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.
45 3: "type checking of structure slots"
47 ANSI specifies that a type mismatch in a structure slot
48 initialization value should not cause a warning.
50 This one might not be fixed for a while because while we're big
51 believers in ANSI compatibility and all, (1) there's no obvious
52 simple way to do it (short of disabling all warnings for type
53 mismatches everywhere), and (2) there's a good portable
54 workaround, and (3) by their own reasoning, it looks as though
55 ANSI may have gotten it wrong. ANSI justifies this specification
57 The restriction against issuing a warning for type mismatches
58 between a slot-initform and the corresponding slot's :TYPE
59 option is necessary because a slot-initform must be specified
60 in order to specify slot options; in some cases, no suitable
62 However, in SBCL (as in CMU CL or, for that matter, any compiler
63 which really understands Common Lisp types) a suitable default
64 does exist, in all cases, because the compiler understands the
65 concept of functions which never return (i.e. has return type NIL).
66 Thus, as a portable workaround, you can use a call to some
67 known-never-to-return function as the default. E.g.
69 (BAR (ERROR "missing :BAR argument")
70 :TYPE SOME-TYPE-TOO-HAIRY-TO-CONSTRUCT-AN-INSTANCE-OF))
72 (DECLAIM (FTYPE (FUNCTION () NIL) MISSING-ARG))
73 (DEFUN REQUIRED-ARG () ; workaround for SBCL non-ANSI slot init typing
74 (ERROR "missing required argument"))
76 (BAR (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
77 (BLETCH (REQUIRED-ARG) :TYPE TRICKY-TYPE-OF-SOME-SORT)
78 (N-REFS-SO-FAR 0 :TYPE (INTEGER 0)))
79 Such code should compile without complaint and work correctly either
80 on SBCL or on any other completely compliant Common Lisp system.
82 b: &AUX argument in a boa-constructor without a default value means
83 "do not initilize this slot" and does not cause type error. But
84 an error may be signalled at read time and it would be good if
90 The "compiling top-level form:" output ought to be condensed.
91 Perhaps any number of such consecutive lines ought to turn into a
92 single "compiling top-level forms:" line.
95 It would be nice if the
97 (during macroexpansion)
98 said what macroexpansion was at fault, e.g.
100 (during macroexpansion of IN-PACKAGE,
101 during macroexpansion of DEFFOO)
104 (I *think* this is a bug. It certainly seems like strange behavior. But
105 the ANSI spec is scary, dark, and deep.. -- WHN)
106 (FORMAT NIL "~,1G" 1.4) => "1. "
107 (FORMAT NIL "~3,1G" 1.4) => "1. "
110 Sometimes (SB-EXT:QUIT) fails with
111 Argh! maximum interrupt nesting depth (4096) exceeded, exiting
112 Process inferior-lisp exited abnormally with code 1
113 I haven't noticed a repeatable case of this yet.
116 The printer doesn't report closures very well. This is true in
120 #<Closure Over Function "DEFUN STRUCTURE-SLOT-ACCESSOR" {134D1A1}>
121 It would be nice to make closures have a settable name slot,
122 and make things like DEFSTRUCT and FLET, which create closures,
123 set helpful values into this slot.
126 And as long as we're wishing, it would be awfully nice if INSPECT could
127 also report on closures, telling about the values of the bound variables.
130 The compiler assumes that any time a function of declared FTYPE
131 doesn't signal an error, its arguments were of the declared type.
132 E.g. compiling and loading
133 (DECLAIM (OPTIMIZE (SAFETY 3)))
134 (DEFUN FACTORIAL (X) (GAMMA (1+ X)))
136 (DECLAIM (FTYPE (FUNCTION (UNSIGNED-BYTE)) FACTORIAL))
138 (COND ((> (FACTORIAL X) 1.0E6)
139 (FORMAT T "too big~%"))
141 (FORMAT T "exactly ~S~%" (FACTORIAL X)))
143 (FORMAT T "approximately ~S~%" (FACTORIAL X)))))
146 will cause the INTEGERP case to be selected, giving bogus output a la
148 This violates the "declarations are assertions" principle.
149 According to the ANSI spec, in the section "System Class FUNCTION",
150 this is a case of "lying to the compiler", but the lying is done
151 by the code which calls FACTORIAL with non-UNSIGNED-BYTE arguments,
152 not by the unexpectedly general definition of FACTORIAL. In any case,
153 "declarations are assertions" means that lying to the compiler should
154 cause an error to be signalled, and should not cause a bogus
155 result to be returned. Thus, the compiler should not assume
156 that arbitrary functions check their argument types. (It might
157 make sense to add another flag (CHECKED?) to DEFKNOWN to
158 identify functions which *do* check their argument types.)
159 (Also, verify that the compiler handles declared function
160 return types as assertions.)
163 The definitions of SIGCONTEXT-FLOAT-REGISTER and
164 %SET-SIGCONTEXT-FLOAT-REGISTER in x86-vm.lisp say they're not
165 supported on FreeBSD because the floating point state is not saved,
166 but at least as of FreeBSD 4.0, the floating point state *is* saved,
167 so they could be supported after all. Very likely
168 SIGCONTEXT-FLOATING-POINT-MODES could now be supported, too.
171 a slew of floating-point-related errors reported by Peter Van Eynde
173 b: SBCL's value for LEAST-POSITIVE-SHORT-FLOAT on the x86 is
174 bogus, and should probably be 1.4012985e-45. In SBCL,
175 (/ LEAST-POSITIVE-SHORT-FLOAT 2) returns a number smaller
176 than LEAST-POSITIVE-SHORT-FLOAT. Similar problems
177 exist for LEAST-NEGATIVE-SHORT-FLOAT, LEAST-POSITIVE-LONG-FLOAT,
178 and LEAST-NEGATIVE-LONG-FLOAT.
179 c: Many expressions generate floating infinity on x86/Linux:
184 PVE's regression tests want them to raise errors. sbcl-0.7.0.5
185 on x86/Linux generates the infinities instead. That might or
186 might not be conforming behavior, but it's also inconsistent,
187 which is almost certainly wrong. (Inconsistency: (/ 1 0.0)
188 should give the same result as (/ 1.0 0.0), but instead (/ 1 0.0)
189 generates SINGLE-FLOAT-POSITIVE-INFINITY and (/ 1.0 0.0)
191 d: (in section12.erg) various forms a la
192 (FLOAT 1 DOUBLE-FLOAT-EPSILON)
193 don't give the right behavior.
196 type safety errors reported by Peter Van Eynde July 25, 2000:
197 k: READ-BYTE is supposed to signal TYPE-ERROR when its argument is
198 not a binary input stream, but instead cheerfully reads from
199 character streams, e.g. (MAKE-STRING-INPUT-STREAM "abc").
202 The debugger LIST-LOCATIONS command doesn't work properly.
203 (How should it work properly?)
206 Compiling and loading
207 (DEFUN FAIL (X) (THROW 'FAIL-TAG X))
209 then requesting a BACKTRACE at the debugger prompt gives no information
210 about where in the user program the problem occurred.
213 Using the pretty-printer from the command prompt gives funny
214 results, apparently because the pretty-printer doesn't know
215 about user's command input, including the user's carriage return
216 that the user, and therefore the pretty-printer thinks that
217 the new output block should start indented 2 or more characters
218 rightward of the correct location.
221 As reported by Winton Davies on a CMU CL mailing list 2000-01-10,
222 and reported for SBCL by Martin Atzmueller 2000-10-20: (TRACE GETHASH)
223 crashes SBCL. In general tracing anything which is used in the
224 implementation of TRACE is likely to have the same problem.
227 As reported by Martin Atzmueller on sbcl-devel 26 Dec 2000,
228 ANSI says that WITH-OUTPUT-TO-STRING should have a keyword
229 :ELEMENT-TYPE, but in sbcl-0.6.9 this is not defined for
230 WITH-OUTPUT-TO-STRING.
233 ANSI says in one place that type declarations can be abbreviated even
234 when the type name is not a symbol, e.g.
235 (DECLAIM ((VECTOR T) *FOOVECTOR*))
236 SBCL doesn't support this. But ANSI says in another place that this
237 isn't allowed. So it's not clear this is a bug after all. (See the
238 e-mail on cmucl-help@cons.org on 2001-01-16 and 2001-01-17 from WHN
242 as pointed out by Dan Barlow on sbcl-devel 2000-07-02:
243 The PICK-TEMPORARY-FILE-NAME utility used by LOAD-FOREIGN uses
244 an easily guessable temporary filename in a way which might open
245 applications using LOAD-FOREIGN to hijacking by malicious users
246 on the same machine. Incantations for doing this safely are
247 floating around the net in various "how to write secure programs
248 despite Unix" documents, and it would be good to (1) fix this in
249 LOAD-FOREIGN, and (2) hunt for any other code which uses temporary
250 files and make it share the same new safe logic.
252 (partially alleviated in sbcl-0.7.9.32 by a fix by Matthew Danish to
253 make the temporary filename less easily guessable)
256 RANDOM-INTEGER-EXTRA-BITS=10 may not be large enough for the RANDOM
257 RNG to be high quality near RANDOM-FIXNUM-MAX; it looks as though
258 the mean of the distribution can be systematically O(0.1%) wrong.
259 Just increasing R-I-E-B is probably not a good solution, since
260 it would decrease efficiency more than is probably necessary. Perhaps
261 using some sort of accept/reject method would be better.
264 Internally the compiler sometimes evaluates
265 (sb-kernel:type/= (specifier-type '*) (specifier-type t))
266 (I stumbled across this when I added an
267 (assert (not (eq type1 *wild-type*)))
268 in the NAMED :SIMPLE-= type method.) '* isn't really a type, and
269 in a type context should probably be translated to T, and so it's
270 probably wrong to ask whether it's equal to the T type and then (using
271 the EQ type comparison in the NAMED :SIMPLE-= type method) return NIL.
272 (I haven't tried to investigate this bug enough to guess whether
273 there might be any user-level symptoms.)
275 In fact, the type system is likely to depend on this inequality not
276 holding... * is not equivalent to T in many cases, such as
277 (VECTOR *) /= (VECTOR T).
280 Inconsistencies between derived and declared VALUES return types for
281 DEFUN aren't checked very well. E.g. the logic which successfully
282 catches problems like
283 (declaim (ftype (function (fixnum) float) foo))
285 (declare (type integer x))
286 (values x)) ; wrong return type, detected, gives warning, good!
288 (declaim (ftype (function (t) (values t t)) bar))
290 (values x)) ; wrong number of return values, no warning, bad!
291 The cause of this is seems to be that (1) the internal function
292 VALUES-TYPES-EQUAL-OR-INTERSECT used to make the check handles its
293 arguments symmetrically, and (2) when the type checking code was
294 written back when when SBCL's code was still CMU CL, the intent
296 (declaim (ftype (function (t) t) bar))
298 (values x x)) ; wrong number of return values; should give warning?
299 not be warned for, because a two-valued return value is considered
300 to be compatible with callers who expects a single value to be
301 returned. That intent is probably not appropriate for modern ANSI
302 Common Lisp, but fixing this might be complicated because of other
303 divergences between auld-style and new-style handling of
304 multiple-VALUES types. (Some issues related to this were discussed
305 on cmucl-imp at some length sometime in 2000.)
308 The facility for dumping a running Lisp image to disk gets confused
309 when run without the PURIFY option, and creates an unnecessarily large
310 core file (apparently representing memory usage up to the previous
311 high-water mark). Moreover, when the file is loaded, it confuses the
312 GC, so that thereafter memory usage can never be reduced below that
316 In sbcl-0.6.11.41 (and in all earlier SBCL, and in CMU
317 CL), out-of-line structure slot setters are horribly inefficient
318 whenever the type of the slot is declared, because out-of-line
319 structure slot setters are implemented as closures to save space,
320 so the compiler doesn't compile the type test into code, but
321 instead just saves the type in a lexical closure and interprets it
323 To exercise the problem, compile and load
324 (cl:in-package :cl-user)
326 (bar (error "missing") :type bar))
329 (loop (setf (foo-bar *foo*) x)))
331 (defvar *bar* (make-bar))
332 (defvar *foo* (make-foo :bar *bar*))
333 (defvar *setf-foo-bar* #'(setf foo-bar))
335 (loop (funcall *setf-foo-bar* x *foo*)))
336 then run (WASTREL1 *BAR*) or (WASTREL2 *BAR*), hit Ctrl-C, and
337 use BACKTRACE, to see it's spending all essentially all its time
338 in %TYPEP and VALUES-SPECIFIER-TYPE and so forth.
339 One possible solution would be simply to give up on
340 representing structure slot accessors as functions, and represent
341 them as macroexpansions instead. This can be inconvenient for users,
342 but it's not clear that it's worse than trying to help by expanding
343 into a horribly inefficient implementation.
344 As a workaround for the problem, #'(SETF FOO) expressions
345 can be replaced with (EFFICIENT-SETF-FUNCTION FOO), where
346 (defmacro efficient-setf-function (place-function-name)
347 (or #+sbcl (and (sb-int:info :function :accessor-for place-function-name)
348 ;; a workaround for the problem, encouraging the
349 ;; inline expansion of the structure accessor, so
350 ;; that the compiler can optimize its type test
351 (let ((new-value (gensym "NEW-VALUE-"))
352 (structure-value (gensym "STRUCTURE-VALUE-")))
353 `(lambda (,new-value ,structure-value)
354 (setf (,place-function-name ,structure-value)
356 ;; no problem, can just use the ordinary expansion
357 `(function (setf ,place-function-name))))
360 There's apparently a bug in CEILING optimization which caused
361 Douglas Crosher to patch the CMU CL version. Martin Atzmueller
362 applied the patches to SBCL and they didn't seem to cause problems
363 (as reported sbcl-devel 2001-05-04). However, since the patches
364 modify nontrivial code which was apparently written incorrectly
365 the first time around, until regression tests are written I'm not
366 comfortable merging the patches in the CVS version of SBCL.
369 (TIME (ROOM T)) reports more than 200 Mbytes consed even for
370 a clean, just-started SBCL system. And it seems to be right:
371 (ROOM T) can bring a small computer to its knees for a *long*
372 time trying to GC afterwards. Surely there's some more economical
373 way to implement (ROOM T).
376 When the compiler inline expands functions, it may be that different
377 kinds of return values are generated from different code branches.
378 E.g. an inline expansion of POSITION generates integer results
379 from one branch, and NIL results from another. When that inline
380 expansion is used in a context where only one of those results
383 (aref *a1* (position x *a2*)))
384 and the compiler can't prove that the unacceptable branch is
385 never taken, then bogus type mismatch warnings can be generated.
386 If you need to suppress the type mismatch warnings, you can
387 suppress the inline expansion,
389 #+sbcl (declare (notinline position)) ; to suppress bug 117 bogowarnings
390 (aref *a1* (position x *a2*)))
391 or, sometimes, suppress them by declaring the result to be of an
394 (aref *a1* (the integer (position x *a2*))))
396 This is not a new compiler problem in 0.7.0, but the new compiler
397 transforms for FIND, POSITION, FIND-IF, and POSITION-IF make it
398 more conspicuous. If you don't need performance from these functions,
399 and the bogus warnings are a nuisance for you, you can return to
400 your pre-0.7.0 state of grace with
401 #+sbcl (declaim (notinline find position find-if position-if)) ; bug 117..
404 as reported by Eric Marsden on cmucl-imp@cons.org 2001-08-14:
405 (= (FLOAT 1 DOUBLE-FLOAT-EPSILON)
406 (+ (FLOAT 1 DOUBLE-FLOAT-EPSILON) DOUBLE-FLOAT-EPSILON)) => T
407 when of course it should be NIL. (He says it only fails for X86,
408 not SPARC; dunno about Alpha.)
410 Also, "the same problem exists for LONG-FLOAT-EPSILON,
411 DOUBLE-FLOAT-NEGATIVE-EPSILON, LONG-FLOAT-NEGATIVE-EPSILON (though
412 for the -negative- the + is replaced by a - in the test)."
414 Raymond Toy comments that this is tricky on the X86 since its FPU
415 uses 80-bit precision internally.
418 Even in sbcl-0.pre7.x, which is supposed to be free of the old
419 non-ANSI behavior of treating the function return type inferred
420 from the current function definition as a declaration of the
421 return type from any function of that name, the return type of NIL
422 is attached to FOO in 120a above, and used to optimize code which
426 As of version 0.pre7.14, SBCL's implementation of MACROLET makes
427 the entire lexical environment at the point of MACROLET available
428 in the bodies of the macroexpander functions. In particular, it
429 allows the function bodies (which run at compile time) to try to
430 access lexical variables (which are only defined at runtime).
431 It doesn't even issue a warning, which is bad.
433 The SBCL behavior arguably conforms to the ANSI spec (since the
434 spec says that the behavior is undefined, ergo anything conforms).
435 However, it would be better to issue a compile-time error.
436 Unfortunately I (WHN) don't see any simple way to detect this
437 condition in order to issue such an error, so for the meantime
438 SBCL just does this weird broken "conforming" thing.
440 The ANSI standard says, in the definition of the special operator
442 The macro-expansion functions defined by MACROLET are defined
443 in the lexical environment in which the MACROLET form appears.
444 Declarations and MACROLET and SYMBOL-MACROLET definitions affect
445 the local macro definitions in a MACROLET, but the consequences
446 are undefined if the local macro definitions reference any
447 local variable or function bindings that are visible in that
449 Then it seems to contradict itself by giving the example
451 (macrolet ((fudge (z)
452 ;The parameters x and flag are not accessible
453 ; at this point; a reference to flag would be to
454 ; the global variable of that name.
455 ` (if flag (* ,z ,z) ,z)))
456 ;The parameters x and flag are accessible here.
460 The comment "a reference to flag would be to the global variable
461 of the same name" sounds like good behavior for the system to have.
462 but actual specification quoted above says that the actual behavior
465 (Since 0.7.8.23 macroexpanders are defined in a restricted version
466 of the lexical environment, containing no lexical variables and
467 functions, which seems to conform to ANSI and CLtL2, but signalling
468 a STYLE-WARNING for references to variables similar to locals might
472 (as reported by Gabe Garza on cmucl-help 2001-09-21)
474 (defun test-pred (x y)
478 (func (lambda () x)))
479 (print (eq func func))
480 (print (test-pred func func))
481 (delete func (list func))))
482 Now calling (TEST-CASE) gives output
485 (#<FUNCTION {500A9EF9}>)
486 Evidently Python thinks of the lambda as a code transformation so
487 much that it forgets that it's also an object.
490 Ideally, uninterning a symbol would allow it, and its associated
491 FDEFINITION and PROCLAIM data, to be reclaimed by the GC. However,
492 at least as of sbcl-0.7.0, this isn't the case. Information about
493 FDEFINITIONs and PROCLAIMed properties is stored in globaldb.lisp
494 essentially in ordinary (non-weak) hash tables keyed by symbols.
495 Thus, once a system has an entry in this system, it tends to live
496 forever, even when it is uninterned and all other references to it
499 141: "pretty printing and backquote"
502 ``(FOO SB-IMPL::BACKQ-COMMA-AT S)
505 * (write '`(, .ala.) :readably t :pretty t)
508 (note the space between the comma and the point)
511 (reported by Jesse Bouwman 2001-10-24 through the unfortunately
512 prominent SourceForge web/db bug tracking system, which is
513 unfortunately not a reliable way to get a timely response from
514 the SBCL maintainers)
515 In the course of trying to build a test case for an
516 application error, I encountered this behavior:
517 If you start up sbcl, and then lay on CTRL-C for a
518 minute or two, the lisp process will eventually say:
519 %PRIMITIVE HALT called; the party is over.
520 and throw you into the monitor. If I start up lisp,
521 attach to the process with strace, and then do the same
522 (abusive) thing, I get instead:
523 access failure in heap page not marked as write-protected
524 and the monitor again. I don't know enough to have the
525 faintest idea of what is going on here.
526 This is with sbcl 6.12, uname -a reports:
527 Linux prep 2.2.19 #4 SMP Tue Apr 24 13:59:52 CDT 2001 i686 unknown
528 I (WHN) have verified that the same thing occurs on sbcl-0.pre7.141
529 under OpenBSD 2.9 on my X86 laptop. Do be patient when you try it:
530 it took more than two minutes (but less than five) for me.
533 ANSI allows types `(COMPLEX ,FOO) to use very hairy values for
534 FOO, e.g. (COMPLEX (AND REAL (SATISFIES ODDP))). The old CMU CL
535 COMPLEX implementation didn't deal with this, and hasn't been
536 upgraded to do so. (This doesn't seem to be a high priority
537 conformance problem, since seems hard to construct useful code
541 Floating point errors are reported poorly. E.g. on x86 OpenBSD
544 debugger invoked on condition of type SB-KERNEL:FLOATING-POINT-EXCEPTION:
545 An arithmetic error SB-KERNEL:FLOATING-POINT-EXCEPTION was signalled.
546 No traps are enabled? How can this be?
547 It should be possible to be much more specific (overflow, division
548 by zero, etc.) and of course the "How can this be?" should be fixable.
550 See also bugs #45.c and #183
553 (reported by Robert E. Brown 2002-04-16)
554 When a function is called with too few arguments, causing the
555 debugger to be entered, the uninitialized slots in the bad call frame
556 seem to cause GCish problems, being interpreted as tagged data even
557 though they're not. In particular, executing ROOM in the
558 debugger at that point causes AVER failures:
561 * (lisp-implementation-version)
567 failed AVER: "(SAP= CURRENT END)"
568 (Christophe Rhodes reports that this doesn't occur on the SPARC, which
569 isn't too surprising since there are many differences in stack
570 implementation and GC conservatism between the X86 and other ports.)
573 In sbcl-0.7.3.11, compiling the (illegal) code
574 (in-package :cl-user)
575 (defmethod prove ((uustk uustk))
578 gives the (not terribly clear) error message
580 ; (during macroexpansion of (DEFMETHOD PROVE ...))
581 ; can't get template for (FROB NIL NIL)
582 The problem seems to be that the code walker used by the DEFMETHOD
583 macro is unhappy with the illegal syntax in the method body, and
584 is giving an unclear error message.
587 The compiler sometimes tries to constant-fold expressions before
588 it checks to see whether they can be reached. This can lead to
589 bogus warnings about errors in the constant folding, e.g. in code
592 (WRITE-STRING (> X 0) "+" "0"))
593 compiled in a context where the compiler can prove that X is NIL,
594 and the compiler complains that (> X 0) causes a type error because
595 NIL isn't a valid argument to #'>. Until sbcl-0.7.4.10 or so this
596 caused a full WARNING, which made the bug really annoying because then
597 COMPILE and COMPILE-FILE returned FAILURE-P=T for perfectly legal
598 code. Since then the warning has been downgraded to STYLE-WARNING,
599 so it's still a bug but at least it's a little less annoying.
601 183: "IEEE floating point issues"
602 Even where floating point handling is being dealt with relatively
603 well (as of sbcl-0.7.5, on sparc/sunos and alpha; see bug #146), the
604 accrued-exceptions and current-exceptions part of the fp control
605 word don't seem to bear much relation to reality. E.g. on
609 debugger invoked on condition of type DIVISION-BY-ZERO:
610 arithmetic error DIVISION-BY-ZERO signalled
611 0] (sb-vm::get-floating-point-modes)
613 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
614 :ROUNDING-MODE :NEAREST
615 :CURRENT-EXCEPTIONS NIL
616 :ACCRUED-EXCEPTIONS (:INEXACT)
619 * (sb-vm::get-floating-point-modes)
620 (:TRAPS (:OVERFLOW :INVALID :DIVIDE-BY-ZERO)
621 :ROUNDING-MODE :NEAREST
622 :CURRENT-EXCEPTIONS (:INEXACT)
623 :ACCRUED-EXCEPTIONS (:INEXACT)
626 188: "compiler performance fiasco involving type inference and UNION-TYPE"
630 (declare (optimize (safety 3)))
631 (declare (optimize (compilation-speed 2)))
632 (declare (optimize (speed 1) (debug 1) (space 1)))
634 (declare (type (integer 0) start))
635 (print (incf start 22))
636 (print (incf start 26))
637 (print (incf start 28)))
639 (declare (type (integer 0) start))
640 (print (incf start 22))
641 (print (incf start 26)))
643 (declare (type (integer 0) start))
644 (print (incf start 22))
645 (print (incf start 26))))))
647 This example could be solved with clever enough constraint
648 propagation or with SSA, but consider
653 The careful type of X is {2k} :-(. Is it really important to be
654 able to work with unions of many intervals?
656 190: "PPC/Linux pipe? buffer? bug"
657 In sbcl-0.7.6, the run-program.test.sh test script sometimes hangs
658 on the PPC/Linux platform, waiting for a zombie env process. This
659 is a classic symptom of buffer filling and deadlock, but it seems
660 only sporadically reproducible.
662 191: "Miscellaneous PCL deficiencies"
663 (reported by Alexey Dejneka sbcl-devel 2002-08-04)
664 a. DEFCLASS does not inform the compiler about generated
665 functions. Compiling a file with
669 (WITH-SLOTS (A-CLASS-X) A
671 results in a STYLE-WARNING:
673 SB-SLOT-ACCESSOR-NAME::|COMMON-LISP-USER A-CLASS-X slot READER|
675 APD's fix for this was checked in to sbcl-0.7.6.20, but Pierre
676 Mai points out that the declamation of functions is in fact
677 incorrect in some cases (most notably for structure
678 classes). This means that at present erroneous attempts to use
679 WITH-SLOTS and the like on classes with metaclass STRUCTURE-CLASS
680 won't get the corresponding STYLE-WARNING.
681 c. the examples in CLHS 7.6.5.1 (regarding generic function lambda
682 lists and &KEY arguments) do not signal errors when they should.
684 201: "Incautious type inference from compound types"
685 a. (reported by APD sbcl-devel 2002-09-17)
687 (LET ((Y (CAR (THE (CONS INTEGER *) X))))
689 (FORMAT NIL "~S IS ~S, Y = ~S"
696 (FOO ' (1 . 2)) => "NIL IS INTEGER, Y = 1"
700 (declare (type (array * (4 4)) x))
702 (setq x (make-array '(4 4)))
703 (adjust-array y '(3 5))
704 (= (array-dimension y 0) (eval `(array-dimension ,y 0)))))
706 * (foo (make-array '(4 4) :adjustable t))
709 205: "environment issues in cross compiler"
710 (These bugs have no impact on user code, but should be fixed or
712 a. Macroexpanders introduced with MACROLET are defined in the null
714 b. The body of (EVAL-WHEN (:COMPILE-TOPLEVEL) ...) is evaluated in
715 the null lexical environment.
716 c. The cross-compiler cannot inline functions defined in a non-null
719 206: ":SB-FLUID feature broken"
720 (reported by Antonio Martinez-Shotton sbcl-devel 2002-10-07)
721 Enabling :SB-FLUID in the target-features list in sbcl-0.7.8 breaks
724 207: "poorly distributed SXHASH results for compound data"
725 SBCL's SXHASH could probably try a little harder. ANSI: "the
726 intent is that an implementation should make a good-faith
727 effort to produce hash-codes that are well distributed
728 within the range of non-negative fixnums". But
729 (let ((hits (make-hash-table)))
732 (let* ((ij (cons i j))
733 (newlist (push ij (gethash (sxhash ij) hits))))
735 (format t "~&collision: ~S~%" newlist))))))
736 reports lots of collisions in sbcl-0.7.8. A stronger MIX function
737 would be an obvious way of fix. Maybe it would be acceptably efficient
738 to redo MIX using a lookup into a 256-entry s-box containing
739 29-bit pseudorandom numbers?
741 211: "keywords processing"
742 a. :ALLOW-OTHER-KEYS T should allow a function to receive an odd
743 number of keyword arguments.
746 (flet ((foo (&key y) (list y)))
747 (list (foo :y 1 :y 2)))
749 issues confusing message
754 ; caught STYLE-WARNING:
755 ; The variable #:G15 is defined but never used.
757 212: "Sequence functions and circular arguments"
758 COERCE, MERGE and CONCATENATE go into an infinite loop when given
759 circular arguments; it would be good for the user if they could be
760 given an error instead (ANSI 17.1.1 allows this behaviour on the part
761 of the implementation, as conforming code cannot give non-proper
762 sequences to these functions. MAP also has this problem (and
763 solution), though arguably the convenience of being able to do
764 (MAP 'LIST '+ FOO '#1=(1 . #1#))
765 might be classed as more important (though signalling an error when
766 all of the arguments are circular is probably desireable).
768 213: "Sequence functions and type checking"
769 a. MAKE-SEQUENCE, COERCE, MERGE and CONCATENATE cannot deal with
770 various complicated, though recognizeable, CONS types [e.g.
771 (CONS * (CONS * NULL))
772 which according to ANSI should be recognized] (and, in SAFETY 3
773 code, should return a list of LENGTH 2 or signal an error)
774 b. MAP, when given a type argument that is SUBTYPEP LIST, does not
775 check that it will return a sequence of the given type. Fixing
776 it along the same lines as the others (cf. work done around
777 sbcl-0.7.8.45) is possible, but doing so efficiently didn't look
778 entirely straightforward.
779 c. All of these functions will silently accept a type of the form
781 whether or not the return value is of this type. This is
782 probably permitted by ANSI (see "Exceptional Situations" under
783 ANSI MAKE-SEQUENCE), but the DERIVE-TYPE mechanism does not
784 know about this escape clause, so code of the form
785 (INTEGERP (CAR (MAKE-SEQUENCE '(CONS INTEGER *) 2)))
786 can erroneously return T.
789 SBCL 0.6.12.43 fails to compile
792 (declare (optimize (inhibit-warnings 0) (compilation-speed 2)))
793 (flet ((foo (&key (x :vx x-p)) (list x x-p)))
796 or a more simple example:
799 (declare (optimize (inhibit-warnings 0) (compilation-speed 2)))
800 (lambda (x) (declare (fixnum x)) (if (< x 0) 0 (1- x))))
802 215: ":TEST-NOT handling by functions"
803 a. FIND and POSITION currently signal errors when given non-NIL for
804 both their :TEST and (deprecated) :TEST-NOT arguments, but by
805 ANSI 17.2 "the consequences are unspecified", which by ANSI 1.4.2
806 means that the effect is "unpredictable but harmless". It's not
807 clear what that actually means; it may preclude conforming
808 implementations from signalling errors.
809 b. COUNT, REMOVE and the like give priority to a :TEST-NOT argument
810 when conflict occurs. As a quality of implementation issue, it
811 might be preferable to treat :TEST and :TEST-NOT as being in some
812 sense the same &KEY, and effectively take the first test function in
814 c. Again, a quality of implementation issue: it would be good to issue a
815 STYLE-WARNING at compile-time for calls with :TEST-NOT, and a
816 WARNING for calls with both :TEST and :TEST-NOT; possibly this
817 latter should be WARNed about at execute-time too.
819 216: "debugger confused by frames with invalid number of arguments"
820 In sbcl-0.7.8.51, executing e.g. (VECTOR-PUSH-EXTEND T), BACKTRACE, Q
821 leaves the system confused, enough so that (QUIT) no longer works.
822 It's as though the process of working with the uninitialized slot in
823 the bad VECTOR-PUSH-EXTEND frame causes GC problems, though that may
824 not be the actual problem. (CMU CL 18c doesn't have problems with this.)
826 217: "Bad type operations with FUNCTION types"
829 * (values-type-union (specifier-type '(function (base-char)))
830 (specifier-type '(function (integer))))
832 #<FUN-TYPE (FUNCTION (BASE-CHAR) *)>
834 It causes insertion of wrong type assertions into generated
838 (let ((f (etypecase x
839 (character #'write-char)
840 (integer #'write-byte))))
843 (character (write-char x s))
844 (integer (write-byte x s)))))
846 Then (FOO #\1 *STANDARD-OUTPUT*) signals type error.
848 (In 0.7.9.1 the result type is (FUNCTION * *), so Python does not
849 produce invalid code, but type checking is not accurate. Similar
850 problems exist with VALUES-TYPE-INTERSECTION.)
853 Sbcl 0.7.9 fails to compile
855 (multiple-value-call #'list
856 (the integer (helper))
859 Type check for INTEGER, the result of which serves as the first
860 argument of M-V-C, is inserted after evaluation of NIL. So arguments
861 of M-V-C are pushed in the wrong order. As a temporary workaround
862 type checking was disabled for M-V-Cs in 0.7.9.13. A better solution
863 would be to put the check between evaluation of arguments, but it
864 could be tricky to check result types of PROG1, IF etc.
866 233: bugs in constraint propagation
869 (declare (optimize (speed 2) (safety 3)))
875 (quux y (+ y 2d0) (* y 3d0)))))
876 (foo 4) => segmentation violation
878 (see usage of CONTINUATION-ASSERTED-TYPE in USE-RESULT-CONSTRAINTS)
882 (declaim (optimize (speed 2) (safety 3)))
884 (if (typep (prog1 x (setq x y)) 'double-float)
887 (foo 1d0 5) => segmentation violation
889 235: "type system and inline expansion"
891 (declaim (ftype (function (cons) number) acc))
892 (declaim (inline acc))
894 (the number (car c)))
897 (values (locally (declare (optimize (safety 0)))
899 (locally (declare (optimize (safety 3)))
902 (foo '(nil) '(t)) => NIL, T.
904 237: "Environment arguments to type functions"
905 a. Functions SUBTYPEP, TYPEP, UPGRADED-ARRAY-ELEMENT-TYPE, and
906 UPGRADED-COMPLEX-PART-TYPE now have an optional environment
907 argument, but they ignore it completely. This is almost
908 certainly not correct.
909 b. Also, the compiler's optimizers for TYPEP have not been informed
910 about the new argument; consequently, they will not transform
911 calls of the form (TYPEP 1 'INTEGER NIL), even though this is
912 just as optimizeable as (TYPEP 1 'INTEGER).
914 238: "REPL compiler overenthusiasm for CLOS code"
916 * (defclass foo () ())
917 * (defmethod bar ((x foo) (foo foo)) (call-next-method))
918 causes approximately 100 lines of code deletion notes. Some
919 discussion on this issue happened under the title 'Three "interesting"
920 bugs in PCL', resulting in a fix for this oververbosity from the
921 compiler proper; however, the problem persists in the interactor
922 because the notion of original source is not preserved: for the
923 compiler, the original source of the above expression is (DEFMETHOD
924 BAR ((X FOO) (FOO FOO)) (CALL-NEXT-METHOD)), while by the time the
925 compiler gets its hands on the code needing compilation from the REPL,
926 it has been macroexpanded several times.
928 A symptom of the same underlying problem, reported by Tony Martinez:
930 (with-input-from-string (*query-io* " no")
932 (simple-type-error () 'error))
934 ; (SB-KERNEL:FLOAT-WAIT)
936 ; note: deleting unreachable code
937 ; compilation unit finished
940 241: "DEFCLASS mysteriously remembers uninterned accessor names."
941 (from tonyms on #lisp IRC 2003-02-25)
942 In sbcl-0.7.12.55, typing
943 (defclass foo () ((bar :accessor foo-bar)))
946 (defclass foo () ((bar :accessor foo-bar)))
947 gives the error message
948 "#:FOO-BAR already names an ordinary function or a macro."
949 So it's somehow checking the uninterned old accessor name instead
950 of the new requested accessor name, which seems broken to me (WHN).
952 242: "WRITE-SEQUENCE suboptimality"
953 (observed from clx performance)
954 In sbcl-0.7.13, WRITE-SEQUENCE of a sequence of type
955 (SIMPLE-ARRAY (UNSIGNED-BYTE 8) (*)) on a stream with element-type
956 (UNSIGNED-BYTE 8) will write to the stream one byte at a time,
957 rather than writing the sequence in one go, leading to severe
958 performance degradation.
960 243: "STYLE-WARNING overenthusiasm for unused variables"
961 (observed from clx compilation)
962 In sbcl-0.7.14, in the presence of the macros
963 (DEFMACRO FOO (X) `(BAR ,X))
964 (DEFMACRO BAR (X) (DECLARE (IGNORABLE X)) 'NIL)
965 somewhat surprising style warnings are emitted for
966 (COMPILE NIL '(LAMBDA (Y) (FOO Y))):
968 ; (LAMBDA (Y) (FOO Y))
970 ; caught STYLE-WARNING:
971 ; The variable Y is defined but never used.
973 245: bugs in disassembler
974 a. On X86 an immediate operand for IMUL is printed incorrectly.
975 b. On X86 operand size prefix is not recognized.
977 248: "reporting errors in type specifier syntax"
978 (TYPEP 1 '(SYMBOL NIL)) says something about "unknown type
982 (defun foo (&key (a :x))
986 does not cause a warning. (BTW: old SBCL issued a warning, but for a
987 function, which was never called!)
990 Compiler does not emit warnings for
992 a. (lambda () (svref (make-array 8 :adjustable t) 1))
995 (list (let ((y (the real x)))
996 (unless (floatp y) (error ""))
1001 (declare (optimize (debug 0)))
1002 (declare (type vector x))
1003 (list (fill-pointer x)
1007 Complex array type does not have corresponding type specifier.
1009 This is a problem because the compiler emits optimization notes when
1010 you use a non-simple array, and without a type specifier for hairy
1011 array types, there's no good way to tell it you're doing it
1012 intentionally so that it should shut up and just compile the code.
1014 Another problem is confusing error message "asserted type ARRAY
1015 conflicts with derived type (VALUES SIMPLE-VECTOR &OPTIONAL)" during
1016 compiling (LAMBDA (V) (VALUES (SVREF V 0) (VECTOR-POP V))).
1018 The last problem is that when type assertions are converted to type
1019 checks, types are represented with type specifiers, so we could lose
1020 complex attribute. (Now this is probably not important, because
1021 currently checks for complex arrays seem to be performed by
1028 (compile nil '(lambda () (aref (make-array 0) 0))) compiles without
1029 warning. Analogous cases with the index and length being equal and
1030 greater than 0 are warned for; the problem here seems to be that the
1031 type required for an array reference of this type is (INTEGER 0 (0))
1032 which is canonicalized to NIL.
1034 DEFUNCT CATEGORIES OF BUGS
1036 These labels were used for bugs related to the old IR1 interpreter.
1037 The # values reached 6 before the category was closed down.