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 implementation of #'+ returns its single argument without
288 type checking, e.g. (+ "illegal") => "illegal".
291 The debugger LIST-LOCATIONS command doesn't work properly.
294 Compiling and loading
295 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
297 then requesting a BACKTRACE at the debugger prompt gives no information
298 about where in the user program the problem occurred.
301 The compiler is supposed to do type inference well enough that
304 ((SIMPLE-ARRAY SINGLE-FLOAT)
306 (DECLARE (TYPE (SIMPLE-ARRAY SINGLE-FLOAT) X))
309 is redundant. However, as reported by Juan Jose Garcia Ripoll for
310 CMU CL, it sometimes doesn't. Adding declarations is a pretty good
311 workaround for the problem for now, but can't be done by the TYPECASE
312 macros themselves, since it's too hard for the macro to detect
313 assignments to the variable within the clause.
314 Note: The compiler *is* smart enough to do the type inference in
315 many cases. This case, derived from a couple of MACROEXPAND-1
316 calls on Ripoll's original test case,
318 (DECLARE (OPTIMIZE SPEED (SAFETY 0)))
319 (COND ((TYPEP A '(SIMPLE-ARRAY SINGLE-FLOAT)) NIL
320 (LET ((LENGTH (ARRAY-TOTAL-SIZE A)))
321 (LET ((I 0) (G2554 LENGTH))
322 (DECLARE (TYPE REAL G2554) (TYPE REAL I))
325 (WHEN (>= I G2554) (GO SB-LOOP::END-LOOP))
326 (SETF (ROW-MAJOR-AREF A I) (- (ROW-MAJOR-AREF A I)))
327 (GO SB-LOOP::NEXT-LOOP)
328 SB-LOOP::END-LOOP))))))
329 demonstrates the problem; but the problem goes away if the TAGBODY
330 and GO forms are removed (leaving the SETF in ordinary, non-looping
331 code), or if the TAGBODY and GO forms are retained, but the
332 assigned value becomes 0.0 instead of (- (ROW-MAJOR-AREF A I)).
335 Paul Werkowski wrote on cmucl-imp@cons.org 2000-11-15
336 I am looking into this problem that showed up on the cmucl-help
337 list. It seems to me that the "implementation specific environment
338 hacking functions" found in pcl/walker.lisp are completely messed
339 up. The good thing is that they appear to be barely used within
340 PCL and the munged environment object is passed to cmucl only
341 in calls to macroexpand-1, which is probably why this case fails.
342 SBCL uses essentially the same code, so if the environment hacking
343 is screwed up, it affects us too.
346 Using the pretty-printer from the command prompt gives funny
347 results, apparently because the pretty-printer doesn't know
348 about user's command input, including the user's carriage return
349 that the user, and therefore the pretty-printer thinks that
350 the new output block should start indented 2 or more characters
351 rightward of the correct location.
354 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
355 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
356 crashes SBCL. In general tracing anything which is used in the
357 implementation of TRACE is likely to have the same problem.
360 As reported by Martin Atzmueller on sbcl-devel 26 Dec 2000,
361 ANSI says that WITH-OUTPUT-TO-STRING should have a keyword
362 :ELEMENT-TYPE, but in sbcl-0.6.9 this is not defined for
363 WITH-OUTPUT-TO-STRING.
366 ANSI says in one place that type declarations can be abbreviated even
367 when the type name is not a symbol, e.g.
368 (DECLAIM ((VECTOR T) *FOOVECTOR*))
369 SBCL doesn't support this. But ANSI says in another place that this
370 isn't allowed. So it's not clear this is a bug after all. (See the
371 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
375 as pointed out by Dan Barlow on sbcl-devel 2000-07-02:
376 The PICK-TEMPORARY-FILE-NAME utility used by LOAD-FOREIGN uses
377 an easily guessable temporary filename in a way which might open
378 applications using LOAD-FOREIGN to hijacking by malicious users
379 on the same machine. Incantations for doing this safely are
380 floating around the net in various "how to write secure programs
381 despite Unix" documents, and it would be good to (1) fix this in
382 LOAD-FOREIGN, and (2) hunt for any other code which uses temporary
383 files and make it share the same new safe logic.
386 Functions are assigned names based on the context in which they're
387 defined. This is less than ideal for the functions which are
388 used to implement CLOS methods. E.g. the output of
389 (DESCRIBE 'PRINT-OBJECT) lists functions like
390 #<FUNCTION "DEF!STRUCT (TRACE-INFO (:MAKE-LOAD-FORM-FUN SB-KERNEL:JUST-DUMP-IT-NORMALLY) (:PRINT-OBJECT #))" {1020E49}>
392 #<FUNCTION "MACROLET ((FORCE-DELAYED-DEF!METHODS NIL #))" {1242871}>
393 It would be better if these functions' names always identified
394 them as methods, and identified their generic functions and
398 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
399 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
400 the mean of the distribution can be systematically O(0.1%) wrong.
401 Just increasing R-I-E-B is probably not a good solution, since
402 it would decrease efficiency more than is probably necessary. Perhaps
403 using some sort of accept/reject method would be better.
406 Internally the compiler sometimes evaluates
407 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
408 (I stumbled across this when I added an
409 (assert (not (eq type1 *wild-type*)))
410 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
411 in a type context should probably be translated to T, and so it's
412 probably wrong to ask whether it's equal to the T type and then (using
413 the EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
414 (I haven't tried to investigate this bug enough to guess whether
415 there might be any user-level symptoms.)
417 In fact, the type system is likely to depend on this inequality not
418 holding... * is not equivalent to T in many cases, such as
419 (VECTOR *) /= (VECTOR T).
422 Inconsistencies between derived and declared VALUES return types for
423 DEFUN aren't checked very well. E.g. the logic which successfully
424 catches problems like
425 (declaim (ftype (function (fixnum) float) foo))
427 (declare (type integer x))
428 (values x)) ; wrong return type, detected, gives warning, good!
430 (declaim (ftype (function (t) (values t t)) bar))
432 (values x)) ; wrong number of return values, no warning, bad!
433 The cause of this is seems to be that (1) the internal function
434 VALUES-TYPES-EQUAL-OR-INTERSECT used to make the check handles its
435 arguments symmetrically, and (2) when the type checking code was
436 written back when when SBCL's code was still CMU CL, the intent
438 (declaim (ftype (function (t) t) bar))
440 (values x x)) ; wrong number of return values; should give warning?
441 not be warned for, because a two-valued return value is considered
442 to be compatible with callers who expects a single value to be
443 returned. That intent is probably not appropriate for modern ANSI
444 Common Lisp, but fixing this might be complicated because of other
445 divergences between auld-style and new-style handling of
446 multiple-VALUES types. (Some issues related to this were discussed
447 on cmucl-imp at some length sometime in 2000.)
450 The facility for dumping a running Lisp image to disk gets confused
451 when run without the PURIFY option, and creates an unnecessarily large
452 core file (apparently representing memory usage up to the previous
453 high-water mark). Moreover, when the file is loaded, it confuses the
454 GC, so that thereafter memory usage can never be reduced below that
458 In sbcl-0.6.11.41 (and in all earlier SBCL, and in CMU
459 CL), out-of-line structure slot setters are horribly inefficient
460 whenever the type of the slot is declared, because out-of-line
461 structure slot setters are implemented as closures to save space,
462 so the compiler doesn't compile the type test into code, but
463 instead just saves the type in a lexical closure and interprets it
465 A proper solution involves deciding whether it's really worth
466 saving space by implementing structure slot accessors as closures.
467 (If it's not worth it, the problem vanishes automatically. If it
468 is worth it, there are hacks we could use to force type tests to
469 be compiled anyway, and even shared. E.g. we could implement
470 an EQUAL hash table mapping from types to compiled type tests,
471 and save the appropriate compiled type test as part of each lexical
472 closure; or we could make the lexical closures be placeholders
473 which overwrite their old definition as a lexical closure with
474 a new compiled definition the first time that they're called.)
475 As a workaround for the problem, #'(SETF FOO) expressions can
476 be replaced with (EFFICIENT-SETF-FUNCTION FOO), where
477 (defmacro efficient-setf-function (place-function-name)
478 (or #+sbcl (and (sb-impl::info :function :accessor-for place-function-name)
479 ;; a workaround for the problem, encouraging the
480 ;; inline expansion of the structure accessor, so
481 ;; that the compiler can optimize its type test
482 (let ((new-value (gensym "NEW-VALUE-"))
483 (structure-value (gensym "STRUCTURE-VALUE-")))
484 `(lambda (,new-value ,structure-value)
485 (setf (,place-function-name ,structure-value)
487 ;; no problem, can just use the ordinary expansion
488 `(function (setf ,place-function-name))))
491 There's apparently a bug in CEILING optimization which caused
492 Douglas Crosher to patch the CMU CL version. Martin Atzmueller
493 applied the patches to SBCL and they didn't seem to cause problems
494 (as reported sbcl-devel 2001-05-04). However, since the patches
495 modify nontrivial code which was apparently written incorrectly
496 the first time around, until regression tests are written I'm not
497 comfortable merging the patches in the CVS version of SBCL.
500 (TIME (ROOM T)) reports more than 200 Mbytes consed even for
501 a clean, just-started SBCL system. And it seems to be right:
502 (ROOM T) can bring a small computer to its knees for a *long*
503 time trying to GC afterwards. Surely there's some more economical
504 way to implement (ROOM T).
507 reported by Martin Atzmueller 2001-06-25; originally from CMU CL bugs
509 (in-package :cl-user)
510 ;;; The following invokes a compiler error.
511 (declaim (optimize (speed 2) (debug 3)))
514 (unwind-protect nil)))
518 The error message in sbcl-0.6.12.42 is
519 internal error, failed AVER:
520 "(COMMON-LISP:EQ (SB!C::TN-ENVIRONMENT SB!C:TN) SB!C::TN-ENV)"
523 When the compiler inline expands functions, it may be that different
524 kinds of return values are generated from different code branches.
525 E.g. an inline expansion of POSITION generates integer results
526 from one branch, and NIL results from another. When that inline
527 expansion is used in a context where only one of those results
530 (aref *a1* (position x *a2*)))
531 and the compiler can't prove that the unacceptable branch is
532 never taken, then bogus type mismatch warnings can be generated.
533 If you need to suppress the type mismatch warnings, you can
534 suppress the inline expansion,
536 #+sbcl (declare (notinline position)) ; to suppress bug 117 bogowarnings
537 (aref *a1* (position x *a2*)))
538 or, sometimes, suppress them by declaring the result to be of an
541 (aref *a1* (the integer (position x *a2*))))
543 This is not a new compiler problem in 0.7.0, but the new compiler
544 transforms for FIND, POSITION, FIND-IF, and POSITION-IF make it
545 more conspicuous. If you don't need performance from these functions,
546 and the bogus warnings are a nuisance for you, you can return to
547 your pre-0.7.0 state of grace with
548 #+sbcl (declaim (notinline find position find-if position-if)) ; bug 117..
551 as reported by Eric Marsden on cmucl-imp@cons.org 2001-08-14:
552 (= (FLOAT 1 DOUBLE-FLOAT-EPSILON)
553 (+ (FLOAT 1 DOUBLE-FLOAT-EPSILON) DOUBLE-FLOAT-EPSILON)) => T
554 when of course it should be NIL. (He says it only fails for X86,
555 not SPARC; dunno about Alpha.)
557 Also, "the same problem exists for LONG-FLOAT-EPSILON,
558 DOUBLE-FLOAT-NEGATIVE-EPSILON, LONG-FLOAT-NEGATIVE-EPSILON (though
559 for the -negative- the + is replaced by a - in the test)."
561 Raymond Toy comments that this is tricky on the X86 since its FPU
562 uses 80-bit precision internally.
565 Even in sbcl-0.pre7.x, which is supposed to be free of the old
566 non-ANSI behavior of treating the function return type inferred
567 from the current function definition as a declaration of the
568 return type from any function of that name, the return type of NIL
569 is attached to FOO in 120a above, and used to optimize code which
573 As of version 0.pre7.14, SBCL's implementation of MACROLET makes
574 the entire lexical environment at the point of MACROLET available
575 in the bodies of the macroexpander functions. In particular, it
576 allows the function bodies (which run at compile time) to try to
577 access lexical variables (which are only defined at runtime).
578 It doesn't even issue a warning, which is bad.
580 The SBCL behavior arguably conforms to the ANSI spec (since the
581 spec says that the behavior is undefined, ergo anything conforms).
582 However, it would be better to issue a compile-time error.
583 Unfortunately I (WHN) don't see any simple way to detect this
584 condition in order to issue such an error, so for the meantime
585 SBCL just does this weird broken "conforming" thing.
587 The ANSI standard says, in the definition of the special operator
589 The macro-expansion functions defined by MACROLET are defined
590 in the lexical environment in which the MACROLET form appears.
591 Declarations and MACROLET and SYMBOL-MACROLET definitions affect
592 the local macro definitions in a MACROLET, but the consequences
593 are undefined if the local macro definitions reference any
594 local variable or function bindings that are visible in that
596 Then it seems to contradict itself by giving the example
598 (macrolet ((fudge (z)
599 ;The parameters x and flag are not accessible
600 ; at this point; a reference to flag would be to
601 ; the global variable of that name.
602 ` (if flag (* ,z ,z) ,z)))
603 ;The parameters x and flag are accessible here.
607 The comment "a reference to flag would be to the global variable
608 of the same name" sounds like good behavior for the system to have.
609 but actual specification quoted above says that the actual behavior
612 (Since 0.7.8.23 macroexpanders are defined in a restricted version
613 of the lexical environment, containing no lexical variables and
614 functions, which seems to conform to ANSI and CLtL2, but signalling
615 a STYLE-WARNING for references to variables similar to locals might
619 (as reported by Gabe Garza on cmucl-help 2001-09-21)
621 (defun test-pred (x y)
625 (func (lambda () x)))
626 (print (eq func func))
627 (print (test-pred func func))
628 (delete func (list func))))
629 Now calling (TEST-CASE) gives output
632 (#<FUNCTION {500A9EF9}>)
633 Evidently Python thinks of the lambda as a code transformation so
634 much that it forgets that it's also an object.
637 The DEFSTRUCT section of the ANSI spec, in the :CONC-NAME section,
638 specifies a precedence rule for name collisions between slot accessors of
639 structure classes related by inheritance. As of 0.7.0, SBCL still
643 Ideally, uninterning a symbol would allow it, and its associated
644 FDEFINITION and PROCLAIM data, to be reclaimed by the GC. However,
645 at least as of sbcl-0.7.0, this isn't the case. Information about
646 FDEFINITIONs and PROCLAIMed properties is stored in globaldb.lisp
647 essentially in ordinary (non-weak) hash tables keyed by symbols.
648 Thus, once a system has an entry in this system, it tends to live
649 forever, even when it is uninterned and all other references to it
653 (reported by Arnaud Rouanet on cmucl-imp 2001-12-18)
654 (defmethod foo ((x integer))
656 (defmethod foo :around ((x integer))
659 Now (FOO 3) should return 3, but instead it returns 4.
662 Pretty-printing nested backquotes doesn't work right, as
663 reported by Alexey Dejneka sbcl-devel 2002-01-13:
665 ``(FOO SB-IMPL::BACKQ-COMMA-AT S)
666 * (lisp-implementation-version)
670 (reported by Jesse Bouwman 2001-10-24 through the unfortunately
671 prominent SourceForge web/db bug tracking system, which is
672 unfortunately not a reliable way to get a timely response from
673 the SBCL maintainers)
674 In the course of trying to build a test case for an
675 application error, I encountered this behavior:
676 If you start up sbcl, and then lay on CTRL-C for a
677 minute or two, the lisp process will eventually say:
678 %PRIMITIVE HALT called; the party is over.
679 and throw you into the monitor. If I start up lisp,
680 attach to the process with strace, and then do the same
681 (abusive) thing, I get instead:
682 access failure in heap page not marked as write-protected
683 and the monitor again. I don't know enough to have the
684 faintest idea of what is going on here.
685 This is with sbcl 6.12, uname -a reports:
686 Linux prep 2.2.19 #4 SMP Tue Apr 24 13:59:52 CDT 2001 i686 unknown
687 I (WHN) have verified that the same thing occurs on sbcl-0.pre7.141
688 under OpenBSD 2.9 on my X86 laptop. Do be patient when you try it:
689 it took more than two minutes (but less than five) for me.
692 (This was once known as IR1-4, but it lived on even after the
693 IR1 interpreter went to the big bit bucket in the sky.)
694 The system accepts DECLAIM in most places where DECLARE would be
695 accepted, without even issuing a warning. ANSI allows this, but since
696 it's fairly easy to mistype DECLAIM instead of DECLARE, and the
697 meaning is rather different, and it's unlikely that the user
698 has a good reason for doing DECLAIM not at top level, it would be
699 good to issue a STYLE-WARNING when this happens. A possible
700 fix would be to issue STYLE-WARNINGs for DECLAIMs not at top level,
701 or perhaps to issue STYLE-WARNINGs for any EVAL-WHEN not at top level.
702 [This is considered an IR1-interpreter-related bug because until
703 EVAL-WHEN is rewritten, which won't happen until after the IR1
704 interpreter is gone, the system's notion of what's a top-level form
705 and what's not will remain too confused to fix this problem.]
708 ANSI allows types `(COMPLEX ,FOO) to use very hairy values for
709 FOO, e.g. (COMPLEX (AND REAL (SATISFIES ODDP))). The old CMU CL
710 COMPLEX implementation didn't deal with this, and hasn't been
711 upgraded to do so. (This doesn't seem to be a high priority
712 conformance problem, since seems hard to construct useful code
716 Floating point errors are reported poorly. E.g. on x86 OpenBSD
719 debugger invoked on condition of type SB-KERNEL:FLOATING-POINT-EXCEPTION:
720 An arithmetic error SB-KERNEL:FLOATING-POINT-EXCEPTION was signalled.
721 No traps are enabled? How can this be?
722 It should be possible to be much more specific (overflow, division
723 by zero, etc.) and of course the "How can this be?" should be fixable.
725 See also bugs #45.c and #183
728 In sbcl-0.7.1.3 on x86, COMPILE-FILE on the file
729 (in-package :cl-user)
732 (defstruct foo bar bletch)
734 (labels ((kidify1 (kid)
742 (declare (inline kid-frob))
745 (the simple-vector (foo-bar perd)))))
747 debugger invoked on condition of type TYPE-ERROR:
748 The value NIL is not of type SB-C::NODE.
749 The location of this failure has moved around as various related
750 issues were cleaned up. As of sbcl-0.7.1.9, it occurs in
751 NODE-BLOCK called by LAMBDA-COMPONENT called by IR2-CONVERT-CLOSURE.
753 (Python LET-converts KIDIFY1 into KID-FROB, then tries to inline
754 expand KID-FROB into %ZEEP. Having partially done it, it sees a call
755 of KIDIFY1, which already does not exist. So it gives up on
756 expansion, leaving garbage consisting of infinished blocks of the
757 partially converted function.)
760 Functions SUBTYPEP, TYPEP, UPGRADED-ARRAY-ELEMENT-TYPE, and
761 UPGRADED-COMPLEX-PART-TYPE should have an optional environment argument.
762 (reported by Alexey Dejneka sbcl-devel 2002-04-12)
765 (reported by Robert E. Brown 2002-04-16)
766 When a function is called with too few arguments, causing the
767 debugger to be entered, the uninitialized slots in the bad call frame
768 seem to cause GCish problems, being interpreted as tagged data even
769 though they're not. In particular, executing ROOM in the
770 debugger at that point causes AVER failures:
773 * (lisp-implementation-version)
779 failed AVER: "(SAP= CURRENT END)"
780 (Christophe Rhodes reports that this doesn't occur on the SPARC, which
781 isn't too surprising since there are many differences in stack
782 implementation and GC conservatism between the X86 and other ports.)
786 (in-package :cl-user)
788 (defmethod permanentize ((uustk uustk))
789 (flet ((frob (hash-table test-for-deletion)
791 (obj-entry.stale? (oe)
792 (destructuring-bind (key . datum) oe
793 (declare (type simple-vector key))
794 (deny0 (void? datum))
795 (some #'stale? key))))
796 (declare (inline frob obj-entry.stale?))
797 (frob (uustk.args-hash->obj-alist uustk)
799 (frob (uustk.hash->memoized-objs-list uustk)
802 in sbcl-0.7.3.11 causes an assertion failure,
805 (AND (NULL (BLOCK-SUCC B))
806 (NOT (BLOCK-DELETE-P B))
807 (NOT (EQ B (COMPONENT-HEAD #)))))"
810 In sbcl-0.7.3.11, compiling the (illegal) code
811 (in-package :cl-user)
812 (defmethod prove ((uustk uustk))
815 gives the (not terribly clear) error message
817 ; (during macroexpansion of (DEFMETHOD PROVE ...))
818 ; can't get template for (FROB NIL NIL)
819 The problem seems to be that the code walker used by the DEFMETHOD
820 macro is unhappy with the illegal syntax in the method body, and
821 is giving an unclear error message.
824 sbcl's treatment of at least macro lambda lists is too permissive;
825 e.g., in sbcl-0.7.3.7:
826 (defmacro foo (&rest rest bar) `(,bar ,rest))
827 (macroexpand '(foo quux zot)) -> (QUUX (QUUX ZOT))
828 whereas section 3.4.4 of the CLHS doesn't allow required parameters
829 to come after the rest argument.
832 The compiler sometimes tries to constant-fold expressions before
833 it checks to see whether they can be reached. This can lead to
834 bogus warnings about errors in the constant folding, e.g. in code
837 (WRITE-STRING (> X 0) "+" "0"))
838 compiled in a context where the compiler can prove that X is NIL,
839 and the compiler complains that (> X 0) causes a type error because
840 NIL isn't a valid argument to #'>. Until sbcl-0.7.4.10 or so this
841 caused a full WARNING, which made the bug really annoying because then
842 COMPILE and COMPILE-FILE returned FAILURE-P=T for perfectly legal
843 code. Since then the warning has been downgraded to STYLE-WARNING,
844 so it's still a bug but at least it's a little less annoying.
846 178: "AVER failure compiling confused THEs in FUNCALL"
847 In sbcl-0.7.4.24, compiling
849 (funcall (the function (the standard-object x))))
852 "(AND (EQ (IR2-CONTINUATION-PRIMITIVE-TYPE 2CONT) FUNCTION-PTYPE) (EQ CHECK T))"
853 This variant compiles OK, though:
854 (defun bug178alternative (x)
855 (funcall (the nil x)))
857 (since 0.7.8.9 it does not signal an error; see also bug 199)
859 183: "IEEE floating point issues"
860 Even where floating point handling is being dealt with relatively
861 well (as of sbcl-0.7.5, on sparc/sunos and alpha; see bug #146), the
862 accrued-exceptions and current-exceptions part of the fp control
863 word don't seem to bear much relation to reality. E.g. on
867 debugger invoked on condition of type DIVISION-BY-ZERO:
868 arithmetic error DIVISION-BY-ZERO signalled
869 0] (sb-vm::get-floating-point-modes)
871 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
872 :ROUNDING-MODE :NEAREST
873 :CURRENT-EXCEPTIONS NIL
874 :ACCRUED-EXCEPTIONS (:INEXACT)
877 * (sb-vm::get-floating-point-modes)
878 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
879 :ROUNDING-MODE :NEAREST
880 :CURRENT-EXCEPTIONS (:INEXACT)
881 :ACCRUED-EXCEPTIONS (:INEXACT)
884 187: "type inference confusion around DEFTRANSFORM time"
885 (reported even more verbosely on sbcl-devel 2002-06-28 as "strange
886 bug in DEFTRANSFORM")
887 After the file below is compiled and loaded in sbcl-0.7.5, executing
888 (TCX (MAKE-ARRAY 4 :FILL-POINTER 2) 0)
889 at the REPL returns an adjustable vector, which is wrong. Presumably
890 somehow the DERIVE-TYPE information for the output values of %WAD is
891 being mispropagated as a type constraint on the input values of %WAD,
892 and so causing the type test to be optimized away. It's unclear how
893 hand-expanding the DEFTRANSFORM would change this, but it suggests
894 the DEFTRANSFORM machinery (or at least the way DEFTRANSFORMs are
895 invoked at a particular phase) is involved.
896 (cl:in-package :sb-c)
897 (eval-when (:compile-toplevel)
898 ;;; standin for %DATA-VECTOR-AND-INDEX
899 (defknown %dvai (array index)
901 (foldable flushable))
902 (deftransform %dvai ((array index)
906 (let* ((atype (continuation-type array))
907 (eltype (array-type-specialized-element-type atype)))
908 (when (eq eltype *wild-type*)
909 (give-up-ir1-transform
910 "specialized array element type not known at compile-time"))
911 (when (not (array-type-complexp atype))
912 (give-up-ir1-transform "SIMPLE array!"))
913 `(if (array-header-p array)
914 (%wad array index nil)
915 (values array index))))
916 ;;; standin for %WITH-ARRAY-DATA
917 (defknown %wad (array index (or index null))
918 (values (simple-array * (*)) index index index)
919 (foldable flushable))
920 ;;; (Commenting out this optimizer causes the bug to go away.)
921 (defoptimizer (%wad derive-type) ((array start end))
922 (let ((atype (continuation-type array)))
923 (when (array-type-p atype)
924 (values-specifier-type
925 `(values (simple-array ,(type-specifier
926 (array-type-specialized-element-type atype))
928 index index index)))))
930 (defun %wad (array start end)
931 (format t "~&in %WAD~%")
932 (%with-array-data array start end))
933 (cl:in-package :cl-user)
935 (declare (type (vector t) v))
936 (declare (notinline sb-kernel::%with-array-data))
937 ;; (Hand-expending DEFTRANSFORM %DVAI here also causes the bug to
941 188: "compiler performance fiasco involving type inference and UNION-TYPE"
942 (In sbcl-0.7.6.10, DEFTRANSFORM CONCATENATE was commented out until this
943 bug could be fixed properly, so you won't see the bug unless you restore
944 the DEFTRANSFORM by hand.) In sbcl-0.7.5.11 on a 700 MHz Pentium III,
948 (declare (optimize (safety 3)))
949 (declare (optimize (compilation-speed 2)))
950 (declare (optimize (speed 1) (debug 1) (space 1)))
951 (let ((fn "if-this-file-exists-the-universe-is-strange"))
952 (load fn :if-does-not-exist nil)
953 (load (concatenate 'string fn ".lisp") :if-does-not-exist nil)
954 (load (concatenate 'string fn ".fasl") :if-does-not-exist nil)
955 (load (concatenate 'string fn ".misc-garbage")
956 :if-does-not-exist nil)))))
958 134.552 seconds of real time
959 133.35156 seconds of user run time
960 0.03125 seconds of system run time
961 [Run times include 2.787 seconds GC run time.]
963 246883368 bytes consed.
964 BACKTRACE from Ctrl-C in the compilation shows that the compiler is
965 thinking about type relationships involving types like
967 (OR (INTEGER 576 576)
978 190: "PPC/Linux pipe? buffer? bug"
979 In sbcl-0.7.6, the run-program.test.sh test script sometimes hangs
980 on the PPC/Linux platform, waiting for a zombie env process. This
981 is a classic symptom of buffer filling and deadlock, but it seems
982 only sporadically reproducible.
984 191: "Miscellaneous PCL deficiencies"
985 (reported by Alexey Dejneka sbcl-devel 2002-08-04)
986 a. DEFCLASS does not inform the compiler about generated
987 functions. Compiling a file with
991 (WITH-SLOTS (A-CLASS-X) A
993 results in a STYLE-WARNING:
995 SB-SLOT-ACCESSOR-NAME::|COMMON-LISP-USER A-CLASS-X slot READER|
997 APD's fix for this was checked in to sbcl-0.7.6.20, but Pierre
998 Mai points out that the declamation of functions is in fact
999 incorrect in some cases (most notably for structure
1000 classes). This means that at present erroneous attempts to use
1001 WITH-SLOTS and the like on classes with metaclass STRUCTURE-CLASS
1002 won't get the corresponding STYLE-WARNING.
1003 c. the examples in CLHS 7.6.5.1 (regarding generic function lambda
1004 lists and &KEY arguments) do not signal errors when they should.
1006 192: "Python treats free type declarations as promises."
1007 b. What seemed like the same fundamental problem as bug 192a, but
1008 was not fixed by the same (APD "more strict type checking
1009 sbcl-devel 2002-08-97) patch:
1010 (DOTIMES (I ...) (DOTIMES (J ...) (DECLARE ...) ...)):
1011 (declaim (optimize (speed 1) (safety 3)))
1012 (defun trust-assertion (i)
1014 (declare (type (mod 4) i)) ; when commented out, behavior changes!
1017 (trust-assertion 6) ; prints nothing unless DECLARE is commented out
1021 193: "unhelpful CLOS error reporting when the primary method is missing"
1023 (defmethod foo :before ((x t)) (print x))
1024 is the only method defined on FOO, the error reporting when e.g.
1026 is relatively unhelpful:
1027 There is no primary method for the generic function
1028 #<STANDARD-GENERIC-FUNCTION FOO (1)>.
1029 with the offending argument nowhere visible in the backtrace. This
1030 continues even if there *are* primary methods, just not for the
1031 specified arg type, e.g.
1032 (defmethod foo ((x character)) (print x))
1033 (defmethod foo ((x string)) (print x))
1034 (defmethod foo ((x pathname)) ...)
1035 In that case it could be very helpful to know what argument value is
1036 falling through the cracks of the defined primary methods, but the
1037 error message stays the same (even BACKTRACE doesn't tell you what the
1038 bad argument value is).
1040 194: "no error from (THE REAL '(1 2 3)) in some cases"
1043 (multiple-value-prog1 (progn (the real '(1 2 3))))
1044 returns (1 2 3) instead of signalling an error. This was fixed by
1045 APD's "more strict type checking patch", but although the fixed
1046 code (in sbcl-0.7.7.19) works (signals TYPE-ERROR) interactively,
1047 it's difficult to write a regression test for it, because
1048 (IGNORE-ERRORS (MULTIPLE-VALUE-PROG1 (PROGN (THE REAL '(1 2 3)))))
1049 still returns (1 2 3).
1051 b. (IGNORE-ERRORS (MULTIPLE-VALUE-PROG1 (PROGN (THE REAL '(1 2 3)))))
1052 returns (1 2 3). (As above, this shows up when writing regression
1053 tests for fixed-ness of part a.)
1054 c. Also in sbcl-0.7.7.9, (IGNORE-ERRORS (THE REAL '(1 2 3))) => (1 2 3).
1056 (null (ignore-errors
1058 (arg2 (identity (the real #(1 2 3)))))
1059 (if (< arg1 arg2) arg1 arg2))))
1061 but putting the same expression inside (DEFUN FOO () ...),
1064 * Actually this entry is probably multiple bugs, as
1065 Alexey Dejneka commented on sbcl-devel 2002-09-03:)
1066 I don't think that placing these two bugs in one entry is
1067 a good idea: they have different explanations. The second
1068 (min 1 nil) is caused by flushing of unused code--IDENTITY
1069 can do nothing with it. So it is really bug 122. The first
1070 (min nil) is due to M-V-PROG1: substituting a continuation
1071 for the result, it forgets about type assertion. The purpose
1072 of IDENTITY is to save the restricted continuation from
1073 inaccurate transformations.
1074 * Alexey Dejneka pointed out that
1075 (IGNORE-ERRORS (IDENTITY (THE REAL '(1 2 3))))
1076 works as it should. Also
1077 (IGNORE-ERRORS (VALUES (THE REAL '(1 2 3))))
1078 works as it should. Perhaps this is another case of VALUES type
1079 intersections behaving in non-useful ways?
1081 199: "hairy FUNCTION types confuse the compiler"
1082 (reported by APD sbcl-devel 2002-09-15)
1084 (EQ NIL (FUNCALL F)))
1087 (DECLARE (TYPE (AND FUNCTION (SATISFIES MUR)) F))
1090 fails to compile, printing
1092 "(AND (EQ (IR2-CONTINUATION-PRIMITIVE-TYPE 2CONT) FUNCTION-PTYPE) (EQ CHECK T))"
1094 APD further reports that this bug is not present in CMUCL.
1096 (this case was fixed in 0.7.8.9; see also bug 178)
1098 201: "Incautious type inference from compound CONS types"
1099 (reported by APD sbcl-devel 2002-09-17)
1101 (LET ((Y (CAR (THE (CONS INTEGER *) X))))
1103 (FORMAT NIL "~S IS ~S, Y = ~S"
1110 (FOO ' (1 . 2)) => "NIL IS INTEGER, Y = 1"
1113 Compiler does not check THEs on unused values, e.g. in
1115 (progn (the real (list 1)) t)
1117 This situation may appear during optimizing away degenerate cases of
1118 certain functions: see bugs 54, 192b.
1120 205: "environment issues in cross compiler"
1121 (These bugs have no impact on user code, but should be fixed or
1123 a. Macroexpanders introduced with MACROLET are defined in the null
1124 lexical environment.
1125 b. The body of (EVAL-WHEN (:COMPILE-TOPLEVEL) ...) is evaluated in
1126 the null lexical environment.
1128 206: ":SB-FLUID feature broken"
1129 (reported by Antonio Martinez-Shotton sbcl-devel 2002-10-07)
1130 Enabling :SB-FLUID in the target-features list in sbcl-0.7.8 breaks
1133 207: "poorly distributed SXHASH results for compound data"
1134 SBCL's SXHASH could probably try a little harder. ANSI: "the
1135 intent is that an implementation should make a good-faith
1136 effort to produce hash-codes that are well distributed
1137 within the range of non-negative fixnums". But
1138 (let ((hits (make-hash-table)))
1141 (let* ((ij (cons i j))
1142 (newlist (push ij (gethash (sxhash ij) hits))))
1144 (format t "~&collision: ~S~%" newlist))))))
1145 reports lots of collisions in sbcl-0.7.8. A stronger MIX function
1146 would be an obvious way of fix. Maybe it would be acceptably efficient
1147 to redo MIX using a lookup into a 256-entry s-box containing
1148 29-bit pseudorandom numbers?
1150 208: "package confusion in PCL handling of structure slot handlers"
1151 In sbcl-0.7.8 compiling and loading
1153 (defstruct foo (slot (error "missing")) :type list :read-only t)
1154 (defmethod print-object ((foo foo) stream) (print nil stream))
1155 causes CERROR "attempting to modify a symbol in the COMMON-LISP
1156 package: FOO-SLOT". (This is fairly bad code, but still it's hard
1157 to see that it should cause symbols to be interned in the CL package.)
1159 211: "keywords processing"
1160 a. :ALLOW-OTHER-KEYS T should allow a function to receive an odd
1161 number of keyword arguments.
1164 (flet ((foo (&key y) (list y)))
1165 (list (foo :y 1 :y 2)))
1167 issues confusing message
1172 ; caught STYLE-WARNING:
1173 ; The variable #:G15 is defined but never used.
1175 212: "Sequence functions and circular arguments"
1176 COERCE, MERGE and CONCATENATE go into an infinite loop when given
1177 circular arguments; it would be good for the user if they could be
1178 given an error instead (ANSI 17.1.1 allows this behaviour on the part
1179 of the implementation, as conforming code cannot give non-proper
1180 sequences to these functions. MAP also has this problem (and
1181 solution), though arguably the convenience of being able to do
1182 (MAP 'LIST '+ FOO '#1=(1 . #1#))
1183 might be classed as more important (though signalling an error when
1184 all of the arguments are circular is probably desireable).
1186 213: "Sequence functions and type checking"
1187 a. MAKE-SEQUENCE, COERCE, MERGE and CONCATENATE cannot deal with
1188 various complicated, though recognizeable, CONS types [e.g.
1189 (CONS * (CONS * NULL))
1190 which according to ANSI should be recognized] (and, in SAFETY 3
1191 code, should return a list of LENGTH 2 or signal an error)
1192 b. MAP, when given a type argument that is SUBTYPEP LIST, does not
1193 check that it will return a sequence of the given type. Fixing
1194 it along the same lines as the others (cf. work done around
1195 sbcl-0.7.8.45) is possible, but doing so efficiently didn't look
1196 entirely straightforward.
1197 c. All of these functions will silently accept a type of the form
1199 whether or not the return value is of this type. This is
1200 probably permitted by ANSI (see "Exceptional Situations" under
1201 ANSI MAKE-SEQUENCE), but the DERIVE-TYPE mechanism does not
1202 know about this escape clause, so code of the form
1203 (INTEGERP (CAR (MAKE-SEQUENCE '(CONS INTEGER *) 2)))
1204 can erroneously return T.
1207 SBCL 0.6.12.43 fails to compile
1210 (declare (optimize (inhibit-warnings 0) (compilation-speed 2)))
1211 (flet ((foo (&key (x :vx x-p)) (list x x-p)))
1214 or a more simple example:
1217 (declare (optimize (inhibit-warnings 0) (compilation-speed 2)))
1218 (lambda (x) (declare (fixnum x)) (if (< x 0) 0 (1- x))))
1220 215: ":TEST-NOT handling by functions"
1221 a. FIND and POSITION currently signal errors when given non-NIL for
1222 both their :TEST and (deprecated) :TEST-NOT arguments, but by
1223 ANSI 17.2 "the consequences are unspecified", which by ANSI 1.4.2
1224 means that the effect is "unpredictable but harmless". It's not
1225 clear what that actually means; it may preclude conforming
1226 implementations from signalling errors.
1227 b. COUNT, REMOVE and the like give priority to a :TEST-NOT argument
1228 when conflict occurs. As a quality of implementation issue, it
1229 might be preferable to treat :TEST and :TEST-NOT as being in some
1230 sense the same &KEY, and effectively take the first test function in
1232 c. Again, a quality of implementation issue: it would be good to issue a
1233 STYLE-WARNING at compile-time for calls with :TEST-NOT, and a
1234 WARNING for calls with both :TEST and :TEST-NOT; possibly this
1235 latter should be WARNed about at execute-time too.
1237 216: "debugger confused by frames with invalid number of arguments"
1238 In sbcl-0.7.8.51, executing e.g. (VECTOR-PUSH-EXTEND T), BACKTRACE, Q
1239 leaves the system confused, enough so that (QUIT) no longer works.
1240 It's as though the process of working with the uninitialized slot in
1241 the bad VECTOR-PUSH-EXTEND frame causes GC problems, though that may
1242 not be the actual problem. (CMU CL 18c doesn't have problems with this.)
1244 217: "Bad type operations with FUNCTION types"
1247 * (values-type-union (specifier-type '(function (base-char)))
1248 (specifier-type '(function (integer))))
1250 #<FUN-TYPE (FUNCTION (BASE-CHAR) *)>
1252 It causes insertion of wrong type assertions into generated
1256 (let ((f (etypecase x
1257 (character #'write-char)
1258 (integer #'write-byte))))
1261 (character (write-char x s))
1262 (integer (write-byte x s)))))
1264 Then (FOO #\1 *STANDARD-OUTPUT*) signals type error.
1266 (In 0.7.9.1 the result type is (FUNCTION * *), so Python does not
1267 produce invalid code, but type checking is not accurate. Similar
1268 problems exist with VALUES-TYPE-INTERSECTION.)
1270 218: "VALUES type specifier semantics"
1271 (THE (VALUES ...) ...) in safe code discards extra values.
1273 (defun test (x y) (the (values integer) (truncate x y)))
1276 219: "DEFINE-COMPILER-MACRO in non-toplevel contexts evaluated at compile-time"
1281 (define-compiler-macro bar (&whole whole)
1282 (declare (ignore whole))
1283 (print "expanding compiler macro")
1286 * (defun baz (x) (bar))
1288 "expanding compiler macro"
1294 Sbcl 0.7.9 fails to compile
1296 (multiple-value-call #'list
1297 (the integer (helper))
1300 Type check for INTEGER is inserted, the result of which serves as
1301 the first argument of M-V-C, is inserted after evaluation of NIL. So
1302 arguments of M-V-C are pushed in the wrong order. As a temporary
1303 workaround type checking was disabled for M-V-Cs in 0.7.9.13. A
1304 better solution would be to put a check between evaluation of
1305 arguments, but it could be tricky to check result types of PROG1, IF
1308 DEFUNCT CATEGORIES OF BUGS
1310 These labels were used for bugs related to the old IR1 interpreter.
1311 The # values reached 6 before the category was closed down.