 (defknown %dvai (array index)
892 (foldable flushable))
893 (deftransform %dvai ((array index)
897 (let* ((atype (continuation-type array))
898 (eltype (array-type-specialized-element-type atype)))
899 (when (eq eltype *wild-type*)
900 (give-up-ir1-transform
901 "specialized array element type not known at compile-time"))
902 (when (not (array-type-complexp atype))
903 (give-up-ir1-transform "SIMPLE array!"))
904 `(if (array-header-p array)
905 (%wad array index nil)
906 (values array index))))
907 ;;; standin for %WITH-ARRAY-DATA
908 (defknown %wad (array index (or index null))
909 (values (simple-array * (*)) index index index)
910 (foldable flushable))
911 ;;; (Commenting out this optimizer causes the bug to go away.)
912 (defoptimizer (%wad derive-type) ((array start end))
913 (let ((atype (continuation-type array)))
914 (when (array-type-p atype)
915 (values-specifier-type
916 `(values (simple-array ,(type-specifier
917 (array-type-specialized-element-type atype))
919 index index index)))))
921 (defun %wad (array start end)
922 (format t "~&in %WAD~%")
923 (%with-array-data array start end))
924 (cl:in-package :cl-user)
926 (declare (type (vector t) v))
927 (declare (notinline sb-kernel::%with-array-data))
928 ;; (Hand-expending DEFTRANSFORM %DVAI here also causes the bug to
932 188: "compiler performance fiasco involving type inference and UNION-TYPE"
933 (In sbcl-0.7.6.10, DEFTRANSFORM CONCATENATE was commented out until this
934 bug could be fixed properly, so you won't see the bug unless you restore
935 the DEFTRANSFORM by hand.) In sbcl-0.7.5.11 on a 700 MHz Pentium III,
939 (declare (optimize (safety 3)))
940 (declare (optimize (compilation-speed 2)))
941 (declare (optimize (speed 1) (debug 1) (space 1)))
942 (let ((fn "if-this-file-exists-the-universe-is-strange"))
943 (load fn :if-does-not-exist nil)
944 (load (concatenate 'string fn ".lisp") :if-does-not-exist nil)
945 (load (concatenate 'string fn ".fasl") :if-does-not-exist nil)
946 (load (concatenate 'string fn ".misc-garbage")
947 :if-does-not-exist nil)))))
949 134.552 seconds of real time
950 133.35156 seconds of user run time
951 0.03125 seconds of system run time
952 [Run times include 2.787 seconds GC run time.]
954 246883368 bytes consed.
955 BACKTRACE from Ctrl-C in the compilation shows that the compiler is
956 thinking about type relationships involving types like
958 (OR (INTEGER 576 576)
969 190: "PPC/Linux pipe? buffer? bug"
970 In sbcl-0.7.6, the run-program.test.sh test script sometimes hangs
971 on the PPC/Linux platform, waiting for a zombie env process. This
972 is a classic symptom of buffer filling and deadlock, but it seems
973 only sporadically reproducible.
975 191: "Miscellaneous PCL deficiencies"
976 (reported by Alexey Dejneka sbcl-devel 2002-08-04)
977 a. DEFCLASS does not inform the compiler about generated
978 functions. Compiling a file with
982 (WITH-SLOTS (A-CLASS-X) A
984 results in a STYLE-WARNING:
986 SB-SLOT-ACCESSOR-NAME::|COMMON-LISP-USER A-CLASS-X slot READER|
988 APD's fix for this was checked in to sbcl-0.7.6.20, but Pierre
989 Mai points out that the declamation of functions is in fact
990 incorrect in some cases (most notably for structure
991 classes). This means that at present erroneous attempts to use
992 WITH-SLOTS and the like on classes with metaclass STRUCTURE-CLASS
993 won't get the corresponding STYLE-WARNING.
994 c. the examples in CLHS 7.6.5.1 (regarding generic function lambda
995 lists and &KEY arguments) do not signal errors when they should.
997 192: "Python treats free type declarations as promises."
998 b. What seemed like the same fundamental problem as bug 192a, but
999 was not fixed by the same (APD "more strict type checking
1000 sbcl-devel 2002-08-97) patch:
1001 (DOTIMES (I ...) (DOTIMES (J ...) (DECLARE ...) ...)):
1002 (declaim (optimize (speed 1) (safety 3)))
1003 (defun trust-assertion (i)
1005 (declare (type (mod 4) i)) ; when commented out, behavior changes!
1008 (trust-assertion 6) ; prints nothing unless DECLARE is commented out
1012 194: "no error from (THE REAL '(1 2 3)) in some cases"
1015 (multiple-value-prog1 (progn (the real '(1 2 3))))
1016 returns (1 2 3) instead of signalling an error. This was fixed by
1017 APD's "more strict type checking patch", but although the fixed
1018 code (in sbcl-0.7.7.19) works (signals TYPE-ERROR) interactively,
1019 it's difficult to write a regression test for it, because
1020 (IGNORE-ERRORS (MULTIPLE-VALUE-PROG1 (PROGN (THE REAL '(1 2 3)))))
1021 still returns (1 2 3).
1023 b. (IGNORE-ERRORS (MULTIPLE-VALUE-PROG1 (PROGN (THE REAL '(1 2 3)))))
1024 returns (1 2 3). (As above, this shows up when writing regression
1025 tests for fixed-ness of part a.)
1026 c. Also in sbcl-0.7.7.9, (IGNORE-ERRORS (THE REAL '(1 2 3))) => (1 2 3).
1028 (null (ignore-errors
1030 (arg2 (identity (the real #(1 2 3)))))
1031 (if (< arg1 arg2) arg1 arg2))))
1033 but putting the same expression inside (DEFUN FOO () ...),
1036 * Actually this entry is probably multiple bugs, as
1037 Alexey Dejneka commented on sbcl-devel 2002-09-03:)
1038 I don't think that placing these two bugs in one entry is
1039 a good idea: they have different explanations. The second
1040 (min 1 nil) is caused by flushing of unused code--IDENTITY
1041 can do nothing with it. So it is really bug 122. The first
1042 (min nil) is due to M-V-PROG1: substituting a continuation
1043 for the result, it forgets about type assertion. The purpose
1044 of IDENTITY is to save the restricted continuation from
1045 inaccurate transformations.
1046 * Alexey Dejneka pointed out that
1047 (IGNORE-ERRORS (IDENTITY (THE REAL '(1 2 3))))
1048 works as it should. Also
1049 (IGNORE-ERRORS (VALUES (THE REAL '(1 2 3))))
1050 works as it should. Perhaps this is another case of VALUES type
1051 intersections behaving in non-useful ways?
1053 199: "hairy FUNCTION types confuse the compiler"
1054 (reported by APD sbcl-devel 2002-09-15)
1056 (EQ NIL (FUNCALL F)))
1059 (DECLARE (TYPE (AND FUNCTION (SATISFIES MUR)) F))
1062 fails to compile, printing
1064 "(AND (EQ (IR2-CONTINUATION-PRIMITIVE-TYPE 2CONT) FUNCTION-PTYPE) (EQ CHECK T))"
1066 APD further reports that this bug is not present in CMUCL.
1068 (this case was fixed in 0.7.8.9; see also bug 178)
1070 201: "Incautious type inference from compound CONS types"
1071 (reported by APD sbcl-devel 2002-09-17)
1073 (LET ((Y (CAR (THE (CONS INTEGER *) X))))
1075 (FORMAT NIL "~S IS ~S, Y = ~S"
1082 (FOO ' (1 . 2)) => "NIL IS INTEGER, Y = 1"
1085 Compiler does not check THEs on unused values, e.g. in
1087 (progn (the real (list 1)) t)
1089 This situation may appear during optimizing away degenerate cases of
1090 certain functions: see bug 192b.
1092 205: "environment issues in cross compiler"
1093 (These bugs have no impact on user code, but should be fixed or
1095 a. Macroexpanders introduced with MACROLET are defined in the null
1096 lexical environment.
1097 b. The body of (EVAL-WHEN (:COMPILE-TOPLEVEL) ...) is evaluated in
1098 the null lexical environment.
1100 206: ":SB-FLUID feature broken"
1101 (reported by Antonio Martinez-Shotton sbcl-devel 2002-10-07)
1102 Enabling :SB-FLUID in the target-features list in sbcl-0.7.8 breaks
1105 207: "poorly distributed SXHASH results for compound data"
1106 SBCL's SXHASH could probably try a little harder. ANSI: "the
1107 intent is that an implementation should make a good-faith
1108 effort to produce hash-codes that are well distributed
1109 within the range of non-negative fixnums". But
1110 (let ((hits (make-hash-table)))
1113 (let* ((ij (cons i j))
1114 (newlist (push ij (gethash (sxhash ij) hits))))
1116 (format t "~&collision: ~S~%" newlist))))))
1117 reports lots of collisions in sbcl-0.7.8. A stronger MIX function
1118 would be an obvious way of fix. Maybe it would be acceptably efficient
1119 to redo MIX using a lookup into a 256-entry s-box containing
1120 29-bit pseudorandom numbers?
1122 208: "package confusion in PCL handling of structure slot handlers"
1123 In sbcl-0.7.8 compiling and loading
1125 (defstruct foo (slot (error "missing")) :type list :read-only t)
1126 (defmethod print-object ((foo foo) stream) (print nil stream))
1127 causes CERROR "attempting to modify a symbol in the COMMON-LISP
1128 package: FOO-SLOT". (This is fairly bad code, but still it's hard
1129 to see that it should cause symbols to be interned in the CL package.)
1131 211: "keywords processing"
1132 a. :ALLOW-OTHER-KEYS T should allow a function to receive an odd
1133 number of keyword arguments.
1136 (flet ((foo (&key y) (list y)))
1137 (list (foo :y 1 :y 2)))
1139 issues confusing message
1144 ; caught STYLE-WARNING:
1145 ; The variable #:G15 is defined but never used.
1147 212: "Sequence functions and circular arguments"
1148 COERCE, MERGE and CONCATENATE go into an infinite loop when given
1149 circular arguments; it would be good for the user if they could be
1150 given an error instead (ANSI 17.1.1 allows this behaviour on the part
1151 of the implementation, as conforming code cannot give non-proper
1152 sequences to these functions. MAP also has this problem (and
1153 solution), though arguably the convenience of being able to do
1154 (MAP 'LIST '+ FOO '#1=(1 . #1#))
1155 might be classed as more important (though signalling an error when
1156 all of the arguments are circular is probably desireable).
1158 213: "Sequence functions and type checking"
1159 a. MAKE-SEQUENCE, COERCE, MERGE and CONCATENATE cannot deal with
1160 various complicated, though recognizeable, CONS types [e.g.
1161 (CONS * (CONS * NULL))
1162 which according to ANSI should be recognized] (and, in SAFETY 3
1163 code, should return a list of LENGTH 2 or signal an error)
1164 b. MAP, when given a type argument that is SUBTYPEP LIST, does not
1165 check that it will return a sequence of the given type. Fixing
1166 it along the same lines as the others (cf. work done around
1167 sbcl-0.7.8.45) is possible, but doing so efficiently didn't look
1168 entirely straightforward.
1169 c. All of these functions will silently accept a type of the form
1171 whether or not the return value is of this type. This is
1172 probably permitted by ANSI (see "Exceptional Situations" under
1173 ANSI MAKE-SEQUENCE), but the DERIVE-TYPE mechanism does not
1174 know about this escape clause, so code of the form
1175 (INTEGERP (CAR (MAKE-SEQUENCE '(CONS INTEGER *) 2)))
1176 can erroneously return T.
1179 SBCL 0.6.12.43 fails to compile
1182 (declare (optimize (inhibit-warnings 0) (compilation-speed 2)))
1183 (flet ((foo (&key (x :vx x-p)) (list x x-p)))
1186 or a more simple example:
1189 (declare (optimize (inhibit-warnings 0) (compilation-speed 2)))
1190 (lambda (x) (declare (fixnum x)) (if (< x 0) 0 (1- x))))
1192 215: ":TEST-NOT handling by functions"
1193 a. FIND and POSITION currently signal errors when given non-NIL for
1194 both their :TEST and (deprecated) :TEST-NOT arguments, but by
1195 ANSI 17.2 "the consequences are unspecified", which by ANSI 1.4.2
1196 means that the effect is "unpredictable but harmless". It's not
1197 clear what that actually means; it may preclude conforming
1198 implementations from signalling errors.
1199 b. COUNT, REMOVE and the like give priority to a :TEST-NOT argument
1200 when conflict occurs. As a quality of implementation issue, it
1201 might be preferable to treat :TEST and :TEST-NOT as being in some
1202 sense the same &KEY, and effectively take the first test function in
1204 c. Again, a quality of implementation issue: it would be good to issue a
1205 STYLE-WARNING at compile-time for calls with :TEST-NOT, and a
1206 WARNING for calls with both :TEST and :TEST-NOT; possibly this
1207 latter should be WARNed about at execute-time too.
1209 216: "debugger confused by frames with invalid number of arguments"
1210 In sbcl-0.7.8.51, executing e.g. (VECTOR-PUSH-EXTEND T), BACKTRACE, Q
1211 leaves the system confused, enough so that (QUIT) no longer works.
1212 It's as though the process of working with the uninitialized slot in
1213 the bad VECTOR-PUSH-EXTEND frame causes GC problems, though that may
1214 not be the actual problem. (CMU CL 18c doesn't have problems with this.)
1216 217: "Bad type operations with FUNCTION types"
1219 * (values-type-union (specifier-type '(function (base-char)))
1220 (specifier-type '(function (integer))))
1222 #<FUN-TYPE (FUNCTION (BASE-CHAR) *)>
1224 It causes insertion of wrong type assertions into generated
1228 (let ((f (etypecase x
1229 (character #'write-char)
1230 (integer #'write-byte))))
1233 (character (write-char x s))
1234 (integer (write-byte x s)))))
1236 Then (FOO #\1 *STANDARD-OUTPUT*) signals type error.
1238 (In 0.7.9.1 the result type is (FUNCTION * *), so Python does not
1239 produce invalid code, but type checking is not accurate. Similar
1240 problems exist with VALUES-TYPE-INTERSECTION.)
1242 218: "VALUES type specifier semantics"
1243 (THE (VALUES ...) ...) in safe code discards extra values.
1245 (defun test (x y) (the (values integer) (truncate x y)))
1248 219: "DEFINE-COMPILER-MACRO in non-toplevel contexts evaluated at compile-time"
1253 (define-compiler-macro bar (&whole whole)
1254 (declare (ignore whole))
1255 (print "expanding compiler macro")
1258 * (defun baz (x) (bar))
1260 "expanding compiler macro"
1266 Sbcl 0.7.9 fails to compile
1268 (multiple-value-call #'list
1269 (the integer (helper))
1272 Type check for INTEGER is inserted, the result of which serves as
1273 the first argument of M-V-C, is inserted after evaluation of NIL. So
1274 arguments of M-V-C are pushed in the wrong order. As a temporary
1275 workaround type checking was disabled for M-V-Cs in 0.7.9.13. A
1276 better solution would be to put a check between evaluation of
1277 arguments, but it could be tricky to check result types of PROG1, IF
1280 222: "environment problems in PCL"
1283 (symbol-macrolet ((x 1))
1285 (macrolet ((ml (form) `(progn ,form ,x)))
1290 debugger invoked on condition of type UNBOUND-VARIABLE:
1291 The variable X is unbound.
1293 223: "(SETF FDEFINITION) and #' semantics broken for wrappers"
1298 (lambda (&rest rest)
1299 (format t "~&about to call ~S on ~S~%" fn rest)
1301 (format t "~&returned from ~S~%" fn)))
1302 (setf (fdefinition 'foo)
1305 does what one would expect, this
1309 (setf (fdefinition 'bar)
1310 (lambda (&rest rest)
1311 (format t "~&about to enter BAR ~S~%" rest)
1313 (format t "~&back from BAR~%"))))
1315 recurses endlessly in sbcl-0.7.9.32. (Or it works if #' and
1316 FDEFINITION are replaced by SYMBOL-FUNCTION.)
1319 SBCL 0.7.8 fails to compile
1320 (localy (declare (optimize (safety 3)))
1321 (ignore-errors (progn (values-list (car (list '(1 . 2)))) t)))
1322 (the LOCALY there is not a typo; any unknown function (e.g. FROB)
1325 226: "AVER failure in COMPILE-FILE of clocc-ansi-test/tests.lisp"
1326 (APD points out that this seems to be another symptom of bug #115.)
1327 sbcl-0.7.9.43 dies with failed AVER "(EQ (TN-PHYSENV TN) TN-ENV)" when
1328 trying to compile clocc-ansi-test/tests.lisp. sbcl-0.7.9.31 was able to
1329 to compile it. A smaller test case exhibiting the same problem is
1330 (declaim (optimize (speed 0) (safety 3) (debug 3)))
1332 (flet ((safe-format (stream string &rest r)
1333 (unless (ignore-errors (progn
1334 (apply #'format stream string r)
1336 (format stream "~&foo ~S" string))))
1338 ((eq my-result :ERROR)
1340 ((ignore-errors (typep condition result))
1341 (safe-format t "~&bar ~S" result))
1343 (safe-format t "~&baz ~S (~A) ~S" condition condition result)))))))
1345 227: "compiler bewilderment with adjustable vectors and COPY-SEQ"
1346 (fixed in sbcl-0.7.9.65)
1348 228: "function-lambda-expression problems"
1349 in sbcl-0.7.9.6x, from the REPL:
1350 * (progn (declaim (inline foo)) (defun foo (x) x))
1352 * (function-lambda-expression #'foo)
1353 (SB-C:LAMBDA-WITH-LEXENV NIL NIL NIL (X) (BLOCK FOO X)), NIL, FOO
1354 but this first return value is not suitable for input to FUNCTION or
1355 COMPILE, as required by ANSI.
1357 DEFUNCT CATEGORIES OF BUGS
1359 These labels were used for bugs related to the old IR1 interpreter.
1360 The # values reached 6 before the category was closed down.